Linaro-mm-sig October 2022

linaro-mm-sig@lists.linaro.org

29 participants
28 discussions

[RFC][PATCH v2 12/31] timers: dma-buf: Use del_timer_shutdown() before freeing timer

by Steven Rostedt

From: "Steven Rostedt (Google)" <rostedt(a)goodmis.org> Before a timer is freed, del_timer_shutdown() must be called. Link: https://lore.kernel.org/all/20220407161745.7d6754b3@gandalf.local.home/ Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: linux-media(a)vger.kernel.org Cc: dri-devel(a)lists.freedesktop.org Cc: linaro-mm-sig(a)lists.linaro.org Signed-off-by: Steven Rostedt (Google) <rostedt(a)goodmis.org> --- drivers/dma-buf/st-dma-fence.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/dma-buf/st-dma-fence.c b/drivers/dma-buf/st-dma-fence.c index fb6e0a6ae2c9..c67b70205b6f 100644 --- a/drivers/dma-buf/st-dma-fence.c +++ b/drivers/dma-buf/st-dma-fence.c @@ -412,7 +412,7 @@ static int test_wait_timeout(void *arg) err = 0; err_free: - del_timer_sync(&wt.timer); + del_timer_shutdown(&wt.timer); destroy_timer_on_stack(&wt.timer); dma_fence_signal(wt.f); dma_fence_put(wt.f); -- 2.35.1

1 year, 6 months

[PATCH] dma-buf: cma_heap: Fix typo in comment

by Mark-PK Tsai

Fix typo in comment. Signed-off-by: Mark-PK Tsai <mark-pk.tsai(a)mediatek.com> --- drivers/dma-buf/heaps/cma_heap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/dma-buf/heaps/cma_heap.c b/drivers/dma-buf/heaps/cma_heap.c index 28fb04eccdd0..cd386ce639f3 100644 --- a/drivers/dma-buf/heaps/cma_heap.c +++ b/drivers/dma-buf/heaps/cma_heap.c @@ -316,7 +316,7 @@ static struct dma_buf *cma_heap_allocate(struct dma_heap *heap, kunmap_atomic(vaddr); /* * Avoid wasting time zeroing memory if the process - * has been killed by by SIGKILL + * has been killed by SIGKILL */ if (fatal_signal_pending(current)) goto free_cma; -- 2.18.0

1 year, 6 months

Re: [syzbot] KASAN: use-after-free Read in task_work_run (2)

