- 17 Nov, 2016 7 commits
-
-
Jani Nikula authored
We've overlooked adding DSI panels to the front of the connector list. This seems to be the right thing to do, and I suspect this might fix some issues, although I currently have no evidence to support this. v2: also git add the comment change Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by:
Jani Nikula <jani.nikula@intel.com> Signed-off-by:
Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1479378548-32695-1-git-send-email-jani.nikula@intel.com
-
Ville Syrjälä authored
It has been suggested that having per-plane modifiers is making life more difficult for userspace, so let's just retire modifier[1-3] and use modifier[0] to apply to the entire framebuffer. Obviosuly this means that if individual planes need different tiling layouts and whatnot we will need a new modifier for each combination of planes with different tiling layouts. For a bit of extra backwards compatilbilty the kernel will allow non-zero modifier[1+] but it require that they will match modifier[0]. This in case there's existing userspace out there that sets modifier[1+] to something non-zero with planar formats. Mostly a cocci job, with a bit of manual stuff mixed in. @@ struct drm_framebuffer *fb; expression E; @@ - fb->modifier[E] + fb->modifier @@ struct drm_framebuffer fb; expression E; @@ - fb.modifier[E] + fb.modifier Cc: Kristian Høgsberg <hoegsberg@gmail.com> Cc: Ben Widawsky <benjamin.widawsky@intel.com> Cc: Rob Clark <robdclark@gmail.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Tomeu Vizoso <tomeu@tomeuvizoso.net> Cc: dczaplejewicz@collabora.co.uk Suggested-by:
Kristian Høgsberg <hoegsberg@gmail.com> Acked-by:
Ben Widawsky <ben@bwidawsk.net> Acked-by:
Daniel Stone <daniels@collabora.com> Acked-by:
Rob Clark <robdclark@gmail.com> Signed-off-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by:
-
Daniel Vetter authored
Sumit still takes care about dma-buf, but we've merged the trees together since way too much overlap. And Gustavo is also part of the drm-misc team to be able to help out. Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Gustavo Padovan <gustavo@padovan.org> Acked-by:
Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by:
Daniel Vetter <daniel.vetter@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20161116140819.31165-2-daniel.vetter@ffwll.ch
-
Chen-Yu Tsai authored
Some dumb VGA DACs are active components which require external power. Add support for specifying a regulator as its power supply. Signed-off-by:
Chen-Yu Tsai <wens@csie.org> Acked-by:
Rob Herring <robh@kernel.org> Signed-off-by:
Archit Taneja <architt@codeaurora.org> Link: http://patchwork.freedesktop.org/patch/msgid/20161116154232.872-2-wens@csie.org
-
Daniel Vetter authored
Again something that's in the drm-misc fold. Cc: Archit Taneja <architt@codeaurora.org> Acked-by:
-
Daniel Vetter authored
KS happened, time to make this official! Well, scripting work to make this all happen is still ongoing, but I'm trying to at least keep the new tree semi-in-sync with the temporary topic/drm-misc branch in the drm-intel.git repo. So for now still no new committers, and existing ones still need to push to topic/drm-misc in drm-intel.git. Big kudos to Jani&Sean for volunteering as co-maintainers! v2: Restrict patterns a bit to avoid all the driver spam. Cc: Daniel Vetter <daniel.vetter@intel.com> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Sean Paul <seanpaul@chromium.org> Acked-by:
Sean Paul <seanpaul@chromium.org> Acked-by:
Jani Nikula <jani.nikula@linux.intel.com> Signed-off-by:
-
Daniel Vetter authored
Looking at the ioctl permission checks I noticed that it's impossible to import gem buffers into a control nodes, and fd2handle/handle2fd also don't work, so no joy with dma-bufs. The only way to do anything with a control node is by drawing stuff into a dumb buffer and displaying that. I suspect control nodes are an entirely unused thing, and a cursory check shows that there does not seem to be any callers of drmOpenControl nor of the other drmOpen functions using DRM_MODE_CONTROL. Since I don't like dead uabi, let's remove it. But since this would be a really big change I think it's better to start out small by simply not registering anything. We can garbage-collect the dead code later on, once we're sure it's really not used anywhere. Acked-by:
Dave Airlie <airlied@gmail.com> Signed-off-by:
-
- 16 Nov, 2016 8 commits
-
-
https://github.com/anholt/linuxDave Airlie authored
This pull request brings in fragment shader threading and ETC1 support for vc4.
