Linaro-mm-sig March 2022

linaro-mm-sig@lists.linaro.org

33 participants
26 discussions

[syzbot] general protection fault in dma_fence_array_first

by syzbot

Hello, syzbot found the following issue on: HEAD commit: 8515d05bf6bc Add linux-next specific files for 20220328 git tree: linux-next console output: https://syzkaller.appspot.com/x/log.txt?x=1694e21b700000 kernel config: https://syzkaller.appspot.com/x/.config?x=530c68bef4e2b8a8 dashboard link: https://syzkaller.appspot.com/bug?extid=5c943fe38e86d615cac2 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=1467313b700000 C reproducer: https://syzkaller.appspot.com/x/repro.c?x=121b7cb9700000 The issue was bisected to: commit 519f490db07e1a539490612f376487f61e48e39c Author: Christian König <christian.koenig(a)amd.com> Date: Fri Mar 11 09:32:26 2022 +0000 dma-buf/sync-file: fix warning about fence containers bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=1058277d700000 final oops: https://syzkaller.appspot.com/x/report.txt?x=1258277d700000 console output: https://syzkaller.appspot.com/x/log.txt?x=1458277d700000 IMPORTANT: if you fix the issue, please add the following tag to the commit: Reported-by: syzbot+5c943fe38e86d615cac2(a)syzkaller.appspotmail.com Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers") general protection fault, probably for non-canonical address 0xdffffc0000000002: 0000 [#1] PREEMPT SMP KASAN KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017] CPU: 1 PID: 3595 Comm: syz-executor814 Not tainted 5.17.0-next-20220328-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:dma_fence_array_first+0x78/0xb0 drivers/dma-buf/dma-fence-array.c:234 Code: 00 00 fc ff df 48 89 fa 48 c1 ea 03 80 3c 02 00 75 43 48 8b 9b 88 00 00 00 48 b8 00 00 00 00 00 fc ff df 48 89 da 48 c1 ea 03 <80> 3c 02 00 75 1b 4c 8b 23 e8 fa a9 e0 fc 4c 89 e0 5b 41 5c c3 45 RSP: 0018:ffffc90003a4fd48 EFLAGS: 00010202 RAX: dffffc0000000000 RBX: 0000000000000010 RCX: 0000000000000000 RDX: 0000000000000002 RSI: ffffffff84980052 RDI: ffff888015c76388 RBP: ffff888015c76300 R08: 0000000000000000 R09: ffff888015c7633b R10: ffffffff8498f6ba R11: 0000000000000000 R12: ffff888015c76300 R13: ffff888015c76690 R14: 00000000c0383e04 R15: 0000000020001840 FS: 0000555556872300(0000) GS:ffff8880b9d00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020001528 CR3: 000000001e82f000 CR4: 00000000003506e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: <TASK> __dma_fence_unwrap_array include/linux/dma-fence-unwrap.h:42 [inline] dma_fence_unwrap_first include/linux/dma-fence-unwrap.h:57 [inline] sync_file_ioctl_fence_info drivers/dma-buf/sync_file.c:414 [inline] sync_file_ioctl+0x248/0x22c0 drivers/dma-buf/sync_file.c:477 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:870 [inline] __se_sys_ioctl fs/ioctl.c:856 [inline] __x64_sys_ioctl+0x193/0x200 fs/ioctl.c:856 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f6aae8951b9 Code: 28 c3 e8 2a 14 00 00 66 2e 0f 1f 84 00 00 00 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffedd290238 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f6aae8951b9 RDX: 0000000020001840 RSI: 00000000c0383e04 RDI: 0000000000000007 RBP: 00007f6aae8591a0 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00007f6aae859230 R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000 </TASK> Modules linked in: ---[ end trace 0000000000000000 ]--- RIP: 0010:dma_fence_array_first+0x78/0xb0 drivers/dma-buf/dma-fence-array.c:234 Code: 00 00 fc ff df 48 89 fa 48 c1 ea 03 80 3c 02 00 75 43 48 8b 9b 88 00 00 00 48 b8 00 00 00 00 00 fc ff df 48 89 da 48 c1 ea 03 <80> 3c 02 00 75 1b 4c 8b 23 e8 fa a9 e0 fc 4c 89 e0 5b 41 5c c3 45 RSP: 0018:ffffc90003a4fd48 EFLAGS: 00010202 RAX: dffffc0000000000 RBX: 0000000000000010 RCX: 0000000000000000 RDX: 0000000000000002 RSI: ffffffff84980052 RDI: ffff888015c76388 RBP: ffff888015c76300 R08: 0000000000000000 R09: ffff888015c7633b R10: ffffffff8498f6ba R11: 0000000000000000 R12: ffff888015c76300 R13: ffff888015c76690 R14: 00000000c0383e04 R15: 0000000020001840 FS: 0000555556872300(0000) GS:ffff8880b9d00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020001528 CR3: 000000001e82f000 CR4: 00000000003506e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 ---------------- Code disassembly (best guess), 4 bytes skipped: 0: df 48 89 fisttps -0x77(%rax) 3: fa cli 4: 48 c1 ea 03 shr $0x3,%rdx 8: 80 3c 02 00 cmpb $0x0,(%rdx,%rax,1) c: 75 43 jne 0x51 e: 48 8b 9b 88 00 00 00 mov 0x88(%rbx),%rbx 15: 48 b8 00 00 00 00 00 movabs $0xdffffc0000000000,%rax 1c: fc ff df 1f: 48 89 da mov %rbx,%rdx 22: 48 c1 ea 03 shr $0x3,%rdx * 26: 80 3c 02 00 cmpb $0x0,(%rdx,%rax,1) <-- trapping instruction 2a: 75 1b jne 0x47 2c: 4c 8b 23 mov (%rbx),%r12 2f: e8 fa a9 e0 fc callq 0xfce0aa2e 34: 4c 89 e0 mov %r12,%rax 37: 5b pop %rbx 38: 41 5c pop %r12 3a: c3 retq 3b: 45 rex.RB --- This report is generated by a bot. It may contain errors. See https://goo.gl/tpsmEJ for more information about syzbot. syzbot engineers can be reached at syzkaller(a)googlegroups.com. syzbot will keep track of this issue. See: https://goo.gl/tpsmEJ#status for how to communicate with syzbot. For information about bisection process see: https://goo.gl/tpsmEJ#bisection syzbot can test patches for this issue, for details see: https://goo.gl/tpsmEJ#testing-patches

