kernel/drivers/dma-buf/reservation.c, branch linux-5.1.y

dma-buf: Discard old fence_excl on retrying get_fences_rcu for realloc

2019-07-28T06:28:36Z

commit f5b07b04e5f090a85d1e96938520f2b2b58e4a8e upstream. If we have to drop the seqcount & rcu lock to perform a krealloc, we have to restart the loop. In doing so, be careful not to lose track of the already acquired exclusive fence. Fixes: fedf54132d24 ("dma-buf: Restart reservation_object_get_fences_rcu() after writes") Signed-off-by: Chris Wilson Cc: Daniel Vetter Cc: Maarten Lankhorst Cc: Christian König Cc: Alex Deucher Cc: Sumit Semwal Cc: stable@vger.kernel.org #v4.10 Reviewed-by: Christian König Link: https://patchwork.freedesktop.org/patch/msgid/20190604125323.21396-1-chris@chris-wilson.co.uk Signed-off-by: Greg Kroah-Hartman

dma-buf: Update reservation shared_count after adding the new fence

2018-10-26T14:42:11Z

We need to serialise the addition of a new fence into the shared list such that the fence is visible before we claim it is there. Otherwise a concurrent reader of the shared fence list will see an uninitialised fence slot before it is set. <4> [109.613162] general protection fault: 0000 [#1] PREEMPT SMP PTI <4> [109.613177] CPU: 1 PID: 1357 Comm: gem_busy Tainted: G U 4.19.0-rc8-CI-CI_DRM_5035+ #1 <4> [109.613189] Hardware name: Dell Inc. XPS 8300 /0Y2MRG, BIOS A06 10/17/2011 <4> [109.613252] RIP: 0010:i915_gem_busy_ioctl+0x146/0x380 [i915] <4> [109.613261] Code: 0b 43 04 49 83 c6 08 4d 39 e6 89 43 04 74 6d 4d 8b 3e e8 5d 54 f4 e0 85 c0 74 0d 80 3d 08 71 1d 00 00 0f 84 bb 00 00 00 31 c0 <49> 81 7f 08 20 3a 2c a0 75 cc 41 8b 97 50 02 00 00 49 8b 8f a8 00 <4> [109.613283] RSP: 0018:ffffc9000044bcf8 EFLAGS: 00010246 <4> [109.613292] RAX: 0000000000000000 RBX: ffffc9000044bdc0 RCX: 0000000000000001 <4> [109.613302] RDX: 0000000000000000 RSI: 00000000ffffffff RDI: ffffffff822474a0 <4> [109.613311] RBP: ffffc9000044bd28 R08: ffff88021e158680 R09: 0000000000000001 <4> [109.613321] R10: 0000000000000040 R11: 0000000000000000 R12: ffff88021e1641b8 <4> [109.613331] R13: 0000000000000003 R14: ffff88021e1641b0 R15: 6b6b6b6b6b6b6b6b <4> [109.613341] FS: 00007f9c9fc84980(0000) GS:ffff880227a40000(0000) knlGS:0000000000000000 <4> [109.613352] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 <4> [109.613360] CR2: 00007f9c9fcb8000 CR3: 00000002247d4005 CR4: 00000000000606e0 Fixes: 27836b641c1b ("dma-buf: remove shared fence staging in reservation object") Testcase: igt/gem_busy/close-race Signed-off-by: Chris Wilson Cc: Christian König Cc: Junwei Zhang Cc: Huang Rui Cc: Sumit Semwal Reviewed-by: Christian König Link: https://patchwork.freedesktop.org/patch/msgid/20181026080302.11507-1-chris@chris-wilson.co.uk

dma-buf: allow reserving more than one shared fence slot

2018-10-25T11:45:07Z

Let's support simultaneous submissions to multiple engines. Signed-off-by: Christian König Reviewed-by: Michel Dänzer Reviewed-by: Junwei Zhang Reviewed-by: Huang Rui Link: https://patchwork.kernel.org/patch/10626149/

dma-buf: remove shared fence staging in reservation object

2018-10-25T11:41:16Z

No need for that any more. Just replace the list when there isn't enough room any more for the additional fence. Signed-off-by: Christian König Reviewed-by: Junwei Zhang Reviewed-by: Huang Rui Link: https://patchwork.kernel.org/patch/10626143/

