CACHE_MODE_1 is a MCR register for all platforms that currently use it
in the Xe driver. Use XE_REG_MCR() when defining it.
Fixes: 8cd7e97597 ("drm/xe: Add missing DG2 lrc workarounds")
Fixes: ff063430ca ("drm/xe/mtl: Add some initial MTL workarounds")
Bspec: 66534, 67788
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patch.msgid.link/20260514-rtp-mcr-check-v3-1-30dd47855fee@intel.com
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Reading debugfs file (/sys/kernel/debug/dri/0/gt*/pf/adverse_events)
with CFI (Control Flow Integrity) enabled, the kernel panics at
xe_gt_debugfs_simple_show+0x82/0xc0.
xe_gt_debugfs_simple_show() declare a function pointer expecting int
return type, but xe_gt_sriov_pf_monitor_print_events() is void return
type, leading to CFI failure and kernel panic.
[507620.973657] CFI failure at xe_gt_debugfs_simple_show+0x82/0xc0 [xe]
(target: xe_gt_sriov_pf_monitor_print_events+0x0/0x130 [xe]; expected
type: 0xd72c7139)
Fix xe_gt_sriov_pf_monitor_print_events() function by updating to return
an int type.
Fixes: 1c99d3d3ed ("drm/xe/pf: Expose PF monitor details via debugfs")
Signed-off-by: Mohanram Meenakshisundaram <mohanram.meenakshisundaram@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patch.msgid.link/20260514174918.1556357-2-mohanram.meenakshisundaram@intel.com
We have plugged-in existing VF print functions into our GT debugfs
show helper as-is, but we missed that the helper expects functions
to return int, while they were defined as void. This can lead to
errors being reported when CFI is enabled.
Fixes: 63d8cb8fe3 ("drm/xe/vf: Expose SR-IOV VF attributes to GT debugfs")
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Mohanram Meenakshisundaram <mohanram.meenakshisundaram@intel.com>
Reviewed-by: Shuicheng Lin <shuicheng.lin@intel.com>
Link: https://patch.msgid.link/20260514155726.7165-1-michal.wajdeczko@intel.com
After removing runtime modification to our drm_driver definitions,
we can make them const as they should be.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patch.msgid.link/20260512183342.3374-8-michal.wajdeczko@intel.com
Instead of updating drm_driver.driver_features in the runtime, we
can use macro which value depends on the CONFIG_DRM_XE_DISPLAY.
And if display support is later disabled by the xe_modparam then we
will clear related bits in the drm_device.driver_features instead.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patch.msgid.link/20260512183342.3374-6-michal.wajdeczko@intel.com
Instead of updating the drm_driver.fbdev_probe field in the runtime,
we can use macro which value depends on the actual Kconfig setup.
The .fbdev_probe hook will not be used by the drm core unless we
also enable a DRIVER_MODESET driver feature flag, and this flag
still depends on the xe_modparam.probe_display parameter.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patch.msgid.link/20260512183342.3374-5-michal.wajdeczko@intel.com
There is no need to include xe_device.h in the xe_display.h header.
Include it in the xe_display.c file instead.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patch.msgid.link/20260512183342.3374-4-michal.wajdeczko@intel.com
Enable SR-IOV support for NVL platforms.
Signed-off-by: Jakub Kolakowski <jakub1.kolakowski@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Piotr Piórkowski <piotr.piorkowski@intel.com>
Cc: Dnyaneshwar Bhadane <dnyaneshwar.bhadane@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Tomasz Lis <tomasz.lis@intel.com>
Link: https://patch.msgid.link/20260331152259.58270-1-jakub1.kolakowski@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
The residency multiplier conversion in get_residency_ms() used the
floating-point literal 1e6 as the divisor of mul_u64_u32_div(). While
the compiler constant-folds this to an integer, using float literals
in kernel code is bad practice since the kernel generally avoids
floating-point operations.
Replace 1e6 with the standard NSEC_PER_MSEC macro from <linux/time64.h>,
which is both self-documenting (ns to ms conversion) and unambiguously
integer. Add the corresponding include rather than relying on
transitive inclusion.