-
git://git.armlinux.org.uk/~rmk/linux-armDave Airlie authored
Fix conncector registration with tda998x. * 'drm-tda998x-mali' of git://git.armlinux.org.uk/~rmk/linux-arm: drm/i2c: tda998x: mali-dp: hdlcd: refactor connector registration
-
git://anongit.freedesktop.org/git/drm-miscDave Airlie authored
Another pile of misc: - Explicit fencing for atomic! Big thanks to Gustavo, Sean, Rob 3x, Brian and anyone else I've forgotten to make this happen. - roll out fbdev helper ops to drivers (Stefan Christ) - last bits of drm_crtc split-up&kerneldoc - some drm_irq.c crtc functions cleanup - prepare_fb helper for cma, works correctly with explicit fencing (Marek Vasut) - misc small patches all over * tag 'drm-misc-next-2016-11-16' of git://anongit.freedesktop.org/git/drm-misc: (51 commits) drm/fence: add out-fences support drm/fence: add fence timeline to drm_crtc drm/fence: add in-fences support drm/bridge: analogix_dp: return error if transfer none byte drm: drm_irq.h header cleanup drm/irq: Unexport drm_vblank_on/off drm/irq: Unexport drm_vblank_count drm/irq: Make drm_vblank_pre/post_modeset internal drm/nouveau: Use drm_crtc_vblank_off/on drm/amdgpu: Use drm_crtc_vblank_on/off for dce6 drm/color: document NULL values and default settings better drm: Drop externs from drm_crtc.h drm: Move tile group code into drm_connector.c drm: Extract drm_mode_config.[hc] Revert "drm: Add aspect ratio parsing in DRM layer" Revert "drm: Add and handle new aspect ratios in DRM layer" drm/print: Move kerneldoc next to definition drm: Consolidate dumb buffer docs drm: Clean up kerneldoc for struct drm_driver drm: Extract drm_drv.h ...
-
Jonas Pfeil authored
FS threading brings performance improvements of 0-20% in glmark2. The validation code checks for thread switch signals and ensures that the registers of the other thread are not touched, and that our clamps are not live across thread switches. It also checks that the threading and branching instructions do not interfere. (Original patch by Jonas, changes by anholt for style cleanup, removing validation the kernel doesn't need to do, and adding the flag for userspace). v2: Minor style fixes from checkpatch. Signed-off-by:
Jonas Pfeil <pfeiljonas@gmx.de> Signed-off-by:
Eric Anholt <eric@anholt.net>
-
Gustavo Padovan authored
Support DRM out-fences by creating a sync_file with a fence for each CRTC that sets the OUT_FENCE_PTR property. We use the out_fence pointer received in the OUT_FENCE_PTR prop to send the sync_file fd back to userspace. The sync_file and fd are allocated/created before commit, but the fd_install operation only happens after we know that commit succeed. v2: Comment by Rob Clark: - Squash commit that adds DRM_MODE_ATOMIC_OUT_FENCE flag here. Comment by Daniel Vetter: - Add clean up code for out_fences v3: Comments by Daniel Vetter: - create DRM_MODE_ATOMIC_EVENT_MASK - userspace should fill out_fences_ptr with the crtc_ids for which it wants fences back. v4: Create OUT_FENCE_PTR properties and remove old approach. v5: Comments by Brian Starkey: - Remove extra fence_get() in atomic_ioctl() - Check ret before iterating on the crtc_state - check ret before fd_install - set fence_state to NULL at the beginning - check fence_state->out_fence_ptr before put_user() - change order of fput() and put_unused_fd() on failure - Add access_ok() check to the out_fence_ptr received - Rebase after fence -> dma_fence rename - Store out_fence_ptr in the drm_atomic_state - Split crtc_setup_out_fence() - return -1 as out_fence with TEST_ONLY flag v6: Comments by Daniel Vetter - Add prepare/unprepare_crtc_signaling() - move struct drm_out_fence_state to drm_atomic.c - mark get_crtc_fence() as static Comments by Brian Starkey - proper set fence_ptr