First of all it was strange to see:
if (allowed) {
...
} else {
D3COLD_ENABLE
}
But besides this misalignment, let's also use the pci
d3cold_allowed useful to us and know that we are not really
allowing d3cold.
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Send uevent in case of gt reset failure. This intimation can be used by
userspace monitoring tool to do the device level reset/reboot
when GT reset fails. udevadm can be used to monitor the uevents.
v2:
- Support only gt failure notification (Rodrigo)
v3
- Rectify the comments in header file.
v4
- Use pci kobj instead of drm kobj for notification.(Rodrigo)
- Cleanup (Badal)
v5
- Add tile id and gt id as additional info provided by uevent.
- Provide code documentation for the uevent. (Rodrigo)
Cc: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Badal Nilawar <badal.nilawar@intel.com>
Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The order: 'offset, mask, val'; is more common in other
drivers and in special in i915, where any dev could copy
a sequence and end up with unexpected behavior.
Done with coccinelle:
@rule1@
expression gt, reg, val, mask, timeout, out, atomic;
@@
- xe_mmio_wait32(gt, reg, val, mask, timeout, out, atomic)
+ xe_mmio_wait32(gt, reg, mask, val, timeout, out, atomic)
spatch -sp_file mmio.cocci *.c *.h compat-i915-headers/intel_uncore.h \
--in-place
v2: Rebased after changes on xe_guc_mcr usage of xe_mmio_wait32.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We cannot have spin locks around xe_irq_reset, since it will
call the intel_display_power_is_enabled() function, and
that needs a mutex lock. Hence causing the undesired
"[ BUG: Invalid wait context ]"
We cannot convert i915's power domain lock to spin lock
due to the nested dependency of non-atomic context waits.
So, let's move the xe_irq_reset functions from the
critical area, while still ensuring that we are protecting
the irq.enabled and ensuring the right serialization
in the irq handlers.
v2: On the first version, I had missed the fact that
irq.enabled is checked on the xe/display glue layer,
and that i915 display code is actually using the irq
spin lock properly. So, this got changed to a version
suggested by Matthew Auld.
v3: do not use lockdep_assert for display glue.
do not save restore irq from inside IRQ or we can
get bogus irq restore warnings
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/463
Suggested-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Sort it by register address to make it easy to update when needed.
v2: Do not create exception for registers with same functionality.
Always sort it.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-11-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Top of DSM contains the WOPCM where kernel driver shouldn't access as
it contains data from other HW agents. Carve it out from the stolen
memory. On a MTL system, the output now matches the expected values:
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-10-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Based on commit 8d8d062be6 ("drm/i915/mtl: Fix MTL stolen memory GGTT
mapping"). For stolen on MTL and beyond, the address in the PTE is the
offset from DSM base. While at it, update the comments explaining each
part of the calculation.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-9-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Integrated graphics 1270 and beyond should set the PTE_LM bit in the PTE
when it's stolen memory. Add a new function, xe_bo_is_stolen_devmem(),
and use it when encoding the PTE.
In some places in the spec the PTE bit is called "Local Memory",
abbreviated as LM, and in others it's called "Device Memory" (DM). Since
we moved away from "Local Memory" and preferred the "vram" terminology,
also rename the macros as DM to follow the name of the new function.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-7-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The output arg is_vram in xe_bo_addr() is unused by several callers.
It's also not what the function is mainly doing. Remove the argument and
let the interested callers to call xe_bo_is_vram().
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-6-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
All the callers pass a NULL vma, so the buffer is always the BO. Remove
the argument and the side effects of dealing with it.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-5-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
in commit 81593af6c8 ("drm/xe: Convert xe_mmio_wait32 to us so we can
stop using wait_for_us.") the mcr semaphore register read was
accidentally switched from waiting for the register to go to 1 to
waiting for the register to go to 0, so we need to flip it back.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Commit 3743040261 ("drm/xe: NULL binding implementation") introduced
the NULL binding implementation, but left a case in which the out value
is_vram is not set and the caller will use whatever was on stack.
Eventually the is_vram out could be removed, but this should at least
fix the current bug.
Fixes: 3743040261 ("drm/xe: NULL binding implementation")
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230726160708.3967790-4-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Bind engines need to use the migration vm, however we don't have any rpm
for such a vm, otherwise the kernel would prevent rpm suspend-resume.
There are two issues here, first is the actual engine create which needs
to touch the lrc, but since that is in VRAM we trigger loads of missing
mem_access asserts. The second issue is when destroying the actual
engine, which requires GuC CT to deregister the context.
v2 (Rodrigo):
- Just use ENGINE_FLAG_VM as the indicator that we need to hold an rpm
ref. This also handles the case in xe_vm_create() where we create
default bind engines.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/499
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/504
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If no operations are generated for VM binds the out-syncs must still be
signaled.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Do not queue the rebind worker directly, rather use the helper
xe_vm_queue_rebind_worker. This ensures we use the correct work queue.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The module parameter should reflect the name of the optional,
experimental and unsafe option, rather than the default one.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
This config is the only real one. If execlist remains in the
code it will forever be experimental and we shouldn't maintain
an uapi like that for that experimental piece of code that
should never be used by real users.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
This allows vram_size > io_size, instead of just clamping the vram size
to the BAR size, now that the driver supports it.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Michael J. Ruhl <michael.j.ruhl@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Mostly the same as i915. We add a new hint for userspace to force an
object into the mappable part of vram.
We also need to tell userspace how large the mappable part is. In Vulkan
for example, there will be two vram heaps for small-bar systems. And
here the size of each heap needs to be known. Likewise the used/avail
tracking needs to account for the mappable part.
We also limit the available tracking going forward, such that we limit
to privileged users only, since these values are system wide and are
technically considered an info leak.
v2 (Maarten):
- s/NEEDS_CPU_ACCESS/NEEDS_VISIBLE_VRAM/ in the uapi. We also no
longer require smem as an extra placement. This is more flexible,
and lets us use this for clear-color surfaces, since we need CPU access
there but we don't want to attach smem, since that effectively disables
CCS from kernel pov.
- Reject clear-color CCS buffers where NEEDS_VISIBLE_VRAM is not set,
instead of migrating it behind the scenes.
v3 (José):
- Split the changes that limit the accounting for perfmon_capable()
into a separate patch.
- Use XE_BO_CREATE_VRAM_MASK.
v4 (Gwan-gyeong Mun):
- Add some kernel-doc for the query bits.
v5:
- One small kernel-doc correction. The cpu_visible_size and
corresponding used tracking are always zero for non
XE_MEM_REGION_CLASS_VRAM.
v6:
- Without perfmon_capable() it likely makes more sense to report as
zero, instead of reporting as used == total size. This should give
similar behaviour as i915 which rather tracks free instead of used.
- Only enforce NEEDS_VISIBLE_VRAM on rc_ccs_cc_plane surfaces when the
device is actually small-bar.
Testcase: igt/tests/xe_query
Testcase: igt/tests/xe_mmap@small-bar
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Filip Hazubski <filip.hazubski@intel.com>
Cc: Carl Zhang <carl.zhang@intel.com>
Cc: Effie Yu <effie.yu@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add the new flag XE_BO_NEEDS_CPU_ACCESS, to force allocating in the
mappable part of vram. If no flag is specified we do a topdown
allocation, to limit the chances of stealing the precious mappable part,
if we don't need it. If this is a full-bar system, then this all gets
nooped.
For kernel users, it looks like xe_bo_create_pin_map() is the central
place which users should call if they want CPU access to the object, so
add the flag there.
We still need to plumb this through for userspace allocations. Also it
looks like page-tables are using pin_map(), which is less than ideal. If
we can already use the GPU to do page-table management, then maybe we
should just force that for small-bar.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Platforms like MTL only have a single tile, but multiple GTs.
Ensure XE_ENGINE_CREATE accepts engine creation on gt1 on such
platforms.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230725003433.1992137-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On MTL and beyond, the GPU performs non-coherent accesses to the PPGTT
page tables. These page tables should be mapped as CPU:WC.
Removes CAT errors triggered by xe_exec_basic@once-basic on MTL:
xe 0000:00:02.0: [drm:__xe_pt_bind_vma [xe]] Preparing bind, with range [1a0000...1a0fff) engine 0000000000000000.
xe 0000:00:02.0: [drm:xe_vm_dbg_print_entries [xe]] 1 entries to update
xe 0000:00:02.0: [drm:xe_vm_dbg_print_entries [xe]] 0: Update level 3 at (0 + 1) [0...8000000000) f:0
xe 0000:00:02.0: [drm] Engine memory cat error: guc_id=2
xe 0000:00:02.0: [drm] Engine memory cat error: guc_id=2
xe 0000:00:02.0: [drm] Timedout job: seqno=4294967169, guc_id=2, flags=0x4
v2:
- Rename to XE_BO_PAGETABLE to make it more clear that this BO is the
pagetable itself, rather than just being bound in the PPGTT. (Lucas)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230725003433.1992137-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The main motivation is with d3cold which will make the suspend and
resume callbacks even more scary, but is useful regardless. We already
have the needed annotation on the acquire side with
xe_device_mem_access_get(), and by adding the annotation on the release
side we should have a lot more confidence that our locking hierarchy is
correct.
v2:
- Move the annotation into both callbacks for better symmetry. Also
don't hold over the entire mem_access_get(); we only need to lockep
to understand what is being held upon entering mem_access_get(), and
how that matches up with locks in the callbacks.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We must use migrate engine for page fault binds in order to avoid a
deadlock as the migrate engine has a reserved BCS instance which cannot
be stuck on a fault. To use the migrate engine the engine argument to
xe_migrate_update_pgtables must be NULL, this was incorrectly wired up
so vm->eng[tile_id] was always being used. Fix this.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Only alloc userptr part of xe_vma for userptrs, this will save on space
in the common BO case.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The callback kicks the worker thus mutually exclusive execution,
combining saves a bit of space in xe_vma.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This will save us a few bytes in the xe_vma structure.
v2: Use hweight8 rather than hweight_long (Rodrigo)
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This list isn't used again, list_del is the proper call.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Combine the userptr, rebind, and destroy links into a union as
the lists these links belong to are mutually exclusive.
v2: Adjust which lists are combined (Thomas H)
v3: Add kernel doc why this is safe (Thomas H), remove related change
of list_del_init -> list_del (Rodrigo)
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If we dont change page sizes we can avoid doing rebinds rather just do a
partial unbind. The algorithm to determine its page size is greedy as we
assume all pages in the removed VMA are the largest page used in the
VMA.
v2: Don't exceed 100 lines
v3: struct xe_vma_op_unmap remove in different patch, remove XXX comment
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We currently have a race between bind engines which can result in
corrupted page tables leading to faults.
A simple example:
bind A 0x0000-0x1000, engine A, has unsatisfied in-fence
bind B 0x1000-0x2000, engine B, no in-fences
exec A uses 0x1000-0x2000
Bind B will pass bind A and exec A will fault. This occurs as bind A
programs the root of the page table in a bind job which is held up by an
in-fence. Bind B in this case just programs a leaf entry of the
structure.
To fix use range-fence utility to track cross bind engine conflicts. In
the above example bind A would insert an dependency into the range-fence
tree with a key of 0x0-0x7fffffffff, bind B would find that dependency
and its bind job would scheduled behind the unsatisfied in-fence and
bind A's job.
Reviewed-by: Maarten Lankhorst<maarten.lankhorst@linux.intel.com>
Co-developed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add generic utility to track range conflicts signaled by a dma-fence.
Tracking implemented via an interval tree. An example use case being
tracking conflicts for pending (un)binds from multiple bind engines. By
being generic ths idea would this could moved to the DRM level and used
in multiple drivers for similar problems.
v2: Make interval tree functions static (CI)
v3: Remove non-static cleanup function (CI)
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Make explicit in the log that execlist submission is used to prevent from
silently using it over GuC submission.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Those look like leftover debug and are not even being used. If they were
real debug/info, they should be using the drm helpers.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Those messages are unnecessary because a generic message is already
produced in case of allocation failure. Besides, this also removes a
misuse of the XE_IOCTL_DBG macro.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Use FIELD_PREP()/FIELD_GET() to encode the tile id into flags. Besides
protecting for eventual overflow it also makes it easier to see a new
flag can't be added as BIT(7).
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230718193924.3084759-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rename XE_VM_FLAGS_64K to XE_VM_FLAG_64K to follow the other names and
s/GT/TILE/ that got missed in commit 08dea76745 ("drm/xe: Move
migration from GT to tile").
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230718193924.3084759-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It looks like bulk_move is set during object construction, but is only
removed on object close, however in various places we might not yet have
an actual fd to close, like on the error paths for the gem_create ioctl,
and also one internal user for the evict_test_run_gt() selftest. Try to
handle those cases by manually resetting the bulk_move. This should
prevent triggering:
WARNING: CPU: 7 PID: 8252 at drivers/gpu/drm/ttm/ttm_bo.c:327
ttm_bo_release+0x25e/0x2a0 [ttm]
v2 (Nirmoy):
- It should be safe to just unconditionally call
__xe_bo_unset_bulk_move() in most places.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Test seems to be failing badly after calling xe_bo_restore_kernel().
Taking a snapshot of the CTB and copying back a potentially old version
seems risky, depending on what might have been inflight. Also it seems
snapshotting the ADS object and copying back results in serious
breakage. Normally when calling xe_bo_restore_kernel() we always fully
restart the GT, which re-intializes such things. We could potentially
skip saving and restoring such objects in xe_bo_evict_all() however
seems quite fragile not to also restart the GT. Try to do that here by
triggering a GT reset.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Acked-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The GPU job will keep the device awake, however assumption here is that
caller of xe_migrate_clear() is also holding mem_access.ref otherwise we
hit the asserts in xe_sa_bo_flush_write() prior to the job construction.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We are calling fairly low level things like xe_bo_restore_kernel() which
expect caller to be holding mem_access.ref. Since we are doing stuff
like evict_all we likely don't want to race with rpm suspend, since that
potentially wants to do the same thing, so just wrap the whole test.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The atomics here might hide potential issues, also rpm core is not
holding any lock when calling our rpm resume callback, so add a dummy lock
with the idea that xe_pm_runtime_resume() is eventually going to be
called when we are holding it. This only needs to happen once and then
lockdep can validate all callers and their locks.
v2: (Thomas Hellström)
- Prefer static lockdep_map instead of full blown mutex.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Acked-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Increase the sensitivity of the ggtt->lock by priming it against
FS_RECLAIM, such that allocating memory while holding will result in
lockdep splats.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The callers should already be holding the mem_access reference, before
calling into this.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Only call access_put after dropping the forcewake. In theory the device
could suspend, but really we want to start asserting that we have a
mem_access.ref when touching mmio.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Any kind of device memory access should first ensure the device is not
suspended, mmio included.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We need keep the device awake when performing any kind of mmio operation.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/279
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The xe_device_mem_access_get() should be all that's needed here and
should now work as expected, without any strange races. In theory should
be no functional changes here.
Reported-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It looks like there is at least one race here, given that the
pm_runtime_suspended() check looks to return false if we are in the
process of suspending the device (RPM_SUSPENDING vs RPM_SUSPENDED). We
later also do xe_pm_runtime_get_if_active(), but since the device is
suspending or has now suspended, this doesn't do anything either.
Following from this we can potentially return from
xe_device_mem_access_get() with the device suspended or about to be,
leading to broken behaviour.
Attempt to fix this by always grabbing the runtime ref when our internal
ref transitions from 0 -> 1. The hard part is then dealing with the
runtime_pm callbacks also calling xe_device_mem_access_get() and
deadlocking, which the pm_runtime_suspended() check prevented.
v2:
- ct->lock looks to be primed with fs_reclaim, so holding that and then
allocating memory will cause lockdep to complain. Now that we
unconditionally grab the mem_access.lock around mem_access_{get,put}, we
need to change the ordering wrt to grabbing the ct->lock, since some of
the runtime_pm routines can allocate memory (or at least that's what
lockdep seems to suggest). Hopefully not a big deal. It might be that
there were already issues with this, just that the atomics where
"hiding" the potential issues.
v3:
- Use Thomas Hellström' idea with tracking the active task that is
executing in the resume or suspend callback, in order to avoid
recursive resume/suspend calls deadlocking on itself.
- Split the ct->lock change.
v4:
- Add smb_mb() around accessing the pm_callback_task for extra safety.
(Thomas Hellström)
v5:
- Clarify the kernel-doc for the mem_access.lock, given that it is quite
strange in what it protects (data vs code). The real motivation is to
aid lockdep. (Rodrigo Vivi)
v6:
- Split out the lock change. We still want this as a lockdep aid but
only for the xe_device_mem_access_get() path. Sticking a lock on the
put() looks be a no-go, also the runtime_put() there is always async.
- Now that the lock is gone move to atomics and rely on the pm code
serialising multiple callers on the 0 -> 1 transition.
- g2h_worker_func() looks to be the next issue, given that
suspend-resume callbacks are using CT, so try to handle that.
v7:
- Add xe_device_mem_access_get_if_ongoing(), and use it in
g2h_worker_func().
v8 (Anshuman):
- Just always grab the rpm, instead of just on the 0 -> 1 transition,
which is a lot clearer and simplifies the code quite a bit.
v9:
- Make sure we also adjust the CT fast-path with if-active.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/258
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Acked-by: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Don't init pcode and restore VRAM objects in vain.
We can rely on primary GT GUC_STATUS to detect whether
card has really lost power even when d3cold is allowed by xe.
Adding d3cold.lost_power flag to avoid pcode init and vram
restoration.
Also cleaning up the TODO code comment.
v2:
- %s/xe_guc_has_lost_power()/xe_guc_in_reset().
- Used existing gt instead of new variable. [Rodrigo]
- Added kernel-doc function comment. [Rodrigo]
- xe_guc_in_reset() return true if failed to get fw.
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230718080703.239343-6-anshuman.gupta@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Adding support to control d3cold by using vram_usages metric from
ttm resource manager.
When root port is capable of d3cold but xe has disallowed d3cold
due to vram_usages above vram_d3ccold_threshol. It is required to
disable d3cold to avoid any resume failure because root port can
still transition to d3cold when all of pcie endpoints and
{upstream, virtual} switch ports will transition to d3hot.
Also cleaning up the TODO code comment.
v2:
- Modify d3cold.allowed in xe_pm_d3cold_allowed_toggle. [Riana]
- Cond changed (total_vram_used_mb < xe->d3cold.vram_threshold)
according to doc comment.
v3:
- Added enum instead of true/false argument in
d3cold_toggle(). [Rodrigo]
- Removed TODO comment. [Rodrigo]
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Badal Nilawar <badal.nilawar@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230718080703.239343-5-anshuman.gupta@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Currently GuCRC is disabled in suspend path for xe.
Rc6 is a prerequiste to enable s0ix and
should not be disabled for s2idle. There is no requirement
to disable GuCRC for S3+.
Remove it from xe_guc_pc_stop, thus removing from suspend path.
Retain the call in other places where xe_guc_pc_stop is
called.
v2: add description and return statement to kernel-doc (Rodrigo)
v3: update commit message (Rodrigo)
v4: add mem_access_get to the gucrc disable function
Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reduce the number of warnings reported by checkpatch.pl from 118 to 48 by
addressing those warnings types:
LEADING_SPACE
LINE_SPACING
BRACES
TRAILING_SEMICOLON
CONSTANT_COMPARISON
BLOCK_COMMENT_STYLE
RETURN_VOID
ONE_SEMICOLON
SUSPECT_CODE_INDENT
LINE_CONTINUATIONS
UNNECESSARY_ELSE
UNSPECIFIED_INT
UNNECESSARY_INT
MISORDERED_TYPE
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Remove almost all existing style issues of type OPEN_BRACE reported
by checkpatch.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Remove almost all existing style issues of type SPACING reported
by checkpatch.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We need to hold vm->lock before the xe_vm_is_closed_or_banned().
Else we get this splat:
[ 802.555227] ------------[ cut here ]------------
[ 802.555234] WARNING: CPU: 33 PID: 3122 at drivers/gpu/drm/xe/xe_vm.h:60
[ 802.555515] CPU: 33 PID: 3122 Comm: xe_exec_fault_m Tainted:
...
[ 802.555709] Call Trace:
[ 802.555714] <TASK>
[ 802.555720] ? __warn+0x81/0x170
[ 802.555737] ? xe_vm_madvise_ioctl+0x2de/0x440 [xe]
Fixes: 9d858b69b0 ("drm/xe: Ban a VM if rebind worker hits an error")
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Brian Welty <brian.welty@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It looks like the single-tile PVC in CI dies during module load when doing
the pcode init. From the logs we try to access the address
0000000000138124 which doesn't map to anything, however 0x138124 also
looks to be the PCODE_MAILBOX register. So looks like the per-tile
mmio register mapping is NULL.