by syzbot

syzbot has found a reproducer for the following issue on: HEAD commit: 88619e77b33d net: stmmac: rk3588: Allow multiple gmac cont.. git tree: bpf console output: https://syzkaller.appspot.com/x/log.txt?x=1646d6f2880000 kernel config: https://syzkaller.appspot.com/x/.config?x=a66c6c673fb555e8 dashboard link: https://syzkaller.appspot.com/bug?extid=9228d6098455bb209ec8 compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2 syz repro: https://syzkaller.appspot.com/x/repro.syz?x=12bc425e880000 C reproducer: https://syzkaller.appspot.com/x/repro.c?x=1126516e880000 Downloadable assets: disk image: https://storage.googleapis.com/syzbot-assets/f8435d5c2c21/disk-88619e77.raw… vmlinux: https://storage.googleapis.com/syzbot-assets/551d8a013e81/vmlinux-88619e77.… kernel image: https://storage.googleapis.com/syzbot-assets/7d3f5c29064d/bzImage-88619e77.… IMPORTANT: if you fix the issue, please add the following tag to the commit: Reported-by: syzbot+9228d6098455bb209ec8(a)syzkaller.appspotmail.com ================================================================== BUG: KASAN: use-after-free in task_work_run+0x1b0/0x270 kernel/task_work.c:178 Read of size 8 at addr ffff8880752b1c18 by task syz-executor361/3766 CPU: 0 PID: 3766 Comm: syz-executor361 Not tainted 6.1.0-rc2-syzkaller-00073-g88619e77b33d #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/11/2022 Call Trace: <TASK> __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106 print_address_description mm/kasan/report.c:284 [inline] print_report+0x15e/0x45d mm/kasan/report.c:395 kasan_report+0xbb/0x1f0 mm/kasan/report.c:495 task_work_run+0x1b0/0x270 kernel/task_work.c:178 exit_task_work include/linux/task_work.h:38 [inline] do_exit+0xb35/0x2a20 kernel/exit.c:820 do_group_exit+0xd0/0x2a0 kernel/exit.c:950 get_signal+0x21a1/0x2430 kernel/signal.c:2858 arch_do_signal_or_restart+0x82/0x2300 arch/x86/kernel/signal.c:869 exit_to_user_mode_loop kernel/entry/common.c:168 [inline] exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203 __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline] syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296 do_syscall_64+0x42/0xb0 arch/x86/entry/common.c:86 entry_SYSCALL_64_after_hwframe+0x63/0xcd RIP: 0033:0x7fb9f674b089 Code: Unable to access opcode bytes at 0x7fb9f674b05f. RSP: 002b:00007fb9f66fb318 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: 0000000000000001 RBX: 00007fb9f67da1a8 RCX: 00007fb9f674b089 RDX: 00000000000f4240 RSI: 0000000000000081 RDI: 00007fb9f67da1ac RBP: 00007fb9f67da1a0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 0000003100000400 R13: 00007fff658570cf R14: 00007fb9f66fb400 R15: 0000000000022000 </TASK> Allocated by task 3766: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:45 kasan_set_track+0x21/0x30 mm/kasan/common.c:52 __kasan_slab_alloc+0x7e/0x80 mm/kasan/common.c:325 kasan_slab_alloc include/linux/kasan.h:201 [inline] slab_post_alloc_hook mm/slab.h:737 [inline] slab_alloc_node mm/slub.c:3398 [inline] kmem_cache_alloc_node+0x2fc/0x400 mm/slub.c:3443 perf_event_alloc.part.0+0x69/0x3bc0 kernel/events/core.c:11625 perf_event_alloc kernel/events/core.c:12174 [inline] __do_sys_perf_event_open+0x4ae/0x32d0 kernel/events/core.c:12272 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd Freed by task 0: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:45 kasan_set_track+0x21/0x30 mm/kasan/common.c:52 kasan_save_free_info+0x2a/0x40 mm/kasan/generic.c:511 ____kasan_slab_free mm/kasan/common.c:236 [inline] ____kasan_slab_free+0x160/0x1c0 mm/kasan/common.c:200 kasan_slab_free include/linux/kasan.h:177 [inline] slab_free_hook mm/slub.c:1724 [inline] slab_free_freelist_hook+0x8b/0x1c0 mm/slub.c:1750 slab_free mm/slub.c:3661 [inline] kmem_cache_free+0xea/0x5b0 mm/slub.c:3683 rcu_do_batch kernel/rcu/tree.c:2250 [inline] rcu_core+0x81f/0x1980 kernel/rcu/tree.c:2510 __do_softirq+0x1f7/0xad8 kernel/softirq.c:571 Last potentially related work creation: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:45 __kasan_record_aux_stack+0xbc/0xd0 mm/kasan/generic.c:481 call_rcu+0x99/0x820 kernel/rcu/tree.c:2798 put_event kernel/events/core.c:5095 [inline] perf_event_release_kernel+0x6f2/0x940 kernel/events/core.c:5210 perf_release+0x33/0x40 kernel/events/core.c:5220 __fput+0x27c/0xa90 fs/file_table.c:320 task_work_run+0x16b/0x270 kernel/task_work.c:179 resume_user_mode_work include/linux/resume_user_mode.h:49 [inline] exit_to_user_mode_loop kernel/entry/common.c:171 [inline] exit_to_user_mode_prepare+0x23c/0x250 kernel/entry/common.c:203 __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline] syscall_exit_to_user_mode+0x19/0x50 kernel/entry/common.c:296 do_syscall_64+0x42/0xb0 arch/x86/entry/common.c:86 entry_SYSCALL_64_after_hwframe+0x63/0xcd Second to last potentially related work creation: kasan_save_stack+0x1e/0x40 mm/kasan/common.c:45 __kasan_record_aux_stack+0xbc/0xd0 mm/kasan/generic.c:481 task_work_add+0x7b/0x2c0 kernel/task_work.c:48 event_sched_out+0xe35/0x1190 kernel/events/core.c:2294 __perf_remove_from_context+0x87/0xc40 kernel/events/core.c:2359 event_function+0x29e/0x3e0 kernel/events/core.c:254 remote_function kernel/events/core.c:92 [inline] remote_function+0x11e/0x1a0 kernel/events/core.c:72 __flush_smp_call_function_queue+0x205/0x9a0 kernel/smp.c:630 __sysvec_call_function_single+0xca/0x4d0 arch/x86/kernel/smp.c:248 sysvec_call_function_single+0x8e/0xc0 arch/x86/kernel/smp.c:243 asm_sysvec_call_function_single+0x16/0x20 arch/x86/include/asm/idtentry.h:657 The buggy address belongs to the object at ffff8880752b17c0 which belongs to the cache perf_event of size 1392 The buggy address is located 1112 bytes inside of 1392-byte region [ffff8880752b17c0, ffff8880752b1d30) The buggy address belongs to the physical page: page:ffffea0001d4ac00 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x752b0 head:ffffea0001d4ac00 order:3 compound_mapcount:0 compound_pincount:0 flags: 0xfff00000010200(slab|head|node=0|zone=1|lastcpupid=0x7ff) raw: 00fff00000010200 0000000000000000 dead000000000122 ffff8880118c23c0 raw: 0000000000000000 0000000080150015 00000001ffffffff 0000000000000000 page dumped because: kasan: bad access detected page_owner tracks the page as allocated page last allocated via order 3, migratetype Unmovable, gfp_mask 0xd20c0(__GFP_IO|__GFP_FS|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_NOMEMALLOC), pid 3754, tgid 3753 (syz-executor361), ts 58662170660, free_ts 58383135648 prep_new_page mm/page_alloc.c:2538 [inline] get_page_from_freelist+0x10b5/0x2d50 mm/page_alloc.c:4287 __alloc_pages+0x1c7/0x5a0 mm/page_alloc.c:5554 alloc_pages+0x1a6/0x270 mm/mempolicy.c:2285 alloc_slab_page mm/slub.c:1794 [inline] allocate_slab+0x213/0x300 mm/slub.c:1939 new_slab mm/slub.c:1992 [inline] ___slab_alloc+0xa91/0x1400 mm/slub.c:3180 __slab_alloc.constprop.0+0x56/0xa0 mm/slub.c:3279 slab_alloc_node mm/slub.c:3364 [inline] kmem_cache_alloc_node+0x189/0x400 mm/slub.c:3443 perf_event_alloc.part.0+0x69/0x3bc0 kernel/events/core.c:11625 perf_event_alloc kernel/events/core.c:12174 [inline] __do_sys_perf_event_open+0x4ae/0x32d0 kernel/events/core.c:12272 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x63/0xcd page last free stack trace: reset_page_owner include/linux/page_owner.h:24 [inline] free_pages_prepare mm/page_alloc.c:1458 [inline] free_pcp_prepare+0x65c/0xd90 mm/page_alloc.c:1508 free_unref_page_prepare mm/page_alloc.c:3386 [inline] free_unref_page+0x19/0x4d0 mm/page_alloc.c:3482 __unfreeze_partials+0x17c/0x1a0 mm/slub.c:2586 qlink_free mm/kasan/quarantine.c:168 [inline] qlist_free_all+0x6a/0x170 mm/kasan/quarantine.c:187 kasan_quarantine_reduce+0x180/0x200 mm/kasan/quarantine.c:294 __kasan_slab_alloc+0x62/0x80 mm/kasan/common.c:302 kasan_slab_alloc include/linux/kasan.h:201 [inline] slab_post_alloc_hook mm/slab.h:737 [inline] slab_alloc_node mm/slub.c:3398 [inline] slab_alloc mm/slub.c:3406 [inline] __kmem_cache_alloc_lru mm/slub.c:3413 [inline] kmem_cache_alloc+0x2ac/0x3c0 mm/slub.c:3422 kmem_cache_zalloc include/linux/slab.h:702 [inline] alloc_buffer_head+0x20/0x140 fs/buffer.c:2899 alloc_page_buffers+0x280/0x790 fs/buffer.c:829 create_empty_buffers+0x2c/0xf20 fs/buffer.c:1543 ext4_block_write_begin+0x10a7/0x15f0 fs/ext4/inode.c:1074 ext4_da_write_begin+0x44c/0xb50 fs/ext4/inode.c:3003 generic_perform_write+0x252/0x570 mm/filemap.c:3753 ext4_buffered_write_iter+0x15b/0x460 fs/ext4/file.c:285 ext4_file_write_iter+0x8b8/0x16e0 fs/ext4/file.c:700 __kernel_write_iter+0x25e/0x730 fs/read_write.c:517 Memory state around the buggy address: ffff8880752b1b00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff8880752b1b80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb >ffff8880752b1c00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ^ ffff8880752b1c80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff8880752b1d00: fb fb fb fb fb fb fc fc fc fc fc fc fc fc fc fc ==================================================================

1 year, 6 months

[PATCH] parport_pc: Remove WCH CH382 PCI-E single parallel port card.