2 years, 9 months

Re: [PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

by nirmoy.das＠linux.intel.com

The content of this message was lost. It was probably cross-posted to multiple lists and previously handled on another list.

2 years, 9 months

[PATCH next,v2] kernel: Add 1 ms delay to init handler to fix s3 resume hang

by Zhenneng Li

This is a workaround for s3 resume hang for r7 340(amdgpu). When we test s3 with r7 340 on arm64 platform, graphics card will hang up, the error message are as follows: Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.599374][ 7] [ T291] amdgpu 0000:02:00.0: fb0: amdgpudrmfb frame buffer device Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.612869][ 7] [ T291] [drm:amdgpu_device_ip_late_init [amdgpu]] *ERROR* late_init of IP block <si_dpm> failed -22 Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.623392][ 7] [ T291] amdgpu 0000:02:00.0: amdgpu_device_ip_late_init failed Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.630696][ 7] [ T291] amdgpu 0000:02:00.0: Fatal error during GPU init Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.637477][ 7] [ T291] [drm] amdgpu: finishing device. On the following hardware: lspci -nn -s 05:00.0 05:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Oland [Radeon HD 8570 / R7 240/340 / Radeon 520 OEM] [1002:6611] (rev 87) Signed-off-by: Zhenneng Li <lizhenneng(a)kylinos.cn> --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 3987ecb24ef4..1eced991b5b2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2903,6 +2903,8 @@ static void amdgpu_device_delayed_init_work_handler(struct work_struct *work) container_of(work, struct amdgpu_device, delayed_init_work.work); int r; + mdelay(1); + r = amdgpu_ib_ring_tests(adev); if (r) DRM_ERROR("ib ring test failed (%d).\n", r); -- 2.25.1

2 years, 9 months

[PATCH 1/2] dma-buf/sync-file: fix logic error in new fence merge code

by Christian König

When the array is empty because everything is signaled we can't use add_fence() to add something because that would filter the signaled fence again. Signed-off-by: Christian König <christian.koenig(a)amd.com> Fixes: 519f490db07e ("dma-buf/sync-file: fix warning about fence containers") --- drivers/dma-buf/sync_file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c index b8dea4ec123b..514d213261df 100644 --- a/drivers/dma-buf/sync_file.c +++ b/drivers/dma-buf/sync_file.c @@ -262,7 +262,7 @@ static struct sync_file *sync_file_merge(const char *name, struct sync_file *a, } if (index == 0) - add_fence(fences, &index, dma_fence_get_stub()); + fences[index++] = dma_fence_get_stub(); if (num_fences > index) { struct dma_fence **tmp; -- 2.25.1

2 years, 9 months

[PATCH] drm/amdgpu: resolve s3 hang for r7340

by Zhenneng Li

This is a workaround for s3 hang for r7340(amdgpu). When we test s3 with r7340 on arm64 platform, graphics card will hang up, the error message are as follows: Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.599374][ 7] [ T291] amdgpu 0000:02:00.0: fb0: amdgpudrmfb frame buffer device Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.612869][ 7] [ T291] [drm:amdgpu_device_ip_late_init [amdgpu]] *ERROR* late_init of IP block <si_dpm> failed -22 Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.623392][ 7] [ T291] amdgpu 0000:02:00.0: amdgpu_device_ip_late_init failed Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.630696][ 7] [ T291] amdgpu 0000:02:00.0: Fatal error during GPU init Mar 4 01:14:11 greatwall-GW-XXXXXX-XXX kernel: [ 1.637477][ 7] [ T291] [drm] amdgpu: finishing device. Change-Id: I5048b3894c0ca9faf2f4847ddab61f9eb17b4823 Signed-off-by: Zhenneng Li <lizhenneng(a)kylinos.cn> --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 3987ecb24ef4..1eced991b5b2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2903,6 +2903,8 @@ static void amdgpu_device_delayed_init_work_handler(struct work_struct *work) container_of(work, struct amdgpu_device, delayed_init_work.work); int r; + mdelay(1); + r = amdgpu_ib_ring_tests(adev); if (r) DRM_ERROR("ib ring test failed (%d).\n", r); -- 2.25.1