During probe the tile count is potentially trimmed, since we don't know
the real count until we actually probe the device. This seems to be
the case for single-tile PVC or similar devices. However it looks like
the gt_count is never adjusted to respect this updated tile count. As a
result when later doing some for_each_gt() loop, like we do for the
pcode, we can get back some GT that maps to some non-existent tile
which hasn't been properly set up, leading to crashes.
Try to fix this by adjusting the gt_count after probing the tiles for
real.
v2: Fix typo so it actually builds
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/383
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Ofir Bitton <obitton@habana.ai>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In various test cases that put the system under a heavy load, we can
sometimes see errors with missed TLB invalidations. In such cases we see
the interrupt arrive for the invalidation from the GuC, however the
actual processing of the completion is pushed onto a workqueue and
handled with all the other CT stuff, which might take longer than
expected. Since we expect TLB invalidations to complete within a
reasonable amount of time (at most ~250ms), and they do seem pretty
critical, allow handling directly from the CT fast-path.
v2 (José):
- Actually use the correct spinlock/unlock_irq, since pending_lock is
grabbed from IRQ.
v3:
- Don't publish the TLB fence on the list until after we fully
initialize it and successfully do the CT send. The list is now only
protected by the spin_lock pending_lock and we can't hold that
across the entire TLB send operation.
v4 (Matt Brost):
- Be careful with racing against fast CT path writing the seqno,
before we have actually published the fence.
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/297
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/320
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/449
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Looks to always to be zero when inspecting the CTB dump.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
To help debugging, sample the current seqno_recv and dump it out if we
encounter a TLB timeout for the fences path.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We might have various kworkers waiting for TLB flushes to complete which
are not tracked with an explicit TLB fence, however at this stage that
will never happen since the CT is already disabled, so make sure we
signal them here under the assumption that we have completed a full GT
reset.
v2:
- We need to use seqno - 1 here. After acquiring ct->lock the seqno is
actually the next users seqno and not the pending one.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Assumption here is that submission is disabled along with CT, and full
GT reset will also nuke TLBs, so should be safe to signal all in-flight
TLB fences, but only after the actual reset so move the placement
slightly.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The fast-path CT could be running as we enter a runtime-suspend or
potentially a GT reset, however here we only use the ct->fast_lock and
not the full ct->lock. Before disabling the CT, also serialise against
the fast_lock to ensure any in-progress work finishes before we start
nuking the CT related stuff. Once we disable ct->enabled and drop the
lock, any new work should fail gracefully, and anything that was in
progress should be finished.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If we are in the middle of a GT reset or similar the CT might be
disabled, such that the CT send fails. However we already incremented
gt->tlb_invalidation.seqno which might lead to warnings, since we
effectively just skipped a seqno:
0000:00:02.0: drm_WARN_ON(expected_seqno != msg[0])
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reserving and checking for space on the g2h side relies on the
fast_lock, and not the CT lock since we need to release space from the
fast CT path. Make sure we hold it when checking for space and reserving
it. The main concern is calling __g2h_release_space() as we are reserving
something and since the info.space and info.g2h_outstanding operations
are not atomic we can get some nonsense values back.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Just checking xe_device_mem_access_ongoing() is not enough, we also need
to hold the reference otherwise the ref can transition from 1 -> 0 as we
enter g2h_read(), leading to warnings. While we can't do a full rpm sync
in the IRQ, we can keep the device awake if the ref is non-zero.
Introduce a new helper for this and set it to work in for the CT
fast-path.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Ensure we load gt->tlb_invalidation.seqno_recv once, and use that for
our seqno checking. The gt->tlb_invalidation_seqno_past is a shared
global variable and can potentially change at any point here. However
the checks here need to operate on a stable version of seqno_recv for
this to make any sense.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
wake_up_all() and wait_event_timeout() already have the correct barriers
as per https://www.kernel.org/doc/Documentation/memory-barriers.txt.
This should ensure that the seqno_recv write can't be re-ordered wrt to
the actual wake_up_all() i.e we get woken up but there is no write. The
reader side with wait_event_timeout() also has the correct barriers.
With that drop the hand rolled smp_wmb(), which is anyway missing some
kind of matching barrier on the reader side.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rather than open coding VM binds and VMA tracking, use the GPUVA
library. GPUVA provides a common infrastructure for VM binds to use mmap
/ munmap semantics and support for VK sparse bindings.
The concepts are:
1) xe_vm inherits from drm_gpuva_manager
2) xe_vma inherits from drm_gpuva
3) xe_vma_op inherits from drm_gpuva_op
4) VM bind operations (MAP, UNMAP, PREFETCH, UNMAP_ALL) call into the
GPUVA code to generate an VMA operations list which is parsed, committed,
and executed.
v2 (CI): Add break after default in case statement.
v3: Rebase
v4: Fix some error handling
v5: Use unlocked version VMA in error paths
v6: Rebase, address some review feedback mainly Thomas H
v7: Fix compile error in xe_vma_op_unwind, address checkpatch
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Not needed so remove it.
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This will help with the GPUVA port as the internals of struct xe_vma
will change.
v2: Update comment around helpers
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.kernel.org>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We cannot recover a VM if a rebind worker hits an error, ban the VM if
happens to ensure we do not attempt to place this VM on the hardware
again.
A follow up will inform the user if this happens.
v2: Return -ECANCELED in exec VM closed or banned, check for closed or
banned within VM lock.
v3: Fix lockdep splat by looking engine outside of vm->lock
v4: Fix error path when engine lookup fails
v5: Add debug message in rebind worker on error, update comments wrt
locking, add xe_vm_close helper
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_vm_create used the IOCTL create flags in a few places rather than the
internal VM flags and this just happened to work as these values
matched. This is risky (and incorrect) as the internal flag values are
free to change. Fix this and use the internal VM flag values.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Since commit ee6d3dd4ed ("driver core: make kobj_type constant.")
the driver core allows the usage of const struct kobj_type.
Take advantage of this to constify the structure definition to prevent
modification at runtime.
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Currently return from xe_gt_sysfs_init() is ignored
and also a failure in xe_gt_sysfs_init() isn't fatal
so make it return void.
V2 :
- add drm_warn in error paths - Himal
- Edit commit message - Nirmoy
Acked-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In the ABI header, GUC_CTB_MSG_MIN_LEN is '1' because
GUC_CTB_HDR_LEN is 1. This aligns with H2G/G2H CTB specification
where all command formats are defined in units of dwords so that '1'
is a dword. Accordingly, GUC_CTB_MSG_MAX_LEN is 256-1 (i.e. 255
dwords). However, h2g_write was incorrectly assuming that
GUC_CTB_MSG_MAX_LEN was in bytes. Fix this.
v3: Fix nit on #define location.(Matt)
v2: By correctly treating GUC_CTB_MSG_MAX_LEN as dwords, it causes
a local array to consume 4x the stack size. Rework the function
to avoid consuming stack even if the action size is large. (Matt)
Signed-off-by: Alan Previn <alan.previn.teres.alexis@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Little by little, make stuff feature complete.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If the default engine m->eng was used, there is no check for idle and
a cpu page-table update may thus happen in parallel with a gpu one.
Don't allow CPU page-table updates with the default engine until
the engine is idle.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230629205134.111849-2-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Newer Intel platforms require that inspect the contents of the GMD_ID
registers very early in the driver initialization process to determine
the IP version (and proper init sequences), of the platform. Move the
general PCI device setup and enablement slightly earlier, before we
start trying to peek at the GMD_ID registers.
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230614205202.3376752-5-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If the GMD_ID register reports a higher media revision ID than we're
expecting, print the media revid, not the graphics revid, in the
debug message.
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230614205202.3376752-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It's legitimate for the media GMD_ID register to read back as 0x0 if
media functionality is fused off or otherwise not present on the
platform. Avoid printing an "unknown media version" error message for
this case.
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230614205202.3376752-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
peek_gmdid() returns the IP version, not the raw value of the GMD_ID
register. Make sure we extract and return the rev_id field as well so
that it can be used to determine the IP steppings properly.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230614205202.3376752-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Current size member of vram struct does not give
complete information as what "size" contains. Does
it contain reserved portions or not. Name it usable
size and accordingly describe other size members as
well.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add sysfs entry to read per tile physical memory
including stolen memory.
V5:
- rename var name and make it part of vram struct - Lucas
V4:
- %s/addr_range/physical_vram_size_byes, make it
user readable name - Joonas/Aravind
- Display in bytes - Joonas/Aravind
V3:
- Exclude DG1, replace sysfs_create_file/files - Aravind
V2:
- Use DEVICE_ATTR_RO - Aravind
- Dont put kobj on sysfs_file_create fail - Himal
- Skip addr_range sysfs create for non dgfx - Himal
Reviewed-by: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With the separation of xe_tile and xe_gt, We now consider
a PCI device (xe_device) to contain one or more tiles (struct xe_tile).
Each tile will contain one or more GTs (struct xe_gt).
So lets align sysfs paths accordingly.
TODO: Currently we have gt0 under tile0 and gt1 under tile1
on multi-tile. This GT indexing still under discussion, when
it is concluded we need to revisit this change.
Reviewed-by: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We have recently introduced tile for each gpu,
so lets add sysfs entry per tile for userspace
to provide required information specific to tile.
V5:
- define ktype as const
V4:
- Reorder headers - Aravind
V3:
- Make API to return void and add drm_warn - Aravind
V2:
- Add logs in failure path
Reviewed-by: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Signed-off-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Using jiffies as a timeout from userspace is weird even if
theoretically exists possiblity of acquiring jiffies via getconf.
Unfortunately this method is unreliable and the returned
value may vary from the one configured in the kernel config.
Now timeout is expressed in nanoseconds and its interpretation depends
on setting DRM_XE_UFENCE_WAIT_ABSTIME flag. Relative timeout (flag
is not set) means fence expire at now() + timeout. Absolute timeout
(flag is set) means that the fence expires at exact point of time.
Passing negative timeout means we will wait "forever" by setting
wait time to MAX_SCHEDULE_TIMEOUT.
Cc: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://lore.kernel.org/r/20230628055141.398036-2-zbigniew.kempczynski@intel.com
Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If !no_engines, then we use copy_from_user to copy to the 'eci' array,
which has XE_HW_ENGINE_MAX_INSTANCE members. The amount of members
copied is given by the user in args->num_engines, so add code to check
that args->num_engines does not exceed XE_HW_ENGINE_MAX_INSTANCE. It's
an unsigned value so there's no need to check for negative values.
Fixes error messages such as:
Buffer overflow detected (54 < 18446744073709551520)!
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230626212221.136640-2-paulo.r.zanoni@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There's this shared machine running xe.ko and I often log in to see my
tmux corrupted by messages such as:
usercopy: Kernel memory overwrite attempt detected to wrapped address (offset 0, size 18446660151965198754)!
I also sometimes see:
kernel BUG at mm/usercopy.c:102!
Someone is running a program that's definitely submitting random
numbers to this ioctl. If you pass width=65535 and
num_placements=32769 then you get a negative 'len', which avoids the
EINVAL check, leading to the bug.
Switch 'len' to u32. It is the result of the multiplication of two u16
numbers, so it won't be able to overflow back into smaller numbers as
an u32.
v2: Make len u32 instead of checking for <=0 (José).
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230626212221.136640-1-paulo.r.zanoni@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The SLPC code has a strict 5ms timeout from when the start command is
queued to when we expect the reply to appear in memory. This works if
the CT channel is empty, but if the channel is busy there might be an
extra delay that causes the process to exceeded the timeout. We see
this issue when a reset occurs while userspace keeps submitting,
because the submission code is re-enabled first and it will start using
the channel to service those submissions.
To fix this, we can simply start SLPC before re-enabling submission.
This has also the benefit of not allowing submissions to go through with
an uninitialized SLPC.
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/375
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230628001642.3170070-1-daniele.ceraolospurio@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With the fake device creation fix in the previous patch,
adding Alderlake P platform in xe_wa_test.
With this, driver is able to run the kunit test for
ADLP properly.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230613174740.786041-2-anusha.srivatsa@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For platform like Alderlake P there are subplatforms and
just Alderlake P. Unlike DG2 in which every flavour is
either a G10,G11 or G12 variant. In this case(Alderlake P/S),
the Kunit test evaluates the subplatform to NONE and is
unable to create a fake device. Removing the condition
in xe_pci_fake_device_init() to support this corner case
so driver can proceed with the unit testing.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230613174740.786041-1-anusha.srivatsa@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Intel Discrete GFX cards gfx may have multiple PCIe endpoints,
they connects to root port via pcie upstream switch port(USP)
and virtual pcie switch port(VSP), sometimes VSP pcie devices
doesn't bind to pcieport driver. Without pcieport driver, pcie PM
comes without any warranty and with unbounded VSP gfx card won't
transition to low power pcie Device and Link states therefore
assert drm_warn on unbounded VSP and disable xe driver
PM support.
v2:
- Disable Xe PCI PM support. [Rodrigo]
v3:
- Changed subject and Rebase.
v4:
- %s/xe_pci_unbounded_bridge_disable_pm/xe_assert_on_unbounded_bridge.
[Rodrigo]
- Use device_set_pm_not_required() instead of dev_pm_ops NULL assignment.
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230524090653.1192566-1-anshuman.gupta@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The main difference is that we don't bounce and sync on eviction, allowing
for pipelined eviction. Moving forward we also need to be careful with
dma mappings which can be released in SYSTEM but may remain in TT.
v2:
- Remove a stale comment (Matthew Brost)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230626181741.32820-5-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The function ttm_bo_move_accel_cleanup() attempts to help pipeline a
move, and in doing so, needs memory allocations which may fail.
Rather than failing in a state where the new resource may freed while
accessed by the copy engine, sync uninterruptible and do a failsafe
cleanup.
v2:
- Don't try to attach the signaled fence on ttm_bo_move_accel_cleanup()
error.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230626181741.32820-4-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When creating a new bo, on the first move the bo->resource is typically
NULL. Our move callback rejected that instructing TTM to create a system
resource. In addition a struct ttm_tt with a page-vector was created,
although not populated with pages. Similarly when the clearing of VRAM
was complete, the system resource was put on a ghost object and freed
using the TTM delayed destroy mechanism.
This is a lot of pointless work. So avoid creating the system resource and
instead change the code to cope with a NULL bo->resource.
v2:
- Add some code comments (Matthew Brost)
v3:
- Fix a dereference of old_mem which might be NULL.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230626181741.32820-3-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When a source system resource had been swapped out, we incorrectly
assumed that we were lacking source data for a move and therefore
cleared the destination instead of swapping in and copying the
swapped-out data. Fix this.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230626181741.32820-2-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add the registers to get C6 residency of MTL SAMedia and
C6 status of MTL gts
v2:
- move register definitions to regs header (Anshuman)
- correct reg definition for mtl rc status
- make idle_status function common (Badal)
v3:
- remove extra line in commit message
- use only media type check in initialization
- use graphics ver check (Anshuman)
v4:
- remove extra lines (Anshuman)
Bspec: 66300
Signed-off-by: Badal Nilawar <badal.nilawar@intel.com>
Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
1) Add a new sysfs directory under devices/gt#/ called gtidle
to contain idle properties of GT such as name, idle_status,
idle_residency_ms
2) Remove forcewake calls for residency counter
v2:
- abstract using function pointers (Anshuman)
- remove forcewake calls for residency counter
- use device_attr (Badal)
- move rc functions to guc_pc
- change name to gt_idle (Rodrigo)
v3:
- return error for drmm_add_action_or_reset
- replace file and functions with gt_idle prefix
to gt_idle_sysfs (Himal)
- use enum for gt idle state
- move multiplier to gt idle and initialize (Anshuman)
- correct doc annotation (Rodrigo)
- remove return variable
- use kobj_gt instead of new gtidle kobj
- move residency_ms to gtidle file
- retain xe_guc_pc prefix for functions in guc_rc file (Michal)
v4:
- fix doc errors in xe_guc_pc file
- change u64 to u32 for reading residency counter
- keep gtidle states generic GT_IDLE_C[0/6] (Anshuman)
v5:
- update commit message to include removal of
forcewake calls (Anshuman)
- return void from sysfs initialization function and add warnings
(Andi)
v6:
- remove extra lines (Anshuman)
Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Anshuman Gupta <anshuman.gupta@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For VRAM allocations the bo->flags can control some characteristics of
the underlying memory, like whether it needs to be contiguous, and in
the future whether it needs to be in the CPU visible portion. Rather use
add_vram() in xe_bo_migrate() which should take care of such things for
us.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Since this is considered an info leak (system wide accounting), rather
hide behind perfmon_capable().
v2:
- Without perfmon_capable() it likely makes more sense to report as zero,
instead of reporting as used == total size. This should give similar
behaviour as i915 which rather tracks free instead of used.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Filip Hazubski <filip.hazubski@intel.com>
Cc: Carl Zhang <carl.zhang@intel.com>
Cc: Effie Yu <effie.yu@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The engine xarray holds a ref to engine, guarded by the lock.
While we do lookup for engine, we need to take the ref inside
the lock to prevent unreffed pointer escaping and
causing potential use-after-free after.
v2: remove branch prediction hint (Thomas)
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230602172732.1001057-1-mika.kuoppala@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Like commit 69a3738ba5 ("drm/i915: Skip applying copy engine fuses"),
do not apply copy engine fuses for platforms where MEML3_EN is not
relevant for determining the presence of the copy engines.
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230613180356.2906441-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When moving between PL_VRAM <-> PL_SYSTEM we have to have use PL_TT in
the middle as a temporary resource for the actual copy. In some GL
workloads it can be seen that once the resource has been moved to the
PL_TT we might have to bail out of the ttm_bo_validate(), before
finishing the final hop. If this happens the resource is left as
TTM_PL_FLAG_TEMPORARY, and when the ttm_bo_validate() is restarted the
current placement is always seen as incompatible, requiring us to
complete the move. However if the BO allows PL_TT as a possible
placement we can end up attempting a PL_TT -> PL_TT move (like when
running out of VRAM) which leads to explosions in xe_bo_move(), like
triggering the XE_BUG_ON(!tile).
Going from TTM_PL_FLAG_TEMPORARY with PL_TT -> PL_VRAM should already
work as-is, so it looks like we only need to worry about PL_TT -> PL_TT
and it looks like we can just treat it as a dummy move, since no real
move is needed.
Reported-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Use the TTM LRU bulk move for BOs tied to a VM. Update the bulk moves
LRU position on every exec.
v2: Bulk move for compute VMs, use WARN rather than BUG
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We only need to try to lock a BO if it's external as non-external BOs
share the dma-resv with the already locked VM. Trying to lock
non-external BOs caused an issue (list corruption) in an uncoming patch
which adds bulk LRU move. Since this code isn't needed, remove it.
v2: New commit message, s/mattthew/matthew/
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With our ref counting scheme long running (LR) engines only close
properly if not persistent, ensure that LR engines are non-persistent.
v2: spell out LR
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For long running (LR) jobs with the DRM scheduler we must return NULL in
run_job which results in signaling the job's finished fence immediately.
This prevents LR jobs from creating infinite dma-fences.
Signaling job's finished fence immediately breaks flow controlling ring
with the DRM scheduler. To work around this, the ring is flow controlled
and written in the exec IOCTL. Signaling job's finished fence
immediately also breaks the TDR which is used in reset / cleanup entity
paths so write a new path for LR entities.
v2: Better commit, white space, remove rmb(), better comment next to
emit_job()
v3 (Thomas): Change LR reference counting, fix working in commit
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add uAPI and implementation for NULL bindings. A NULL binding is defined
as writes dropped and read zero. A single bit in the uAPI has been added
which results in a single bit in the PTEs being set.
NULL bindings are intendedd to be used to implement VK sparse bindings,
in particular residencyNonResidentStrict property.
v2: Fix BUG_ON shown in VK testing, fix check patch warning, fix
xe_pt_scan_64K, update __gen8_pte_encode to understand NULL bindings,
remove else if vma_addr
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Suggested-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
ENOTSUPP is not a standard Unix error should use
EOPNOTSUPP instead.
v2: Update commit description (Aravind)
Reviewed-by: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Signed-off-by: Janga Rahul Kumar <janga.rahul.kumar@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The GuC can't access addresses above GUC_GGTT_TOP, so any GuC-accessible
objects can't be mapped above that offset. Instead of checking each
object to see if GuC may access it or not before mapping it, we just
limit the GGTT size to GUC_GGTT_TOP. This wastes a bit of address space
(about ~18 MBs, which is in addition to what already removed at the bottom
of the GGTT), but it is a good tradeoff to keep the code simple.
The in-code comment has also been updated to explain the limitation.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20230615002521.2587250-1-daniele.ceraolospurio@intel.com/
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This adds a handful of workarounds that apply to production steppings of
MTL:
- Wa_14018575942
- Wa_22016670082
- Wa_14017856879
- Wa_18019271663
Wa_22016670082 is currently only applied to the primary GT at the
moment, but may need to be extended to the media GT in the future if a
pending update to the workaround database gets finalized.