No functional change.
Assisted-by: Claude:claude-opus-4.6
Reviewed-by: Nitin Gote <nitin.r.gote@intel.com>
Link: https://patch.msgid.link/20260511153307.223435-1-shuicheng.lin@intel.com
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
The error path in xe_gsc_init_post_hwconfig() explicitly frees a BO
allocated with xe_managed_bo_create_pin_map() via
xe_bo_unpin_map_no_vm(). Since the managed BO already has a devm
cleanup action registered, this causes a double-free when devm
unwinds during probe failure.
Remove the explicit free and let devm handle it, consistent with
all other xe_managed_bo_create_pin_map() callers.
Fixes: 2e5d47fe78 ("drm/xe/uc: Use managed bo for HuC and GSC objects")
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Assisted-by: Claude:claude-opus-4.6
Link: https://patch.msgid.link/20260511154134.223696-1-shuicheng.lin@intel.com
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
We only expect and handle the GT_MI_USER_INTERRUPT from the
engines, there is no point in enabling other interrupts, like
GT_CONTEXT_SWITCH_INTERRUPT, if we don't intent to handle them.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patch.msgid.link/20260511172838.2299-3-michal.wajdeczko@intel.com
To workaround some corner case hardware limitations, new programming
note for the memory based interrupt handler suggests to assume that
some status bytes, like GT_MI_USER_INTERRUPT and GUC_INTR_GUC2HOST,
are always set. Update our interrupt handler to follow the new rules.
Bspec: 53672
Fixes: a6581ebe76 ("drm/xe/vf: Introduce Memory Based Interrupts Handler")
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patch.msgid.link/20260511172838.2299-2-michal.wajdeczko@intel.com
During recent GGTT refactoring we missed to drop now unused field
from the xe_tile. Drop it now.
Fixes: e904c56ba6 ("drm/xe: Rewrite GGTT VF initialization")
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Maarten Lankhorst <dev@lankhorst.se>
Link: https://patch.msgid.link/20260510205605.642-1-michal.wajdeczko@intel.com
To perform copy, based on whether the platform supports service copy
engines, either MEM_COPY or XY_FAST_COPY_BLT instruction is used.
Length of both the instructions is same today and so they use a common
define EMIT_COPY_DW.
This is not true for the future platforms. Implement separate functions
which return the length of the instruction to help in preparing for it.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patch.msgid.link/20260511123746.616662-8-balasubramani.vivekanandan@intel.com
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Implement a function to return the length of the MEM_SET instruction.
This is to prepare for future platforms where the length of MEM_SET
instruction is expected to change.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patch.msgid.link/20260511123746.616662-7-balasubramani.vivekanandan@intel.com
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Implement a function which returns the length of XY_FAST_COLOR_BLT
instruction instead of hardcoding it inside the emit_clear_main_copy.
In future platforms, the length of this instruction is expected to
change and this patch helps in preparing for it.
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patch.msgid.link/20260511123746.616662-6-balasubramani.vivekanandan@intel.com
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
The FIXME says that xe_engine_snapshot_print.. is accessing persistent
driver data, unlike what the FIXME says that it does. Drop the FIXME
since the current code is not going to access the hardware while
dumping.
More details about this patch:
https://patchwork.freedesktop.org/patch/703884/?series=161407&rev=1
The starting two feedbacks make sense and the original patch is wrong
in adding those changes, but the last feedback is the one which
highlights the point.
Signed-off-by: Shekhar Chauhan <shekhar.chauhan@intel.com>
Reviewed-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Link: https://patch.msgid.link/20260512055508.1380191-1-shekhar.chauhan@intel.com
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Do not allow SYS_MEM_LAT_MEASURE_EN bit to be set in SYS_MEM_LAT_MEASURE
register. Doing so can cause memory corruption.
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patch.msgid.link/20260430161459.2892545-5-ashutosh.dixit@intel.com
Add val arg to xe_oa_is_valid_config_reg so that register values can also
be verified, in addition to register address. Value verification is needed
to implement MERTOA Wa_14026779378.