fence_state array - isolate fence_idx increment - improve error handling v7: Comments by Daniel Vetter - remove prefix from internal functions - make out_fence_ptr an s64 pointer - degrade DRM_INFO to DRM_DEBUG_ATOMIC when put_user fail - fix doc issues - filter out OUT_FENCE_PTR == NULL and do not fail in this case - add complete_crtc_signalling() - krealloc fence_state on demand Comment by Brian Starkey - remove unused crtc_state arg from get_out_fence() v8: Comment by Brian Starkey - cancel events before check for !fence_state - convert a few lefovers u64 types for out_fence_ptr - fix memleak by assign fence_state earlier after realloc - proper accout num_fences in case of error v9: Comment by Brian Starkey - memset last position of fence_state after krealloc Comments by Sean Paul - pass install_fds in complete_crtc_signaling() instead of ret - put_user(-1, fence_ptr) when decoding props v10: Comment by Brian Starkey - remove unneeded num_fences increment on error path - kfree fence_state after installing fences fd v11: rebase against latest drm-misc v12: rebase again against latest drm-misc Signed-off-by:Gustavo Padovan <gustavo.padovan@collabora.co.uk> Reviewed-by: Brian Starkey <brian.starkey@arm.com> (v10) Reviewed-by:
-
Gustavo Padovan authored
Create one timeline context for each CRTC to be able to handle out-fences and signal them. It adds a few members to struct drm_crtc: fence_context, where we store the context we get from fence_context_alloc(), the fence seqno and the fence lock, that we pass in fence_init() to be used by the fence. v2: Comment by Daniel Stone: - add BUG_ON() to fence_to_crtc() macro v3: Comment by Ville Syrjälä - Use more meaningful name as crtc timeline name v4: Comments by Brian Starkey - Use even more meaninful name for the crtc timeline - add doc for timeline_name Comment by Daniel Vetter - use in-line style for comments - rebase after fence -> dma_fence rename v5: Comment by Daniel Vetter - Add doc for drm_crtc_fence_ops v6: Comment by Chris Wilson - Move fence_to_crtc to drm_crtc.c - Move export of drm_crtc_fence_ops to drm_crtc_internal.h - rebase against latest drm-misc Signed-off-by: -
Gustavo Padovan authored
There is now a new property called IN_FENCE_FD attached to every plane state that receives sync_file fds from userspace via the atomic commit IOCTL. The fd is then translated to a fence (that may be a fence_array subclass or just a normal fence) and then used by DRM to fence_wait() for all fences in the sync_file to signal. So it only commits when all framebuffers are ready to scanout. v2: Comments by Daniel Vetter: - remove set state->fence = NULL in destroy phase - accept fence -1 as valid and just return 0 - do not call fence_get() - sync_file_fences_get() already calls it - fence_put() if state->fence is already set, in case userspace set the property more than once. v3: WARN_ON if fence is set but state has no FB v4: Comment from Maarten Lankhorst - allow set fence with no related fb v5: rename FENCE_FD to IN_FENCE_FD v6: Comments by Daniel Vetter: - rename plane_state->in_fence back to "fence" - re-introduce WARN_ON if fence set but no fb - rebase after fence -> dma_fence rename v7: Comments by Brian Starkey - set state->fence to NULL when duplicating the state - fail if IN_FENCE_FD was already set v8: rebase against latest drm-misc Signed-off-by:Brian Starkey <brian.starkey@arm.com> Reviewed-by:
Robert Foss <robert.foss@collabora.com> [danvet: Rebase onto extracted drm_mode_config.[hc].] Signed-off-by:
-
Jianqun Xu authored
