| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAG48ez1cqqYYp2s2Wc_84ksqrvXEcq0WeT6KPtRVzWaoJnS3eA@mail.gmail.com>
Date: Tue, 7 Jan 2020 17:22:50 +0100
From: Jann Horn <jannh@...gle.com>
To: Gerd Hoffmann <kraxel@...hat.com>,
Daniel Vetter <daniel.vetter@...ll.ch>,
Chia-I Wu <olvaffe@...il.com>, dri-devel@...ts.freedesktop.org,
David Airlie <airlied@...ux.ie>,
virtualization@...ts.linux-foundation.org
Cc: kernel list <linux-kernel@...r.kernel.org>
Subject: locking warnings in drm/virtio code
Hi!
My development VM (KVM guest, virtio graphics) is throwing warnings
when I start up X while running a build from Linus' tree with lockdep
turned on. I tried to bisect it, and it looks like at least the
"suspicious RCU usage" one started triggering in commit
889165ad6190556ffe4a8fa6b0e486f1c25589d8 ("drm/virtio: pass gem
reservation object to ttm init").
Slightly shortened dmesg output follows:
[ 37.706151] =============================
[ 37.706605] WARNING: suspicious RCU usage
[ 37.707089] 5.5.0-rc5+ #605 Not tainted
[ 37.707522] -----------------------------
[ 37.708015] include/linux/dma-resv.h:247 suspicious
rcu_dereference_protected() usage!
[ 37.708899]
[ 37.708899] other info that might help us debug this:
[ 37.708899]
[ 37.709856]
[ 37.709856] rcu_scheduler_active = 2, debug_locks = 1
[ 37.710771] 3 locks held by Xorg/1869:
[ 37.711266] #0: ffff8880a976fa48 (crtc_ww_class_acquire){+.+.}, at:
drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:949)
[ 37.712372] #1: ffff8880b32e00a8 (crtc_ww_class_mutex){+.+.}, at:
drm_modeset_lock (linux/drivers/gpu/drm/drm_modeset_lock.c:258
linux/drivers/gpu/drm/drm_modeset_lock.c:342
linux/drivers/gpu/drm/drm_modeset_lock.c:338)
[ 37.713415] #2: ffff8880b32eb6c0 (&(&vgvq->qlock)->rlock){+.+.},
at: virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:381)
[ 37.714650]
[ 37.714650] stack backtrace:
[ 37.715140] CPU: 0 PID: 1869 Comm: Xorg Not tainted 5.5.0-rc5+ #605
[ 37.715889] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
BIOS 1.12.0-1 04/01/2014
[ 37.718253] Call Trace:
[ 37.718596] dump_stack (linux/lib/dump_stack.c:120)
[ 37.719011] dma_resv_add_excl_fence
(linux/./include/linux/dma-resv.h:246
linux/drivers/dma-buf/dma-resv.c:273)
[...]
[ 37.720278] virtio_gpu_array_add_fence
(linux/drivers/gpu/drm/virtio/virtgpu_gem.c:230 (discriminator 3))
[ 37.720925] virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:391)
[...]
[ 37.724065] virtio_gpu_cursor_plane_update
(linux/drivers/gpu/drm/virtio/virtgpu_plane.c:240)
[ 37.724656] drm_atomic_helper_commit_planes
(linux/drivers/gpu/drm/drm_atomic_helper.c:2444 (discriminator 2))
[ 37.725324] vgdev_atomic_commit_tail
(linux/drivers/gpu/drm/virtio/virtgpu_display.c:332)
[ 37.725835] commit_tail (linux/drivers/gpu/drm/drm_atomic_helper.c:1612)
[ 37.726204] drm_atomic_helper_commit
(linux/drivers/gpu/drm/drm_atomic_helper.c:1824)
[ 37.726729] drm_atomic_helper_update_plane
(linux/drivers/gpu/drm/drm_atomic_helper.c:2858)
[ 37.727391] drm_mode_cursor_universal (linux/drivers/gpu/drm/drm_plane.c:909)
[...]
[ 37.729585] drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:995)
[...]
[ 37.731491] drm_ioctl_kernel (linux/drivers/gpu/drm/drm_ioctl.c:792)
[...]
[ 37.733510] drm_ioctl (linux/./include/linux/uaccess.h:152
linux/drivers/gpu/drm/drm_ioctl.c:887)
[...]
[ 37.737651] do_vfs_ioctl (linux/fs/ioctl.c:47 linux/fs/ioctl.c:545
linux/fs/ioctl.c:732)
[...]