by Zhang Xincheng

WCH CH382L PCI-E adapter with 1 parallel port has been included inside parport_serial. Signed-off-by: Zhang Xincheng <zhangxincheng(a)uniontech.com> --- drivers/parport/parport_pc.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/drivers/parport/parport_pc.c b/drivers/parport/parport_pc.c index 7c45927e2131..cf0cefe38e90 100644 --- a/drivers/parport/parport_pc.c +++ b/drivers/parport/parport_pc.c @@ -2613,7 +2613,6 @@ enum parport_pc_pci_cards { netmos_9901, netmos_9865, quatech_sppxp100, - wch_ch382l, }; @@ -2677,7 +2676,6 @@ static struct parport_pc_pci { /* netmos_9901 */ { 1, { { 0, -1 }, } }, /* netmos_9865 */ { 1, { { 0, -1 }, } }, /* quatech_sppxp100 */ { 1, { { 0, 1 }, } }, - /* wch_ch382l */ { 1, { { 2, -1 }, } }, }; static const struct pci_device_id parport_pc_pci_tbl[] = { @@ -2769,8 +2767,6 @@ static const struct pci_device_id parport_pc_pci_tbl[] = { /* Quatech SPPXP-100 Parallel port PCI ExpressCard */ { PCI_VENDOR_ID_QUATECH, PCI_DEVICE_ID_QUATECH_SPPXP_100, PCI_ANY_ID, PCI_ANY_ID, 0, 0, quatech_sppxp100 }, - /* WCH CH382L PCI-E single parallel port card */ - { 0x1c00, 0x3050, 0x1c00, 0x3050, 0, 0, wch_ch382l }, { 0, } /* terminate list */ }; MODULE_DEVICE_TABLE(pci, parport_pc_pci_tbl); -- 2.20.1

1 year, 6 months

[PATCH v4] drm/sched: Fix kernel NULL pointer dereference error

by Arvind Yadav

-This is purely a timing issue. Here, sometimes Job free is happening before the job is done. To fix this issue moving 'dma_fence_cb' callback from job(struct drm_sched_job) to scheduler fence (struct drm_sched_fence). -Added drm_sched_fence_set_parent() and drm_sched_fence_clear_parent() functions to move fence handling into sched_fence.c and this just cleanup. BUG: kernel NULL pointer dereference, address: 0000000000000088 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 0 P4D 0 Oops: 0000 [#1] PREEMPT SMP NOPTI CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1 Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018 RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched] Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66 0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0 RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087 RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018 RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000 RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708 FS: 0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0 Call Trace: <IRQ> drm_sched_job_done_cb+0x12/0x20 [gpu_sched] dma_fence_signal_timestamp_locked+0x7e/0x110 dma_fence_signal+0x31/0x60 amdgpu_fence_process+0xc4/0x140 [amdgpu] gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu] amdgpu_irq_dispatch+0xb7/0x210 [amdgpu] amdgpu_ih_process+0x86/0x100 [amdgpu] amdgpu_irq_handler+0x24/0x60 [amdgpu] __handle_irq_event_percpu+0x4b/0x190 handle_irq_event_percpu+0x15/0x50 handle_irq_event+0x39/0x60 handle_edge_irq+0xaf/0x210 __common_interrupt+0x6e/0x110 common_interrupt+0xc1/0xe0 </IRQ> <TASK> Signed-off-by: Arvind Yadav <Arvind.Yadav(a)amd.com> --- Changes in v2: Moving 'dma_fence_cb' callback from job(struct drm_sched_job) to scheduler fence(struct drm_sched_fence) instead of adding NULL check for s_fence. Changes in v3: Added drm_sched_fence_set_parent() function(and others *_parent_cb) in sched_fence.c. Moved parent fence intilization and callback installation into this (this just cleanup). Changes in v4 : Add drm_sched_fence_clear_parent() function in sched_fence.c. and done the changes as per review comments. --- drivers/gpu/drm/scheduler/sched_fence.c | 64 +++++++++++++++++++++++++ drivers/gpu/drm/scheduler/sched_main.c | 53 ++++---------------- include/drm/gpu_scheduler.h | 10 +++- 3 files changed, 81 insertions(+), 46 deletions(-) diff --git a/drivers/gpu/drm/scheduler/sched_fence.c b/drivers/gpu/drm/scheduler/sched_fence.c index 7fd869520ef2..68343614f9ed 100644 --- a/drivers/gpu/drm/scheduler/sched_fence.c +++ b/drivers/gpu/drm/scheduler/sched_fence.c @@ -78,6 +78,70 @@ static void drm_sched_fence_free_rcu(struct rcu_head *rcu) kmem_cache_free(sched_fence_slab, fence); } +/** + * drm_sched_fence_parent_cb - the callback for a done job + * @f: fence + * @cb: fence callbacks + */ +static void drm_sched_fence_parent_cb(struct dma_fence *f, struct dma_fence_cb *cb) +{ + struct drm_sched_fence *s_fence = container_of(cb, struct drm_sched_fence, + cb); + struct drm_gpu_scheduler *sched = s_fence->sched; + + atomic_dec(&sched->hw_rq_count); + atomic_dec(sched->score); + + dma_fence_get(&s_fence->finished); + drm_sched_fence_finished(s_fence); + dma_fence_put(&s_fence->finished); + wake_up_interruptible(&sched->wake_up_worker); +} + +/** + * drm_sched_fence_clear_parent - Remove callbacks from pending list + * @s_fence: pointer to the fence + * + * Remove callbacks from pending list and clear the parent fence. + */ +bool drm_sched_fence_clear_parent(struct drm_sched_fence *s_fence) +{ + if (s_fence->parent && + dma_fence_remove_callback(s_fence->parent, &s_fence->cb)) { + dma_fence_put(s_fence->parent); + s_fence->parent = NULL; + return true; + } + + return false; +} + +/** + * drm_sched_fence_set_parent - set the parent fence and add the callback + * @s_fence: pointer to the fence + * fence: pointer to the hw fence + * + * Set the parent fence and install the