OOB workarounds will need to be implemented separately in future patches
for Wa_14016712196, Wa_16018063123, and Wa_18013179988.
Reviewed-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Link: https://lore.kernel.org/r/20230608181217.2385932-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It's not possible for the condition checking if we're running on
platform without geometry pipeline to ever be true, since
gt->fuse_topo.g_dss_mask is an array.
It also breaks the build:
../drivers/gpu/drm/xe/xe_rtp.c:183:50: error: address of array 'gt->fuse_topo.g_dss_mask' will always evaluate to 'true' [-Werror,-Wpointer-bool-conversion]
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230523135020.345596-2-michal@hardline.pl
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Using uninitialized variables leads to undefined behavior.
Moreover, it causes the compiler to complain with:
../drivers/gpu/drm/xe/xe_vm.c:3265:40: error: variable 'vma' is uninitialized when used here [-Werror,-Wuninitialized]
../drivers/gpu/drm/xe/xe_rtp.c:118:36: error: variable 'i' is uninitialized when used here [-Werror,-Wuninitialized]
../drivers/gpu/drm/xe/xe_mocs.c:449:3: error: variable 'flags' is uninitialized when used here [-Werror,-Wuninitialized]
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230523135020.345596-1-michal@hardline.pl
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
gt_count is only being incremented when initializing the primary GT;
since the media GT sets the ID directly, gt_count is not incremented
again, resulting in an incorrect count on MTL. Use autoincrement while
assigning the media GTs ID to ensure gt_count is correct on MTL and
other future platforms with standalone media.
Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Link: https://lore.kernel.org/r/20230613094232.3703549-1-riana.tauro@intel.com
[mattrope: Tweaked commit message to focus on gt_count importance]
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
RPSTAT1 is an sgunit register and thus doesn't need forcewake.
MTL_MIRROR_TARGET_WP1 is within an "always on" power domain and thus
doesn't require any forcewake to ensure the register is powered
up and usable. When GT is RC6 the actual frequency reported will be 0.
v2:
- Add bspec index (Anshuman)
- %s/GEN12_RPSTAT1/GT_PERF_STATUS as per bspec
v3: Update Fixes tag
Bspec: 51837, 67651
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Badal Nilawar <badal.nilawar@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230609024954.987039-1-badal.nilawar@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
One of the messages was printed without 0x prefix, so it was not clear
if it was decimal or hex: make sure to add the prefix by using %#x.
While at it, normalize the other messages in the same function to follow
the same pattern.
Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230611222447.2837573-3-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Remove the leftover TODO by renameing the functions to use xe prefix.
Since the static __gen8_pte_encode() already has a double score,
just remove the prefix.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230611222447.2837573-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
XE_PTE_FLAG_READ_ONLY is specific to struct xe_vma, move it from xe_bo.h
to xe_vm_types.h to reflect that.
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This define is for internal PTE flags rather than fields in the hardware
PTEs, rename as such. This will help in an upcoming patch to avoid
further confusion.
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
A mix of the system unbound wq and Xe ordered wq was used for the
rebind, only use the Xe ordered wq. This will ensure only 1 rebind is
occuring at a time providing a somewhat clunky work around for short
comings in TTM wrt to memory contention. Once the TTM memory contention
is resolved we should be able to use a dedicated non-ordered workqueue.
Also add helper to queue rebind worker to avoid using wrong workqueue
going forward.
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
A corner exists where a userptr may have no mapping when analyze VM is
called, handle this case.
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We need to flush render caches before fence signalling, where we might
release the memory for reuse. We can't rely on userspace doing this,
so flush render caches after the batch, but before user fence- and
dma_fence signalling.
Copy the cache flush from i915, but omit PIPE_CONTROL_FLUSH_L3, since it
should be implied by the other flushes. Also omit
PIPE_CONTROL_TLB_INVALIDATE since there should be no apparent need to
invalidate TLB after batch completion.
v2:
- Update Makefile for OOB WA.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Tested-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com> #1
Reported-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/291
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/291
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For scratch table mode we need to cover the case where a scratch PTE might
have been pre-fetched and cached and used instead of that of the newly
bound vma.
For compute vms, invalidate TLB globally using GuC before signalling
bind complete. For !long-running vms, invalidate TLB at batch start.
Also document how TLB invalidation works.
v2:
- Fix a pointer to the comment about TLB invalidation (Jose Souza).
- Add a bool to the vm whether we want to invalidate TLB at batch start.
- Invalidate TLB also on BCS- and video engines at batch start where
needed.
- Use BIT() macro instead of explicit shift.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Tested-by: José Roberto de Souza <jose.souza@intel.com> #v1
Reported-by: José Roberto de Souza <jose.souza@intel.com> #v1
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/291
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/291
Acked-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If RING_MAX_NONPRIV_SLOTS denotes the maximum number of whitelisting
slots, then it makes sense to refuse going above it.
v2:
- Use xe_gt_err() instead of drm_err() for more detailed info in the
error message. (Matt)
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230609143815.302540-3-gustavo.sousa@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
All other parameters can be extracted from a single struct xe_hw_engine
reference. This removes redundancy and simplifies the code.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230609143815.302540-2-gustavo.sousa@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The HXG fields are DW1 not DW0, fix this.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Acked-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Wa_14015795083 was already implemented for DG2 and PVC, but the
workaround database has been updated to extend it to more platforms. It
should now apply to all platforms with graphics versions 12.00 - 12.60,
as well as A-step of Xe_LPG (12.70 / 12.71).
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://lore.kernel.org/r/20230602231054.1306865-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The resizing of the PCI BAR is a best effort feature. If it is
not available, it should not fail the driver probe.
Rework the resize to not exit on failure.
Fixes: 7f075300a3 ("drm/xe: Simplify rebar sizing")
Acked-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Although PVC is currently the only platform that needs us to program a
GuC register with the index of an uncached MOCS entry, it's likely other
platforms will need this in the future. Rather than hardcoding PVC's
index into the register header, we should just pull the appropriate
index from gt->mocs.uc_index to future-proof the code.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230602235210.1314028-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_mocs_init_early doesn't touch the hardware, it just sets up internal
software state. There's no need to perform this step in the "forcewake
held" region. Moving the init earlier will also make the uc_index
values available earlier which will be important for an upcoming GuC
init patch.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230602235210.1314028-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reformat the GuC register header according to the same rules used by
other register headers:
- Register definitions are ordered by offset
- Value of #define's start on column 49
- Lowercase used for hex values
No functional change.
This header has some things that aren't directly related to register
definitions (e.g., number of doorbells, doorbell info structure, GuC
interrupt vector layout, etc. These items have been moved to the bottom
of the header.
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230602235210.1314028-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
DRM_ERROR() has been deprecated in favor of pr_err(). However, we should
prefer to use xe_gt_err() or drm_err() whenever possible so we get gt-
or device-specific output with the error message.
v2:
- Prefer drm_err() over pr_err(). (Matt, Jani)
v3:
- Prefer xe_gt_err() over drm_err() when possible. (Matt)
v4:
- Use the already available dev variable instead of xe->drm as
parameter to drm_err(). (Matt)
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Haridhar Kalvala <haridhar.kalvala@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230601194419.1179609-1-gustavo.sousa@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that tiles and GTs are handled separately and other prerequisite
changes are in place, we're ready to re-enable the media GT.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-31-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that a higher GT count can result from either multiple tiles (with
one GT each) or an extra media GT within the root tile, we need to
update the query code slightly to stop looking at tile_count.
FIXME: As noted previously, we need to decide on a formal direction for
exposing tiles and/or GTs to userspace.
v2:
- Drop num_gt() function in favor of stored xe->info.gt_count. (Brian)
v3:
- Keep XE_QUERY_GT_TYPE_REMOTE around for now. Userspace probably
doesn't actually need this, and we may remove it in the future, but
for now let's avoid changing uapi. (Brian)
Cc: Brian Welty <brian.welty@intel.com>
Reviewed-by: Brian Welty <brian.welty@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-30-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Allow xe_device_get_gt() and for_each_gt() to operate as expected on
platforms with standalone media.
FIXME: We need to figure out a consistent ID scheme for GTs. This patch
keeps the pre-existing behavior of 0/1 being the GT IDs for both PVC
(multi-tile) and MTL (multi-GT), but depending on the direction we
decide to go with uapi, we may change this in the future (e.g., to
return 0/1 on PVC and 0/2 on MTL). Or if we decide we only need to
expose tiles to userspace and not GTs, we may not even need ID numbers
for the GTs anymore.
v2:
- Restructure a bit to make the assertions more clear.
- Clarify in commit message that the goal here is to preserve existing
behavior; UAPI-visible changes may be introduced in the future once
we settle on what we really want.
v3:
- Store total GT count in xe_device for ease of lookup. (Brian)
- s/(id__++)/(id__)++/ (Gustavo)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Cc: Brian Welty <brian.welty@intel.com>
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-29-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Updates to the GGTT can happen when there are no in-flight jobs keeping
the hardware awake. If the GT is powered down when invalidation is
requested, we will not be able to communicate with the GuC (or MMIO) and
the invalidation request will go missing. Explicitly grab GT forcewake
to ensure the GT and GuC are powered up during the TLB invalidation.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-28-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The GGTT is part of the tile and is shared by the primary and media GTs
on platforms with a standalone media architecture. However each of
these GTs has its own TLBs caching the page table lookups, and each
needs to be invalidated separately.
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-27-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The majority of xe_gt_irq_postinstall() is really focused on the
hardware engine interrupts; other GT-related interrupts such as the GuC
are enabled/disabled independently. Renaming the function and making it
truly GT-specific will make it more clear what the intended focus is.
Disabling/masking of other interrupts (such as GuC interrupts) is
unnecessary since that has already happened during the irq_reset stage,
and doing so will become harmful once the media GT is re-enabled since
calls to xe_gt_irq_postinstall during media GT initialization would
incorrectly disable the primary GT's GuC interrupts.
Also, since this function is called from gt_fw_domain_init(), it's not
necessary to also call it earlier during xe_irq_postinstall; just
xe_irq_resume to handle runtime resume should be sufficient.
v2:
- Drop unnecessary !gt check. (Lucas)
- Reword some comments about enable/unmask for clarity. (Lucas)
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-26-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The xe_irq_postinstall() never actually gets called after installing the
interrupt handler. This oversight seems to get papered over due to the
fact that the (misnamed) xe_gt_irq_postinstall does more than it really
should and gets called in the middle of the GT initialization. The
callstack for postinstall is also a bit muddled with top-level device
interrupt enablement happening within platform-specific functions called
from the per-tile xe_gt_irq_postinstall() function.
Clean this all up by adding the missing call to xe_irq_postinstall()
after installing the interrupt handler and pull top-level irq enablement
up to xe_irq_postinstall where we'd expect it to be.
The xe_gt_irq_postinstall() function is still a bit misnamed here; an
upcoming patch will refocus its purpose and rename it.
v2:
- Squash in patch to actually call xe_irq_postinstall() after
installing the interrupt handler.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-25-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Although primary and media GuC share a single interrupt enable bit, they
each have distinct bits in the mask register. Although we always enable
interrupts for the primary GuC before the media GuC today (and never
disable either of them), this might not always be the case in the
future, so use a RMW when updating the mask register to ensure the other
GuC's mask doesn't get clobbered.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-24-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Our only use of GUnit interrupts is to handle ASLE backlight operations
that are reported as GUnit GSE interrupts. Move the enable/disable of
these interrupts to a more sensible place, in the same area where we
expect display interrupt code to be added by future patches.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-23-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
IRQ delivery and handling needs to be handled on a per-tile basis. Note
that this is true even for the "GT interrupts" relating to engines and
GuCs --- the interrupts relating to both GTs get raised through a single
set of registers in the tile's sgunit range.
On true multi-tile platforms, interrupts on remote tiles are internally
forwarded to the root tile; the first thing the top-level interrupt
handler should do is consult the root tile's instance of
DG1_MSTR_TILE_INTR to determine which tile(s) had interrupts. This
register is also responsible for enabling/disabling top-level reporting
of any interrupts to the OS. Although this register technically exists
on all tiles, it should only be used on the root tile.
The (mis)use of struct xe_gt as a target for MMIO operations in the
driver makes the code somewhat confusing since we wind up needing a GT
pointer to handle programming that's unrelated to the GT. To mitigate
this confusion, all of the xe_gt structures used solely as an MMIO
target in interrupt code are renamed to 'mmio' so that it's clear that
the structure being passed does not necessarily relate to any specific
GT (primary or media) that we might be dealing with interrupts for.
Reworking the driver's MMIO handling to not be dependent on xe_gt is
planned as a future patch series.
Note that GT initialization code currently calls xe_gt_irq_postinstall()
in an attempt to enable the HWE interrupts for the GT being initialized.
Unfortunately xe_gt_irq_postinstall() doesn't really match its name and
does a bunch of other stuff unrelated to the GT interrupts (such as
enabling the top-level device interrupts). That will be addressed in
future patches.
v2:
- Clarify commit message with explanation of why DG1_MSTR_TILE_INTR is
only used on the root tile, even though it's an sgunit register that
is technically present in each tile's MMIO space. (Aravind)
- Also clarify that the xe_gt used as a target for MMIO operations may
or may not relate to the GT we're dealing with for interrupts.
(Lucas)
Cc: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Aravind Iddamsetty <aravind.iddamsetty@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-22-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This media_gt pointer isn't actually allocated yet. Future patches will
start hooking it up at appropriate places in the code, and then creation
of the media GT will be added once those infrastructure changes are in
place.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-20-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In preparation for re-adding media GT support, switch the primary GT
within the tile to a dynamic allocation.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-19-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that tiles and GTs are handled separately, extra_gts[] doesn't
really provide any useful information that we can't just infer directly.
The primary GT of the root tile and of the remote tiles behave the same
way and don't need independent handling.
When we re-add support for media GTs in a future patch, the presence of
media can be determined from MEDIA_VER() (i.e., >= 13) and media's GSI
offset handling is expected to remain constant for all forseeable future
platforms, so it won't need to be provided in a definition structure
either.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-18-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The VRAM ID is always the tile ID; there's no need to track it
separately within a GT.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-17-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There are a bunch of places in the driver where we need to perform
non-GT MMIO against the platform's primary tile (display code, top-level
interrupt enable/disable, driver initialization, etc.). Rename
'to_gt()' to 'xe_primary_mmio_gt()' to clarify that we're trying to get
a primary MMIO handle for these top-level operations.
In the future we need to move away from xe_gt as the target for MMIO
operations (most of which are completely unrelated to GT).
v2:
- s/xe_primary_mmio_gt/xe_root_mmio_gt/ for more consistency with how
we refer to tile 0. (Lucas)
v3:
- Tweak comment on xe_root_mmio_gt(). (Lucas)
Acked-by: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-16-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Migration primarily focuses on the memory associated with a tile, so it
makes more sense to track this at the tile level (especially since the
driver was already skipping migration operations on media GTs).
Note that the blitter engine used to perform the migration always lives
in the tile's primary GT today. In theory that could change if media
GTs ever start including blitter engines in the future, but we can
extend the design if/when that happens in the future.
v2:
- Fix kunit test build
- Kerneldoc parameter name update
v3:
- Removed leftover prototype for removed function. (Gustavo)
- Remove unrelated / unwanted error handling change. (Gustavo)
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-15-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Since memory and address spaces are a tile concept rather than a GT
concept, we need to plumb tile-based handling through lots of
memory-related code.
Note that one remaining shortcoming here that will need to be addressed
before media GT support can be re-enabled is that although the address
space is shared between a tile's GTs, each GT caches the PTEs
independently in their own TLB and thus TLB invalidation should be
handled at the GT level.
v2:
- Fix kunit test build.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-13-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On platforms with VRAM, the VRAM is associated with the tile, not the
GT.
v2:
- Unsquash the GGTT handling back into its own patch.
- Fix kunit test build
v3:
- Tweak the "FIXME" comment to clarify that this function will be
completely gone by the end of the series. (Lucas)
v4:
- Move a few changes that were supposed to be part of the GGTT patch
back to that commit. (Gustavo)
v5:
- Kerneldoc parameter name fix.
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-11-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The GGTT exists at the tile level. When a tile contains multiple GTs,
they share the same GGTT.
v2:
- Include some changes that were mis-squashed into the VRAM patch.
(Gustavo)
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-9-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Each tile has its own register region in the BAR, containing instances
of all registers for the platform. In contrast, the multiple GTs within
a tile share the same MMIO space; there's just a small subset of
registers (the GSI registers) which have multiple copies at different
offsets (0x0 for primary GT, 0x380000 for media GT). Move the register
MMIO region size/pointers to the tile structure, leaving just the GSI
offset information in the GT structure.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-7-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
As we start splitting tile handling out from GT handling, we'll need to
be able to iterate over tiles separately from GTs. This iterator will
be used in upcoming patches.
v2:
- s/(id__++)/(id__)++/ (Gustavo)
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-6-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rather than a backpointer to the xe_device, a GT should have a
backpointer to its tile (which can then be used to lookup the device if
necessary).
The gt_to_xe() helper macro (which moves from xe_gt.h to xe_gt_types.h)
can and should still be used to jump directly from an xe_gt to
xe_device.
v2:
- Fix kunit test build
- Move a couple changes to the previous patch. (Lucas)
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Create a new xe_tile structure to begin separating the concept of "tile"
from "GT." A tile is effectively a complete GPU, and a GT is just one
part of that. On platforms like MTL, there's only a single full GPU
(tile) which has its IP blocks provided by two GTs. In contrast, a
"multi-tile" platform like PVC is basically multiple complete GPUs
packed behind a single PCI device.
For now, just create xe_tile as a simple wrapper around xe_gt. The
items in xe_gt that are truly tied to the tile rather than the GT will
be moved in future patches. Support for multiple GTs per tile (i.e.,
the MTL standalone media case) will also be re-introduced in a future
patch.
v2:
- Fix kunit test build
- Move hunk from next patch to use local tile variable rather than
direct xe->tiles[id] accesses. (Lucas)
- Mention compute in kerneldoc. (Rodrigo)
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Xe incorrectly conflates the concept of 'tile' and 'GT.' Since MTL's
media support is not yet functioning properly, let's just disable it
completely for now while we fix the fundamental driver design. Support
for media GTs on platforms like MTL will be re-added later.
v2:
- Drop some unrelated code cleanup that didn't belong in this patch.
(Lucas)
v3:
- Drop unnecessary xe_gt.h include. (Gustavo)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20230601215244.678611-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It looks like the driver only wants to track one vma for each external
object per vm. However it looks like bo_has_vm_references_locked() will
ignore any vma that is marked as vma->destroyed (not actually destroyed
yet). If we then mark our externally tracked vma as destroyed and then
create a new vma for the same object and vm, we can have two externally
tracked vma for the same object and vm. When the destroy actually
happens it tries to move the external tracking to a different vma, but
in this case it is already being tracked, leading to double list add
errors. It should be safe to simply drop the destroyed check in
bo_has_vm_references(), since the actual destroy will switch the
external tracking to the next available vma.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/290
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
__emit_job_gen12_render_compute() masks some PIPE_CONTROL bits that
do not exist in platforms without render engine.
So here replacing the PVC check by something more generic that will
support any future platforms without render engine.
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The _io_offset helper function is returning an offset into the GPU
address space. Using the CPU address offset (io_) is not correct.
Rename to reflect usage.
Update to use GPU offset information.
Update PT dma_offset to use the helper
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The current method of sizing GT device memory is not quite right.
Update the algorithm to use the relevant HW information and offsets
to set up the sizing correctly.
Update the stolen memory sizing to reflect the changes, and to be
GT specific.
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
"Right sizing" the PCI BAR is not necessary. If rebar is needed
size to the maximum available.
Preserve the force_vram_bar_size sizing.
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The _total_vram_size helper is device based and is not complete.
Teach the helper to be tile aware and add the ability to size
DG1 correctly.
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Lockdep is unhappy about ggtt->lock -> runtime_pm, where it seems
to think this can somehow get inverted. The ggtt->lock looks like a
potentially sensitive driver lock, so likely a sensible move to never
call the runtime_pm routines while holding it. Actually it looks like
d3cold wants to grab this, so perhaps this can indeed deadlock.
v2:
- Don't forget about xe_gt_tlb_invalidation_vma(), which now needs
explicit access_get.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Seems to be a sensitive lock, where ct->lock looks to be primed with
fs_reclaim, so holding that and then allocating memory will cause
lockdep to complain. We need to change the ordering wrt to grabbing the
ct->lock and potentially grabbing the runtime_pm, since some of the
runtime_pm routines can allocate memory (or at least that's what lockdep
seems to suggest).
