Move the driver specific fbdev fb calls to the display parent
interface. Reuse the existing struct intel_display_bo_interface, as this
is mostly about gem objects.
Put everything behind IS_ENABLED(CONFIG_DRM_FBDEV_EMULATION) to catch
configuration issues at build or link time.
v2: Rebase
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com> # v1
Link: https://patch.msgid.link/a6bb24909a58181cfc41b91a4c6538a181d27158.1774950508.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
The struct drm_device *drm parameter of intel_fbdev_fb_fill_info() is
redundant, as you can obtain the same information via obj->dev. Drop the
drm parameter, and move obj as first (or the context) parameter.
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Link: https://patch.msgid.link/6c617dc41c07b0e3e9af01b82d0d1dadbd4760fc.1774950508.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Unify the behaviour with i915. Similar to commit b63c6b9b7f
("drm/i915/fbdev: fix link failure without FBDEV emulation"). The
difference is that unlike i915, xe doesn't build
display/intel_fbdev_fb.c conditional to CONFIG_DRM_FBDEV_EMULATION.
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Link: https://patch.msgid.link/245880a591fe5267a1c06bfbbcc42a4dd9f2e4ee.1774950508.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
This far using crtc_state->pipe_src as borders for Selective Update area
haven't caused visible problems as drm_rect_width(crtc_state->pipe_src) ==
crtc_state->hw.adjusted_mode.crtc_hdisplay and
drm_rect_height(crtc_state->pipe_src) ==
crtc_state->hw.adjusted_mode.crtc_vdisplay when pipe scaling is not
used. On the other hand using pipe scaling is forcing full frame updates and all the
Selective Update area calculations are skipped. Now this improper usage of
crtc_state->pipe_src is causing following warnings:
<4> [7771.978166] xe 0000:00:02.0: [drm] drm_WARN_ON_ONCE(su_lines % vdsc_cfg->slice_height)
after WARN_ON_ONCE was added by commit:
"drm/i915/dsc: Add helper for writing DSC Selective Update ET parameters"
These warnings are seen when DSC and pipe scaling are enabled
simultaneously. This is because on full frame update SU area is improperly
set as pipe_src which is not aligned with DSC slice height.
Fix these by creating local rectangle using
crtc_state->hw.adjusted_mode.crtc_hdisplay and
crtc_state->hw.adjusted_mode.crtc_vdisplay. Use this local rectangle as
borders for SU area.
Fixes: d6774b8c3c ("drm/i915: Ensure damage clip area is within pipe area")
Cc: <stable@vger.kernel.org> # v6.0+
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Link: https://patch.msgid.link/20260327114553.195285-1-jouni.hogander@intel.com
Replace the naked intel_uncore_read*() with intel_de_read*()
in the MCHBAR code.
v2: Rebase due to intel_uncore_read64_2x32()
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-12-ville.syrjala@linux.intel.com
intel_de_read64_2x32_volatile() is a complex beast because
it needs to deal with volatile register values. For simpler
cases we can simply do a pair normal intel_de_read()s.
My main reason for hating overuse of intel_de_read64_2x32_volatile()
is that it makes register tracepoints confusing. It always
does three accesses in the somewhat weird udw,ldw,udw order,
confusing the reader of the trace. Much more clear if we just
observe the two reads in the natural little endian order.
We also have no non-volatile use case where the LDW and UDW
are stored in non-consecutive registers, so we can just pass
along a single register offset.
v2: Put the function arguments on one line (Jani)
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-10-ville.syrjala@linux.intel.com
The raison d'etre of intel_de_read64_2x32() is that it can
handle registers where volatile values are split across two
registers. I don't like that it's being used needlessly.
Rename it to intel_de_read64_2x32_volatile() to make it
more clear when it should be used.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-9-ville.syrjala@linux.intel.com
We'll need a replacement for intel_uncore_read16() in order to
untangle intel_mchbar_read16() from uncore. As with the 8 bit
counterpart this doesn't need to work on modern platforms
so we can forgo all the DMC wakelock stuff and whatnot.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-8-ville.syrjala@linux.intel.com
Replace all the naked intel_uncore_read*() accesses to MCHBAR
registers with the dedicated intel_mchbar_read*().
v2: Rebase due to the intel_mchbar_read64_2x32() rename
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-7-ville.syrjala@linux.intel.com
We are doing a few accesses to MCHBAR registers with intel_de_read().