callback for a done job. + */ +void drm_sched_fence_set_parent(struct drm_sched_fence *s_fence, + struct dma_fence *fence) +{ + int r; + + if (s_fence->parent && + dma_fence_remove_callback(s_fence->parent, &s_fence->cb)) + dma_fence_put(s_fence->parent); + + /* We keep the reference of the parent fence here. */ + swap(s_fence->parent, fence); + dma_fence_put(fence); + + r = dma_fence_add_callback(s_fence->parent, &s_fence->cb, + drm_sched_fence_parent_cb); + if (r == -ENOENT) + drm_sched_fence_parent_cb(NULL, &s_fence->cb); +} + /** * drm_sched_fence_free - free up an uninitialized fence * diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 4cc59bae38dd..30597d9a949f 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -253,13 +253,12 @@ drm_sched_rq_select_entity_fifo(struct drm_sched_rq *rq) /** * drm_sched_job_done - complete a job - * @s_job: pointer to the job which is done + * @s_fence: pointer to the fence of a done job * * Finish the job's fence and wake up the worker thread. */ -static void drm_sched_job_done(struct drm_sched_job *s_job) +static void drm_sched_job_done(struct drm_sched_fence *s_fence) { - struct drm_sched_fence *s_fence = s_job->s_fence; struct drm_gpu_scheduler *sched = s_fence->sched; atomic_dec(&sched->hw_rq_count); @@ -273,18 +272,6 @@ static void drm_sched_job_done(struct drm_sched_job *s_job) wake_up_interruptible(&sched->wake_up_worker); } -/** - * drm_sched_job_done_cb - the callback for a done job - * @f: fence - * @cb: fence callbacks - */ -static void drm_sched_job_done_cb(struct dma_fence *f, struct dma_fence_cb *cb) -{ - struct drm_sched_job *s_job = container_of(cb, struct drm_sched_job, cb); - - drm_sched_job_done(s_job); -} - /** * drm_sched_dependency_optimized - test if the dependency can be optimized * @@ -504,11 +491,7 @@ void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad) */ list_for_each_entry_safe_reverse(s_job, tmp, &sched->pending_list, list) { - if (s_job->s_fence->parent && - dma_fence_remove_callback(s_job->s_fence->parent, - &s_job->cb)) { - dma_fence_put(s_job->s_fence->parent); - s_job->s_fence->parent = NULL; + if (drm_sched_fence_clear_parent(s_job->s_fence)) { atomic_dec(&sched->hw_rq_count); } else { /* @@ -560,7 +543,6 @@ EXPORT_SYMBOL(drm_sched_stop); void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery) { struct drm_sched_job *s_job, *tmp; - int r; /* * Locking the list is not required here as the sched thread is parked @@ -575,16 +557,10 @@ void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery) if (!full_recovery) continue; - if (fence) { - r = dma_fence_add_callback(fence, &s_job->cb, - drm_sched_job_done_cb); - if (r == -ENOENT) - drm_sched_job_done(s_job); - else if (r) - DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n", - r); - } else - drm_sched_job_done(s_job); + if (fence) + drm_sched_fence_set_parent(s_job->s_fence, fence); + else + drm_sched_job_done(s_job->s_fence); } if (full_recovery) { @@ -1008,7 +984,6 @@ static bool drm_sched_blocked(struct drm_gpu_scheduler *sched) static int drm_sched_main(void *param) { struct drm_gpu_scheduler *sched = (struct drm_gpu_scheduler *)param; - int r; sched_set_fifo_low(current); @@ -1049,22 +1024,12 @@ static int drm_sched_main(void *param) drm_sched_fence_scheduled(s_fence); if (!IS_ERR_OR_NULL(fence)) { - s_fence->parent = dma_fence_get(fence); - /* Drop for original kref_init of the fence */ - dma_fence_put(fence); - - r = dma_fence_add_callback(fence, &sched_job->cb, - drm_sched_job_done_cb); - if (r == -ENOENT) - drm_sched_job_done(sched_job); - else if (r) - DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n", - r); + drm_sched_fence_set_parent(s_fence, fence); } else { if (IS_ERR(fence)) dma_fence_set_error(&s_fence->finished, PTR_ERR(fence)); - drm_sched_job_done(sched_job); + drm_sched_job_done(s_fence); } wake_up(&sched->job_scheduled); diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 1f7d9dd1a444..5066729c15ce 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -281,6 +281,10 @@ struct drm_sched_fence { * @owner: job owner for debugging */ void *owner; + /** + * @cb: callback + */ + struct dma_fence_cb cb; }; struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f); @@ -300,7 +304,6 @@ struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f); * be scheduled further. * @s_priority: the priority of the job. * @entity: the entity to which this job belongs. - * @cb: the callback for the parent fence in s_fence. * * A job is created by the driver using drm_sched_job_init(), and * should call drm_sched_entity_push_job() once it wants the scheduler @@ -325,7 +328,6 @@ struct drm_sched_job { atomic_t karma; enum drm_sched_priority s_priority; struct drm_sched_entity *entity; - struct dma_fence_cb cb; /** * @dependencies: * @@ -559,6 +561,10 @@ void drm_sched_fence_free(struct drm_sched_fence *fence); void drm_sched_fence_scheduled(struct drm_sched_fence *fence); void drm_sched_fence_finished(struct drm_sched_fence *fence); +bool drm_sched_fence_clear_parent(struct drm_sched_fence *s_fence); +void drm_sched_fence_set_parent(struct drm_sched_fence *s_fence, + struct dma_fence *fence); + unsigned long drm_sched_suspend_timeout(struct drm_gpu_scheduler *sched); void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched, unsigned long remaining); -- 2.25.1