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There is no mention that migrate_copy() will skip copying the CCS aux
state for all types of vram -> vram transfers. Currently we don't need
such a facility but might be surprising if we ever do.
v2: (Lucas):
- s/lmem/vram/ in the commit message
- Tidy up the code a bit; use one emit_copy_ccs()
v3:
- Reword the commit message
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
No need to allocate extra pages for this if we know flat-ccs AUX state
is not even possible, like for normal system memory objects.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Modify the xe_migrate_copy() function somewhat to explicitly allow
copying of data between two buffer objects including system memory
buffer objects. Update the migrate test accordingly.
v2:
- Check that buffer object sizes match when copying (Matthew Auld)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The TTM resource cursor was set up incorrectly.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Port Wa_14014475959 to xe_wa fixing its condition. The workaround should
only be applied on the primary GT, not on media. So just checking by
MTL platform is not enough: checking GT is of the right type is also
needed.
Since the GRAPHICS_STEP() does checks the GT type, we could leave the
first check as a platform one: it'd would be easier to understand and
not go out of sync with the graphics_ip_map[] in
drivers/gpu/drm/xe/xe_pci.c. However it also means that new platforms
using the same IP wouldn't match. Prefer using the IP version.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-22-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When running rules on MTL and beyond that have media as a standalone GT,
the rule should only match if the gt passed as parameter match the
version/range/stepping that the rule is checking. This allows
workarounds affecting only the media GT to be applied only on that GT
and vice-versa.
For platforms before MTL, the GT will not be of media type, even if it
includes media engines. Make sure to cover that case by checking if the
platforma has standalone media.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-21-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Port Wa_1509372804 to xe_wa so it's reported as active.
v2: Match workaround database, starting from A0 stepping (Matt Roper)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-20-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Wa_16015675438 and Wa_18020744125 apply to DG2 using the same action and
conditions. Add both to the oob rules so they are both reported as
active. Note that previously they were not checking by platform or IP
version, hence making them not future-proof. Those workarounds should
only be active in PVC and DG2, besides the check for "no render engine".
v2: From current WA database, Wa_16015675438 applies to all DG2
subplatforms except G11. Migrate condition to use subplatform and
remove G11 from the match (Matt Roper)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-19-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Wa_22012727170 and Wa_22012727685 apply to DG2 using the same action and
conditions. Add both to the oob rules so they are both reported as
active.
Do not Wa_22012727170 to PVC and MTL since only early A* steppings are
affected.
v2: Remove DG2_G10 from Wa_22012727685 to match current WA database
(Matt Roper)
v3: GRAPHICS_STEP(A0, FOREVER) can be left alone for DG2 as this means
all steppings
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-18-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Port Wa_16011777198 to xe_wa so it's reported as active.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-17-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Wa_14012197797 and Wa_22011391025 apply to DG2 using the same action.
They apply to slightly different conditions. Add both to the oob rules
so they are both reported as active.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-16-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Port Wa_16011759253 to oob. Wa_22011383443, that has the same action,
doesn't need to be ported as it targets early PVC steppings.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-15-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Let xe_guc.c start using XE_WA() for workarounds, starting from a simple
one: Wa_22012773006. It's also changed to start with graphics version
12, since that is the first supported by xe.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-14-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There are WAs that, due to their nature, cannot be applied from a
central place like xe_wa.c. Those are peppered around the rest of the
code, as needed. Now they have a new name: "out-of-band workarounds".
These workarounds have their names and rules still grouped in xe_wa.c,
inside the xe_wa_oob array, which is generated at compile time by
xe_wa_oob.rules and the hostprog xe_gen_wa_oob. The code generation
guarantees that the header xe_wa_oob.h contains the IDs for the
workarounds that match the index in the table. This way the runtime
checks that are spread throughout the code are simple tests against the
bitmap saved during initialization.
v2: Fix prev_name tracking not working when it's empty, i.e. when there
is more than 1 continuation rule.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-13-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When doing out-of-tree builds with O= or KBUILD_OUTPUT=, it's important
to also add the directory where the target is saved. Otherwise any file
generated by the build system may not be available for other targets
depending on it.
The $(obj) is added automatically when building the entire kernel,
but it's not added when M=drivers/gpu/drm/xe is added.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-12-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add a separate struct to hold entries in a table that has no action
associated with each of them. The goal is that the caller in future can
set a per-context callback, or just use the active entry marking
feature.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-11-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Start differentiating the media and graphics stepping as it will be
important for MTL. Note that RTP is still not prepared to handle the
different types of GT, i.e. checking for graphics version/range/stepping
on a media gt or vice versa still matches regardless of the gt being
passed as parameter. Changing it to accommodate MTL is left for a future
patch.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-10-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rename the RTP match in order to prepare the code base to check for the
media version. Up until MTL, the graphics vs media distinction wrt to
stepping was not ver relevant as they were the same GT. However, with
MTL this is no longer true.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-9-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add a "workarounds" node in debugfs that can dump all the active
workarounds using the information recorded by rtp infra when those
workarounds were processed.
v2: move workarounds to be reported per-GT
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-8-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Allocate the data to track workarounds on each gt of the device,
and pass that to RTP so the active workarounds are tracked.
Even if the workarounds available until now are mostly device
or platform centric, with the different IP versions for media and
graphics starting with MTL, it's possible that some workarounds
need to be applied only on select GTs. Also, given the workaround
database is per IP block, for tracking purposes there is no need to
differentiate the workarounds per engine class. Hence the bitmask
to track active workarounds can be tracked per GT.
v2: Move the tracking from per-device to per-GT basis (Matt Roper)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-7-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add the metadata in struct xe_rtp_process_ctx, to be set by
xe_rtp_process_ctx_enable_active_tracking(), so rtp knows how to mark
the active entries while processing the table. This can be used by the
WA infra to record what are the active workarounds.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-6-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The xe_rtp_process() function and xe_rtp_entry depend on the
save-restore struct. In future it will be desired to process rtp rules,
regardless of adding them to a save-restore. Rename the struct and
function so the intent is clear and the name is freed for future uses.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-5-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that rule_matches() always has an xe pointer, replace the XE_WARN_ON
with the more appropriate drm_warn().
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-4-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The selection between hwe and gt is exposed to the outside of rtp, by
the xe_rtp_process() function. However it doesn't make seense from the
caller point of view to pass a hwe and a gt as argument since the gt
should always be the one containing the hwe.
This clarifies the interface by separating the context creation into an
initializer. The initializer then passes the correct value and there
should never be a case with hwe and gt set: when hwe is passed, the gt
is the one containing it. Internally the functions continue receiving
the argument separately.
v2: Leave the device-only context to a separate patch if they are indeed
needed later
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-3-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It was missing one digit, so not showing up as a proper WA number. Add
the missing number and annotate it with a FIXME as there are more to be
implemented to consider this WA done: ensure CS is stop before doing a
reset, wait for pending.
Also, this WA applies to platforms up to graphics version 1270 (with the
exception of MTL A*, that are not supported in xe). Fix platform check.
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/284
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230526164358.86393-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Generally !has_flatccs implies that a platform has AuxCCS compression
and thus needs to invalidate the AuxCCS TLB. However PVC is a special
case because it has no compression of either type (FlatCCS or AuxCCS)
so we should avoid writing to non-existent AuxCCS registers.
Reviewed-by: Haridhar Kalvala <haridhar.kalvala@intel.com>
Link: https://lore.kernel.org/r/20230524192635.673293-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Padding and reserved fields are declared such that they must be
zeroed, so verify that they're all zero in the respective ioctl
functions.
Derived from original patch by mlankhorst.
v2:
Removed extensions checks where there were none originally. (José)
Moved extraneous parentheses to the correct places. (Lucas)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Christopher Snowhill <kode54@gmail.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The iommu_dma_map_sg() function ensures iova allocation doesn't
cross dma segment boundary. It does so by padding some sg elements.
This can cause overflow, ending up with sg->length being set to 0.
Avoid this by halving the maximum segment size (rounded down to
PAGE_SIZE).
Specify maximum segment size for sg elements by using
sg_alloc_table_from_pages_segment() to allocate sg_table.
v2: Use correct max segment size in dma_set_max_seg_size() call
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Reviewed-by: Bruce Chang <yu.bruce.chang@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For platforms with GMD_ID registers, the IP stepping should be
determined from the 'revid' field of those registers rather than from
the PCI revid.
The hardware teams have indicated that they plan to keep the revid =>
stepping mapping consistent across all GMD_ID platforms, with major
steppings (A0, B0, C0, etc.) having revids that are multiples of 4, and
minor steppings (A1, A2, A3, etc.) taking the intermediate values. For
now we'll trust that hardware follows through on this plan; if they have
to change direction in the future (e.g., they wind up needing something
like an "A4" that doesn't fit this scheme), we can add a GMD_ID-based
lookup table when the time comes.
v2:
- Set xe->info.platform before finding stepping; the pre-GMD_ID code
relies on this value to pick a lookup table.
v3:
- Also set xe->info.subplatform before picking the stepping for
pre-GMD_ID lookup.
Reviewed-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Link: https://lore.kernel.org/r/20230524185952.666158-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Let's make sure we give the driver a valid workqueue.
While at it, also make sure to call destroy_workqueue() only if the
workqueue is a valid one. That is necessary because xe_device_destroy()
is indirectly called as part of the cleanup process of a failed
xe_device_create().
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230518215651.502159-3-gustavo.sousa@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Otherwise no cleanup is actually done if we branch to err_put.
This works for now: currently we do know that, once inside
xe_device_destroy(), ttm_device_init() was successful so we can safely
call ttm_device_fini(); and, for xe->ordered_wq, there is an upcoming
commit to check its value before calling destroy_workqueue().
However, we might need change this in the future if we have more
initializers called that can fail in a way that we can not know which
one was it once inside xe_device_destroy().
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230518215651.502159-2-gustavo.sousa@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The function drm_dev_put() should also be called if xe_device_probe()
fails.
v2:
- Improve commit message. (Lucas)
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230519194802.578182-1-gustavo.sousa@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
drivers/gpu/drm/xe/xe_guc_submit_types.h:47: warning: cannot understand
function prototype: 'struct guc_submit_parallel_scratch '
drivers/gpu/drm/xe/xe_devcoredump_types.h:38: warning: Function
parameter or member 'ct' not described in 'xe_devcoredump_snapshot'
CI doesn't appear to be running BAT anymore, assuming this is caused by
the CI.Hooks now failing due to above warnings.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Both GUC_HOST_INTERRUPT and MED_GUC_HOST_INTERRUPT can pass
additional payload data to the GuC but this capability is not
used by the firmware yet.
Stop using value mandated by legacy GuC interrupt register and
use default notify value (zero) instead.
Bspec: 49813, 63363
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This GuC register can be moved together with the rest of the
GuC register definitions and be named in a similar way.
v2: fix placement
Bspec: 63363
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com> #v1
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There are multiple kind of config prints and with the upcoming
devcoredump there will be another layer. Let's limit the config
to the top level functions and leave the clean-up work for the
compilers so we don't create a spider-web of configs.
No functional change. Just a preparation for the devcoredump.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Let's continue to add our existent simple logs to devcoredump one
by one. Any format change should come on follow-up work.
v2: remove unnecessary, and now duplicated, dma_fence annotation. (Matthew)
v3: avoid for_each with faulty_engine since that can be already freed at
the time of the read/free. Instead, iterate in the full array of
hw_engines. (Kasan)
Cc: Francois Dugast <francois.dugast@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
The goal is to allow for a snapshot capture to be taken at the time
of the crash, while the print out can happen at a later time through
the exposed devcoredump virtual device.
v2: Addressing these Matthew comments:
- Handle memory allocation failures.
- Do not use GFP_ATOMIC on cases like debugfs prints.
- placement of @reg doc.
- identation issues.
v3: checkpatch
v4: Rebase and get back to GFP_ATOMIC only.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Let's start to move our existent logs to devcoredump one by
one. Any format change should come on follow-up work.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
The goal is to allow for a snapshot capture to be taken at the time
of the crash, while the print out can happen at a later time through
the exposed devcoredump virtual device.
v2: Handle memory allocation failures. (Matthew)
Do not use GFP_ATOMIC on cases like debugfs prints. (Matthew)
v3: checkpatch
v4: pending_list allocation needs to be atomic because of the
spin_lock. (Matthew)
get back to GFP_ATOMIC only. (lockdep).
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
These structs and definitions are only used for the guc_submit
and they were added specifically for the parallel submission.
While doing that also delete the unused struct guc_wq_item.
v2: checkpatch fixes.
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Let's start to move our existent logs to devcoredump one by
one. Any format change should come on follow-up work.
v2: Rebase and add the dma_fence locking annotation here.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
The goal is to allow for a snapshot capture to be taken at the time
of the crash, while the print out can happen at a later time through
the exposed devcoredump virtual device.
v2: Handle memory allocation failures. (Matthew)
Do not use GFP_ATOMIC on cases like debugfs prints. (Matthew)
v3: checkpatch fixes
v4: Do not use atomic in the g2h_worker_func (Matthew)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
No functional change here. The goal is to have a clear split between
the mapped portions of the CTB and the static information, so we can
easily capture snapshots that will be used for later read out with
the devcoredump infrastructure.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Unfortunately devcoredump infrastructure does not provide and
interface for us to force the device removal upon the pci_remove
time of our device.
The devcoredump is linked at the device level, so when in use
it will prevent the module removal, but it doesn't prevent the
call of the pci_remove callback. This callback cannot fail
anyway and we end up clearing and freeing the entire pci device.
Hence, after we removed the pci device, we shouldn't allow any
read or free operations to avoid segmentation fault.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
The goal is to use devcoredump infrastructure to report error states
captured at the crash time.
The error state will contain useful information for GPU hang debug, such
as INSTDONE registers and the current buffers getting executed, as well
as any other information that helps user space and allow later replays of
the error.
The proposal here is to avoid a Xe only error_state like i915 and use
a standard dev_coredump infrastructure to expose the error state.
For our own case, the data is only useful if it is a snapshot of the
time when the GPU crash has happened, since we reset the GPU immediately
after and the registers might have changed. So the proposal here is to
have an internal snapshot to be printed out later.
Also, usually a subsequent GPU hang can be only a cause of the initial
one. So we only save the 'first' hang. The dev_coredump has a delayed
work queue where it remove the coredump and free all the data within a
few moments of the error. When that happens we also reset our capture
state and allow further snapshots.
Right now this infra only print out the time of the hang. More information
will be migrated here on subsequent work. Also, in order to organize the
dump better, the goal is to propose dev_coredump changes itself to allow
multiple files and different controls. But for now we start Xe usage of
it without any dependency on dev_coredump core changes.
v2: Add dma_fence annotation for capture that might happen during long
running. (Thomas and Matt)
Use xe->drm.primary->index on drm_info msg. (Jani)
v3: checkpatch fixes
v4: Fix building and locking issues found by Francois.
Actually let's kill all of the locking in here. gt_reset serialization
already guarantee that there will be only one capture at the same time.
Also, the devcoredump has its own locking to protect the free and reads
and drivers don't need to duplicate it.
Besides this, the dma_fence locking was pushed to a following patch
since it is not needed in this one.
Fix a use after free identified by KASAN: Do not stash the faulty_engine
since that will be freed somewhere else.
v5: Fix Uptime - ktime_get_boottime actually returns the Uptime. (Francois)
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Francois Dugast <francois.dugast@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
While debugging GT related problems, it's good to know which GT was
reporting problems. Introduce helper macros to allow prefix GT logs
with GT identifier. We will use them in upcoming patches.
v2: use xe_ prefix (Lucas)
v3: use correct include
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This is allowed and encouraged by the dma-fencing rules. This along with
allowing compute VMs to export dma-fences on binds will result in a
simpler compute UMD.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Binds are not long running jobs thus we can export dma-fences even if a
VM is in compute mode.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Move xe_register_pci_driver() and xe_unregister_pci_driver() to
init_funcs to make sure that exit functions are also called when
xe_register_pci_driver() fails.
Note that this also allows adding init functions to be run after
xe_register_pci_driver().
v2:
- Move functions to init_funcs instead of having a special case for
xe_register_pci_driver(). (Jani)
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There is not much of a benefit from using that macro as of now and it
hurts grepability or other ways of cross-referencing.
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Alderlake P uses TGL HuC and it was not added together with ADL-S,
because it was failing for unrelated reasons. Now that those are fixed,
allow it to load HuC.
# cat /sys/kernel/debug/dri/0/gt0/uc/huc_info
HuC firmware: i915/tgl_huc.bin
status: RUNNING
version: wanted 0.0, found 7.9
uCode: 589504 bytes
RSA: 256 bytes
HuC status: 0x00090001
Reviewed-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Link: https://lore.kernel.org/r/20230512233649.3218736-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
ADL-N is pretty much the same as ADL-P (i.e., Xe_LP graphics + Xe_M
media + Xe_LPD display). However unlike ADL-P, there's no GuC hwconfig
support so the "tgl" GuC firmware should be loaded (i.e., the same
situation as ADL-S).
Acked-by: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Ravi Kumar Vodapalli <ravi.kumar.vodapalli@intel.com>
Link: https://lore.kernel.org/r/20230419213703.3993439-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Setup the mapping from PCI revid to IP stepping for ADL-P (and its RPL-P
subplatform) in case this information becomes important for implementing
workarounds.
Bspec: 55376
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230419213703.3993439-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Checking seqno_recv >= seqno looks like it will incorrectly report true
when the seqno has wrapped (not unlikely given
TLB_INVALIDATION_SEQNO_MAX). Calling xe_gt_tlb_invalidation_wait() might
then return before the flush has been completed by the GuC.
Fix this by treating a large negative delta as an indication that the
seqno has wrapped around. Similar to how we treat a large positive delta
as an indication that the seqno_recv must have wrapped around, but in
that case the seqno has likely also signalled.
It looks like we could also potentially make the seqno use the full
32bits as supported by the GuC.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fix the indent to align with open parenthesis, following the coding
style.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230508225322.2692066-5-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rename the address field to "addr" rather than "reg" so it's easier to
understand what it is.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230508225322.2692066-4-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Convert all the callers to deal with xe_mmio_*() using struct xe_reg
instead of plain u32. In a few places there was also a rename
s/reg/reg_val/ when dealing with the value returned so it doesn't get
mixed up with the register address.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230508225322.2692066-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With multiple active VMs, under memory pressure, it is possible that
ttm_bo_validate() run into -EDEADLK in ttm_mem_evict_wait_busy() and
return -ENOMEM.
Until ttm properly handles locking in such scenarios, best thing the
driver can do is unwind the lock and retry.
Update xe_exec_begin to retry validating BOs with a timeout upon
-ENOMEM.
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With multiple active VMs, under memory pressure, it is possible that
ttm_bo_validate() run into -EDEADLK in ttm_mem_evict_wait_busy() and
return -ENOMEM.
Until ttm properly handles locking in such scenarios, best thing the
driver can do is unwind the lock and retry.
Update preempt worker to retry validating BOs with a timeout upon
-ENOMEM.
v2: revert retry timeout upon -EAGAIN (Thomas)
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_file_close cleanups the xarrays but forgets
to destroy them causing a memleak in xarray internals.
Found with kmemleak.
Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Christoph Manszewski <christoph.manszewski@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Set atomic in xe_mmio_wait32() otherwise we would
be scheduling in atomic context.
Fixes: 7dc9b92dcf ("drm/xe: Remove i915_utils dependency from xe_pcode.")
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
guc_mmio_regset_write() had a flags for the registers to be added to the
GuC's regset list. The only register actually using that was RCU_MODE,
but it was setting the flags to a bogus value. From
struct xe_guc_fwif.h,
#define GUC_REGSET_MASKED BIT(0)
#define GUC_REGSET_MASKED_WITH_VALUE BIT(2)
#define GUC_REGSET_RESTORE_ONLY BIT(3)
Cross checking with i915, the only flag to set in RCU_MODE is
GUC_REGSET_MASKED. That can be done automatically from the register, as
long as the definition is correct.
Add the XE_REG_OPTION_MASKED annotation to RCU_MODE and kill the "flags"
field in guc_mmio_regset_write(): guc_mmio_regset_write_one() can decide
that based on the register being passed.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230429062332.354139-3-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The rest of the driver doesn't really support PVC before B0 stepping.
Drop the special handling in xe_guc.c.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230504073250.1436293-4-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On i915 the "see comment about Wa_22013088509" referred to the comment
in the graphics version >= 11 branch, where there were more details
about it. From the platforms supported by xe, only PVC needs
Wa_22013088509, but as the comment says, it's simpler to do it for all
platforms as there is no downside. Bring the missing comment over from
i915 and reword it to fit xe better.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230504073250.1436293-3-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Additional programming annotated with Wa_<number> should be reserved to
those that have a official workaround. Just pointing to a bug or
additional reference can be done with something else. Copy what i915
does and refer to it as "hsdes: ....".
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230504073250.1436293-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This is required at the minimum for the DRM UAPI to function from
32-bit userspace with a 64-bit kernel.