Use the dedicated intel_mchbar_read() instead.
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-6-ville.syrjala@linux.intel.com
The intel_mchbar_read*() functions should only be used for
accessing MCHBAR registers. Warn if someone tries to use
them for other registers.
I suppose we could even have a dedicated type for MCHBAR
registers. But that is true for many other special register
types as well, and so far we haven't bothered adding any
special types apart from i915_mcr_reg_t.
v2: Print the register offset (Jani)
Mention i915_mcr_reg_t (Jani)
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-5-ville.syrjala@linux.intel.com
MCHBAR registers are a bit special in that:
- we access them through the mirror
- the mirror is read only on HSW+
- the mirror requires the actual MCHBAR to be enabled in device 0:0.0
- the mirror is gone on MTL+
So I'd prefer to treat MCHBAR registers as a bit special in
the code as well, and do all accesses to them via dedicated
functions. Prodive such functions in the form of
intel_mchbar_read*().
v2: Put the function arguments on one line
No intel_uncore_read64() on xe, use intel_uncore_read64_2x32()
Name the new function intel_mchbar_read64_2x32() as well
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260331154259.24600-3-ville.syrjala@linux.intel.com
Move vlv_iosf_sb_reg.h to include/drm/intel/vlv_iosf_sb_regs.h. Use
_regs.h suffix to align better with other register headers. Move enum
vlv_iosf_sb_unit there as well, breaking the final include tie related
to IOSF sideband between display and i915 core.
With this, we can completely remove the xe compat vls_iosf_sb*.h
headers.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/41b060b0d6453de39ca775eab10ee12b25c45b7d.1774957233.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Convert the VLV sideband static inline wrappers into real
functions. This will help the follow-up work of moving the VLV sideband
to the display parent interface.
The downside is that we'll have to build vlv_sideband.c as part of xe
build, to avoid a plethora of stubs.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/57740dc3a820cb5fc1cfcd28e4be58b2cb48020d.1774957233.git.jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Implement the POST_LT_ADJ_REQ sequence, which should be used
to further fine tune the link if TPS4 is not supported.
The POST_LT_ADJ_REQ sequence will be performed after
the normal link training has succeeded.
Only the final hop between the last LTTPR and DPRX will
perform the POST_LT_ADJ_REQ adjustment. The earlier hops
will use TPS4 instead since it's mandatory for LTTPRs.
The sequence will terminate when the sink clears the
"in progress" flag, the vswing/pre-emphasis values have
changed six times, or the vswing/pre-emphasis values have
remained unchanged for 200 ms.
Tested-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260330202629.28616-2-ville.syrjala@linux.intel.com
The pipe_src dump doesn't really fit in with the rest
of the guys (pixel_rate, port_clock, min_cdclk) included
on the same line. Split pipe_src onto its own line,
next to the related pfit stuff.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260325135849.12603-4-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Looks like I missed the drm_dp_enhanced_frame_cap() in the ivb/hsw CPU
eDP code when I introduced crtc_state->enhanced_framing. Fix it up so
that the state we program to the hardware is guaranteed to match what
we computed earlier.
Cc: stable@vger.kernel.org
Fixes: 3072a24c77 ("drm/i915: Introduce crtc_state->enhanced_framing")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260325135849.12603-3-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Apparently I forgot about the pipe min_voltage_level when I
decoupled the CDCLK calculations from modesets. Even if the
CDCLK frequency doesn't need changing we may still need to
bump the voltage level to accommodate an increase in the
port clock frequency.
Currently, even if there is a full modeset, we won't notice the
need to go through the full CDCLK calculations/programming,
unless the set of enabled/active pipes changes, or the
pipe/dbuf min CDCLK changes.
Duplicate the same logic we use the pipe's min CDCLK frequency
to also deal with its min voltage level.