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patch.msgid.link/20260430161459.2892545-4-ashutosh.dixit@intel.com
Wa_14026746987 implies that only XE_OAM_FORMAT_MPEC8u32_B8_C8 and not
XE_OAM_FORMAT_MPEC8u64_B8_C8 can be supported for MERTOA unit.
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patch.msgid.link/20260430161459.2892545-3-ashutosh.dixit@intel.com
Retry doesn't work here, since bo will be freed on error, leading to
UAF. However, now that we do the alloc & init before the attach, we can
now combine this as one unit and have the init do the alloc for us. This
should make the retry safe.
Reported by Sashiko.
v2: Fix up the error unwind (CI)
Closes: https://sashiko.dev/#/patchset/20260506184332.86743-2-matthew.auld%40intel.com
Fixes: eb289a5f6c ("drm/xe: Convert xe_dma_buf.c for exhaustive eviction")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.18+
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patch.msgid.link/20260508102635.149172-4-matthew.auld@intel.com
There look to be some nasty races here when triggering the
invalidate_mappings hook:
1) We do xe_bo_alloc() followed by the attach, before the actual full bo
init step in xe_dma_buf_init_obj(). However the bo is visible on the
attachments list after the attach. This is bad since exporter driver,
say amdgpu, can at any time call back into our invalidate_mappings hook,
with an empty/bogus bo, leading to potential bugs/crashes.
2) Similar to 1) but here we get a UAF, when the invalidate_mappings
hook is triggered. For example, we get as far as xe_bo_init_locked()
but this fails in some way. But here the bo will be freed on error, but
we still have it attached from dma-buf pov, so if the
invalidate_mappings is now triggered then the bo we access is gone and
we trigger UAF and more bugs/crashes.
To fix this, move the attach step until after we actually have a fully
set up buffer object. Note that the bo is not published to userspace
until later, so not sure what the comment "Don't publish the bo
until we have a valid attachment", is referring to.
We have at least two different customers reporting hitting a NULL ptr
deref in evict_flags when importing something from amdgpu, followed by
triggering the evict flow. Hit rate is also pretty low, which would
hint at some kind of race, so something like 1) or 2) might explain
this.
v2:
- Shuffle the order of the ops slightly (no functional change)
- Improve the comment to better explain the ordering (Matt B)
Assisted-by: Gemini:gemini-3 #debug
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/7903
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/4055
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Acked-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Link: https://patch.msgid.link/20260508102635.149172-3-matthew.auld@intel.com
In a multi-queue use case, when a job is running on the secondary queue,
the CTX_TIMESTAMP does not reflect the queues run ticks. Instead, we use
the QUEUE TIMESTAMP to check how long the job ran. For user space to see
the run ticks for a secondary queue, whitelist the QUEUE_TIMESTAMP
register.
Compute PR: https://github.com/intel/compute-runtime/pull/923
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-24-umesh.nerlige.ramappa@intel.com
Each queue in a multi queue group has a dedicated timestamp counter. Use
this QUEUE TIMESTAMP register to capture the start timestamp for the
job.
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-23-umesh.nerlige.ramappa@intel.com
If a queue is currently active on the CS, query the QUEUE TIMESTAMP
register to get an up to date value of the runtime. To do so, ensure
that the primary queue is active and then check if the secondary queue
is executing on the CS.
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-21-umesh.nerlige.ramappa@intel.com
We need to define more helpers that read engine ID specific register, so
move that logic outside of get_ctx_timestamp().
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-20-umesh.nerlige.ramappa@intel.com
In secondary queue LRCs, the QUEUE TIMESTAMP register is saved and
restored allowing us to view the individual queue run times. Add helpers
to read this value from the LRC.
BSpec: 73988
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-19-umesh.nerlige.ramappa@intel.com
For an LRC belonging to the secondary queue, in order to check if its
context group is active, we need to check the LRC of the primary queue.
In addition to that we want to compare the secondary queue position to
CSMQDEBUG register to check if the queue itself is active.
To do so, store primary LRC and position information in the LRC.
A note on references involved:
- In general the Queue takes a ref on its LRC.