Signed-off-by: Christopher Snowhill <kode54@gmail.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Allow xe_bo_addr without lock to print debug information, such
as from xe_analyze_vm.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
copy cs instructions that dont have a explict MOCS field will use this
default MOCS value.
v2:
- move to xe_hw_engine.c
- remove BLIT_CCTL auxiliary macros
- removed MASKED_REG
v3:
- rebased
v4:
- process workaround in hwe->reg_lrc
v5:
- add a new function and call it from xe_gt_record_default_lrcs()
because hwe->reg_lrc is initialized later
BSpec: 45807
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
CS instructions that dont have a explicit MOCS field will use this
default MOCS value.
To do this, it was necessary to initialize part of the mocs earlier
and add new function that loads another array of rtp entries set
during run-time.
This is still missing to handle of mocs read for platforms with
HAS_L3_CCS_READ(aka PVC).
v2:
- move to xe_hw_engine.c
- remove CMD_CCTL auxiliary macros
v3:
- rebased
Bspec: 45826
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Things defined in <linux/init.h> are not really used by that header.
Replace that with <linux/types.h>, to have bool and u32 available.
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230503174922.252111-1-gustavo.sousa@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Raptorlake-P was tested and it is working as the same as Alderlake-P.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reading the GMD_ID register on platforms before that register
became available is not reliable. The assumption was that since the
register was not allocated, it would return 0. But on PVC for example it
returns garbage (or a very specific number), triggering the following
error:
xe 0000:8c:00.0: [drm] *ERROR* Hardware reports unknown media version 1025.55
Fix it by stop relying on the value returned by that registers on
platforms before GMD_ID. Instead this relies on the graphics description
struct being already pre-set on the device: this can only ever be true
for platforms before the GMD_ID support. In that case, GMD_ID is skipped
and the hardcoded values are used. This should also help on early
bring-up in case the GMD_ID returns something not expected and we need to
temporarily hardcode values. With this, PVC doesn't trigger the error
and goes straight to:
xe 0000:8c:00.0: [drm:xe_display_info_init [xe]] No display IP, skipping
xe 0000:8c:00.0: [drm:xe_pci_probe [xe]] XE_PVC 0bd5:002f dgfx:1 gfx:Xe_HPC (12.60) media:none (0.00) dma_m_s:52 tc:2
xe 0000:8c:00.0: [drm:xe_pci_probe [xe]] Stepping = (G:C0, M:**, D:**, B:B3)
Fixes: 5822bba943 ("drm/xe: Select graphics/media descriptors from GMD_ID")
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://lore.kernel.org/r/20230427184408.1340988-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The macros to handle the RTP tables are very scary, but shouldn't be
used outside of the header adding the infra. Move it to a separate
header and make sure it's only included when it can be.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-11-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that struct xe_reg and struct xe_reg_mcr are types that can be used
by xe, convert more of the driver to use them. Some notes about the
conversions:
- The RTP tables don't need the MASKED flags anymore in the
actions as that information now comes from the register
definition
- There is no need for the _XE_RTP_REG/_XE_RTP_REG_MCR macros
and the register types on RTP infra: that comes from the
register definitions.
- When declaring the RTP entries, there is no need anymore to
undef XE_REG and friends: the RTP macros deal with removing
the cast where needed due to not being able to use a compound
statement for initialization in the tables
- The index in the reg-sr xarray is the register offset only.
Otherwise we wouldn't catch mistakes about adding both a
MCR-style and normal-style registers. For that, the register
is now also part of the entry, so the options can be compared
to check for compatible entries.
In order to be able to accomplish this, some improvements are needed on
the RTP macros. Change its implementation to concentrate on "pasting a prefix
to each argument" rather than the more general "call any macro for each
argument". Hopefully this will avoid trying to extend this infra and
making it more complex. With the use of tuples for building the
arguments, it's not possible to pass additional register fields and
using xe_reg in the RTP tables.
xe_mmio_* still need to be converted, from u32 to xe_reg, but that is
left for another change.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-10-lucas.demarchi@intel.com
Link: https://lore.kernel.org/r/20230427223256.1432787-6-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Go over all registers used in xe_rtp tables and mark the registers as
masked if they were passed a XE_RTP_ACTION_FLAG(MASKED_REG) flag.
This will allow the flag to be removed in future when xe_rtp starts
using the real xe_reg_t type.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-9-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
These should replace the _MMIO() and MCR_REG() from i915, with the goal
of being more extensible, allowing to pass the additional fields for
struct xe_reg and struct xe_reg_mcr. Replace all uses of _MMIO() and
MCR_REG() in xe.
Since the RTP, reg-save-restore and WA infra are not ready to use the
new type, just undef the macro like was done for the i915 types
previously. That conversion will come later.
v2: Remove MEDIA_SOFT_SCRATCH_COUNT/MEDIA_SOFT_SCRATCH re-added by
mistake (Matt Roper)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-8-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Stop using i915 types for registers. Use our own types. Differently from
i915, this will keep under the register definition the knowledge for the
different types of registers. For now, the "flags"/"options" are mcr and
masked, although only the former is being used.
Additionally MCR registers have their own type. The only place that
should really look inside a xe_mcr_reg_t is that code dealing with the
steering and using other APIs when the register is MCR has been a source
of problem in the past.
Most of the driver is agnostic to the register differences since they
either use the definition from the header or already call the correct
MCR_REG()/_MMIO() macros. By embeding the struct xe_reg inside the
struct it's also possible to guarantee the compiler will break if
using RANDOM_MCR_REG.reg is attempted, since now the u32 is inside the
inner struct.
v2:
- Deep a dedicated type for MCR registers to avoid misuse
(Matt Roper, Jani)
- Drop the typedef and just use a struct since it's not an opaque type
(Jani)
- Add more kernel-doc
v3:
- Use only 22 bits for the register address since all the platforms
supported so far have only 4MB of MMIO per tile (Matt Roper)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-7-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Clarify a few things related to the PAT programming, particularly on
MTL:
- The register type doesn't change depending on the GT - what
happens is that media GT writes to other set of registers that
are not MCR
- Remove "UNICAST": otherwise it's confusing why it's not using
MCR registers with the unicast function variant
Also, there isn't much reason to keep those parts as macros: promote
them to proper functions and let the compiler inline if it sees fit.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-5-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Convert the macro declarations to the equivalent GENMASK and
and bitfield prep for all registers.
v2 (Matt Roper):
- Fix wrong conversion of RPM_CONFIG0_CRYSTAL_CLOCK_FREQ_MASK
- Reorder fields of XEHP_SLICE_UNIT_LEVEL_CLKGATE for consistency
- Simplify CTC_SOURCE_* by only defining CTC_SOURCE_DIVIDE_LOGIC
as REG_BIT(0)
v3: Also remove DOP_CLOCK_GATE_ENABLE that is unused and wrongly defined
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-4-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The defines for the registers were brought over from i915 while
bootstrapping the driver. As xe supports TGL and later only, it doesn't
make sense to keep the GEN* prefixes and suffixes in the registers: TGL
is graphics version 12, previously called "GEN12". So drop the prefix
everywhere.
v2:
- Also drop _TGL suffix and reword commit message as suggested
by Matt Roper. While at it, rename VSUNIT_CLKGATE_DIS_TGL to
VSUNIT_CLKGATE2_DIS with the additional "2", so it doesn't clash
with the define for the other register
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-3-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cleanup GuC register declarations by converting them to use REG_FIELD,
REG_BIT and REG_GENMASK. While converting, also reorder the bitfields
so they follow the convention of declaring the higher bits first.
v2:
- Drop unused HUC_LOADING_AGENT_VCR and DMA_ADDRESS_SPACE_GTT (Matt Roper)
- Simplify HUC_LOADING_AGENT_GUC define (Matt Roper)
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://lore.kernel.org/r/20230427223256.1432787-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Commit 7aaec3a623 ("drm/xe: Let's return last value read on
xe_mmio_wait32.") mentions that we should return the last value read,
but we never actually return it. This breaks display which depends on
the value being actually returned where needed.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fixes: 7aaec3a623 ("drm/xe: Let's return last value read on xe_mmio_wait32.")
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/257
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There's no good reason to keep the GuC registers outside the regs/
directory: move the header with GuC registers under that.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
That register is a completely different register, it's not the same as
SOFT_SCRATCH for GEN11 and beyond. Rename to to the same name as the
bspec uses, including the new variant for media. Also, move the
definitions to the guc header.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There was both BLT_DEPTH_32 and XY_FAST_COLOR_BLT_DEPTH_32 - also add
the prefix to the first to make it clear this is about the FAST_**COPY**
operation. While at it, remove the GEN9_ prefix.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Follow up commits will mass-remove the gen prefix/suffix. For GEN6_RC0
and GEN6_RC6 that would make the variable too short and easy to
conflict. So, add "GT_" prefix that is also part of the register name.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rename the following defines to lose the GEN* prefixes since they don't
make sense for xe:
GEN8_PTE_SHIFT -> XE_PTE_SHIFT
GEN8_PAGE_SIZE -> XE_PAGE_SIZE
GEN8_PTE_MASK -> XE_PTE_MASK
GEN8_PDE_SHIFT -> XE_PDE_SHIFT
GEN8_PDES -> XE_PDES
GEN8_PDE_MASK -> XE_PDE_MASK
GEN8_64K_PTE_SHIFT -> XE_64K_PTE_SHIFT
GEN8_64K_PAGE_SIZE -> XE_64K_PAGE_SIZE
GEN8_64K_PTE_MASK -> XE_64K_PTE_MASK
GEN8_64K_PDE_MASK -> XE_64K_PDE_MASK
GEN8_PDE_PS_2M -> XE_PDE_PS_2M
GEN8_PDPE_PS_1G -> XE_PDPE_PS_1G
GEN8_PDE_IPS_64K -> XE_PDE_IPS_64K
GEN12_GGTT_PTE_LM -> XE_GGTT_PTE_LM
GEN12_USM_PPGTT_PTE_AE -> XE_USM_PPGTT_PTE_AE
GEN12_PPGTT_PTE_LM -> XE_PPGTT_PTE_LM
GEN12_PDE_64K -> XE_PDE_64K
GEN12_PTE_PS64 -> XE_PTE_PS64
GEN8_PAGE_PRESENT -> XE_PAGE_PRESENT
GEN8_PAGE_RW -> XE_PAGE_RW
PTE_READ_ONLY -> XE_PTE_READ_ONLY
Keep an XE_ prefix to make sure we don't mix the defines for the CPU
(e.g. PAGE_SIZE) with the ones fro the GPU).
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On xe_hw_engine_print_state we were printing:
value_of(0x510) + 4 instead of
value_of(0x514) as desired.
So, let's properly define a RING_EXECLIST_SQ_CONTENTS_HI
register to fix the issue and also to avoid other issues
like that.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
The main issue with buddy allocator was fixed, but then
we ended up on other issues, so we need to step back and rethink
our strategy with D3cold. So, let's update the comment with a
todo list so we don't get tempted in removing it before we are
really ready.
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Riana Tauro <riana.tauro@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
ttm_global_init() imposes this limitation.
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Build without CONFIG_PM_SLEEP (such as for riscv) was failing due
to unused xe_pci_runtime_* functions.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When the device is runtime suspended, reading some of the sysfs
entries under device/gt#/ causes a resume error
This is due to the ordering of pm_runtime and forcewake calls.
Reorder to wake up using xe_pm_runtime_get and then forcewake
v2: add goto statements (Rodrigo)
Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_resize_vram_bar() function is already printing the status of bar
resizing. It has prints covering both success and failure.
There is no need of additional prints in the caller which were not so
easily to follow.
Modified all BAR size prints to consistently print the size in MiB.
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The write of GFX_FLSH_CNTL_GEN6 was inherited from the i915 codebase
where it was used to force a flush of the write-combine buffer in cases
where the GSM/GGTT were mapped as WC. Since Xe never uses WC mappings
of the GGTT, this register write is unnecessary. Furthermore, this
register was removed on Xe_HP-based platforms, so this write winds up
clobbering an unrelated register.
v2:
- Also drop GFX_FLSH_CNTL_GEN6 from the register file now that it's no
longer used. (Lucas)
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230418230247.3802438-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_mmio_rmw32 was failing to invert the passed in mask, resulting in a
register update that wasn't the expected RMW operation. Fortunately the
impact of this mistake was limited, since this function isn't heavily
used in Xe right now; this will mostly fix some GuC PM interrupt
unmasking.
v2:
- Rename parameters as 'clr' and 'set' to clarify semantics. (Lucas)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230421145006.10940-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If looking at the initial engine dump we should expect this to match
XE_FENCE_INITIAL_SEQNO - 1.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Doesn't look like we are accounting for seqno wrap. Just use
__dma_fence_is_later() like we already do for xe_hw_fence_signaled().
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The 'clear' field for register save/restore entries was being placed in
the value bits of the register rather than the mask bits; make sure it
gets shifted into the mask bits.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230419224909.4000920-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It looks like when tlb_invalidation.seqno overflows
TLB_INVALIDATION_SEQNO_MAX, we start counting again from one, as per
send_tlb_invalidation(). This is also inline with initial value we give
it in xe_gt_tlb_invalidation_init(). When calculating the
expected_seqno we should also take this into account.
While we are here also print out the values if we ever trigger the
warning.
v2 (José):
- drm_WARN_ON() is preferred over plain WARN_ON(), since it gives
information on the originating device.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/248
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add missing device info for Rocketlake.
While at it, also set the value for IS_ROCKETLAKE
macro which is right now set to 0.
v2: Also add abox_mask to the device info(Lucas)
v3: rebase
v4: Set IS_ROCKETLAKE (Anusha)
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Tested-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>(v2)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
CI keeps triggering:
xe 0000:03:00.0: [drm] Restricting VRAM size to PCI resource size
(0x400000000->0x3fa000000)
Due to usable_size vs vram_size differences. However, we only want to
trigger the drm_warn() to let developers know that the system they are
using is going clamp the VRAM size to match the IO size, where they can
likely only use 256M of VRAM. Once we properly support small-bar we can
revisit this.
v2 (Lucas): Drop the TODO for now
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rocketlake uses TGL GuC and HuC
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
PCODE_WRITE_MIN_FREQ_TABLE is only applicable to platforms with an LLC.
Change the discrete GPU check to an LLC check instead; this take care of
skipping not only the discrete platforms, but also integrated platforms
like MTL that do not have an LLC.
Fixes MTL dmesg error:
xe 0000:00:02.0: [drm] *ERROR* PCODE Mailbox failed: 1 Illegal Command
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230410183910.2696628-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Some driver initialization is conditional on the presence of an LLC.
Add an extra feature flag to support this.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230410183910.2696628-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Replace 'bool' fields with single bits to allow the various device
feature flags to pack more tightly.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230410183910.2696628-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
A VM had a runtime PM ref, a engine can't be created without a VM, and
the engine holds a ref to the VM thus this is unnecessary. Beyond that
taking a ref in the engine create IOCTL and dropping it in the destroy
IOCTL is wrong as a user doesn't have to call the destroy IOCTL (e.g.
they can just kill the process or close the driver FD). If a user does
this PM refs are leaked.
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The media GT requires a valid gt->kernel_bb_pool during driver probe to
allocate the WA and NOOP batchbuffers used to record default context
images. Dynamically allocate the bb_pools so that the primary and media
GT can use the same pool during driver init.
The media GT still shouldn't be need the USM pool, so only hook up the
kernel_bb_pool for now.
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230410200229.2726648-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The wait_event_timeout() on g2h_fence.wq which is declared on
stack can return before the wake_up() gets called, resulting in a
stack out of bound access when wake_up() accesses the g2h_fene.wq.
Do not declare g2h_fence related wait_queue_head_t on stack.
Fixes the below KASAN BUG and associated kernel crashes.
BUG: KASAN: stack-out-of-bounds in do_raw_spin_lock+0x6f/0x1e0
Read of size 4 at addr ffff88826252f4ac by task kworker/u128:5/467
CPU: 25 PID: 467 Comm: kworker/u128:5 Tainted: G U 6.3.0-rc4-xe #1
Workqueue: events_unbound g2h_worker_func [xe]
Call Trace:
<TASK>
dump_stack_lvl+0x64/0xb0
print_report+0xc2/0x600
kasan_report+0x96/0xc0
do_raw_spin_lock+0x6f/0x1e0
_raw_spin_lock_irqsave+0x47/0x60
__wake_up_common_lock+0xc0/0x150
dequeue_one_g2h+0x20f/0x6a0 [xe]
g2h_worker_func+0xa9/0x180 [xe]
process_one_work+0x527/0x990
worker_thread+0x2d1/0x640
kthread+0x174/0x1b0
ret_from_fork+0x29/0x50
</TASK>
Tested-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Bruce Chang <yu.bruce.chang@intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This stopped working now that TTM treats moving a pinned object through
ttm_bo_validate() as an error, for the general case. Add some new
routines to handle the new special casing needed for suspend-resume.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/244
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Tested-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that most of the characteristics of a device are associated with the
graphics and media IPs, the remaining contents of xe_device_desc can be
cleaned up a bit:
* 'gt' is unused; drop it
* DEV_INFO_FOR_EACH_FLAG only covers two flags and is only used in this
one file; drop the unnecessary macro complexity
* Convert .has_4tile to a single bitfield bit so that it can be packed
with the other feature flags
* Move 'platform' lower in the structure for better packing
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-10-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add a simple KUnit test to ensure that the hardware engine lists for
GMD_ID IP definitions are sensible (i.e., no graphics engines defined
for the media IP and vice versa).
Only the IP descriptors for GMD_ID platforms are checked for now.
Presumably the engine lists on older pre-GMD_ID platforms shouldn't be
changing. We can extend the KUnit testing in the future if we decide we
want to check those as well.
v2:
- Add missing 'const' in xe_call_for_each_media_ip to avoid compiler
warning.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-9-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If graphics_desc and media_desc are not specified in a platform's
xe_device_desc, treat this as an indication that the IP version should
be determined from the hardware's GMD_ID register.
Note that leaving media_desc unset for a platform that simply doesn't
have the IP (e.g., PVC) is also okay --- a read of the GMD_ID register
offset will be attempted, but since there's no register at that location
a value of '0' will be returned, effectively disabling media support.
Mapping of version -> IP description is done via a table lookup; this
table will be re-used in future patches for some KUnit testing.
v2:
- Drop dummy structures. NULL can be safely used for both the GMD_ID
cases and the "media not present case."
- Use a table-based lookup of GMD_ID versions rather than a simple
switch statement; the table will allow us to easily perform kunit
testing of all the IP descriptors.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-8-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Printing the name, along with the IP version number, will help reduce
confusion about which IP is present on a platform.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-7-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The total number of GTs supported by a platform should be one primary
GT, one media GT (if media version >= 13), and a number of remote tile
GTs dependent on the graphics IP present. Express this more clearly in
the device setup.
Note that xe->info.tile_count is inaccurately named; the rest of the
driver treats this as the GT count, not just the tile count. This
will need to be cleaned up at some point down the road.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-6-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Break the top-level platform_engine_mask field into separate
hw_engine_mask fields in the graphics and media structures. Since
hardware has more flexibility to mix-and-match IP versions going
forward, this allows each IP to list exactly which engines it provides;
the final per-GT engine list can then be constructured from those:
* On platforms without a standalone media GT (i.e., media IP versions
prior to 13), the primary GT's engine list is the union of the
graphics IP's engine list and the media IP's engine list.
* Otherwise, GT0's engine list is the graphics IP's engine list.
* For GT1 and beyond, the type of GT determines which IP's engine list
is used.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-5-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Most of the traits currently in the device descriptor structures are
either tied to the graphics IP or should be inferred from the graphics
IP. This becomes important on MTL and beyond where IP versions are
supposed to be detected from the hardware's GMD_ID registers rather than
mapped from PCI devid.
Engine masks are left where they are for now; they'll be dealt with
separately in a future patch.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Set require_force_probe explicitly in each platform's description
structure rather than embedding it within the FOO_FEATURES macros. Even
though we expect all platforms currently supported by the Xe driver to
be under force_probe protection, this will help prepare for some other
upcoming restructuring.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rather than storing all characteristics for an entire platform in the
xe_device_desc structure, create secondary graphics and media structures
to hold traits and feature flags specific to those IPs. This will
eventually allow us to assign the graphics and media characteristics at
runtime based on the contents of the relevant GMD_ID registers.
For now, just move the IP versions into the new structures to keep
things simple. Other IP-specific fields will migrate to these
structures in future patches.
Note that there's one functional change introduced by this: previously
PVC was recognized as media version 12.60. That's technically true, but
in practice the media engines are fused off on all production hardware.