Note that the 'allow_voltage_level_decrease' stuff isn't
really useful here since the min voltage level can only
change during a full modeset. But I think sticking to the
same approach in the three similar parts (pipe min cdclk,
pipe min voltage level, dbuf min cdclk) is a good idea.
Cc: stable@vger.kernel.org
Tested-by: Mikhail Rudenko <mike.rudenko@gmail.com>
Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/15826
Fixes: ba91b9eecb ("drm/i915/cdclk: Decouple cdclk from state->modeset")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260325135849.12603-2-ville.syrjala@linux.intel.com
Reviewed-by: Michał Grzelak <michal.grzelak@intel.com>
Enable PIPEDMC_ERROR interrupt bit for display version 35+.
Changes in v2:
- Move PIPEDMC_ERROR interrupt enable to a separate patch (Ville Syrjala)
Changes in v3:
- Add Bspec reference (Suraj Kandpal)
Changes in v4:
- Add TODO to check the enablement of PIPEDMC_ERROR for PTL (Suraj Kandpal)
Changes in v5:
- Move TODO to the beginning of pipedmc_interrupt_mask() (Suraj Kandpal)
Bspec: 70296
Signed-off-by: Dibin Moolakadan Subrahmanian <dibin.moolakadan.subrahmanian@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://patch.msgid.link/20260327064453.2241523-3-dibin.moolakadan.subrahmanian@intel.com
On display version 35+ PIPEDMC_ATS_FAULT and
PIPEDMC_GTT_FAULT interrupt bits are no longer defined.
Update the interrupt mask to drop these.
Changes in v2:
- Move PIPEDMC_ERROR interrupt enable to a separate patch (Ville Syrjala)
Changes in v3:
- Add Bspec reference (Suraj Kandpal)
Changes in v4:
- Move bspec position in commit message(Suraj Kandpal)
Bspec: 70296
Signed-off-by: Dibin Moolakadan Subrahmanian <dibin.moolakadan.subrahmanian@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Signed-off-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://patch.msgid.link/20260327064453.2241523-2-dibin.moolakadan.subrahmanian@intel.com
Hi Dave and Sima,
Here goes our late, final drm-xe-next PR towards 7.1. We just purgeable
BO uAPI in today, hence the late pull.
In the big things we have:
- Add support for purgeable buffer objects
Thanks,
Matt
UAPI Changes:
- Add support for purgeable buffer objects (Arvind, Himal)
Driver Changes:
- Remove useless comment (Maarten)
- Issue GGTT invalidation under lock in ggtt_node_remove (Brost, Fixes)
- Fix mismatched include guards in header files (Shuicheng)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/acX4fWxPkZrrfwnT@gsse-cloud1.jf.intel.com
We're mixing up the GT FIFO debug checks (overflows and such)
with RMbus unclaimed register checks. The two are quite different
things as RMbus is only relevant for display registers, and the
GT FIFO only relevant for GT registers.
Split the GT FIFO debugs out from the unclaimed register logic
and just do the checks during forcewake_get() and early init.
That is still sufficient to detect if any errors have happened.
Any errors would anyway be caused by overflowing the FIFO
rather than accessing specific registers, so trying to figure
out exactly when the error happened isn't particularly useful.
To fix such issues we'd rather have to do something to slow down
the rate at which registers are accessed (eg. increase
GT_FIFO_NUM_RESERVED_ENTRIES or something).
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260323101609.8391-3-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
The live_forcewake_domains selftest doesn't really test anything
particularly sensible. It only runs on platforms that have RMbus
unclaimer error detection, but that only catches display registers
which the test doesn't even access.
I suppose if we really wanted to we might try to make the test
exercise the GT FIFO instead by writing GT registers as fast
as possible, and then checking GTFIFODBG to see if the FIFO has
overflowed. But dunno if there's much point in that. I think a
GT FIFO overflow might even be fatal to the machine.
So in its current for the test doesn't really make sense,
and it's in the way of moving all the RMbus noclaim stuff
to the display driver side. So let's just get rid of it.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260323101609.8391-2-ville.syrjala@linux.intel.com
Acked-by: Jani Nikula <jani.nikula@intel.com>
TGL/ADL DSI can be configured to place the clock lane into
LP state during LPM, if otherwise configured for continuous
HS clock.