1 year, 6 months

[PATCH v3] drm/sched: Fix kernel NULL pointer dereference error

by Arvind Yadav

-This is purely a timing issue. Here, sometimes Job free is happening before the job is done. To fix this issue moving 'dma_fence_cb' callback from job(struct drm_sched_job) to scheduler fence (struct drm_sched_fence). - Added drm_sched_fence_set_parent() function(and others *_parent_cb) in sched_fence.c. Moved parent fence intilization and callback installation into this (this just cleanup). BUG: kernel NULL pointer dereference, address: 0000000000000088 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 0 P4D 0 Oops: 0000 [#1] PREEMPT SMP NOPTI CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1 Arvind : [dma_fence_default_wait _START] timeout = -1 Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018 RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched] Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66 0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0 RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087 RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018 RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000 RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708 FS: 0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0 Call Trace: <IRQ> drm_sched_job_done_cb+0x12/0x20 [gpu_sched] dma_fence_signal_timestamp_locked+0x7e/0x110 dma_fence_signal+0x31/0x60 amdgpu_fence_process+0xc4/0x140 [amdgpu] gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu] amdgpu_irq_dispatch+0xb7/0x210 [amdgpu] amdgpu_ih_process+0x86/0x100 [amdgpu] amdgpu_irq_handler+0x24/0x60 [amdgpu] __handle_irq_event_percpu+0x4b/0x190 handle_irq_event_percpu+0x15/0x50 handle_irq_event+0x39/0x60 handle_edge_irq+0xaf/0x210 __common_interrupt+0x6e/0x110 common_interrupt+0xc1/0xe0 </IRQ> <TASK> Signed-off-by: Arvind Yadav <Arvind.Yadav(a)amd.com> --- Changes in v2: Moving 'dma_fence_cb' callback from job(struct drm_sched_job) to scheduler fence(struct drm_sched_fence) instead of adding NULL check for s_fence. Changes in v3: Added drm_sched_fence_set_parent() function(and others *_parent_cb) in sched_fence.c. Moved parent fence intilization and callback installation into this (this just cleanup). --- drivers/gpu/drm/scheduler/sched_fence.c | 53 +++++++++++++++++++++++++ drivers/gpu/drm/scheduler/sched_main.c | 38 +++++------------- include/drm/gpu_scheduler.h | 12 +++++- 3 files changed, 72 insertions(+), 31 deletions(-) diff --git a/drivers/gpu/drm/scheduler/sched_fence.c b/drivers/gpu/drm/scheduler/sched_fence.c index 7fd869520ef2..f6808f363261 100644 --- a/drivers/gpu/drm/scheduler/sched_fence.c +++ b/drivers/gpu/drm/scheduler/sched_fence.c @@ -77,6 +77,59 @@ static void drm_sched_fence_free_rcu(struct rcu_head *rcu) if (!WARN_ON_ONCE(!fence)) kmem_cache_free(sched_fence_slab, fence); } +/** + * drm_sched_job_done_cb - the callback for a done job + * @f: fence + * @cb: fence callbacks + */ +static void drm_sched_job_done_cb(struct dma_fence *f, struct dma_fence_cb *cb) +{ + struct drm_sched_fence *s_fence = container_of(cb, struct drm_sched_fence, + cb); + struct drm_gpu_scheduler *sched = s_fence->sched; + + atomic_dec(&sched->hw_rq_count); + atomic_dec(sched->score); + + dma_fence_get(&s_fence->finished); + drm_sched_fence_finished(s_fence); + dma_fence_put(&s_fence->finished); + wake_up_interruptible(&sched->wake_up_worker); +} + +int drm_sched_fence_add_parent_cb(struct dma_fence *fence, + struct drm_sched_fence *s_fence) +{ + return dma_fence_add_callback(fence, &s_fence->cb, + drm_sched_job_done_cb); +} + +bool drm_sched_fence_remove_parent_cb(struct drm_sched_fence *s_fence) +{ + return dma_fence_remove_callback(s_fence->parent, + &s_fence->cb); +} + +/** + * drm_sched_fence_set_parent - set the parent fence and add the callback + * fence: pointer to the hw fence + * @s_fence: pointer to the fence + * + * Set the parent fence and intall the callback for a done job. + */ +int drm_sched_fence_set_parent(struct dma_fence *fence, + struct drm_sched_fence *s_fence) +{ + if (s_fence->parent && + dma_fence_remove_callback(s_fence->parent, &s_fence->cb)) + dma_fence_put(s_fence->parent); + + s_fence->parent = dma_fence_get(fence); + /* Drop for original kref_init of the fence */ + dma_fence_put(fence); + return dma_fence_add_callback(fence, &s_fence->cb, + drm_sched_job_done_cb); +} /** * drm_sched_fence_free - free up an uninitialized fence diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 4cc59bae38dd..cfb52e15f5b0 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -253,13 +253,12 @@ drm_sched_rq_select_entity_fifo(struct drm_sched_rq *rq) /** * drm_sched_job_done - complete a job - * @s_job: pointer to the job which is done + * @s_fence: pointer to the fence of a done job * * Finish the job's fence and wake up the worker thread. */ -static void drm_sched_job_done(struct drm_sched_job *s_job) +static void drm_sched_job_done(struct drm_sched_fence *s_fence) { - struct drm_sched_fence *s_fence = s_job->s_fence; struct drm_gpu_scheduler *sched = s_fence->sched; atomic_dec(&sched->hw_rq_count); @@ -273,18 +272,6 @@ static void drm_sched_job_done(struct drm_sched_job *s_job) wake_up_interruptible(&sched->wake_up_worker); } -/** - * drm_sched_job_done_cb - the callback for a done job - * @f: fence - * @cb: fence callbacks - */ -static void drm_sched_job_done_cb(struct dma_fence *f, struct dma_fence_cb *cb) -{ - struct drm_sched_job *s_job = container_of(cb, struct drm_sched_job, cb); - - drm_sched_job_done(s_job); -} - /** * drm_sched_dependency_optimized - test if the dependency can be optimized * @@ -505,8 +492,7 @@ void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad) list_for_each_entry_safe_reverse(s_job, tmp, &sched->pending_list, list) { if (s_job->s_fence->parent && - dma_fence_remove_callback(s_job->s_fence->parent, - &s_job->cb)) { + drm_sched_fence_remove_parent_cb(s_job->s_fence)) { dma_fence_put(s_job->s_fence->parent); s_job->s_fence->parent = NULL; atomic_dec(&sched->hw_rq_count); @@ -576,15 +562,14 @@ void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery) continue; if (fence) { - r = dma_fence_add_callback(fence, &s_job->cb, - drm_sched_job_done_cb); + r = drm_sched_fence_add_parent_cb(fence, s_job->s_fence); if (r == -ENOENT) - drm_sched_job_done(s_job); + drm_sched_job_done(s_job->s_fence); else if (r) DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n", r); } else - drm_sched_job_done(s_job); + drm_sched_job_done(s_job->s_fence); } if (full_recovery) { @@ -1049,14 +1034,9 @@ static int drm_sched_main(void *param) drm_sched_fence_scheduled(s_fence); if (!IS_ERR_OR_NULL(fence)) { - s_fence->parent = dma_fence_get(fence); - /* Drop for original kref_init of the fence */ - dma_fence_put(fence); - - r = dma_fence_add_callback(fence, &sched_job->cb, - drm_sched_job_done_cb); + r = drm_sched_fence_set_parent(fence, s_fence); if (r == -ENOENT) - drm_sched_job_done(sched_job); + drm_sched_job_done(s_fence); else if (r) DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n", r); @@ -1064,7 +1044,7 @@ static int drm_sched_main(void *param) if (IS_ERR(fence)) dma_fence_set_error(&s_fence->finished, PTR_ERR(fence)); - drm_sched_job_done(sched_job); + drm_sched_job_done(s_fence); } wake_up(&sched->job_scheduled); diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 1f7d9dd1a444..7258e2fa195f 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -281,6 +281,10 @@ struct drm_sched_fence { * @owner: job owner for debugging */ void *owner; + /** + * @cb: callback + */ + struct dma_fence_cb cb; }; struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f); @@ -300,7 +304,6 @@ struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f); * be scheduled further. * @s_priority: the priority of the job. * @entity: the entity to which this job belongs. - * @cb: the callback for the parent fence in s_fence. * * A job is created by the driver using drm_sched_job_init(), and * should call drm_sched_entity_push_job() once it wants the scheduler @@ -325,7 +328,6 @@ struct drm_sched_job { atomic_t karma; enum drm_sched_priority s_priority; struct drm_sched_entity *entity; - struct dma_fence_cb cb; /** * @dependencies: * @@ -559,6 +561,12 @@ void drm_sched_fence_free(struct drm_sched_fence *fence); void drm_sched_fence_scheduled(struct drm_sched_fence *fence); void drm_sched_fence_finished(struct drm_sched_fence *fence); +int drm_sched_fence_add_parent_cb(struct dma_fence *fence, + struct drm_sched_fence *s_fence); +bool drm_sched_fence_remove_parent_cb(struct drm_sched_fence *s_fence); +int drm_sched_fence_set_parent(struct dma_fence *fence, + struct drm_sched_fence *s_fence); + unsigned long drm_sched_suspend_timeout(struct drm_gpu_scheduler *sched); void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched, unsigned long remaining); -- 2.25.1