By simply not assigning a media IP structure to PVC it will effectively
be treated as IP version 0.0 now (which the rest of the driver should
treat as non-existent media).
v2:
- Split the new structures out to their own header. This will ease the
addition of KUnit tests later.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230406235621.1914492-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When debugging issues related to GuC/HuC, it's important to know what is
the firmware version being used. The version from the filename can't be
relied upon, also because it normally only contains the major version
(except for the ones under experimental support).
Log the version from the blob after reading the CSS header. Example:
xe 0000:03:00.0: [drm] Using GuC firmware (70.5) from i915/dg2_guc_70.bin
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20230405224725.1993719-1-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If CCS0 was fused we did not write this register thus CCS engine were
not enabled resulting in driver load failures.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Update the logic to autoselect GuC/HuC for the platforms with the
following improvements:
- Document what is the firmware file that is expected to be
loaded and what is checked from blob headers
- When the platform is under force-probe it's desired to enforce
the full-version requirement so the correct firmware is used
before widespread adoption and backward-compatibility
commitments
- Directory from which we expect firmware blobs to be available in
upstream linux-firmware repository depends on the platform: for
the ones supported by i915 it uses the i915/ directory, but the ones
expected to be supported by xe, it's on the xe/ directory. This
means that for platforms in the intersection, the firmware is
loaded from a different directory, but that is not much important
in the firmware repo and it avoids firmware duplication.
- Make the table with the firmware definitions clearly state the
versions being expected. Now with macros to select the version it's
possible to choose between full-version/major-version for GuC and
full-version/no-version for HuC. These are similar to the macros used
in i915, but implemented in a slightly different way to avoid
duplicating the macros for each firmware/type and functionality,
besides adding the support for different directories.
- There is no check added regarding force-probe since xe should
reuse the same firmware files published for i915 for past
platforms. This can be improved later with additional
kunit checking against a hardcoded list of platforms that
falls in this category.
- As mentioned in the TODO, the major version fallback was not
implemented before as currently each platform only supports one
major. That can be easily added later.
- GuC version for MTL and PVC were updated to 70.6.4, using the exact
full version, while the
After this the GuC firmware used by PVC changes to pvc_guc_70.5.2.bin
since it's using a file not published yet.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Link: https://lore.kernel.org/r/20230324051754.1346390-4-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In order to avoid mistakes when populating the workarounds, it's good to
be able to test if the entries added are all compatible for a certain
platform. The platform itself is not needed as long as we create fake
devices with enough configuration for the RTP helpers to process the
tables. Common mistakes that can be avoided:
- Entries clashing the bitfields being updated
- Register type being mixed (MCR vs regular / masked vs regular)
- Unexpected errors while adding the reg_sr entry
To test, inject a duplicate entry in gt_was, but with platform == tigerlake
rather than the currenct graphics version check:
{ XE_RTP_NAME("14011059788"),
XE_RTP_RULES(PLATFORM(TIGERLAKE)),
XE_RTP_ACTIONS(SET(GEN10_DFR_RATIO_EN_AND_CHICKEN, DFR_DISABLE))
},
This produces the following result:
$ ./tools/testing/kunit/kunit.py run \
--kunitconfig drivers/gpu/drm/xe/.kunitconfig xe_wa
[14:18:02] Starting KUnit Kernel (1/1)...
[14:18:02] ============================================================
[14:18:02] ==================== xe_wa (1 subtest) =====================
[14:18:02] ======================== xe_wa_gt =========================
[14:18:02] [drm:xe_reg_sr_add] *ERROR* Discarding save-restore reg 9550 (clear: 00000200, set: 00000200, masked: no): ret=-22
[14:18:02] # xe_wa_gt: ASSERTION FAILED at drivers/gpu/drm/xe/tests/xe_wa_test.c:116
[14:18:02] Expected gt->reg_sr.errors == 0, but
[14:18:02] gt->reg_sr.errors == 1 (0x1)
[14:18:02] [FAILED] TIGERLAKE (B0)
[14:18:02] [PASSED] DG1 (A0)
[14:18:02] [PASSED] DG1 (B0)
...
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://lore.kernel.org/r/20230401085151.1786204-8-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add some basic unit tests for rtp. This is intended to prove the
functionality of the rtp itself, like coalescing entries, rejecting
non-disjoint values, etc.
Contrary to the other tests in xe, this is a unit test to test the
sw-side only, so it can be executed on any machine - it doesn't interact
with the real hardware. Running it produces the following output:
$ ./tools/testing/kunit/kunit.py run --raw_output-kunit \
--kunitconfig drivers/gpu/drm/xe/.kunitconfig xe_rtp
...
[01:26:27] Starting KUnit Kernel (1/1)...
KTAP version 1
1..1
KTAP version 1
# Subtest: xe_rtp
1..1
KTAP version 1
# Subtest: xe_rtp_process_tests
ok 1 coalesce-same-reg
ok 2 no-match-no-add
ok 3 no-match-no-add-multiple-rules
ok 4 two-regs-two-entries
ok 5 clr-one-set-other
ok 6 set-field
[drm:xe_reg_sr_add] *ERROR* Discarding save-restore reg 0001 (clear: 00000001, set: 00000001, masked: no): ret=-22
ok 7 conflict-duplicate
[drm:xe_reg_sr_add] *ERROR* Discarding save-restore reg 0001 (clear: 00000003, set: 00000000, masked: no): ret=-22
ok 8 conflict-not-disjoint
[drm:xe_reg_sr_add] *ERROR* Discarding save-restore reg 0001 (clear: 00000002, set: 00000002, masked: no): ret=-22
[drm:xe_reg_sr_add] *ERROR* Discarding save-restore reg 0001 (clear: 00000001, set: 00000001, masked: yes): ret=-22
ok 9 conflict-reg-type
# xe_rtp_process_tests: pass:9 fail:0 skip:0 total:9
ok 1 xe_rtp_process_tests
# Totals: pass:9 fail:0 skip:0 total:9
ok 1 xe_rtp
...
Note that the ERRORs in the kernel log are expected since it's testing
incompatible entries.
v2:
- Use parameterized table for tests (Michał Winiarski)
- Move everything to the xe_rtp_test.ko and only add a few exports to the
right namespace
- Add more tests to cover FIELD_SET, CLR, partially true rules, etc
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst<maarten.lankhorst@linux.intel.com> # v1
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://lore.kernel.org/r/20230401085151.1786204-7-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When there's an entry that is dropped when xe_reg_sr_add(), there's
not much we can do other than reporting the error - it's for certain a
driver issue or conflicting workarounds/tunings. Save the number of
errors to be used later by kunit to report where it happens.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230401085151.1786204-6-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Instead of requiring tests to initialize a fake device an keep it in
sync with xe_pci.c when it's platform-dependent, export a function from
xe_pci.c to be used and piggy back on the device info creation. For
simpler tests that don't need any specific platform and just need a fake
xe device to pass around, xe_pci_fake_device_init_any() can be used.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230401085151.1786204-5-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Instead of simply using EXPORT_SYMBOL() to export the functions needed
in xe.ko to be be called across modules, use EXPORT_SYMBOL_IF_KUNIT()
which will export the symbol under the EXPORTED_FOR_KUNIT_TESTING
namespace.
This avoids accidentally "leaking" these functions and letting them be
called from outside the kunit tests. If these functiosn are accidentally
called from another module, they receive a modpost error like below:
ERROR: modpost: module XXXXXXX uses symbol
xe_ccs_migrate_kunit from namespace EXPORTED_FOR_KUNIT_TESTING,
but does not import it.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Link: https://lore.kernel.org/r/20230401085151.1786204-4-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Move code out of xe_pci.[ch] into tests/*.[ch], like is done in other
similar compilation units. Even if this is not part of "tests for
xe_pci.c", they are functions exported and required by other tests. It's
better not to clutter the module headers and sources with them.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230401085151.1786204-3-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Extract the part setting up from xe->info from xe_pci_probe() into its
own function. This pairs nicely with the display counterpart, avoids
info initialization to be placed elsewhere and helps future
improvements to build fake devices for tests.
While at it, normalize the names a little bit: the _get() suffix may be
mistaken by lock-related operation, so rename the function to
"find_subplatform()". Also rename the variable to subplatform_desc to
make it easier to understand, even if longer.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230401085151.1786204-2-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Although our only multi-tile platform today (PVC) doesn't support
display, it's possible that some future multi-tile platform will.
If/when this happens, display interrupts (both traditional display and
ASLE backlight interrupts raised as a Gunit interrupt) should be
delivered to the primary tile. Save away tile0's master_ctl value so
that it can still be used for display interrupt handling after the GT
loop.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-9-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Although the hardware team has set aside some register bits for extra
media engines, no platform supported by the Xe driver today has VCS4-7
or VECS2-3. Drop the corresponding code (which was already commented
out); we can bring it back easily enough if such engines show up on a
future platform.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-8-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The remaining "gen11_*" IRQ functions are common to all platforms
supported by the Xe driver. Drop the unnecessary prefix.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-7-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Platforms supported by the Xe driver handle top-level interrupts in one
of two ways:
- Xe_LP platforms only have a "graphics master" register and lack a
"master tile" register, so top-level interrupt detection and
enable/disable happens in the graphics master.
- Xe_LP+ (aka DG1) and beyond have a "master tile" interrupt register
that controls the enable/disable of top-level interrupts and must
also be consulted to determine which tiles have received interrupts
before the driver moves on the process the graphics master register.
For functions that are only relevant to the first set of platforms,
rename the function prefix to Xe_LP since "gen11" doesn't make sense in
the Xe driver. Also add some comments briefly describing the two
top-level handlers.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-6-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Any interrupt registers that were introduced by platforms i915
considered to be "gen11" or "gen12" are present on all platforms that
the Xe driver supports; drop the unnecessary prefixes.
While working in the area, also convert a few open-coded bit
manipulations over to REG_BIT and REG_FIELD_GET notation.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-5-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Rodrigo: removed display. That was later squashed to the xe Display patch]
It's no longer necessary to wrap these operations in macros; a simple
function will suffice. Also switch to function names that more clearly
describe what operation is being performed: unmask_and_enable() and
mask_and_disable().
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For cases where IRQ_INIT and IRQ_RESET are used, the relevant interrupt
registers are always consecutive and ordered ISR, IMR, IIR, IER. Adding
helpers to look these up from a base offset will let us eliminate some
of the CPP pasting and simplify other upcoming patches.
v2:
- s/_REGS/_OFFSET/ for consistency. (Lucas)
- Move IMR/IIR/IER helpers into xe_irq.c; they aren't needed anywhere
else. (Lucas)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
"Gen" terminology should be avoided in the Xe driver and "gen3" refers
to platforms that are 9 (!!) graphics generations earlier than the
oldest supported by the Xe driver, so this prefix really doesn't make
sense.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230401002106.588656-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Currently, unload pvc driver will generate a null dereference
and the call stack is as below.
[ 4850.618000] Call Trace:
[ 4850.620740] <TASK>
[ 4850.623134] ttm_bo_cleanup_memtype_use+0x3f/0x50 [ttm]
[ 4850.628661] ttm_bo_release+0x154/0x2c0 [ttm]
[ 4850.633317] ? drm_buddy_fini+0x62/0x80 [drm_buddy]
[ 4850.638487] ? __kmem_cache_free+0x27d/0x2c0
[ 4850.643054] ttm_bo_put+0x38/0x60 [ttm]
[ 4850.647190] xe_gem_object_free+0x1f/0x30 [xe]
[ 4850.651945] drm_gem_object_free+0x1e/0x30 [drm]
[ 4850.656904] ggtt_fini_noalloc+0x9d/0xe0 [xe]
[ 4850.661574] drm_managed_release+0xb5/0x150 [drm]
[ 4850.666617] drm_dev_release+0x30/0x50 [drm]
[ 4850.671209] devm_drm_dev_init_release+0x3c/0x60 [drm]
There are a couple issues, but the main one is due to TTM has only
one TTM_PL_TT region, but since pvc has 2 tiles and tries to setup
1 TTM_PL_TT each tile. The second will overwrite the first one.
During unload time, the first tile will reset the TTM_PL_TT manger
and when the second tile is trying to free Bo and it will generate
the null reference since the TTM manage is already got reset to 0.
The fix is to use one global TTM_PL_TT manager.
v2: make gtt mgr global and change the name to sys_mgr
Cc: Stuart Summers <stuart.summers@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Vivi, Rodrigo <rodrigo.vivi@intel.com>
Signed-off-by: Bruce Chang <yu.bruce.chang@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Platform order in enum xe_platform started to be used by some parts of
the code, like the GuC/HuC firmware loading logic. The order itself is
not very important, but it's better to follow a convention: as was
documented in the comment above the enum, reorder the platforms by
graphics version. While at it, remove the gen terminology.
v2:
- Use "graphics version" instead of chronological order (Matt Roper)
- Also change pciidlist to follow the same order
- Remove "gen" from comments around enum xe_platform
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230331230902.1603294-1-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In xe_migrate functions, use proper vram io offset of the
tiles while calculating addresses.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In xe_migrate_sanity_kunit test, use correct expected value as
the expected value was not only used for the xe_migrate_clear(),
but also for the xe_migrate_copy() operation.
v2: Add 'Fixes' tag and update commit text
Fixes: 11a2407ed5 ("drm/xe: Stop accepting value in xe_migrate_clear")
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In xe_migrate_sanity_kunit test, use proper batch base address
by considering usm case.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The rev field is always 0 so it ends up never used. In i915 it was
introduced because of CML: up to rev 5 it reuses the guc and huc
firmware blobs from KBL. After that there is a specific firmware for
that platform. This can be reintroduced later if ever needed.
With the removal of revid the packed attribute in
uc_fw_platform_requirement, which is there only for reducing the space
these tables take, can also be removed since it has even more limited
usefulness: currently there's only padding of 2 bytes. Remove the
attribute to avoid the unaligned access.
$ pahole -C uc_fw_platform_requirement build64/drivers/gpu/drm/xe/xe_uc_fw.o
struct uc_fw_platform_requirement {
enum xe_platform p; /* 0 4 */
const struct uc_fw_blob blob; /* 4 10 */
/* size: 16, cachelines: 1, members: 2 */
/* padding: 2 */
/* last cacheline: 16 bytes */
};
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230324051754.1346390-2-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The MI_BATCH_BUFFER_END is already added automatically by
__xe_bb_create_job(); including it in the construction of the workaround
batchbuffer results in an unnecessary duplicate.
Link: https://lore.kernel.org/r/20230329173334.4015124-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We should WARN (not BUG) when creating a job if the batchbuffer does not
have sufficient space and padding. The hardware prefetch requirements
should also be considered.
Link: https://lore.kernel.org/r/20230329173334.4015124-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The hardware prefetches several cachelines of data from batchbuffers
before they are parsed. This prefetching only stops when the parser
encounters an MI_BATCH_BUFFER_END instruction (or a nested
MI_BATCH_BUFFER_START), so we must ensure that there is enough padding
at the end of the batchbuffer to prevent the prefetcher from running
past the end of the allocation and potentially faulting.
Bspec: 45717
Link: https://lore.kernel.org/r/20230329173334.4015124-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add some error messages describing the problem when
xe_gt_record_default_lrcs fails.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In general, this is due to FW load failure, should just report
error and fail the probe so that user can easily retry again.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Bruce Chang <yu.bruce.chang@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The tables are only used within this file; there's no reason for them
not to be static.
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230327175824.2967914-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Get rid of some of the duplication here. In a future patch we need to
also consider [fpfn, lpfn], so better adjust in only one place.
Suggested-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
So we don't have to keep repeating VRAM0 | VRAM1. Also if there are ever
more instances, then we have less places to update.
Suggested-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Starting with MTL, the number of entries in the PAT table increased to
16. The register offset jumped between index 7 and index 8, so a slight
adjustment is needed to ensure the PAT_INDEX macros select the proper
offset for the upper half of the table.
Note that although there are 16 registers in the hardware, the driver is
currently only asked to program the first 5, and we leave the rest at
their hardware default values. That means we don't actually touch the
upper half of the PAT table in the driver today and this patch won't
have any functional effect [yet].
Bspec: 44235
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230324210415.2434992-7-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Re-sync our MTL PAT table with the bspec. 1-way coherency should only
be set on table entry 3. We do not want an incorrect setting here to
accidentally paper over other bugs elsewhere in the driver.
Bspec: 45101
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230324210415.2434992-6-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The PAT_INDEX registers are MCR registers on some platforms and unicast
on others. On MTL the handling even varies between GTs: the primary GT
uses MCR registers while the media GT uses unicast registers. Let's add
proper MCR programming on the relevant platforms/GTs.
Given that we PAT tables to change pretty regularly on future platforms,
we'll make PAT programming an exception to the usual model of assuming
new platforms should inherit the previous platform's behavior. Instead
we'll raise a warning if the current platform isn't handled in the
if/else ladder. This should help prevent subtle cache misbehavior if we
forget to add the table for a new platform.
Bspec: 66534, 67609, 67788
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230324210415.2434992-4-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Provide per-platform tables of PAT values rather than per-platform
functions. This will simplify the handling of unicast vs MCR registers
in the upcoming patches.
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230324210415.2434992-3-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
PAT handling is growing in complexity and will continue to do so in
upcoming platforms. Separate it out to a dedicated file to keep things
tidy.
The code is moved as-is here (aside from a few unused #define's that are
just dropped); further changes will come in future patches.
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://lore.kernel.org/r/20230324210415.2434992-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rather waiting for the VM to be destroyed (all refs to VM go to zero),
drop the fault mode counts when the VM is closed in xe_vm_close_and_put.
This avoids a window where user space can create a faulting VM, close
it, and a subsequent creation of a non-faulting VM fails.
v2 (Lucas): Drop VLK reference in commit message
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Suggested-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Intel Vulkan driver needs to know what is the maximum priority to fill
a device info struct for applications.
Right now we getting this information by creating a engine and setting
priorities from min to high to know what is the maximum priority for
running process but this leads to info messages to be printed to
dmesg:
xe 0000:03:00.0: [drm] Ioctl argument check failed at drivers/gpu/drm/xe/xe_engine.c:178: value == DRM_SCHED_PRIORITY_HIGH && !capable(CAP_SYS_NICE)
It does not cause any harm but when executing a test suite like
crucible it causes thousands of those messages to be printed.
So here adding one more property to drm_xe_query_config to fetch the
max engine priority.
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Alderlake S uses TGL HuC.
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230323224651.1187366-3-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Follow the new direction of firmware and add macro
support for loading unversioned HuC. Keep HuC
versioned loading support as well for platforms
that fall under force_probe support
Add check to ensure driver does not do any version check
for HuC if going through unversioned load.
v2: unversioned firmware to be the default for platforms
not under force_probe. Maintain versioned firmware macro support
for platforms under force-probe protection.
v3: Minor style and naming adjustments (Lucas)
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230323224651.1187366-2-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Within a class the GuC will hault scheduling if the head of the queue
can't be scheduled the queue will block. This can lead to deadlock if
BCS0-7 all have faults and another engine on BCS0-7 is at head of the
GuC scheduling queue as the migration engine used to fix tthe fault will
be blocked. To work around this set the migration engine to the highest
priority when servicing page faults.
v2 (Maarten): Set priority to kernel once at creation
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Brian Welty <brian.welty@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rather than using the passed in GT, use the BO's GT determine dma_offset
when programming PTEs as these two GT's could differ (i.e. mapping a BO
from a remote GT). The BO's GT is correct GT to use as this where BO
resides, while the passed in GT is where the mapping is created.
v2:
(Thomas) - Kernel doc, extra new line
(CI) - Rebase to tip
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Make sure we pass along the correct errors.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Render / compute engines have additional caches (not just TLBs) that
need to be invalidated each batch, reinstate these invalidations in ring
ops.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Suggested-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_guc_ct_fast_path() is called from an irq context, and cannot lock
the mutex used by xe_device_mem_access_ongoing().
Fortunately it is easy to fix, and the atomic guarantees are good enough
to ensure xe->mem_access.hold_rpm is set before last ref is dropped.
As far as I can tell, the runtime ref in device access should be
killable, but don't dare to do it yet.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Zero-length arrays as fake flexible arrays are deprecated and we are
moving towards adopting C99 flexible-array members instead.
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Copy this from i915. We need .compatible for vram -> vram transfers, so
they don't just get nooped by ttm, if need to move something from
mappable to non-mappble or vice versa. The .intersects is needed for
eviction, to determine if a victim resource is worth eviction. e.g if we
need mappable space there is no point in evicting a resource that has
zero mappable pages.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Replace the allocation code with the i915 version. This simplifies the
code a little, and importantly we get the accounting at the mgr level,
which is useful for debug (and maybe userspace), plus per resource
tracking so we can easily check if a resource is using one or pages in
the mappable part of vram (useful for eviction), or if the resource is
completely within the mappable portion (useful for checking if the
resource can be safely CPU mapped).
v2: Fix missing PAGE_SHIFT
v3: (Gwan-gyeong Mun)
- Fix incorrect usage of ilog2(mm.chunk_size).