Hook that up. VBT tells us whether this should be done.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260326111814.9800-6-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
TGL/ADL DSI can be configured to fill all BLLPs with blanking
packets. Currently we enable that always, but the VBT actually
tells us whether this is desired or not. Hook that up.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260326111814.9800-5-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
The DSI 'clock_stop' parameter is a boolean, so use a real
'bool' for it. And pimp the debug print while at it.
Note that we also remove the incorrect negation of the value
in the debug print. That has been there since the code was
introduced in commit 2ab8b458c6 ("drm/i915: Add support for
Generic MIPI panel driver"). An earlier version of the patch
https://lore.kernel.org/intel-gfx/1397454507-10273-5-git-send-email-shobhit.kumar@intel.com/
got it right, but looks like it got fumbled while dealing
with other review comments.
v2: Highlight the removal of the '!' (Jani)
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260326111814.9800-4-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
eotp == "End of Transmission Packet". Drop the redundant
extra 'p' from 'eotp_pkt', and make the thing a boolean
while at it.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260326111814.9800-3-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Stop adjusting the horizontal timing values based on the
compression ratio in command mode. Bspec seems to be telling
us to do this only in video mode, and this is also how the
Windows driver does things.
This should also fix a div-by-zero on some machines because
the adjusted htotal ends up being so small that we end up with
line_time_us==0 when trying to determine the vtotal value in
command mode.
Note that this doesn't actually make the display on the
Huawei Matebook E work, but at least the kernel no longer
explodes when the driver loads.
Cc: stable@vger.kernel.org
Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/12045
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20260326111814.9800-2-ville.syrjala@linux.intel.com
Fixes: 53693f02d8 ("drm/i915/dsi: account for DSC in horizontal timings")
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Userspace passes canonical (sign-extended) GPU addresses where bits 63:48
mirror bit 47. The internal GPUVM uses non-canonical form (upper bits
zeroed), so passing raw canonical addresses into GPUVM lookups causes
mismatches for addresses above 128TiB.
Strip the sign extension with xe_device_uncanonicalize_addr() at the
top of xe_vm_madvise_ioctl(). Non-canonical addresses are unaffected.
Fixes: ada7486c56 ("drm/xe: Implement madvise ioctl for xe")
Suggested-by: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-13-arvind.yadav@intel.com
Hook the madvise_purgeable() handler into the madvise IOCTL now that all
supporting infrastructure is complete:
- Core purge implementation (patch 3)
- BO state tracking and helpers (patches 1-2)
- Per-VMA purgeable state tracking (patch 6)
- Shrinker integration for memory reclamation (patch 10)
This final patch enables userspace to use the
DRM_XE_VMA_ATTR_PURGEABLE_STATE madvise type to mark buffers as
WILLNEED/DONTNEED and receive the retained status indicating whether
buffers were purged.
The feature was kept disabled in earlier patches to maintain
bisectability and ensure all components are in place before exposing to
userspace.
Userspace can detect kernel support for purgeable BOs by checking the
DRM_XE_QUERY_CONFIG_FLAG_HAS_PURGING_SUPPORT flag in the query_config
response.
Suggested-by: Matthew Brost <matthew.brost@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-12-arvind.yadav@intel.com
Encapsulate TTM purgeable flag updates and shrinker page accounting
into helper functions to prevent desynchronization between the TTM
tt->purgeable flag and the shrinker's page bucket counters.
Without these helpers, direct manipulation of xe_ttm_tt->purgeable
risks forgetting to update the corresponding shrinker counters,
leading to incorrect memory pressure calculations.
Update purgeable BO state to PURGED after successful shrinker purge
for DONTNEED BOs.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-11-arvind.yadav@intel.com
Don't allow exporting BOs marked DONTNEED or PURGED as dma-bufs.
DONTNEED BOs can have their contents discarded at any time, making
the exported dma-buf unusable for external devices. PURGED BOs have
no backing store and are permanently invalid.
Return -EBUSY for DONTNEED BOs (temporary purgeable state) and
-EINVAL for purged BOs (permanent, no backing store).