1 year, 6 months

[PATCH v3] dma-buf: Proactively round up to kmalloc bucket size

by Kees Cook

Instead of discovering the kmalloc bucket size _after_ allocation, round up proactively so the allocation is explicitly made for the full size, allowing the compiler to correctly reason about the resulting size of the buffer through the existing __alloc_size() hint. Cc: Sumit Semwal <sumit.semwal(a)linaro.org> Cc: linux-media(a)vger.kernel.org Cc: dri-devel(a)lists.freedesktop.org Cc: linaro-mm-sig(a)lists.linaro.org Reviewed-by: Christian König <christian.koenig(a)amd.com> Signed-off-by: Kees Cook <keescook(a)chromium.org> --- v3: added reviewed-by, rebase to v6.1-rc1 v2: https://lore.kernel.org/lkml/20220923202822.2667581-9-keescook@chromium.org/ --- drivers/dma-buf/dma-resv.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/dma-buf/dma-resv.c b/drivers/dma-buf/dma-resv.c index e3885c90a3ac..1c76aed8e262 100644 --- a/drivers/dma-buf/dma-resv.c +++ b/drivers/dma-buf/dma-resv.c @@ -98,12 +98,17 @@ static void dma_resv_list_set(struct dma_resv_list *list, static struct dma_resv_list *dma_resv_list_alloc(unsigned int max_fences) { struct dma_resv_list *list; + size_t size; - list = kmalloc(struct_size(list, table, max_fences), GFP_KERNEL); + /* Round up to the next kmalloc bucket size. */ + size = kmalloc_size_roundup(struct_size(list, table, max_fences)); + + list = kmalloc(size, GFP_KERNEL); if (!list) return NULL; - list->max_fences = (ksize(list) - offsetof(typeof(*list), table)) / + /* Given the resulting bucket size, recalculated max_fences. */ + list->max_fences = (size - offsetof(typeof(*list), table)) / sizeof(*list->table); return list; -- 2.34.1

1 year, 6 months

[PATCH v2] drm/sched: Fix kernel NULL pointer dereference error

by Arvind Yadav

This is purely a timing issue. Here, sometimes Job free is happening before the job is done. To fix this issue moving 'dma_fence_cb' callback from job(struct drm_sched_job) to scheduler fence (struct drm_sched_fence). BUG: kernel NULL pointer dereference, address: 0000000000000088 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 0 P4D 0 Oops: 0000 [#1] PREEMPT SMP NOPTI CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1 Arvind : [dma_fence_default_wait _START] timeout = -1 Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018 RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched] Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66 0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0 RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087 RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018 RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000 RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708 FS: 0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0 Call Trace: <IRQ> drm_sched_job_done_cb+0x12/0x20 [gpu_sched] dma_fence_signal_timestamp_locked+0x7e/0x110 dma_fence_signal+0x31/0x60 amdgpu_fence_process+0xc4/0x140 [amdgpu] gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu] amdgpu_irq_dispatch+0xb7/0x210 [amdgpu] amdgpu_ih_process+0x86/0x100 [amdgpu] amdgpu_irq_handler+0x24/0x60 [amdgpu] __handle_irq_event_percpu+0x4b/0x190 handle_irq_event_percpu+0x15/0x50 handle_irq_event+0x39/0x60 handle_edge_irq+0xaf/0x210 __common_interrupt+0x6e/0x110 common_interrupt+0xc1/0xe0 </IRQ> <TASK> Signed-off-by: Arvind Yadav <Arvind.Yadav(a)amd.com> --- Changes in v2: Moving 'dma_fence_cb' callback from job(struct drm_sched_job) to scheduler fence(struct drm_sched_fence) instead of adding NULL check for s_fence. --- drivers/gpu/drm/scheduler/sched_main.c | 23 +++++++++++------------ include/drm/gpu_scheduler.h | 6 ++++-- 2 files changed, 15 insertions(+), 14 deletions(-) diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 4cc59bae38dd..62d8eca05b92 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -253,13 +253,12 @@ drm_sched_rq_select_entity_fifo(struct drm_sched_rq *rq) /** * drm_sched_job_done - complete a job - * @s_job: pointer to the job which is done + * @s_fence: pointer to the fence of a done job * * Finish the job's fence and wake up the worker thread. */ -static void drm_sched_job_done(struct drm_sched_job *s_job) +static void drm_sched_job_done(struct drm_sched_fence *s_fence) { - struct drm_sched_fence *s_fence = s_job->s_fence; struct drm_gpu_scheduler *sched = s_fence->sched; atomic_dec(&sched->hw_rq_count); @@ -280,9 +279,9 @@ static void drm_sched_job_done(struct drm_sched_job *s_job) */ static void drm_sched_job_done_cb(struct dma_fence *f, struct dma_fence_cb *cb) { - struct drm_sched_job *s_job = container_of(cb, struct drm_sched_job, cb); + struct drm_sched_fence *s_fence = container_of(cb, struct drm_sched_fence, cb); - drm_sched_job_done(s_job); + drm_sched_job_done(s_fence); } /** @@ -506,7 +505,7 @@ void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad) list) { if (s_job->s_fence->parent && dma_fence_remove_callback(s_job->s_fence->parent, - &s_job->cb)) { + &s_job->s_fence->cb)) { dma_fence_put(s_job->s_fence->parent); s_job->s_fence->parent = NULL; atomic_dec(&sched->hw_rq_count); @@ -576,15 +575,15 @@ void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery) continue; if (fence) { - r = dma_fence_add_callback(fence, &s_job->cb, + r = dma_fence_add_callback(fence, &s_job->s_fence->cb, drm_sched_job_done_cb); if (r == -ENOENT) - drm_sched_job_done(s_job); + drm_sched_job_done(s_job->s_fence); else if (r) DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n", r); } else - drm_sched_job_done(s_job); + drm_sched_job_done(s_job->s_fence); } if (full_recovery) { @@ -1053,10 +1052,10 @@ static int drm_sched_main(void *param) /* Drop for original kref_init of the fence */ dma_fence_put(fence); - r = dma_fence_add_callback(fence, &sched_job->cb, + r = dma_fence_add_callback(fence, &s_fence->cb, drm_sched_job_done_cb); if (r == -ENOENT) - drm_sched_job_done(sched_job); + drm_sched_job_done(s_fence); else if (r) DRM_DEV_ERROR(sched->dev, "fence add callback failed (%d)\n", r); @@ -1064,7 +1063,7 @@ static int drm_sched_main(void *param) if (IS_ERR(fence)) dma_fence_set_error(&s_fence->finished, PTR_ERR(fence)); - drm_sched_job_done(sched_job); + drm_sched_job_done(s_fence); } wake_up(&sched->job_scheduled); diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 1f7d9dd1a444..9a96d49dfd75 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -281,6 +281,10 @@ struct drm_sched_fence { * @owner: job owner for debugging */ void *owner; + /** + * @cb: callback + */ + struct dma_fence_cb cb; }; struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f); @@ -300,7 +304,6 @@ struct drm_sched_fence *to_drm_sched_fence(struct dma_fence *f); * be scheduled further. * @s_priority: the priority of the job. * @entity: the entity to which this job belongs. - * @cb: the callback for the parent fence in s_fence. * * A job is created by the driver using drm_sched_job_init(), and * should call drm_sched_entity_push_job() once it wants the scheduler @@ -325,7 +328,6 @@ struct drm_sched_job { atomic_t karma; enum drm_sched_priority s_priority; struct drm_sched_entity *entity; - struct dma_fence_cb cb; /** * @dependencies: * -- 2.25.1