- Fix calculation when checking for impossible allocation sizes, also
check much earlier.
v4: (Gwan-gyeong Mun)
- Fix calculation when extending the [fpfn, lpfn] range due to the
roundup_pow_of_two().
v5: (Gwan-gyeong Mun)
- Move the check for running out of mappable VRAM to before doing any of
the roundup_pow_of_two().
v6: (Jani)
- Stop abusing BUG_ON(). We can easily just use WARN_ON() here and
return a proper error to the caller, which is much nicer if we ever
trigger these.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Although xe_migrate_clear() has a value argument, currently the driver
is only passing 0 at all the places this function is invoked with the
exception the kunit tests are using the parameter to validate this
function with different values.
xe_migrate_clear() is failing on platforms with link copy engines
because xe_migrate_clear() via emit_clear() is using the blitter
instruction XY_FAST_COLOR_BLT to clear the memory. But this instruction
is not supported by link copy engine.
So the solution is to use the alternate instruction MEM_SET when
platform contains link copy engine. But MEM_SET instruction accepts only
8-bit value for setting whereas the value agrument of xe_migrate_clear()
is 32-bit.
So instead of spreading this limitation around all invocations of
xe_migrate_clear() and causing more confusion, it was decided to not
accept any value itself as driver does not really need this currently.
All the kunit tests are adapted as per the new function prototype.
This will be followed by a patch to add support for link copy engines.
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When the GuC wopcm base and size registers are populated by BIOS/IFWI,
validate the parameters against the maximum allowed wopcm size.
Bpsec: 44982
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Hopefully not needed anymore. We can add a .compatible() hook once we
need to differentiate between mappable and non-mappable vram. If the
allocation is not contiguous then the start value is kind of
meaningless, so rather just mark as invalid.
In upstream, TTM wants to eventually remove the ttm_resource.start
usage.
References: 544432703b ("drm/ttm: Add new callbacks to ttm res mgr")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In order to improve readability and make it more future proof,
split the engine type from the graphics/platform checks.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230317223441.3891073-1-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
First step towards supporting small-bar is to track the io_size for
vram. We can longer assume that the io_size == vram size. This way we
know how much is CPU accessible via the BAR, and how much is not.
Effectively giving us a two tiered vram, where in some later patches we
can support different allocation strategies depending on if the memory
needs to be CPU accessible or not.
Note as this stage we still clamp the vram size to the usable vram size.
Only in the final patch do we turn this on for real, and allow distinct
io_size and vram_size.
v2: (Lucas):
- Improve the commit message, plus improve the kernel-doc for the
io_size to give a better sense of what it actually is.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If all compute engines of a vm in compute mode are idle,
defer a rebind to the next exec to avoid the VM unnecessarily trying
to make memory resident and compete with other VMs for available
memory space.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add a test parameter to force GPU page-table updates with the migrate
test and test both CPU- and GPU updates. Also provide some timing
results.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Causes an early 32-bit wrap and may thus help CI catch wrapping errors
that may otherwise not show early enough.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Introduce xe_engine_is_idle, and replace the static function in
xe_migrate.c.
The latter had two flaws. First the seqno == 1 test might return a false
true value each time the seqno counter wrapped, Second, the
cur_seqno == next_seqno test would never return true.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The migrate test currently supports only GPU pagetable updates and
will thus break if we fix the CPU pagetable update selection.
Fix the migrate test first.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Don't wait for GPU to be able to update page-tables using CPU. Putting
ourselves to sleep may be more of a problem than using GPU for
page-table updates. Also allow the vm to be NULL since the migrate
kunit test uses NULL for vm.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Also for HW fences, write the initial seqno - 1 to the HW completed
seqno to initialize.
v2:
- Use __dma_fence_is_later() to compare hw fence seqnos. (Matthew Auld)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The GuC/HuC firmware files used by Xe drivers are the same as
used by i915. Use the already-known location to find those
firmware files, for a couple of reasons:
1. Avoid having the same firmware placed on two different
places on MODULE_FIRMWARE(), if both 915 and xe drivers
are compiled;
2. Having firmware files located on different locations may end
creating bigger initramfs, as the same files will be copied
twice my mkinitrd/dracut/...;
3. this is the place where those firmware files are located at
https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git
Upstream doesn't expect them to have on other places;
4. When built with display support, DMC firmware will be
loaded from i915/ directory. It is very confusing to have
some firmware files on a different place for the same driver.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Lucas de Marchi <lucas.demarchi@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: David Airlie <airlied@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
[ Mostly agree with the direction of "use the firmware blobs from
upstream at their current location for these platforms". Previous
directory was not wrong as the plan was to have it handled in the
upstream firmware repo. For future platforms the location can be
changed if the support is only in xe ]
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230310081338.3275583-1-mauro.chehab@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It's just a local variable.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Synchronize LRC workarounds for graphics version 1200 with i915 up to
commit 7cdae9e9ee ("drm/i915: Move DG2 tuning to the right function").
These were probably missed for TGL/RKL before because in i915 it uses a
!IS_DG1() condition. Avoid a similar issue by just checking the
graphics version 1200 since DG1 is 1210.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-14-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add the one missing workaround for ADL-P when comparing to i915 up to
commit 7cdae9e9ee ("drm/i915: Move DG2 tuning to the right function").
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-13-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Synchronize with i915 the DG2 lrc workarounds as of
commit 4d14d7717f ("drm/i915/selftest: Fix ktime_get() and h/w access
order").
A few simplifications were done when the WA should be applied to some
steps of a subplatform and all the steppings of the other subplatforms.
In this case, it was simply applied to all the steppings, which only
means applying it to a few more A* steppings.
The implementation of the workaround 16011186671 triggers a bug in the
RTP infra: it's not possible to set the flag the usual way when having
multiple actions in the entry. This may be fixed later, but for now it's
sufficient to just set the flag directly without the helper macro.
v2: Fix 14014947963 to use FIELD_SET (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-12-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Synchronize with i915 the DG2 tunings as of
commit 4d14d7717f ("drm/i915/selftest: Fix ktime_get() and h/w access
order").
Contrary to the tuning "gang timer" for TGL, there is no quick
justification for why the read back is disabled in i915. Keep it
with that flag for now. That can be tentatively removed later when the
read values are checked.
v2: Use XEHP_FF_MODE2 instead of GEN12_FF_MODE2 (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-11-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Synchronize with i915 the DG2 gt workarounds as of
commit 4d14d7717f ("drm/i915/selftest: Fix ktime_get() and h/w access
order").
A few simplifications were done when the WA should be applied to
some steps of a subplatform and all the steppings of the other
subplatforms. This happened with Wa_1509727124, Wa_22012856258 and a few
others. In figure the pre-production steppings will be removed, so this
can be already simplified a little bit.
v2:
- Make 1308578152 conditional on first gslice fused off
- Add the missing Wa_1608949956/Wa_14010198302 (Matt Roper)
v3:
- Do not duplicate the implementation of 18019627453 since it's
already covered by other WA numbers in graphics versions 1200 and
1210
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-10-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Synchronize with i915 the DG2 gt workarounds as of
commit 4d14d7717f ("drm/i915/selftest: Fix ktime_get() and h/w access
order").
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-9-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Sync PVC engine workarounds with i915.
v2: Remove 16016694945. It was added by mistake. It's a GT workaround,
already present in the GT table (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-8-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Synchronize with i915 the PVC gt workarounds as of committ
commit 4d14d7717f ("drm/i915/selftest: Fix ktime_get() and h/w
access order").
v2: Add masked flag to XEHPC_LNCFMISCCFGREG0 (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-7-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that number of platforms is growing, it's getting hard to know the
workarounds for each platform. Split the entries inside the same table
so the workarounds checking IP version are listed first, followed by
each platform. Next step when it grows too much is to split in smaller
tables.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-6-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add debugfs entry to dump the final tables with register save-restore
information.
For the workarounds, this has a format a little bit different than when the
values are applied because we don't want to read the values from the HW
when dumping via debugfs. For whitelist it just re-uses the print
function added for when the whitelist is being built.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-5-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Besides printing the various register save-restore, it's also useful to
know the register being allowed/denied access from unprivileged batch
buffers. Print them during device probe.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-4-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add match helper to detect when the first gslice is fused off, as needed
by future workarounds.
v2:
- Add warning if called on a platform without geometry pipeline
(Matt Roper)
- Hardcode 4 as the number of gslices, which matches all the currently
supported platforms. PVC doesn't have geometry pipeline and
shouldn't use this function (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230314003012.2600353-2-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_ttm_stolen_cpu_inaccessible() was originally meant to just cover the
case where stolen is not directly CPU accessible on some older
integrated platforms, and as such a GGTT mapping was also required for
CPU access (as per the check in xe_bo_create_pin_map_at()).
However with small-bar systems on dgfx we have one more case where
stolen is also inaccessible, however here we don't have any fallback
GGTT mode for CPU access. Fix the check in xe_bo_create_pin_map_at() to
make this distinction clear. In such a case the later vmap() will fail
anyway.
v2: fix kernel-doc warning
v3: Simplify further and remove cpu_inaccessible()
Suggested-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Looks to have been introduced in some very recent changes, in-between
merging the driver wide s/lmem/vram/.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Be careful about having const in the compound literal initialization to
keep the initializers in rodata. Here, the impact is 1.8k of mutable
data moved to rodata.
add/remove: 0/1 grow/shrink: 0/0 up/down: 0/-1804 (-1804)
Data old new delta
__compound_literal 1804 - -1804
Total: Before=42425, After=40621, chg -4.25%
add/remove: 0/0 grow/shrink: 1/0 up/down: 1804/0 (1804)
RO Data old new delta
__compound_literal 7696 9500 +1804
Total: Before=138535, After=140339, chg +1.30%
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230309121746.479146-1-jani.nikula@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The PVC pre-production A* steppings are not going to be supported in xe
driver - the steppings are important for the WAs and since we are not
adding the pre-productions ones, there is no need to add the stepping.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This makes it easier when printing the register-save-restore values
to know what is the engine.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The dump function was originally added with the idea that it could be
re-used both for printing the reg-sr data and saving it to pass to GuC
via ADS. This was not used by the GuC integration, so remove it now to
give place to a new debug.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This allows to create WA/tuning rules that match the first engine that
is either of compute or render class. This matters for platforms that
don't have a render engine and that may have arbitrary compute engines
fused off: some register programming need to be added to one of those
engines.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Match functions are generally useful for other parts of the code (e.g.
xe_tuning.c). Move and rename the single one available to create a place
where similar match functions can be added.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Due to how xe_dss_mask_t is implemented, the type is a pointer. Since
this is only used for looking up the bits, make it const so it can be
used together with a const gt passed around.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Replace the inline function with a _Generic() so gt_to_xe() can work
with a const struct xe_gt*, which leads to a const struct xe *.
This allows a const gt being passed around and when the xe device is
needed, compiler won't issue a warning that calling gt_to_xe() would
discard the const. Rather, just propagate the const to the xe pointer
being returned.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add a sentence about the initialization so it's clear for newcomers how
to tweak the init functions for new platforms.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Some register ranges with replication type L3BANK were missing from the
driver table. The following warning was triggering when adding a
workaround touching the register 0xb188:
xe 0000:03:00.0: Did not find MCR register 0xb188 in any MCR steering table
Add the L3BANK ranges according to the spec.
v2:
- Fix typo in one of the ranges: s/0x00BCFF/0x008CFF/ (Matt Roper)
- Add termination rule in the init function for L3BANK (Matt Roper)
Bspec: 66534
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For Xe_HP platforms that can have multiple CCS engines, the
presence/absence of each CCS is inferred by the presence/absence of any
DSS in the corresponding quadrant of the GT's DSS mask.
This handling is only needed on platforms that can have more than one
CCS. The CCS is never fused off on platforms like MTL that can only
have one.
v2:
- Add extra warnings to try to catch mistakes where the register counts
in get_num_dss_regs() are updated without corresponding updates to
the register parameters passed to load_dss_mask(). (Lucas)
- Add kerneldoc for xe_gt_topology_has_dss_in_quadrant() and clarify
why we care about quadrants of the DSS space. (Lucas)
- Ensure CCS engine counting treats engine mask as 64-bit. (Lucas)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230309005530.3140173-2-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The single function to handle fuse registers for all types of engines is
becoming a bit long and hard to follow (and we haven't even added the
compute engines yet). Let's split it into dedicated functions for each
engine class.
v2:
- Add note about BCS0 always being present. (Bala)
- Add forcewake assertion to read_copy_fuses. (Bala)
Cc: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230309005530.3140173-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This seems to be the preferred nomenclature in xe. Currently we are
intermixing vram and lmem, which is confusing.
v2 (Gwan-gyeong Mun & Lucas):
- Rather apply to the entire driver
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Acked-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Checking whether a single engine's register save/restore entries
overflow the expected/pre-allocated GuC ADS regset area isn't terribly
useful; we actually want to check whether the combined entries from all
engines on the GT overflow the regset space.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20230308005509.2975663-1-matthew.d.roper@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The address set by firmware in GEN12_DSMBASE in driver initialization
doesn't mean "anything above that and until end of lmem is part of DSM".
In fact, there may be a few KB that is not part of DSM on the end of
lmem. How large is that space is platform-dependent, but since it's
always less than the DSM granularity, it can be simplified by simply
aligning the size down.
Suggested-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It appears we don't hold a memory access reference for the accesses in
this test, which may results in printed warnings and possibly the GT
not woken up for the memory accesses.
Add a memory access reference around the test.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We currently don't have any tests that explicitly depends on this
config option, so remove that build dependency.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When this register was added in xe for some workarounds, it was copied
from i915 before the registers got changed to add the MCR annotation.
The register 0xe4f4 is MCR since gen8, long before any GPU supported by
the xe driver. Replace all occurrences with the right register.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230306165757.633796-1-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The following warning shows up for TGL:
[drm:xe_reg_sr_add [xe]] *ERROR* Discarding save-restore reg 6604 (clear: 00ff0000, set: 00040000, masked: no): ret=-22
[drm:xe_reg_sr_add [xe]] *ERROR* Discarding save-restore reg 6604 (clear: 00ff0000, set: 00040000, masked: no): ret=-22
That is because the same register is being set both by the WAs and the
tunings. Like was done in i915, prefer the tuning over the workaround
since that is applicable for more platforms. Also fix the tuning: it
was incorrectly using the MCR version of the register, but that only
became true in XEHP.
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/233
Reported-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20230306212450.803557-1-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
XE_MAX_DSS_FUSE_REGS was being used to calculate the size of
xe_eu_mask_t while it should use XE_MAX_EU_FUSE_REGS.
There are no know issues about this but fixing it anyways.
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fix typo as noticed by Matt Roper:
git grep -l persitent | xargs sed -i 's/persitent/persistent/g'
... and then fix coding style issues.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://lore.kernel.org/r/20230302013411.3262608-2-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There's inconsistent use of mutex_init(), in xe_device_create(), with
several of them never calling mutex_destroy() in xe_device_destroy().
Migrate all of them to drmm_mutex_init(), so the destroy part is
automatically called.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://lore.kernel.org/r/20230225002138.1759016-2-lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In the depths of ttm, when allocating the vma node this should result in
-ENOSPC it seems. However we should probably rather reject as part of
our own ioctl sanity checking, and then treat as programmer error in the
lower levels.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With small-bar we likely want to annotate all the kernel users that
require CPU access with vram. If xe_bo_create_pin_map() is the central
place for that then we should have a central place to annotate.
This also simplifies the code and fixes what appears to be a double
xe_bo_put(hwe->hwsp) in the error handling.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The HAS_RENDER_L3CC is set unconditionally so there's no need to keep it
as a dedicated flag. For error checking purposes, we can just make sure
the 'table' field is initialized properly.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Suggested-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reprogramming the LNCF MOCS registers on render domain reset is not
intended to be regular driver programming, but rather the implementation
of a specific workaround (Wa_1607983814). This workaround no longer
applies on Xe_HP any beyond, so we can expect that these registers, like
the rest of the LNCF/LBCF registers, will maintain their values through
all engine resets. We should only add these registers to the GuC's
save/restore list on platforms that need the workaround.
Furthermore, xe_mocs_init_engine() appears to be another attempt to
satisfy this same workaround. This is unnecessary on the Xe driver
since even on platforms where the workaround is necessary, all
single-engine resets are initiated by the GuC and thus the GuC will take
care of saving/restoring these registers. The only host-initiated
resets we have in Xe are full GT resets which will already
(re)initialize these registers as part of the regular xe_mocs_init()
flow.
v2:
- Add needs_wa_1607983814() so that calculate_regset_size() doesn't
overallocate regset space when the workaround isn't needed. (Lucas)
- On platforms affected by Wa_1607983814, only add the LNCF MOCS
registers to the render engine's GuC save/restore list; resets of
other engines don't need to save/restore these.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It was incorrectly using dg2_mocs for now.
v2 (MattR):
- Use REG_GENMASK/REG_FIELD_PREP for bitfields
- Add bspec references
Bspec: 45101, 45410, 63882
Signed-off-by: Philippe Lecluse <philippe.lecluse@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The DG1 branch needlessly assigns uc_index twice. Drop the second
instance.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The values in the xe_mocs_info_index enum only match old pre-gen12
hardware not supported by the Xe driver.
The only usage of this enum was to set a default value for
info->unused_entries_index, but this is unnecessary since every platform
in the subsequent switch statement sets a proper platform-specific value
(and the XE_MOCS_PTE default doesn't even make sense since the hardware
dropped the "use PAT settings" capability in gen12).
v2:
- Add a check that unusued_entries_index is non-zero; even for
platforms where this is a valid table entry, it's never the one we
want this value assigned to. (Lucas)
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
RKL should use the same "gen12" MOCS handling as TGL/ADL-S/ADL-P.
Bspec: 45101
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
TGL/RKL/ADLS/ADLP are all supposed to use the same MOCS table, with
values defined in the bspec. Any entries listed in the bspec as
reserved/error/undefined should always be initialized to the most cached
and least coherent setting possible so that any userspace accidentally
referencing those undefined entries will only experience an increase in
coherency if spec updates down the road start defining real values.
The TGL and gen12 table entries defined in the driver today are
identical except that the TGL includes one additional (incorrect)
setting for table index 1. Furthermore, the TGL-specific initialization
does not define a dedicated value for info->unused_entries_index, so
this incorrect table entry 1 also gets used to populate the MOCS
registers for all reserved/unused table entries. This incorrect
behavior is a holdover from i915 where the platform was enabled with an
incorrect setting and by the time we noticed, it was too late to fix the
table without breaking ABI compatibility (and on TGL we did indeed have
some buggy userspace that was referencing the 'reserved' entry 1).
Since the Xe driver starts fresh with a clean slate on ABI, there's no
need to repeat the mistakes of i915 here.
v2:
- Reword/clarify commit message. (Lucas)
Bspec: 45101
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reduce the use of i915_reg_defs.h so it can be encapsulated in a single
place.
1) If it was being included by mistake, remove
2) If it was included for FIELD_GET()/FIELD_PREP()/GENMASK() and the
like, just include <linux/bitfield.h>
3) If it was included to be able to define additional registers, move
the registers to the relavant headers (regs/xe_regs.h or
regs/xe_gt_regs.h)
v2:
- Squash commit fixing i915_reg_defs.h include and with the one
introducing regs/xe_reg_defs.h
- Remove more cases of i915_reg_defs.h being used when all it was
needed was linux/bitfield.h (Matt Roper)
- Move some registers to the corresponding regs/*.h file (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[Rodrigo squashed here the removal of the i915 include]
The only thing really needed is the base offset, MCHBAR_MIRROR_BASE_SNB.
Remove the include and just define it inplace.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Move a few defines from xe_guc_pc.c to the right register, now that
there is one: xe_gt_regs.h.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Copy the macros used by xe in i915_reg.h to regs/xe_regs.h. A minimal
cleanup is done while copying so they adhere minimally to the coding
style. Further reordering and cleaning is left for later.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Copy the macros used by xe in intel_gpu_commands.h to
regs/xe_gpu_commands.h. PIPE_CONTROL_3D_ENGINE_FLAGS and
PIPE_CONTROL_3D_ARCH_FLAGS were already defined in
drivers/gpu/drm/xe/xe_ring_ops.c and only used there. So let that define
to be used instead of also adding to the new header.
v2: Let PIPE_CONTROL_3D_ENGINE_FLAGS/PIPE_CONTROL_3D_ARCH_FLAGS in the
only .c that uses it instead of redefining (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Create regs/xe_lrc_layout.h file with all the offsets used by the xe
driver. Eventually the xe driver may use a different way to define them
since it doesn't supported below gen12.
v2: Rename file to intel_lrc_layout.h since it's not really about
registers (Matt Roper)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Create regs/xe_gt_regs.h file with all the registers and bit
definitions used by the xe driver. Eventually the registers may be
defined in a different way and since xe doesn't supported below gen12,
the number of registers touched is much smaller, so create a new header.