Reference from drm_dp_aux description (about transfer): Upon success, the implementation should return the number of payload bytes that were transferred, or a negative error-code on failure. Helpers propagate errors from the .transfer() function, with the exception of the -EBUSY error, which causes a transaction to be retried. On a short, helpers will return -EPROTO to make it simpler to check for failure. The analogix_dp_transfer will return num_transferred, but if there is none byte been transferred, the return value will be 0, which means success, we should return error-code if transfer none byte. for (retry = 0; retry < 32; retry++) { err = aux->transfer(aux, &msg); if (err < 0) { if (err == -EBUSY) continue; goto unlock; } } Cc: zain wang <wzz@rock-chips.com> Reviewed-by:Jianqun Xu <jay.xu@rock-chips.com> Signed-off-by:
-
- 15 Nov, 2016 25 commits
-
-
Dave Airlie authored
Merge tag 'sunxi-drm-for-4.10' of https://git.kernel.org/pub/scm/linux/kernel/git/mripard/linux into drm-next sun4i-drm changes for 4.10 Support for the Allwinner A31 SoC display engine using the sun4i-drm driver. * tag 'sunxi-drm-for-4.10' of https://git.kernel.org/pub/scm/linux/kernel/git/mripard/linux: drm/sun4i: Add a few formats drm/sun4i: Add compatible strings for A31/A31s display pipelines drm/sun4i: Add compatible string for A31/A31s TCON (timing controller) drm/sun4i: tcon: Move SoC specific quirks to a DT matched data structure drm/sun4i: sun6i-drc: Support DRC on A31 and A31s
-
git://linuxtv.org/pinchartl/mediaDave Airlie authored
rcar-du -next branch. * 'drm/next/du' of git://linuxtv.org/pinchartl/media: drm: rcar-du: Fix LVDS start sequence on Gen3 drm: rcar-du: Fix H/V sync signal polarity configuration drm: rcar-du: Fix display timing controller parameter drm: rcar-du: Fix dot clock routing configuration drm: rcar-du: Add R8A7796 support drm: rcar-du: Add R8A7792 support drm: rcar-du: Simplify and fix probe error handling drm: rcar-du: Fix crash in encoder failure error path drm: rcar-du: Remove memory allocation error message drm: rcar-du: Remove test for impossible error condition drm: rcar-du: Bring HDMI encoder comments in line with the driver drm: rcar-du: Constify node argument to rcar_du_lvds_connector_init() video: of: Constify node argument to display timing functions
-
Daniel Vetter authored
- Drop extern for functions, it's noise. - Move&consolidate drm.ko internal parts into drm-internal.h. Reviewed-by:
Alex Deucher <alexander.deucher@amd.com> Signed-off-by:
-
Daniel Vetter authored
Only remaining use was in amdgpu, and trivial to convert over to drm_crtc_vblank_* variants. Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Christian König <christian.koenig@amd.com> Reviewed-by:
-
Daniel Vetter authored
No one outside of drm_irq.c should ever need this. The correct way to implement get_vblank_count for hw lacking a vblank counter is drm_vblank_no_hw_counter. Fix this up in mtk, which is the only offender left over. Cc: CK Hu <ck.hu@mediatek.com> Cc: Philipp Zabel <p.zabel@pengutronix.de> Acked-by:
Philipp Zabel <p.zabel@pengutronix.de> Reviewed-by:
-
Daniel Vetter authored
Now that all drivers are switched over to drm_crtc_vblank_on/off we can relegate pre/post_modeset to the purely drm_irq.c internal role of supporting old ums userspace. As usual switch to the drm_legacy_ prefix to make it clear this is for old drivers only. v2: Rebase on top of Thierry's s/int crtc/unsigned int pipe/ changes. Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Alex Deucher <alexander.deucher@amd.com> Reviewed-by:
-
Daniel Vetter authored
With atomic nv50+ is already converted over to them, but the old display code is still using it. Found in a 2 year old patch I have lying around to un-export these old helpers! v2: Drop the hand-rolled versions from resume/suspend code. Now that crtc callbacks do this, we don't need a special case for s/r anymore. v3: Remove unused variables. v4: Don't remove drm_crtc_vblank_off from suspend paths, non-atomic nouveau still needs that. But still switch to drm_crtc_vblank_off since drm_vblank_off will disappear. Cc: Mario Kleiner <mario.kleiner.de@gmail.com> Cc: Ben Skeggs <bskeggs@redhat.com> Reviewed-by:
-
Daniel Vetter authored
This old code pattern was reintroduced in Author: Ken Wang <Qingqing.Wang@amd.com> Date: Tue Jan 19 14:03:24 2016 +0800 drm/amdgpu: add display controller implementation for si v10 Realign it with later display code. Tsk, tsk for massive copypasting ;-) Cc: Christian König <christian.koenig@amd.com> Cc: Ken Wang <Qingqing.Wang@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Acked-by: Christian König <christian.koenig@amd.com> for the whole series. Reviewed-by: -
Daniel Vetter authored
Brought up in a discussion for enabling gamma on fsl-dcu. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Meng Yi <meng.yi@nxp.com> Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by:
Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by:
-
Daniel Vetter authored
Just noise. Reviewed-by:
Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by:
-
Daniel Vetter authored
And also put the overview section into the KMS Properties part of the docs, instead of randomly-placed within the helpers - this is part of the uabi. With this patch I think drm_crtc.[hc] is cleaned up and entirely documented. Reviewed-by:
-
Daniel Vetter authored
And shuffle the kernel-doc structure a bit since drm_crtc.[hc] now only contains CRTC-related functions and structures. v2: - rebase onto drm-misc - don't forget to move drm_mode_config_cleanup. - move 2 internal decls under the right heading (Chris) Reviewed-by:
-
Ville Syrjälä authored
This reverts commit 6dffd431. Adding new mode flags willy nilly breaks existing userspace. We need to coordinate this better, potentially with a new client cap that only exposes the aspect ratio flags when userspace is prepared for them (similar to what we do with stereo 3D modes). This also broke things so that we would always send out VIC==0 in the AVI infoframe unless the user specified an aspect ratio via the mode flags. And the automagic RGB full vs. limited range handling was similartly broken as the user mode would never match any CEA mode. Cc: Shashank Sharma <shashank.sharma@intel.com> Cc: Lin, Jia <lin.a.jia@intel.com> Cc: Akashdeep Sharma <akashdeep.sharma@intel.com> Cc: Jim Bride <jim.bride@linux.intel.com> Cc: Jose Abreu <Jose.Abreu@synopsys.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Emil Velikov <emil.l.velikov@gmail.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by:
-
Ville Syrjälä authored
This reverts commit a68362fe. Adding new mode flags willy nilly breaks existing userspace. We need to coordinate this better, potentially with a new client cap that only exposes the aspect ratio flags when userspace is prepared for them (similar to what we do with stereo 3D modes). Cc: Shashank Sharma <shashank.sharma@intel.com> Cc: Lin, Jia <lin.a.jia@intel.com> Cc: Akashdeep Sharma <akashdeep.sharma@intel.com> Cc: Jim Bride <jim.bride@linux.intel.com> Cc: Jose Abreu <Jose.Abreu@synopsys.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Emil Velikov <emil.l.velikov@gmail.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by:
-
Daniel Vetter authored
kerneldoc expects the comment next to definitions, otherwise it can't pick up exported vs. internal stuff. This fixes a warning from the doc build done with: $ make DOCBOOKS="" htmldocs Fixes: d8187177 ("drm: add helper for printing to log or seq_file") Cc: Rob Clark <robdclark@gmail.com> Cc: Sean Paul <seanpaul@chromium.org> Reviewed-by:
-
Daniel Vetter authored
Put the callback docs into struct drm_driver, and the small overview into a DOC comment. Reviewed-by:
-
Daniel Vetter authored
Just cleans up what's there, still plenty missing. Reviewed-by:
-
Daniel Vetter authored
I want to move dumb buffer documentation into the right vfuncs, and for that I first need to be able to pull that into kerneldoc without having to clean up all of drmP.h. Also, header-splitting is nice. While at it shuffle all the function declarations for drm_drv.c into the right spots, and drop the kerneldoc for drm_minor_acquire/release since it's only used internally. v2: Keep all existing copyright notices (Chris). Reviewed-by:
-
Daniel Vetter authored
Would be great if everony could add $ make DOCBOOKS="" htmldocs to their build scripts to catch these. 0day should also report them, not sure why it failed to spot this. Fixes: f54d1867 ("dma-buf: Rename struct fence to dma_fence") Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Gustavo Padovan <gustavo.padovan@collabora.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Christian König <christian.koenig@amd.com> Reviewed-by:
-
Daniel Vetter authored
Would be great if everony could add $ make DOCBOOKS="" htmldocs to their build scripts to catch these. 0day should also report them, not sure why it failed to spot this. Fixes: b42fe9ca ("drm/i915: Split out i915_vma.c") Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by:
-
Daniel Vetter authored
Just code movement, doc cleanup will follow up later. v2: Keep all the copyright notices. Reviewed-by:
-
Marek Vasut authored
Add missing forward declaration for struct drm_plane and drm_plane_state, which causes the following warning in the VC4 driver (can be replicated by building using bcm2835_defconfig): In file included from drivers/gpu/drm/vc4/vc4_drv.c:18:0: include/drm/drm_fb_cma_helper.h:45:13: warning: ‘struct drm_plane_state’ declared inside parameter list will not be visible outside of this definition or declaration struct drm_plane_state *state); ^~~~~~~~~~~~~~~ include/drm/drm_fb_cma_helper.h:44:34: warning: ‘struct drm_plane’ declared inside parameter list will not be visible outside of this definition or declaration int drm_fb_cma_prepare_fb(struct drm_plane *plane, Signed-off-by:Marek Vasut <marex@denx.de> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by:
-
Dan Carpenter authored
drm_dev_alloc() never returns NULL, it only returns error pointers on error. Fixes: 0a886f59 ("drm: zte: add initial vou drm driver") Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
Rob Clark authored
kbuild spotted this error, with drm/msm patches that add a new modeset-lock in the driver and driver built as a module: ERROR: "crtc_ww_class" [drivers/gpu/drm/msm/msm.ko] undefined! Really the only reason for crtc_ww_class not being internal to drm_modeset_lock.c is that drm_modeset_lock_init() was static-inline (for no particularly good reason). Fix that, and move crtc_ww_class into drm_modeset_lock.c. Signed-off-by:
-
Chris Wilson authored
The current code is subject to a race where we may try to acquire a reference on a stale fence: [13703.335118] WARNING: CPU: 1 PID: 14975 at ./include/linux/kref.h:46 i915_gem_object_wait+0x1a3/0x1c0 [13703.335184] Modules linked in: [13703.335202] CPU: 1 PID: 14975 Comm: gem_concurrent_ Not tainted 4.9.0-rc4+ #26 [13703.335216] Hardware name: / , BIOS PYBSWCEL.86A.0027.2015.0507.1758 05/07/2015 [13703.335233] ffffc90002f5bcc8 ffffffff812807de 0000000000000000 0000000000000000 [13703.335257] ffffc90002f5bd08 ffffffff81073811 0000002e80000000 ffff88026bf7c780 [13703.335279] 7fffffffffffffff 0000000000000001 ffff88027045a550 ffff88026bf7c780 [13703.335301] Call Trace: [13703.335316] [<ffffffff812807de>] dump_stack+0x4d/0x6f [13703.335331] [<ffffffff81073811>] __warn+0xc1/0xe0 [13703.335343] [<ffffffff810738e8>] warn_slowpath_null+0x18/0x20 [13703.335355] [<ffffffff813ac443>] i915_gem_object_wait+0x1a3/0x1c0 [13703.335367] [<ffffffff813ae8ec>] i915_gem_set_domain_ioctl+0xcc/0x330 [13703.335386] [<ffffffff813534ab>] drm_ioctl+0x1cb/0x410 [13703.335400] [<ffffffff813ae820>] ? i915_gem_obj_prepare_shmem_write+0x1d0/0x1d0 [13703.335416] [<ffffffff8135359b>] ? drm_ioctl+0x2bb/0x410 [13703.335429] [<ffffffff8117d32f>] do_vfs_ioctl+0x8f/0x5c0 [13703.335442] [<ffffffff8117d89c>] SyS_ioctl+0x3c/0x70 [13703.335456] [<ffffffff815a07a4>] entry_SYSCALL_64_fastpath+0x17/0x98 [13703.335558] ---[ end trace fd24176416ba6981 ]--- [13703.382778] general protection fault: 0000 [#1] SMP [13703.382802] Modules linked in: [13703.382816] CPU: 1 PID: 14967 Comm: gem_concurrent_ Tainted: G W 4.9.0-rc4+ #26 [13703.382828] Hardware name: / , BIOS PYBSWCEL.86A.0027.2015.0507.1758 05/07/2015 [13703.382841] task: ffff880275458000 task.stack: ffffc90002f18000 [13703.382849] RIP: 0010:[<ffffffff813b3534>] [<ffffffff813b3534>] i915_gem_request_retire+0x2b4/0x320 [13703.382870] RSP: 0018:ffffc90002f1bbc8 EFLAGS: 00010293 [13703.382878] RAX: dead000000000200 RBX: ffff88026bf7dce8 RCX: dead000000000100 [13703.382887] RDX: dead000000000100 RSI: ffff88026bf7c930 RDI: ffff88026bf7dd00 [13703.382897] RBP: ffffc90002f1bbf8 R08: 00000000ffffffff R09: ffff88026b89a000 [13703.382905] R10: 0000000000000001 R11: ffff88026bbe8fe0 R12: ffff88026bf7c000 [13703.382913] R13: ffff880275af8000 R14: ffff88026bf7c180 R15: dead000000000200 [13703.382922] FS: 00007f89e787d740(0000) GS:ffff88027fd00000(0000) knlGS:0000000000000000 [13703.382934] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [13703.382942] CR2: 00007f9053d2e000 CR3: 000000026d414000 CR4: 00000000001006e0 [13703.382951] Stack: [13703.382958] ffff880275413000 ffffc90002f1bde8 ffff880275af8000 ffff880274e8a600 [13703.382976] ffff880276a06000 ffffc90002f1bde8 ffffc90002f1bc38 ffffffff813b48c5 [13703.382995] ffffc90002f1bc00 ffffc90002f1bde8 ffff88026972a440 0000000000000000 [13703.383021] Call Trace: [13703.383032] [<ffffffff813b48c5>] i915_gem_request_alloc+0xa5/0x350 [13703.383043] [<ffffffff813a17c3>] i915_gem_do_execbuffer.isra.41+0x7b3/0x18b0 [13703.383055] [<ffffffff813b144c>] ? i915_gem_object_get_sg+0x25c/0x2b0 [13703.383065] [<ffffffff813b1d4d>] ? i915_gem_object_get_page+0x1d/0x50 [13703.383076] [<ffffffff813b28cc>] ? i915_gem_pwrite_ioctl+0x66c/0x6d0 [13703.383086] [<ffffffff813a2c25>] i915_gem_execbuffer2+0x95/0x1e0 [13703.383096] [<ffffffff813534ab>] drm_ioctl+0x1cb/0x410 [13703.383105] [<ffffffff813a2b90>] ? i915_gem_execbuffer+0x2d0/0x2d0 [13703.383117] [<ffffffff810c3df0>] ? hrtimer_start_range_ns+0x1a0/0x310 [13703.383128] [<ffffffff8117d32f>] do_vfs_ioctl+0x8f/0x5c0 [13703.383140] [<ffffffff810c60e8>] ? SyS_timer_settime+0x118/0x1a0 [13703.383150] [<ffffffff8117d89c>] SyS_ioctl+0x3c/0x70 [13703.383162] [<ffffffff815a07a4>] entry_SYSCALL_64_fastpath+0x17/0x98 [13703.383172] Code: 49 39 c6 48 8d 70 e8 48 8d 5f e8 75 16 eb 47 48 8d 43 18 48 8b 53 18 48 89 de 49 39 c6 48 8d 5a e8 74 33 48 8b 56 08 48 8b 46 10 <48> 89 42 08 48 89 10 f6 46 38 01 48 89 4e 08 4c 89 7e 10 74 cf [13703.383557] RIP [<ffffffff813b3534>] i915_gem_request_retire+0x2b4/0x320 [13703.383570] RSP <ffffc90002f1bbc8> [13703.383586] ---[ end trace fd24176416ba6982 ]--- This is fixed by using the kref_get_unless_zero() as a full memory barrier to validate the fence is still the current exclusive fence before returning it back to the caller. (Note the fix only requires using dma_fence_get_rcu() and correct handling, but we may as well use the helper rather than inline equivalent code.) Note: Issue can only be hit with the i915 driver. Signed-off-by:
-