1 year, 6 months

[PATCH v6 00/21] Move all drivers to a common dma-buf locking convention

by Dmitry Osipenko

Hello, This series moves all drivers to a dynamic dma-buf locking specification. From now on all dma-buf importers are made responsible for holding dma-buf's reservation lock around all operations performed over dma-bufs in accordance to the locking specification. This allows us to utilize reservation lock more broadly around kernel without fearing of a potential deadlocks. This patchset passes all i915 selftests. It was also tested using VirtIO, Panfrost, Lima, Tegra, udmabuf, AMDGPU and Nouveau drivers. I tested cases of display+GPU, display+V4L and GPU+V4L dma-buf sharing (where appropriate), which covers majority of kernel drivers since rest of the drivers share same or similar code paths. Changelog: v6: - Added r-b from Michael Ruhl to the i915 patch. - Added acks from Sumit Semwal and updated commit message of the "Move dma_buf_vmap() to dynamic locking specification" patch like was suggested by Sumit. - Added "!dmabuf" check to dma_buf_vmap_unlocked() to match the locked variant of the function, for consistency. v5: - Added acks and r-bs that were given to v4. - Changed i915 preparation patch like was suggested by Michael Ruhl. The scope of reservation locking is smaller now. v4: - Added dma_buf_mmap() to the "locking convention" documentation, which was missed by accident in v3. - Added acks from Christian König, Tomasz Figa and Hans Verkuil that they gave to couple v3 patches. - Dropped the "_unlocked" postfix from function names that don't have the locked variant, as was requested by Christian König. - Factored out the per-driver preparations into separate patches to ease reviewing of the changes, which is now doable without the global dma-buf functions renaming. - Factored out the dynamic locking convention enforcements into separate patches which add the final dma_resv_assert_held(dmabuf->resv) to the dma-buf API functions. v3: - Factored out dma_buf_mmap_unlocked() and attachment functions into aseparate patches, like was suggested by Christian König. - Corrected and factored out dma-buf locking documentation into a separate patch, like was suggested by Christian König. - Intel driver dropped the reservation locking fews days ago from its BO-release code path, but we need that locking for the imported GEMs because in the end that code path unmaps the imported GEM. So I added back the locking needed by the imported GEMs, updating the "dma-buf attachment locking specification" patch appropriately. - Tested Nouveau+Intel dma-buf import/export combo. - Tested udmabuf import to i915/Nouveau/AMDGPU. - Fixed few places in Etnaviv, Panfrost and Lima drivers that I missed to switch to locked dma-buf vmapping in the drm/gem: Take reservation lock for vmap/vunmap operations" patch. In a result invalidated the Christian's r-b that he gave to v2. - Added locked dma-buf vmap/vunmap functions that are needed for fixing vmappping of Etnaviv, Panfrost and Lima drivers mentioned above. I actually had this change stashed for the drm-shmem shrinker patchset, but then realized that it's already needed by the dma-buf patches. Also improved my tests to better cover these code paths. v2: - Changed locking specification to avoid problems with a cross-driver ww locking, like was suggested by Christian König. Now the attach/detach callbacks are invoked without the held lock and exporter should take the lock. - Added "locking convention" documentation that explains which dma-buf functions and callbacks are locked/unlocked for importers and exporters, which was requested by Christian König. - Added ack from Tomasz Figa to the V4L patches that he gave to v1. Dmitry Osipenko (21): dma-buf: Add unlocked variant of vmapping functions dma-buf: Add unlocked variant of attachment-mapping functions drm/gem: Take reservation lock for vmap/vunmap operations drm/prime: Prepare to dynamic dma-buf locking specification drm/armada: Prepare to dynamic dma-buf locking specification drm/i915: Prepare to dynamic dma-buf locking specification drm/omapdrm: Prepare to dynamic dma-buf locking specification drm/tegra: Prepare to dynamic dma-buf locking specification drm/etnaviv: Prepare to dynamic dma-buf locking specification RDMA/umem: Prepare to dynamic dma-buf locking specification misc: fastrpc: Prepare to dynamic dma-buf locking specification xen/gntdev: Prepare to dynamic dma-buf locking specification media: videobuf2: Prepare to dynamic dma-buf locking specification media: tegra-vde: Prepare to dynamic dma-buf locking specification dma-buf: Move dma_buf_vmap() to dynamic locking specification dma-buf: Move dma_buf_attach() to dynamic locking specification dma-buf: Move dma_buf_map_attachment() to dynamic locking specification dma-buf: Move dma_buf_mmap() to dynamic locking specification dma-buf: Document dynamic locking convention media: videobuf2: Stop using internal dma-buf lock dma-buf: Remove obsoleted internal lock Documentation/driver-api/dma-buf.rst | 6 + drivers/dma-buf/dma-buf.c | 214 +++++++++++++++--- drivers/gpu/drm/armada/armada_gem.c | 8 +- drivers/gpu/drm/drm_client.c | 4 +- drivers/gpu/drm/drm_gem.c | 24 ++ drivers/gpu/drm/drm_gem_dma_helper.c | 6 +- drivers/gpu/drm/drm_gem_framebuffer_helper.c | 6 +- drivers/gpu/drm/drm_gem_ttm_helper.c | 9 +- drivers/gpu/drm/drm_prime.c | 6 +- drivers/gpu/drm/etnaviv/etnaviv_gem_prime.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_object.c | 14 ++ .../drm/i915/gem/selftests/i915_gem_dmabuf.c | 16 +- drivers/gpu/drm/lima/lima_sched.c | 4 +- drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c | 4 +- drivers/gpu/drm/panfrost/panfrost_dump.c | 4 +- drivers/gpu/drm/panfrost/panfrost_perfcnt.c | 6 +- drivers/gpu/drm/qxl/qxl_object.c | 17 +- drivers/gpu/drm/qxl/qxl_prime.c | 4 +- drivers/gpu/drm/tegra/gem.c | 17 +- drivers/infiniband/core/umem_dmabuf.c | 7 +- .../common/videobuf2/videobuf2-dma-contig.c | 22 +- .../media/common/videobuf2/videobuf2-dma-sg.c | 19 +- .../common/videobuf2/videobuf2-vmalloc.c | 17 +- .../platform/nvidia/tegra-vde/dmabuf-cache.c | 6 +- drivers/misc/fastrpc.c | 6 +- drivers/xen/gntdev-dmabuf.c | 8 +- include/drm/drm_gem.h | 3 + include/linux/dma-buf.h | 17 +- 29 files changed, 323 insertions(+), 155 deletions(-) -- 2.37.3