[ 37.740419] ksys_ioctl (linux/./include/linux/file.h:43 linux/fs/ioctl.c:750)
[ 37.740797] __x64_sys_ioctl (linux/fs/ioctl.c:754)
[...]
[ 37.750537] ------------[ cut here ]------------
[ 37.751063] WARNING: CPU: 0 PID: 1869 at
drivers/dma-buf/dma-resv.c:277 dma_resv_add_excl_fence
(linux/drivers/dma-buf/dma-resv.c:277)
[ 37.752227] CPU: 0 PID: 1869 Comm: Xorg Not tainted 5.5.0-rc5+ #605
[ 37.752963] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
BIOS 1.12.0-1 04/01/2014
[ 37.753918] RIP: 0010:dma_resv_add_excl_fence
(linux/drivers/dma-buf/dma-resv.c:277)
[ 37.754516] Code: 56 84 83 c6 05 98 4b 0b 02 01 e8 ba 0a 19 ff e9 1d
ff ff ff 49 8d 7d 60 be ff ff ff ff e8 ca 55 18 ff 85 c0 0f 85 18 fd
ff ff <0f> 0b e9 11 fd ff ff 49 8d 7d 60 be ff ff ff ff e8 ad 55 18 ff
85
All code
========
0: 56 push %rsi
1: 84 83 c6 05 98 4b test %al,0x4b9805c6(%rbx)
7: 0b 02 or (%rdx),%eax
9: 01 e8 add %ebp,%eax
b: ba 0a 19 ff e9 mov $0xe9ff190a,%edx
10: 1d ff ff ff 49 sbb $0x49ffffff,%eax
15: 8d 7d 60 lea 0x60(%rbp),%edi
18: be ff ff ff ff mov $0xffffffff,%esi
1d: e8 ca 55 18 ff callq 0xffffffffff1855ec
22: 85 c0 test %eax,%eax
24: 0f 85 18 fd ff ff jne 0xfffffffffffffd42
2a:* 0f 0b ud2 <-- trapping instruction
2c: e9 11 fd ff ff jmpq 0xfffffffffffffd42
31: 49 8d 7d 60 lea 0x60(%r13),%rdi
35: be ff ff ff ff mov $0xffffffff,%esi
3a: e8 ad 55 18 ff callq 0xffffffffff1855ec
3f: 85 .byte 0x85
Code starting with the faulting instruction
===========================================
0: 0f 0b ud2
2: e9 11 fd ff ff jmpq 0xfffffffffffffd18
7: 49 8d 7d 60 lea 0x60(%r13),%rdi
b: be ff ff ff ff mov $0xffffffff,%esi
10: e8 ad 55 18 ff callq 0xffffffffff1855c2
15: 85 .byte 0x85
[ 37.756590] RSP: 0018:ffff8880a976f498 EFLAGS: 00010246
[ 37.757264] RAX: 0000000000000000 RBX: 0000000000000001 RCX: 0000000000000000
[ 37.758041] RDX: 0000000000000007 RSI: ffff8880b21b4988 RDI: ffff8880ad39a0e4
[ 37.758754] RBP: ffff88823acb6380 R08: fffffbfff0a098c8 R09: fffffbfff0a098c8
[ 37.759535] R10: fffffbfff0a098c7 R11: ffffffff8504c63f R12: ffff8880b21b4800
[ 37.760318] R13: ffff8880b21b4928 R14: ffff8880b21b49e0 R15: dffffc0000000000
[ 37.761165] FS: 00007f1e79690f00(0000) GS:ffff8880bb200000(0000)
knlGS:0000000000000000
[ 37.762010] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 37.762653] CR2: 00007f1e7b012000 CR3: 00000000b31f8000 CR4: 00000000000006b0
[ 37.766715] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 37.767455] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 37.768337] Call Trace:
[ 37.771029] virtio_gpu_array_add_fence
(linux/drivers/gpu/drm/virtio/virtgpu_gem.c:230 (discriminator 3))
[ 37.771571] virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:391)
[...]
[ 37.774631] virtio_gpu_cursor_plane_update
(linux/drivers/gpu/drm/virtio/virtgpu_plane.c:240)
[ 37.775275] drm_atomic_helper_commit_planes
(linux/drivers/gpu/drm/drm_atomic_helper.c:2444 (discriminator 2))
[ 37.775880] vgdev_atomic_commit_tail
(linux/drivers/gpu/drm/virtio/virtgpu_display.c:332)
[ 37.776446] commit_tail (linux/drivers/gpu/drm/drm_atomic_helper.c:1612)
[ 37.776867] drm_atomic_helper_commit
(linux/drivers/gpu/drm/drm_atomic_helper.c:1824)
[ 37.777456] drm_atomic_helper_update_plane
(linux/drivers/gpu/drm/drm_atomic_helper.c:2858)
[ 37.778052] drm_mode_cursor_universal (linux/drivers/gpu/drm/drm_plane.c:909)
[...]