dma-buf: Move BUG_ON from _add_shared_fence to _add_shared_inplace

2018-07-16T14:50:37Z

Fixes the BUG_ON spuriously triggering under the following circumstances: * reservation_object_reserve_shared is called with shared_count == shared_max - 1, so obj->staged is freed in preparation of an in-place update. * reservation_object_add_shared_fence is called with the first fence, after which shared_count == shared_max. * reservation_object_add_shared_fence is called with a follow-up fence from the same context. In the second reservation_object_add_shared_fence call, the BUG_ON triggers. However, nothing bad would happen in reservation_object_add_shared_inplace, since both fences are from the same context, so they only occupy a single slot. Prevent this by moving the BUG_ON to where an overflow would actually happen (e.g. if a buggy caller didn't call reservation_object_reserve_shared before). v2: * Fix description of breaking scenario (Christian König) * Add bugzilla reference Cc: stable@vger.kernel.org Bugzilla: https://bugs.freedesktop.org/106418 Reviewed-by: Chris Wilson # v1 Reviewed-by: Christian König # v1 Signed-off-by: Michel Dänzer Signed-off-by: Sumit Semwal Link: https://patchwork.freedesktop.org/patch/msgid/20180704151405.10357-1-michel@daenzer.net

locking: Implement an algorithm choice for Wound-Wait mutexes

2018-07-03T07:44:36Z

The current Wound-Wait mutex algorithm is actually not Wound-Wait but Wait-Die. Implement also Wound-Wait as a per-ww-class choice. Wound-Wait is, contrary to Wait-Die a preemptive algorithm and is known to generate fewer backoffs. Testing reveals that this is true if the number of simultaneous contending transactions is small. As the number of simultaneous contending threads increases, Wait-Wound becomes inferior to Wait-Die in terms of elapsed time. Possibly due to the larger number of held locks of sleeping transactions. Update documentation and callers. Timings using git://people.freedesktop.org/~thomash/ww_mutex_test tag patch-18-06-15 Each thread runs 100000 batches of lock / unlock 800 ww mutexes randomly chosen out of 100000. Four core Intel x86_64: Algorithm #threads Rollbacks time Wound-Wait 4 ~100 ~17s. Wait-Die 4 ~150000 ~19s. Wound-Wait 16 ~360000 ~109s. Wait-Die 16 ~450000 ~82s. Cc: Ingo Molnar Cc: Jonathan Corbet Cc: Gustavo Padovan Cc: Maarten Lankhorst Cc: Sean Paul Cc: David Airlie Cc: Davidlohr Bueso Cc: "Paul E. McKenney" Cc: Josh Triplett Cc: Thomas Gleixner Cc: Kate Stewart Cc: Philippe Ombredanne Cc: Greg Kroah-Hartman Cc: linux-doc@vger.kernel.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Co-authored-by: Peter Zijlstra Signed-off-by: Thomas Hellstrom Acked-by: Peter Zijlstra (Intel) Acked-by: Ingo Molnar

Merge tag 'drm-misc-next-2018-02-13' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