1 year, 6 months

[PATCH AUTOSEL 5.10 07/22] udmabuf: Set ubuf->sg = NULL if the creation of sg table fails

by Sasha Levin

From: Vivek Kasireddy <vivek.kasireddy(a)intel.com> [ Upstream commit d9c04a1b7a15b5e74b2977461d9511e497f05d8f ] When userspace tries to map the dmabuf and if for some reason (e.g. OOM) the creation of the sg table fails, ubuf->sg needs to be set to NULL. Otherwise, when the userspace subsequently closes the dmabuf fd, we'd try to erroneously free the invalid sg table from release_udmabuf resulting in the following crash reported by syzbot: general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] PREEMPT SMP KASAN KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007] CPU: 0 PID: 3609 Comm: syz-executor487 Not tainted 5.19.0-syzkaller-13930-g7ebfc85e2cd7 #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/22/2022 RIP: 0010:dma_unmap_sgtable include/linux/dma-mapping.h:378 [inline] RIP: 0010:put_sg_table drivers/dma-buf/udmabuf.c:89 [inline] RIP: 0010:release_udmabuf+0xcb/0x4f0 drivers/dma-buf/udmabuf.c:114 Code: 48 89 fa 48 c1 ea 03 80 3c 02 00 0f 85 2b 04 00 00 48 8d 7d 0c 4c 8b 63 30 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <0f> b6 14 02 48 89 f8 83 e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 e2 RSP: 0018:ffffc900037efd30 EFLAGS: 00010246 RAX: dffffc0000000000 RBX: ffffffff8cb67800 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffffffff84ad27e0 RDI: 0000000000000000 RBP: fffffffffffffff4 R08: 0000000000000005 R09: 0000000000000000 R10: 0000000000000000 R11: 000000000008c07c R12: ffff88801fa05000 R13: ffff888073db07e8 R14: ffff888025c25440 R15: 0000000000000000 FS: 0000555555fc4300(0000) GS:ffff8880b9a00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fc1c0ce06e4 CR3: 00000000715e6000 CR4: 00000000003506f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: <TASK> dma_buf_release+0x157/0x2d0 drivers/dma-buf/dma-buf.c:78 __dentry_kill+0x42b/0x640 fs/dcache.c:612 dentry_kill fs/dcache.c:733 [inline] dput+0x806/0xdb0 fs/dcache.c:913 __fput+0x39c/0x9d0 fs/file_table.c:333 task_work_run+0xdd/0x1a0 kernel/task_work.c:177 ptrace_notify+0x114/0x140 kernel/signal.c:2353 ptrace_report_syscall include/linux/ptrace.h:420 [inline] ptrace_report_syscall_exit include/linux/ptrace.h:482 [inline] syscall_exit_work kernel/entry/common.c:249 [inline] syscall_exit_to_user_mode_prepare+0x129/0x280 kernel/entry/common.c:276 __syscall_exit_to_user_mode_work kernel/entry/common.c:281 [inline] syscall_exit_to_user_mode+0x9/0x50 kernel/entry/common.c:294 do_syscall_64+0x42/0xb0 arch/x86/entry/common.c:86 entry_SYSCALL_64_after_hwframe+0x63/0xcd RIP: 0033:0x7fc1c0c35b6b Code: 0f 05 48 3d 00 f0 ff ff 77 45 c3 0f 1f 40 00 48 83 ec 18 89 7c 24 0c e8 63 fc ff ff 8b 7c 24 0c 41 89 c0 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 35 44 89 c7 89 44 24 0c e8 a1 fc ff ff 8b 44 RSP: 002b:00007ffd78a06090 EFLAGS: 00000293 ORIG_RAX: 0000000000000003 RAX: 0000000000000000 RBX: 0000000000000007 RCX: 00007fc1c0c35b6b RDX: 0000000020000280 RSI: 0000000040086200 RDI: 0000000000000006 RBP: 0000000000000007 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000293 R12: 000000000000000c R13: 0000000000000003 R14: 00007fc1c0cfe4a0 R15: 00007ffd78a06140 </TASK> Modules linked in: ---[ end trace 0000000000000000 ]--- RIP: 0010:dma_unmap_sgtable include/linux/dma-mapping.h:378 [inline] RIP: 0010:put_sg_table drivers/dma-buf/udmabuf.c:89 [inline] RIP: 0010:release_udmabuf+0xcb/0x4f0 drivers/dma-buf/udmabuf.c:114 Reported-by: syzbot+c80e9ef5d8bb45894db0(a)syzkaller.appspotmail.com Cc: Gerd Hoffmann <kraxel(a)redhat.com> Signed-off-by: Vivek Kasireddy <vivek.kasireddy(a)intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20220825063522.801264-1-vivek.… Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/dma-buf/udmabuf.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index b624f3d8f0e6..e359c5c6c4df 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -118,17 +118,20 @@ static int begin_cpu_udmabuf(struct dma_buf *buf, { struct udmabuf *ubuf = buf->priv; struct device *dev = ubuf->device->this_device; + int ret = 0; if (!ubuf->sg) { ubuf->sg = get_sg_table(dev, buf, direction); - if (IS_ERR(ubuf->sg)) - return PTR_ERR(ubuf->sg); + if (IS_ERR(ubuf->sg)) { + ret = PTR_ERR(ubuf->sg); + ubuf->sg = NULL; + } } else { dma_sync_sg_for_cpu(dev, ubuf->sg->sgl, ubuf->sg->nents, direction); } - return 0; + return ret; } static int end_cpu_udmabuf(struct dma_buf *buf, -- 2.35.1

1 year, 6 months

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

Linaro-mm-sig October 2022