- In addition, for multi-queue,
a. Primary Queue takes a ref for each Secondary LRC.
b. Each Secondary Queue takes a ref to the Primary Queue
In the current patch, each LRC in the queue group is storing a pointer
to primary LRC. Both primary and secondary LRCs are freed only when
primary queue is destroyed. At this time, all secondary queues are
already destroyed, so there is no one using secondary LRCs. We should be
good without taking any additional references.
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-18-umesh.nerlige.ramappa@intel.com
xe exec queue code is using a check to see if a class of engines support
multi queue. This check is also needed by other code, so move it to
xe_gt and export it for others.
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-17-umesh.nerlige.ramappa@intel.com
Add a timestamp in milliseconds to the LRC snapshot to make it easier to
reason about how long the LRC has been running and the average duration
of each job.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-15-umesh.nerlige.ramappa@intel.com
Use the 64 bit value when available for the context timestamp in the LRC
snapshot.
Suggested-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Link: https://patch.msgid.link/20260507162016.3888309-14-umesh.nerlige.ramappa@intel.com
If a reset (GT or engine) happens during EU stall data sampling, all the
EU stall registers can get reset to 0. This will result in EU stall data
buffers' read and write pointer register values to be out of sync with
the cached values. This will result in read() returning invalid data. To
prevent this, check the value of a EU stall base register. If it is zero,
it indicates a reset may have happened that wiped the register to zero.
If this happens, return ENODEV from read() upon which the user space
should disable and enable EU stall data sampling or close the fd and
open a new fd for a new EU stall data collection session.
This patch has been tested by running two IGT tests simultaneously
xe_eu_stall and xe_exec_reset.
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Acked-by: Felix Degrood <felix.j.degrood@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patch.msgid.link/6935698d70b6c7c347ff8b138209d601030c2c9f.1776200094.git.harish.chegondi@intel.com
Factor the repeated CGP_SYNC action build-and-send sequence into a new
helper guc_exec_queue_send_cgp_sync(). Drop the redundant guc parameter
from __register_exec_queue_group() since it can be derived via
exec_queue_to_guc(q). Remove xe_guc_exec_queue_group_add() which is now
identical to the helper and replace its call site directly.
No functional change.
Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260508194428.61819-6-niranjana.vishwanathapura@intel.com
The 'killed_or_banned_or_wedged = true' assignment is redundant
since the variable is never read after that point.
Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260508194428.61819-5-niranjana.vishwanathapura@intel.com
The blitter engines' MEM_COPY and MEM_SET instructions were added as
part of the same hardware change that introduced service copy engines
(i.e., BCS1-BCS8) which is why the driver checks for service copy engine
presence when deciding whether to use these instructions or the older
XY_* instructions. However when making this decision the driver should
consider which engines are part of the hardware architecture, not which
engines are present/usable on the current device. For graphics IP
versions that architecturally include service copy engines (i.e.,
everything Xe2 and later, plus PVC's Xe_HPC) we should use MEM_SET and
MEM_COPY even in if all of the service copy engines wind up getting
fused off. I.e., we need to decide based on whether the platform's
graphics descriptor contains these engines, rather than whether the
usable engine mask contains them. This logic got broken when
gt->info.__engine_mask was removed, although in practice that mistake
has been harmless so far because there haven't been any hardware
SKUs that fuse off all of the service copy engines yet.
Replace the incorrect has_service_copy_support() function with a GT
feature flag that tracks more accurately whether the new blitter
instructions are usable. In addition to fixing incorrect logic if all
service copies are fused off, the flag also makes it more obvious what
the calling code is trying to do; previously it wasn't terribly obvious
why "has service copy engines" was being used as the condition for using
different instructions on all copy engine types.
The new feature flag is named 'has_xe2_blt_instructions' because we
expect this flag to be set for all Xe2 and later platforms (i.e.,
everything officially supported by the Xe driver). Technically there's
also one Xe1-era platform (PVC) that supports these engines/instructions
and will set this flag, but this still seems to be the most clear and
understandable name for the flag.