The export path now checks the BO's purgeable state before creating
the dma-buf, preventing external devices from accessing memory that
may be purged at any time.
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-10-arvind.yadav@intel.com
Don't allow new CPU mmaps to BOs marked DONTNEED or PURGED.
DONTNEED BOs can have their contents discarded at any time, making
CPU access undefined behavior. PURGED BOs have no backing store and
are permanently invalid.
Return -EBUSY for DONTNEED BOs (temporary purgeable state) and
-EINVAL for purged BOs (permanent, no backing store).
The mmap offset ioctl now checks the BO's purgeable state before
allowing userspace to establish a new CPU mapping. This prevents
the race where userspace gets a valid offset but the BO is purged
before actual faulting begins.
Existing mmaps (established before DONTNEED) may still work until
pages are purged, at which point CPU faults fail with SIGBUS.
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-9-arvind.yadav@intel.com
Prevent marking imported or exported dma-bufs as purgeable.
External devices may be accessing these buffers without our
knowledge, making purging unsafe.
Check drm_gem_is_imported() for buffers created by other
drivers and obj->dma_buf for buffers exported to other
drivers. Silently skip these BOs during madvise processing.
This follows drm_gem_shmem's purgeable implementation and
prevents data corruption from purging actively-used shared
buffers.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-8-arvind.yadav@intel.com
Track purgeable state per-VMA instead of using a coarse shared
BO check. This prevents purging shared BOs until all VMAs across
all VMs are marked DONTNEED.
Add xe_bo_all_vmas_dontneed() to check all VMAs before marking
a BO purgeable. Add xe_bo_recheck_purgeable_on_vma_unbind() to
handle state transitions when VMAs are destroyed - if all
remaining VMAs are DONTNEED the BO can become purgeable, or if
no VMAs remain it transitions to WILLNEED.
The per-VMA purgeable_state field stores the madvise hint for
each mapping. Shared BOs can only be purged when all VMAs
unanimously indicate DONTNEED.
This prevents the bug where unmapping the last VMA would incorrectly
flip a DONTNEED BO back to WILLNEED. The enum-based state check
preserves BO state when no VMAs remain, only updating when VMAs provide
explicit hints.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-7-arvind.yadav@intel.com
Add purge checking to vma_lock_and_validate() to block new mapping
operations on purged BOs while allowing cleanup operations to proceed.
Purged BOs have their backing pages freed by the kernel. New
mapping operations (MAP, PREFETCH, REMAP) must be rejected with
-EINVAL to prevent GPU access to invalid memory. Cleanup
operations (UNMAP) must be allowed so applications can release
resources after detecting purge via the retained field.
REMAP operations require mixed handling - reject new prev/next
VMAs if the BO is purged, but allow the unmap portion to proceed
for cleanup.
The check_purged flag in struct xe_vma_lock_and_validate_flags
distinguishes between these cases: true for new mappings (must reject),
false for cleanup (allow).
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-6-arvind.yadav@intel.com
Block CPU page faults to buffer objects marked as purgeable (DONTNEED)
or already purged. Once a BO is marked DONTNEED, its contents can be
discarded by the kernel at any time, making access undefined behavior.
Return VM_FAULT_SIGBUS immediately to fail consistently instead of
allowing erratic behavior where access sometimes works (if not yet
purged) and sometimes fails (if purged).
For DONTNEED BOs:
- Block new CPU faults with SIGBUS to prevent undefined behavior.
- Existing CPU PTEs may still work until TLB flush, but new faults
fail immediately.
For PURGED BOs:
- Backing store has been reclaimed, making CPU access invalid.
- Without this check, accessing existing mmap mappings would trigger
xe_bo_fault_migrate() on freed backing store, causing kernel hangs
or crashes.
The purgeable check is added to both CPU fault paths:
- Fastpath (xe_bo_cpu_fault_fastpath): Returns VM_FAULT_SIGBUS
immediately under dma-resv lock, preventing attempts to
migrate/validate DONTNEED/purged pages.
- Slowpath (xe_bo_cpu_fault): Returns -EFAULT under drm_exec lock,
converted to VM_FAULT_SIGBUS.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20260326130843.3545241-5-arvind.yadav@intel.com