[ 37.780252] drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:995)
[...]
[ 37.782269] drm_ioctl_kernel (linux/drivers/gpu/drm/drm_ioctl.c:792)
[...]
[ 37.783146] drm_ioctl (linux/./include/linux/uaccess.h:152
linux/drivers/gpu/drm/drm_ioctl.c:887)
[...]
[ 37.785736] do_vfs_ioctl (linux/fs/ioctl.c:47 linux/fs/ioctl.c:545
linux/fs/ioctl.c:732)
[...]
[ 37.788715] ksys_ioctl (linux/./include/linux/file.h:43 linux/fs/ioctl.c:750)
[...]
[ 37.801595] irq event stamp: 829986
[ 37.802060] hardirqs last enabled at (829985):
trace_hardirqs_on_thunk (linux/arch/x86/entry/thunk_64.S:41)
[ 37.803168] hardirqs last disabled at (829986):
trace_hardirqs_off_thunk (linux/arch/x86/entry/thunk_64.S:42)
[ 37.804277] softirqs last enabled at (829984): __do_softirq
(linux/./arch/x86/include/asm/preempt.h:26 linux/kernel/softirq.c:320)
[ 37.805293] softirqs last disabled at (829969): irq_exit
(linux/kernel/softirq.c:373 linux/kernel/softirq.c:413)
[ 37.806258] ---[ end trace 328e17e8d7e3e2ae ]---
[ 37.806846]
[ 37.807090] =============================
[ 37.807557] WARNING: suspicious RCU usage
[ 37.808024] 5.5.0-rc5+ #605 Tainted: G W
[ 37.808686] -----------------------------
[ 37.809151] include/linux/dma-resv.h:94 suspicious
rcu_dereference_protected() usage!
[ 37.810088]
[ 37.810088] other info that might help us debug this:
[ 37.810088]
[ 37.811071]
[ 37.811071] rcu_scheduler_active = 2, debug_locks = 1
[ 37.811910] 3 locks held by Xorg/1869:
[ 37.812381] #0: ffff8880a976fa48 (crtc_ww_class_acquire){+.+.}, at:
drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:949)
[ 37.813479] #1: ffff8880b32e00a8 (crtc_ww_class_mutex){+.+.}, at:
drm_modeset_lock (linux/drivers/gpu/drm/drm_modeset_lock.c:258
linux/drivers/gpu/drm/drm_modeset_lock.c:342
linux/drivers/gpu/drm/drm_modeset_lock.c:338)
[ 37.814358] #2: ffff8880b32eb6c0 (&(&vgvq->qlock)->rlock){+.+.},
at: virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:381)
[ 37.815482]
[ 37.815482] stack backtrace:
[ 37.815989] CPU: 0 PID: 1869 Comm: Xorg Tainted: G W
5.5.0-rc5+ #605
[ 37.816852] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
BIOS 1.12.0-1 04/01/2014
[ 37.817804] Call Trace:
[ 37.818066] dump_stack (linux/lib/dump_stack.c:120)
[ 37.818412] dma_resv_add_excl_fence
(linux/./include/linux/dma-resv.h:93
linux/drivers/dma-buf/dma-resv.c:279)
[ 37.819019] virtio_gpu_array_add_fence
(linux/drivers/gpu/drm/virtio/virtgpu_gem.c:230 (discriminator 3))
[ 37.819586] virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:391)
[...]
[ 37.822629] virtio_gpu_cursor_plane_update
(linux/drivers/gpu/drm/virtio/virtgpu_plane.c:240)
[ 37.823219] drm_atomic_helper_commit_planes
(linux/drivers/gpu/drm/drm_atomic_helper.c:2444 (discriminator 2))
[ 37.823792] vgdev_atomic_commit_tail
(linux/drivers/gpu/drm/virtio/virtgpu_display.c:332)
[ 37.824310] commit_tail (linux/drivers/gpu/drm/drm_atomic_helper.c:1612)
[ 37.824706] drm_atomic_helper_commit
(linux/drivers/gpu/drm/drm_atomic_helper.c:1824)
[ 37.825235] drm_atomic_helper_update_plane
(linux/drivers/gpu/drm/drm_atomic_helper.c:2858)
[ 37.825864] drm_mode_cursor_universal (linux/drivers/gpu/drm/drm_plane.c:909)
[...]