Fixes: 61549a2ee5 ("drm/xe: Drop __engine_mask")
Cc: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Link: https://patch.msgid.link/20260507-xe2_copy-v1-1-26506381b821@intel.com
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Stolen memory requires physically contiguous allocations for display
scanout and compressed framebuffers. The stolen memory manager was
sharing the gpu_buddy allocator backend with the VRAM manager, but
buddy manages non-contiguous power-of-two blocks making it a poor fit.
Stolen memory also has fundamentally different allocation patterns:
- Allocation sizes are not power-of-two. Since buddy rounds up to the
next power-of-two block size, a ~17MB request can fail even with
~22MB free, because the free space is fragmented across non-fitting
power-of-two blocks.
- Hardware restrictions prevent using the first 4K page of stolen for
certain allocations (e.g., FBC). The display code sets fpfn=1 to
enforce this, but when fpfn != 0, gpu_buddy enables
GPU_BUDDY_RANGE_ALLOCATION mode which disables the try_harder
coalescing path, further reducing allocation success.
This combination caused FBC compressed framebuffer (CFB) allocation
failures on platforms like NVL/PTL. In case of NVL where stolen memory
is ~56MB and the initial plane framebuffer consumes ~34MB at probe time,
leaving ~22MB for subsequent allocations.
Use ttm_range_man_init_nocheck() to set up a drm_mm-backed TTM resource
manager for stolen memory. This reuses the TTM core's ttm_range_manager
callbacks, avoiding duplicate implementations.
Tested on NVL with a 4K DP display: stolen_mm shows a single ~22MB
contiguous free hole after initial plane framebuffer allocation, and
FBC successfully allocates its CFB from that region. The corresponding
IGT was previously skipped and now passes.
v2:
- Clarify that stolen memory requires contiguous allocations (Matt B)
- Properly handle xe_ttm_resource_visible() for stolen instead of
unconditionally returning true (Matt A)
v3:
- Rebase
- Fix xe_display_bo_fbdev_prefer_stolen() to compare in pages, since
ttm_range_manager stores stolen->size in pages not bytes (Matt A)
v4:
- Add kernel-doc for struct xe_ttm_stolen_mgr (Matt B)
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/7631
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Suggested-by: Matthew Auld <matthew.auld@intel.com>
Assisted-by: GitHub Copilot:claude-sonnet-4.6
Signed-off-by: Sanjay Yadav <sanjay.kumar.yadav@intel.com>
Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Acked-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patch.msgid.link/20260422125502.3088222-2-sanjay.kumar.yadav@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
xe_bo_recompute_purgeable_state() walks all VMAs of a BO to determine
whether the BO can be made purgeable. This makes VMA create/destroy and
madvise updates O(n) in the number of mappings.
Replace the walk with BO-local counters protected by the BO dma-resv
lock:
- vma_count tracks the number of VMAs mapping the BO.
- willneed_count tracks active WILLNEED holders, including WILLNEED
VMAs and active dma-buf exports for non-imported BOs.
A DONTNEED BO is promoted back to WILLNEED on a 0->1 transition of
willneed_count. A BO is demoted to DONTNEED on a 1->0 transition only
when it still has VMAs, preserving the previous behaviour where a BO
with no mappings keeps its current madvise state.
PURGED remains terminal, preserving the existing "once purged, always
purged" rule.
Fixes: 4f44961eab ("drm/xe/vm: Prevent binding of purged buffer objects")
v2:
- Use early return for imported BOs in all four helpers to avoid
nesting (Matt B).
- Group purgeability state into a purgeable sub-struct on struct
xe_bo (Matt B).
- Reword xe_bo_willneed_put_locked() kernel-doc to explain that a 1->0
transition means all remaining active VMAs are DONTNEED (Matt B).
v3:
- Move DONTNEED/PURGED reject from vma_lock_and_validate() into
xe_vma_create(), gated on attr->purgeable_state == WILLNEED.
Fixes vm_bind bypass and partial-unbind rejection on DONTNEED
BOs (Matt B).
- Drop .check_purged from MAP and REMAP; keep it for PREFETCH and
add a comment why (Matt B).