The definitions themselves are direct copy from the
gt/intel_gt_regs.h file, just sorting the registers by address.
Cleaning those up and adhering to a common coding style is left for
later.
v2: Make the change to MCR_REG location in a separate patch to go
through the i915 branch (Matt Roper / Rodrigo)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Create regs/xe_engine_regs.h file with all the registers and bit
definitions used by the xe driver. Eventually the registers may be
defined in a different way and since xe doesn't supported below gen12,
the number of registers touched is much smaller, so create a new header.
The definitions themselves are direct copy from the
gt/intel_engine_regs.h file, just sorting the registers by address.
Cleaning those up and adhering to a common coding style is left for
later.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Sort includes and split them in blocks:
1) .h corresponding to the .c. Example: xe_bb.c should have a "#include
"xe_bb.h" first.
2) #include <linux/...>
3) #include <drm/...>
4) local includes
5) i915 includes
This is accomplished by running
`clang-format --style=file -i --sort-includes drivers/gpu/drm/xe/*.[ch]`
and ignoring all the changes after the includes. There are also some
manual tweaks to split the blocks.
v2: Also sort includes in headers
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Starting with MTL, the GT forcewake ack register moved from 0x130044 to
0xDFC. We expect this change to carry forward to future platforms as
well, so forcewake initialization should use an IP version check instead
of matching the MTL platform specifically.
The (re)definition of FORCEWAKE_ACK_GT_MTL in the forcewake file is also
unnecessary; we can take the definition that already exists in the
dedicated register header.
Bspec: 65031, 64629
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
During early generations of Intel GPUs, hardware engines would sometimes
move to new MMIO offsets from one platform/generation to the next.
These days engines the hardware teams put more effort into ensuring that
engines stay at consistent locations; even major design changes (like
the introduction of standalone media) keep the MMIO locations of the
engines constant.
Since all platforms supported by the Xe driver are new enough to have a
single MMIO offset for each engine (and since our crystal ball says that
these offsets are very unlikely to change again in the foreseeable
future), we can simplify the driver's engine definitions and remove the
gen-based MMIO bases.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In order to avoid -Werror=missing-prototypes, add the prototypes
in a separate tests/<test-name>_test.h file that is included by both
the implementation (tests/xe_<testname>.c, injected in xe.ko) and the
kunit module (tests/xe_<testname>_test.c -> xe-<testname>-test.ko).
v2: Add header and don't add ifdef to files that are already not built
when not using kunit (Matt Auld)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
In local_pci_probe() the core kernel increments the rpm for the device,
just before calling into the probe hook. If the driver/device supports
runtime pm it is then meant to put this ref during probe (like we do in
xe_pm_runtime_init()). However when removing the device we then also
need to take the reference back, otherwise the ref that is put in
pci_device_remove() will be unbalanced when for example unloading the
driver, leading to warnings like:
[ 3808.596345] xe 0000:03:00.0: Runtime PM usage count underflow!
Fix this by incrementing the rpm ref when removing the device.
v2:
- Improve the terminology in the commit message; s/drop/put/ etc (Lucas & Rodrigo)
- Also call pm_runtime_forbid(dev) (Rodrigo)
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/193
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
i915_regs.h is not needed, particularly in a header file. What is needed
is i915_reg_defs.h for use of _MMIO() and similar macros.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Use the more common "call cc-disable-warning" way to disable warnings.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
media_verx100 supersedes the info from media_ver. Leave media_ver in the
struct xe_device_desc, used in xe_pci.c since it's easier to define
common parts of the platforms like that. However all the rest of the
driver should be using media_verx100 that is more future proof.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/216
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Make xe_wait_user_fence.c include xe_wait_user_fence.h so it doesn't
rely on indirect includes and also doesn't fail the build due to missing
prototype for xe_wait_user_fence_ioctl().
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fix the following warning:
../drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c:55: warning: Function
parameter or member 'xe' not described in
'xe_ttm_stolen_cpu_inaccessible'
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_gt_topology_dss_group_mask and xe_gt_topology_count_dss are probably
leftover from initial implementation - they are not called from
anywhere. Remove those functions.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
LRC tunings were added after the gt ones and didn't add the call
in xe_gt_record_default_lrcs() to process them like is done for
workarounds. Add such a function and call it from
xe_gt_record_default_lrcs().
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
A few static functions not being declared like that break the build with
W=1, like e.g.
cc1: all warnings being treated as errors
make[2]: *** [../scripts/Makefile.build:250: drivers/gpu/drm/xe/xe_gt.o] Error 1
../drivers/gpu/drm/xe/xe_guc.c:240:6: error: no previous prototype for ‘guc_write_params’ [-Werror=missing-prototypes]
240 | void guc_write_params(struct xe_guc *guc)
| ^~~~~~~~~~~~~~~~
Make them static.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There are also some reserved fields in here which are not currently
cleared when handing back to userspace. Otherwise we might run into
issues if we later wish to use them.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Lucas De Marchi lucas.demarchi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The driver should still be functional with small-bar, just that the vram
size is clamped to the BAR size (until we add proper support for tiered
vram). For stolen vram we shouldn't iomap anything if the BAR size
doesn't also contain the stolen portion, since on discrete the stolen
portion is always at the end of normal vram. Stolen should still be
functional, just that allocating CPU visible io memory will always
return an error.
v2 (Lucas)
- Mention in the commit message that stolen vram is always as the end
of normal vram, which is why stolen in not mappable on small-bar
systems.
- Just make xe_ttm_stolen_inaccessible() return true for such cases.
Also rename to xe_ttm_stolen_cpu_inaccessible to better describe
that we are talking about direct CPU access. Plus add some
kernel-doc.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/209
Reported-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Make sure we properly release the forcewake ref on all error paths.
v2(Lucas):
- Make it less verbose and just fold the unimplemented options into
the default. The exact return value doesn't seem to matter for the
corresponding IGT.
- Replace the user triggerable WARN() with drm_dbg().
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
(!(gt->info.engine_mask & BIT(i))) cases are already
handled in the init function. And these masks are not
modified between the init and the prune.
Suggested-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
The list of GTs got splitted a while back between GT1
and GT2 on TGL.
References: https://patchwork.freedesktop.org/patch/388414/
CC: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Carlos Santa <carlos.santa@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
ret is not initialized in mcr_lock() when running in platforms with
graphics IP version < 1270, this could cause drm_WARN_ON_ONCE()
to hit eventually(what just happened to me).
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Just like there is support for multiple rules per entry in an rtp table,
also support multiple actions. This makes it easier to add support for
workarounds that need to change multiple registers. It also makes it
slightly more readable as now the action part resembles the rule part.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Entry flags is meant for the whole entry, including the rule
evaluation. Action flags are for flags applied to the register or
action being taken. Since there's only one action per entry, the
distinction was not important and a u8 was spared. However more and more
workarounds are needing multiple actions. This prepares for multiple
action support.
Right now there are these action flags:
- XE_RTP_ACTION_FLAG_MASKED_REG: register in the action is a masked
register
- XE_RTP_ACTION_FLAG_ENGINE_BASE: the engine base should be added to
the register in order to form the real address
And this entry flag:
- XE_RTP_ENTRY_FLAG_FOREACH_ENGINE: the rules should be evaluated for
each engine on the gt. It also automatically implies
XE_RTP_ACTION_FLAG_ENGINE_BASE.
Since there are likely not that many rules, reduce n_rules to u8 so the
overall entry size doesn't increase more than needed.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
It's true that the struct records the register and the value (in form of
2 masks) to restore, but it also records more fields important to
the application of workarounds/tuning, etc. One important part is what
is the macro used to record these fields: SET/CLR/WR/FIELD_SET/etc.
Thinking of the table as a set of rules + actions is more intuitive than
rules + regval.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Like detailed in commit 927dfdd09d ("drm/i915/dg2: Add SQIDI
steering"), some registers are expected to have the selector
initialized just once and never set to anything else. For xe, the
registers with SQIDI replication type (SF and MCFG) were missing,
resulting in warnings like:
[ 410.685565] xe 0000:03:00.0: Did not find MCR register 0x8724 in any MCR steering table
While adding these registers, abstract the handling for
"dg2_gam_ranges", moving them together with SF/MCFG to a dedicated
table. This also avoids that range to be checked for platforms other
than DG2. For DG2, this is the new steering output:
# cat /sys/kernel/debug/dri/0/gt0/steering
...
IMPLICIT steering: group=0x0, instance=0x0
0x000b00 - 0x000bff
0x001000 - 0x001fff
0x004000 - 0x004aff
0x008700 - 0x0087ff
0x00c800 - 0x00cfff
0x00f000 - 0x00ffff
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
There is already a BUILD_BUG_ON() check to make sure the size follow the
number of steering types. Also make sure the right index is being used
for each steering type.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The function pointer is already present as match_func, inside
struct xe_rtp_rule and handled as so instead of inside rtp_regval as
originally thought out when this was written.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_tuning.c should include xe_tuning.h, not xe_wa.h
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add missing "multicast" word and adapt/wrap the rest of the sentence.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Adding a debugfs dump of GGTT was useful for some debugging I did,
and easy to add. Might be useful for others too.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now that we issue TLB invalidations on unbinds and rebind from execs we
no longer need to issue TLB invalidations from the ring operations.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
If we add an TLB invalidation fence for rebinds issued from execs we
should be able to drop the TLB invalidation from the ring operations.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Rather than alias supports_usm to ASIS support, add an explicit
variable to indicate ASID support.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
This means we are in the middle of a GT reset and no need to do TLB
invalidation so just signal invalidation fence immediately.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
If a VM unbind hits an error, do not issue a TLB invalidation and
propagate the error the invalidation fence.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Make lockdep happy as we required to hold the GGTT when calling
xe_ggtt_map_bo.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Only the GuC should be issuing TLB invalidations if it is enabled. Part
of this patch is sanitize the device on driver unload to ensure we do
not send GuC based TLB invalidations during driver unload.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
If an bind operation fails we need to report it via the async fence.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
If the platform supports range based TLB invalidations use them. Hide
these details in the xe_gt_tlb_invalidation layer.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
This will help implementing range based TLB invalidations.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
This will help with TLB invalidation as the ASID in TLB invalidate
should be zero for platforms that do not support a ASID.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Endless fences are not good, add a TDR to cleanup any invalidation
fences which have not received an invalidation message within a timeout
period.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
This will help debug issues with TLB invalidation fences.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This gets tricky as we can't do the TLB invalidation until the unbind
operation is done on the hardware and we can't signal the unbind as
complete until the TLB invalidation is done. To work around this we
create an unbind fence which does a TLB invalidation after unbind is
done on the hardware, signals on TLB invalidation completion, and this
fence is installed in the BO dma-resv slot and installed in out-syncs
for the unbind operation.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Suggested-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com
Suggested-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fence will be signaled when TLB invalidation completion.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Suggested-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
TLB invalidations no longer just restricted to USM, move the variables
to own sub-structure.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
TLB invalidation is used by more than USM (page faults) so break this
code out into its own file.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We can't currently do this due to TLB invalidation done handler
expecting the seqno being received in-order, with the fast-path a TLB
invalidation done could pass one being processed in the slow-path in an
extreme corner case. Remove TLB invalidation done from the fast-path for
now and in a follow up reenable this once the TLB invalidation done
handler can deal with out of order seqno.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Scratch page is in VRAM, and therefore requires 64K GTT layout. In GGTT
world this just means having 16 consecutive entries, with 64K GTT
alignment for the GTT address of the first entry (also matching physical
alignment). However to keep things simple just dump it into system
memory, like we already do for ppGTT. While we are here, also give it
known default value.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Spec says we need to use 64K VRAM pages for GGTT on platforms like DG2.
In GGTT this just means aligning the GTT address to 64K and ensuring
that we have 16 consecutive entries each pointing to the respective 4K
entry. We already ensure we have 64K pages underneath, so it's just a
case of forcing the GTT alignment.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On DG2 when running the xe_vm IGT, the kernel generates loads of CAT
errors and GT resets (sometimes at least). On small-bar systems seems
to trigger a lot more easily (maybe due to difference in allocation
strategy). Appears to be related to scratch, since we seem to use the
64K TLB hint on scratch entries, even though the scratch page is a 4K
vram page. Bumping the scratch page size and physical alignment seems
to fix it. Or at least we no longer hit:
[ 148.872683] xe 0000:03:00.0: [drm] Engine memory cat error: guc_id=0
[ 148.872701] xe 0000:03:00.0: [drm] Engine memory cat error: guc_id=0
[ 148.875108] WARNING: CPU: 0 PID: 953 at drivers/gpu/drm/xe/xe_guc_submit.c:797
However to keep things simple, so we don't have to deal with 64K TLB
hints, just move the scratch page into system memory on platforms that
require 64K VRAM pages.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We need to ensure we don't leak the contents to userspace.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On DGFX this blows up if can call this with a system memory object:
XE_BUG_ON(!mem_type_is_vram(place->mem_type) && place->mem_type != XE_PL_STOLEN);
If we consider dpt it looks like we can already in theory hit this, if
we run out of vram and stolen vram. It at least seems reasonable to
allow calling this on any object which supports CPU access.
Note this also changes the behaviour with stolen VRAM and suspend, such
that we no longer attempt to migrate stolen objects into system memory.
However nothing in stolen should ever need to be restored (same on
integrated), so should be fine. Also on small-bar systems the stolen
portion is pretty much always non-CPU accessible, and currently pinned
objects use plain memcpy when being moved, which doesn't play nicely.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
I saw a flicker when booting xe, and it's very likely that the original
FB was not mapped at the same place when inheriting, fix it.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The comparison with < 0 suggests that the memory device access
should be signed to handle underflow. This makes it work more reliably.
As a result, the max refcount is now S32_MAX instead.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This is intended to get some properties that are of interest of UMDs
like the ban state.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Engine property get uAPI will be added, so to avoid ambiguity here
renaming XE_ENGINE_PROPERTY_X to XE_ENGINE_SET_PROPERTY_X.
No changes in behavior.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This aligns with other drivers and fixes build failure when
CONFIG_PM_SLEEP is not set, such as on RISC-V.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Some tests are meant to run only on real hardware. Skip those,
if no device was found.
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When adding the frequency management, Meteor Lake platform
was left behind. Handling it properly now.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Cc: Francois Dugast <francois.dugast@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
At present the interrupts are enabled while initializing the last GT.
But this is incorrect for a Multi-GT platform, as root GT initialization
will fail with interrupt disabled. Interrupts are required for
the GuC submission triggered during initialization.
Enable the interrupt during the root GT initialization.
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Some of the tests may benefit from running with ARCH=um, forgoing any
additional setup on the CI build side. Add min config for that.
Tested with:
./tools/testing/kunit/kunit.py build \
--kunitconfig drivers/gpu/drm/xe/.kunitconfig \
--jobs $(nproc) \
--build_dir build_kunit
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab@kernel.org>
mem_type field was added in commit d8b52a02cb ("drm/xe: Implement
stolen memory.") to designate the TTM memory type for that mgr. Add
kernel-doc with its description.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Fix the following error while building for 32b:
In file included from ../drivers/gpu/drm/xe/xe_gt.c:6:
../drivers/gpu/drm/xe/xe_gt.c: In function ‘gt_ttm_mgr_init’:
../include/linux/minmax.h:20:35: error: comparison of distinct pointer types lacks a cast [-Werror]
20 | (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
| ^~
Cast it to u64 so size of the second operand matches the first one when
building it for 32 bits.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Leave the types as u64, but cast the pointers to unsigned long before
assigning so the compiler doesn't throw warning about casting a pointer
to integer of different size.
Also, size_t should use %zu, not %ld.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
writeq() and readq() and other functions working on 64 bit variables
are not provided by 32b arch. For that it's needed to choose between
linux/io-64-nonatomic-hi-lo.h and linux/io-64-nonatomic-lo-hi.h,
spliting the read/write in 2 accesses. For xe driver, it doesn't matter
much, so just choose one and include in xe_mmio.h.
This also removes some ifdef CONFIG_64BIT we had around because of the
missing 64bit functions.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Either use the proper format or cast up to 64b depending on the case.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Make sure it's 64bit value in both 32b and 64b arch.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
As also cause issue on PVC, moving back
to what we did before stolen was
introduced
Signed-off-by: Philippe Lecluse <philippe.lecluse@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On DG2 we are now getting:
[ 104.456607] xe 0000:03:00.0: [drm] *ERROR* PCODE timeout, retrying with preemption disabled
Looks like we just need to invert the error check for
xe_pcode_try_request(), which returns zero on success.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
On DG1, BAR2 is not reliable for reporting Vram size, need to use GSMBASE.
Simplify xe_mmio_total_vram_size to report vram size and usable size.
Signed-off-by: Philippe Lecluse <philippe.lecluse@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The bo_create ioctl relied on the internal ordering of memory regions to
be the same, make sure we don't allocate stolen instead of VRAM0.
Also remove a debug warning left in.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Philippe Lecluse <philippe.lecluse1@gmail.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
SRIOV has a use case of GuC MMIO send / recv, add a function for it.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Philippe Lecluse <philippe.lecluse1@gmail.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Now aligns with the xe_guc_ct_send naming.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Philippe Lecluse <philippe.lecluse1@gmail.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Starting in 70.6.* GuC firmware the CSS header includes the submission
version, pull this from the CSS header. Prior 70.* versions accidentally
omitted this informatio so hard code to the correct values. This
information will be used by VFs when communicating with the PF.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Philippe Lecluse <philippe.lecluse1@gmail.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The blob doesn't fully support this yet, so fake for now to ensure our
driver load order is correct.
Once the blob supports pulling gt->info.engine_mask from the blob, this
patch can be removed.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This adds support for stolen memory, with the same allocator as
vram_mgr. This allows us to skip a whole lot of copy-paste,
by re-using parts of xe_ttm_vram_mgr.
The stolen memory may be bound using VM_BIND, so it performs like any
other memory region.
We should be able to map a stolen BO directly using the physical memory
location instead of through GGTT even on old platforms, but I don't know
what the effects are on coherency.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
When a job is inflight we may access memory to read the hardware seqno.
All user jobs have VM open which has a ref but kernel jobs do not
require VM so it is possible to not have memory ref. To avoid this, take
a memory ref on kernel job creation.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Expand xe_mmio_wait32 to accept atomic and then use
that directly when possible, and create own routine to
wait for the pcode status.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Another clean-up towards killing the usage of i915_utils.h
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
We can run the bit operation locally without yet another macro.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
We don't need any macro for a simple check we can do explicitly
and clear.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
To make it simpler, all of the status checks also waits and
times out.
Also, no ktime precision is needed in this case, and we
can use usleep_range because we are not in atomic paths here.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Rather than a constant check on proto and wait not busy,
let's wait for the expected success and then check the
protocol afterwards.
With this, we can now use the regular xe_mmio_wait32
and kill this local need for the wait_for.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Possible now that the wait function returns the last read value.
So we can remove the users of i915's wait_for one by one...
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
This is already useful because it avoids some extra reads
where registers might have changed after the timeout decision.
But also, it will be important to end the kill of i915's wait_for.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Then, move the i915_utils.h include to its user.
The overall goal is to kill all the usages of the i915_utils
stuff.
Yes, wait_for also depends on <linux/delay.h>, so they go
together to where it is needed. It will be likely needed
anyway directly for udelay or usleep_range.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Xe, is a new driver for Intel GPUs that supports both integrated and
discrete platforms starting with Tiger Lake (first Intel Xe Architecture).
The code is at a stage where it is already functional and has experimental
support for multiple platforms starting from Tiger Lake, with initial
support implemented in Mesa (for Iris and Anv, our OpenGL and Vulkan
drivers), as well as in NEO (for OpenCL and Level0).
The new Xe driver leverages a lot from i915.
As for display, the intent is to share the display code with the i915
driver so that there is maximum reuse there. But it is not added
in this patch.
This initial work is a collaboration of many people and unfortunately
the big squashed patch won't fully honor the proper credits. But let's
get some git quick stats so we can at least try to preserve some of the
credits:
Co-developed-by: Matthew Brost <matthew.brost@intel.com>
Co-developed-by: Matthew Auld <matthew.auld@intel.com>
Co-developed-by: Matt Roper <matthew.d.roper@intel.com>
Co-developed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Co-developed-by: Francois Dugast <francois.dugast@intel.com>
Co-developed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Co-developed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Co-developed-by: Philippe Lecluse <philippe.lecluse@intel.com>
Co-developed-by: Nirmoy Das <nirmoy.das@intel.com>
Co-developed-by: Jani Nikula <jani.nikula@intel.com>
Co-developed-by: José Roberto de Souza <jose.souza@intel.com>
Co-developed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Co-developed-by: Dave Airlie <airlied@redhat.com>
Co-developed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Co-developed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Co-developed-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