2018-02-15T23:29:27Z

drm-misc-next for 4.17: UAPI Changes: - drm/vc4: Expose performance counters to userspace (Boris) Cross-subsystem Changes: - MAINTAINERS: Linus to maintain panel-arm-versatile in -misc (Linus) Core Changes: - Only use swiotlb when necessary (Chunming) Driver Changes: - drm/panel: Add support for ARM Versatile panels (Linus) - pl111: Improvements around versatile panel support (Linus) ---------------------------------------- Tagged on 2018-02-06: drm-misc-next for 4.17: UAPI Changes: - Validate mode flags + type (Ville) - Deprecate unused mode flags PIXMUX, BCAST (Ville) - Deprecate unused mode types BUILTIN, CRTC_C, CLOCK_C, DEFAULT (Ville) Cross-subsystem Changes: - MAINTAINERS: s/Daniel/Maarten/ for drm-misc (Daniel) Core Changes: - gem: Export gem functions for drivers to use (Samuel) - bridge: Introduce bridge timings in drm_bridge (Linus) - dma-buf: Allow exclusive fence to be bundled in fence array when calling reservation_object_get_fences_rcu (Christian) - dp: Add training pattern 4 and HBR3 support to dp helpers (Manasi) - fourcc: Add alpha bit to formats to avoid driver format LUTs (Maxime) - mode: Various cleanups + add new device-wide .mode_valid hook (Ville) - atomic: Fix state leak when non-blocking commits fail (Leo) NOTE: IIRC, this was cross-picked to -fixes so it might fall out - crc: Allow polling on the data fd (Maarten) Driver Changes: - bridge/vga-dac: Add THS8134* support (Linus) - tinydrm: Various MIPI DBI improvements/cleanups (Noralf) - bridge/dw-mipi-dsi: Cleanups + use create_packet helper (Brian) - drm/sun4i: Add Display Engine frontend support (Maxime) - drm/sun4i: Add zpos support + increase num planes from 2 to 4 (Maxime) - various: Use drm_mode_get_hv_timing() to fill plane clip rectangle (Ville) - stm: Add 8-bit clut support, add dsi phy v1.31 support, +fixes (Phillipe) Cc: Boris Brezillon Cc: Chunming Zhou Cc: Samuel Li Cc: Linus Walleij Cc: Noralf Trønnes Cc: Brian Norris Cc: Maxime Ripard Cc: Ville Syrjala Cc: Christian König Cc: Manasi Navare Cc: Philippe Cornu Cc: Leo (Sunpeng) Li Cc: Daniel Vetter Cc: Maarten Lankhorst * tag 'drm-misc-next-2018-02-13' of git://anongit.freedesktop.org/drm/drm-misc: (115 commits) drm/radeon: only enable swiotlb path when need v2 drm/amdgpu: only enable swiotlb alloc when need v2 drm: add func to get max iomem address v2 drm/vc4: Expose performance counters to userspace drm: Print the pid when debug logging an ioctl error. drm/stm: ltdc: remove non-alpha color formats on layer 2 for older hw drm/stm: ltdc: add non-alpha color formats drm/bridge/synopsys: dsi: Add 1.31 version support drm/bridge/synopsys: dsi: Add read feature drm/pl111: Support multiple endpoints on the CLCD drm/pl111: Support variants with broken VBLANK drm/pl111: Support variants with broken clock divider drm/pl111: Handle the Versatile RGB/BGR565 mode drm/pl111: Properly detect the ARM PL110 variants drm/panel: Add support for ARM Versatile panels drm/panel: Device tree bindings for ARM Versatile panels drm/bridge: Rename argument from crtc to bridge drm/crc: Add support for polling on the data fd. drm/sun4i: Use drm_mode_get_hv_timing() to populate plane clip rectangle drm/rcar-du: Use drm_mode_get_hv_timing() to populate plane clip rectangle ...

dma-buf: make returning the exclusive fence optional

2018-01-23T17:14:26Z

Change reservation_object_get_fences_rcu to make the exclusive fence pointer optional. If not specified the exclusive fence is put into the fence array as well. This is helpful for a couple of cases where we need all fences in a single array. Reviewed-by: Chunming Zhou Acked-by: Daniel Vetter Signed-off-by: Christian König Signed-off-by: Alex Deucher Link: https://patchwork.freedesktop.org/patch/msgid/20180110125341.3618-1-christian.koenig@amd.com

dma-buf: fix reservation_object_wait_timeout_rcu once more v2

2018-01-23T15:46:39Z

We need to set shared_count even if we already have a fence to wait for. v2: init i to -1 as well Signed-off-by: Christian König Cc: stable@vger.kernel.org Tested-by: Lyude Paul Reviewed-by: Lyude Paul Reviewed-by: Chris Wilson Signed-off-by: Alex Deucher Link: https://patchwork.freedesktop.org/patch/msgid/20180122200003.6665-1-christian.koenig@amd.com

dma-buf: try to replace a signaled fence in reservation_object_add_shared_inplace

2017-11-14T17:01:45Z

The amdgpu issue to also need signaled fences in the reservation objects should be fixed by now. Optimize the handling by replacing a signaled fence when adding a new shared one. Signed-off-by: Christian König Reviewed-by: Chris Wilson Signed-off-by: Alex Deucher Link: https://patchwork.freedesktop.org/patch/msgid/20171114142436.1360-2-christian.koenig@amd.com