- Skip BO validation in vma_lock_and_validate() for non-WILLNEED
VMA remnants so cleanup/remap paths do not repopulate
DONTNEED/PURGED BOs.
Suggested-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260506132027.2556046-1-arvind.yadav@intel.com
Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
UAPI Changes:
- Support medium/low power modes in amdxdna.
- Support limiting frequency in ivpu.
- Document license for drm core uAPI headers.
- Add the following DRM formats: P230, Y7, XYYY2101010, T430,
XVUY210101010.
Cross-subsystem Changes:
- Add and improve dt-bindings.
- Remove unused dma-fence-array's signal_on_any support.
Core Changes:
- Do not call drop_master on file close if not master.
- Convert drm-bridge and drm/atomic to use drm_printf_indent.
- Remove the extra call to drm_connector_attach_encoder after
drm_bridge_connector_init().
- Assorted docbook updates.
Driver Changes:
- Bugfixes in amdxdna, ivpu, mipi-dsi, imagination, nouveau, panthor,
bridge/analogix_dp, ipv3, lontium-lt8912b, verisilicon, tve200,
etnaviv, panel/focaltech-ota7290b, panel/jadard-jd9365da-h3,
bridge/ite-it6263, renesas, xlnx, bridge/cdns-dsi, gma500,
bridge/microchip-lvds, mgag200.
- Add support for MStar TSUMU88ADT3-LF-1 bridge.
- Add support for WaveShare 7, Novatek NT35532, Startek KD070HDFLD092,
ChipWealth CH13726A AMOLED, Team Source Display TST070WSNE-196C,
Displaytech DT050BTFT-PTS panels.
- Improve mipi-dsi shutdown and convert a panasonic panel to use the
mipi-dsi wrappers.
- Allowing dumping vbios over debugfs in GSP-RM mode.
- Update maintainers for ivpu, add reviewer for drm-bridge code
and update maintainers for LT8912B DRM HDMI bridge.
- Add test pattern support to bridge/ti-sn65dsi83.
- Convert vmwgfx to vblank timers.
- Add power management to sysfb drm drivers to allow suspend/resume.
- Support the aforementioned new drm formats in xlnx/qynqmp.
- Fix panel Kconfig dependencies.
- Add carveout support for debugging and bringup to amxdna.
- Add support for long command tx via videobuffer in bridge/tc358768.
-----BEGIN PGP SIGNATURE-----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=kWIv
-----END PGP SIGNATURE-----
Merge tag 'drm-misc-next-2026-05-07' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-next
drm-misc-next for v7.2-rc1:
UAPI Changes:
- Support medium/low power modes in amdxdna.
- Support limiting frequency in ivpu.
- Document license for drm core uAPI headers.
- Add the following DRM formats: P230, Y7, XYYY2101010, T430,
XVUY210101010.
Cross-subsystem Changes:
- Add and improve dt-bindings.
- Remove unused dma-fence-array's signal_on_any support.
Core Changes:
- Do not call drop_master on file close if not master.
- Convert drm-bridge and drm/atomic to use drm_printf_indent.
- Remove the extra call to drm_connector_attach_encoder after
drm_bridge_connector_init().
- Assorted docbook updates.
Driver Changes:
- Bugfixes in amdxdna, ivpu, mipi-dsi, imagination, nouveau, panthor,
bridge/analogix_dp, ipv3, lontium-lt8912b, verisilicon, tve200,
etnaviv, panel/focaltech-ota7290b, panel/jadard-jd9365da-h3,
bridge/ite-it6263, renesas, xlnx, bridge/cdns-dsi, gma500,
bridge/microchip-lvds, mgag200.
- Add support for MStar TSUMU88ADT3-LF-1 bridge.
- Add support for WaveShare 7, Novatek NT35532, Startek KD070HDFLD092,
ChipWealth CH13726A AMOLED, Team Source Display TST070WSNE-196C,
Displaytech DT050BTFT-PTS panels.
- Improve mipi-dsi shutdown and convert a panasonic panel to use the
mipi-dsi wrappers.
- Allowing dumping vbios over debugfs in GSP-RM mode.