[ 37.830815] drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:995)
[...]
[ 37.832818] drm_ioctl_kernel (linux/drivers/gpu/drm/drm_ioctl.c:792)
[...]
[ 37.833832] drm_ioctl (linux/./include/linux/uaccess.h:152
linux/drivers/gpu/drm/drm_ioctl.c:887)
[...]
[ 37.836672] do_vfs_ioctl (linux/fs/ioctl.c:47 linux/fs/ioctl.c:545
linux/fs/ioctl.c:732)
[...]
[ 37.839627] ksys_ioctl (linux/./include/linux/file.h:43 linux/fs/ioctl.c:750)
[...]
[ 37.849656] =====================================
[ 37.850165] WARNING: bad unlock balance detected!
[ 37.850764] 5.5.0-rc5+ #605 Tainted: G W
[ 37.851307] -------------------------------------
[ 37.851955] Xorg/1869 is trying to release lock
(reservation_ww_class_mutex) at:
[ 37.852803] virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:391)
[ 37.853618] but there are no more locks to release!
[ 37.854243]
[ 37.854243] other info that might help us debug this:
[ 37.855038] 3 locks held by Xorg/1869:
[ 37.855421] #0: ffff8880a976fa48 (crtc_ww_class_acquire){+.+.}, at:
drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:949)
[ 37.856452] #1: ffff8880b32e00a8 (crtc_ww_class_mutex){+.+.}, at:
drm_modeset_lock (linux/drivers/gpu/drm/drm_modeset_lock.c:258
linux/drivers/gpu/drm/drm_modeset_lock.c:342
linux/drivers/gpu/drm/drm_modeset_lock.c:338)
[ 37.857490] #2: ffff8880b32eb6c0 (&(&vgvq->qlock)->rlock){+.+.},
at: virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:381)
[ 37.858743]
[ 37.858743] stack backtrace:
[ 37.859186] CPU: 0 PID: 1869 Comm: Xorg Tainted: G W
5.5.0-rc5+ #605
[ 37.863137] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
BIOS 1.12.0-1 04/01/2014
[ 37.864083] Call Trace:
[ 37.864342] dump_stack (linux/lib/dump_stack.c:120)
[ 37.864805] lock_release (linux/kernel/locking/lockdep.c:4243
linux/kernel/locking/lockdep.c:4503)
[...]
[ 37.866717] __mutex_unlock_slowpath
(linux/./include/asm-generic/atomic-instrumented.h:836
linux/./include/asm-generic/atomic-long.h:28
linux/kernel/locking/mutex.c:1237)
[...]
[ 37.868513] virtio_gpu_queue_fenced_ctrl_buffer
(linux/drivers/gpu/drm/virtio/virtgpu_vq.c:391)
[...]
[ 37.871574] virtio_gpu_cursor_plane_update
(linux/drivers/gpu/drm/virtio/virtgpu_plane.c:240)
[ 37.872225] drm_atomic_helper_commit_planes
(linux/drivers/gpu/drm/drm_atomic_helper.c:2444 (discriminator 2))
[ 37.872839] vgdev_atomic_commit_tail
(linux/drivers/gpu/drm/virtio/virtgpu_display.c:332)
[ 37.873406] commit_tail (linux/drivers/gpu/drm/drm_atomic_helper.c:1612)
[ 37.873818] drm_atomic_helper_commit
(linux/drivers/gpu/drm/drm_atomic_helper.c:1824)
[ 37.874408] drm_atomic_helper_update_plane
(linux/drivers/gpu/drm/drm_atomic_helper.c:2858)
[ 37.875001] drm_mode_cursor_universal (linux/drivers/gpu/drm/drm_plane.c:909)
[...]
[ 37.877216] drm_mode_cursor_common (linux/drivers/gpu/drm/drm_plane.c:995)
[...]
[ 37.879145] drm_ioctl_kernel (linux/drivers/gpu/drm/drm_ioctl.c:792)
[...]
[ 37.880035] drm_ioctl (linux/./include/linux/uaccess.h:152
linux/drivers/gpu/drm/drm_ioctl.c:887)
[...]
[ 37.882646] do_vfs_ioctl (linux/fs/ioctl.c:47 linux/fs/ioctl.c:545
linux/fs/ioctl.c:732)
[...]
[ 37.885407] ksys_ioctl (linux/./include/linux/file.h:43 linux/fs/ioctl.c:750)
Powered by blists - more mailing lists