- Update maintainers for ivpu, add reviewer for drm-bridge code
and update maintainers for LT8912B DRM HDMI bridge.
- Add test pattern support to bridge/ti-sn65dsi83.
- Convert vmwgfx to vblank timers.
- Add power management to sysfb drm drivers to allow suspend/resume.
- Support the aforementioned new drm formats in xlnx/qynqmp.
- Fix panel Kconfig dependencies.
- Add carveout support for debugging and bringup to amxdna.
- Add support for long command tx via videobuffer in bridge/tc358768.
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patch.msgid.link/f73f342d-6efb-416d-81b0-1716bdd98d5f@linux.intel.com
TC358768 has two ways to send DSI commands: 1) buffer the payload data
into registers (DSICMD_WDx), which supports up to 8 bytes of payload, 2)
buffer the payload data into the video buffer, which supports up to 1024
bytes of payload.
The driver currently supports method 1).
Add support for transmitting long DSI commands (more than 8 bytes, up to
1024 bytes) using the video buffer. This mode can only be used before
the actual video transmission is enabled, i.e. the initial configuration.
Original version from Parth Pancholi <parth.pancholi@toradex.com>
Tested-by: João Paulo Gonçalves <joao.goncalves@toradex.com> # Toradex Verdin AM62
Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Link: https://patch.msgid.link/20260311-tc358768-v2-7-e75a99131bd5@ideasonboard.com
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Sending long commands using the video buffer (to be implemented in
following patches) requires setting TC358768_DATAFMT and
TC358768_DSITX_DT registers for command transfer. The same registers
also need to be configured properly for video transfer.
The long commands will be sent between the bridge's pre_enable() and
enable(), and currently we configure the registers for video transfer in
pre_enable(). Thus, they would be overwritten by the long command
transfer code.
To prevent that from happening, set those registers for video transfer
in enable(), not in pre_enable().
Based on code from Parth Pancholi <parth.pancholi@toradex.com>
Tested-by: João Paulo Gonçalves <joao.goncalves@toradex.com> # Toradex Verdin AM62
Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Link: https://patch.msgid.link/20260311-tc358768-v2-6-e75a99131bd5@ideasonboard.com
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Currently the driver ignores MIPI_DSI_MODE_LPM and always uses HS mode.
Add code to enable HS mode in pre_enable() only if MIPI_DSI_MODE_LPM is
not set, and always enable HS mode in enable() for video transmission.
Tested-by: João Paulo Gonçalves <joao.goncalves@toradex.com> # Toradex Verdin AM62
Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Link: https://patch.msgid.link/20260311-tc358768-v2-5-e75a99131bd5@ideasonboard.com
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The driver prints a warning if MIPI_DSI_CLOCK_NON_CONTINUOUS is set, and
falls back to continuous clock mode. This was added in commit
fbc5a90e82 ("drm/bridge: tc358768: Disable non-continuous clock mode").
However, there have been multiple changes to the driver since then, and
at least in my setup, non-continuous clock mode works: I can see an
image on the panel, and I can see the clock lanes being non-continuous
with an oscilloscope.
So, let's enable MIPI_DSI_CLOCK_NON_CONTINUOUS support.
Cc: Dmitry Osipenko <digetx@gmail.com>
Tested-by: João Paulo Gonçalves <joao.goncalves@toradex.com> # Toradex Verdin AM62
Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Acked-by: Dmitry Osipenko <digetx@gmail.com>
Link: https://patch.msgid.link/20260311-tc358768-v2-4-e75a99131bd5@ideasonboard.com
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Some registers can only be written indirectly, using DSI_CONFW register.
We don't have many uses for those registers (in fact, only DSI_CONTROL
is currently written), but the code to do those writes inline is a bit
confusing.
Add a new function, tc358768_confw_update_bits() which can be used to
write the bits indirectly. Only DSI_CONTROL is currently supported.
Tested-by: João Paulo Gonçalves <joao.goncalves@toradex.com> # Toradex Verdin AM62
Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Link: https://patch.msgid.link/20260311-tc358768-v2-3-e75a99131bd5@ideasonboard.com
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