2 years, 9 months

[PATCH 1/2] dma-buf: add dma_fence_unwrap

by Christian König

Add a general purpose helper to deep dive into dma_fence_chain/dma_fence_array structures and iterate over all the fences in them. This is useful when we need to flatten out all fences in those structures. Signed-off-by: Christian König <christian.koenig(a)amd.com> --- Documentation/driver-api/dma-buf.rst | 6 + drivers/dma-buf/Makefile | 1 + drivers/dma-buf/selftests.h | 1 + drivers/dma-buf/st-dma-fence-unwrap.c | 279 ++++++++++++++++++++++++++ include/linux/dma-fence-unwrap.h | 99 +++++++++ 5 files changed, 386 insertions(+) create mode 100644 drivers/dma-buf/st-dma-fence-unwrap.c create mode 100644 include/linux/dma-fence-unwrap.h diff --git a/Documentation/driver-api/dma-buf.rst b/Documentation/driver-api/dma-buf.rst index 2cd7db82d9fe..7209500f08c8 100644 --- a/Documentation/driver-api/dma-buf.rst +++ b/Documentation/driver-api/dma-buf.rst @@ -194,6 +194,12 @@ DMA Fence Chain .. kernel-doc:: include/linux/dma-fence-chain.h :internal: +DMA Fence unwrap +~~~~~~~~~~~~~~~~ + +.. kernel-doc:: include/linux/dma-fence-unwrap.h + :internal: + DMA Fence uABI/Sync File ~~~~~~~~~~~~~~~~~~~~~~~~ diff --git a/drivers/dma-buf/Makefile b/drivers/dma-buf/Makefile index 511805dbeb75..4c9eb53ba3f8 100644 --- a/drivers/dma-buf/Makefile +++ b/drivers/dma-buf/Makefile @@ -12,6 +12,7 @@ dmabuf_selftests-y := \ selftest.o \ st-dma-fence.o \ st-dma-fence-chain.o \ + st-dma-fence-unwrap.o \ st-dma-resv.o obj-$(CONFIG_DMABUF_SELFTESTS) += dmabuf_selftests.o diff --git a/drivers/dma-buf/selftests.h b/drivers/dma-buf/selftests.h index 97d73aaa31da..851965867d9c 100644 --- a/drivers/dma-buf/selftests.h +++ b/drivers/dma-buf/selftests.h @@ -12,4 +12,5 @@ selftest(sanitycheck, __sanitycheck__) /* keep first (igt selfcheck) */ selftest(dma_fence, dma_fence) selftest(dma_fence_chain, dma_fence_chain) +selftest(dma_fence_unwrap, dma_fence_unwrap) selftest(dma_resv, dma_resv) diff --git a/drivers/dma-buf/st-dma-fence-unwrap.c b/drivers/dma-buf/st-dma-fence-unwrap.c new file mode 100644 index 000000000000..d821faaebe93 --- /dev/null +++ b/drivers/dma-buf/st-dma-fence-unwrap.c @@ -0,0 +1,279 @@ +// SPDX-License-Identifier: MIT + +/* + * Copyright (C) 2022 Advanced Micro Devices, Inc. + */ + +#include <linux/dma-fence-unwrap.h> +#if 0 +#include <linux/kernel.h> +#include <linux/kthread.h> +#include <linux/mm.h> +#include <linux/sched/signal.h> +#include <linux/slab.h> +#include <linux/spinlock.h> +#include <linux/random.h> +#endif + +#include "selftest.h" + +#define CHAIN_SZ (4 << 10) + +static struct kmem_cache *slab_fences; + +static inline struct mock_fence { + struct dma_fence base; + spinlock_t lock; +} *to_mock_fence(struct dma_fence *f) { + return container_of(f, struct mock_fence, base); +} + +static const char *mock_name(struct dma_fence *f) +{ + return "mock"; +} + +static void mock_fence_release(struct dma_fence *f) +{ + kmem_cache_free(slab_fences, to_mock_fence(f)); +} + +static const struct dma_fence_ops mock_ops = { + .get_driver_name = mock_name, + .get_timeline_name = mock_name, + .release = mock_fence_release, +}; + +static struct dma_fence *mock_fence(void) +{ + struct mock_fence *f; + + f = kmem_cache_alloc(slab_fences, GFP_KERNEL); + if (!f) + return NULL; + + spin_lock_init(&f->lock); + dma_fence_init(&f->base, &mock_ops, &f->lock, 0, 0); + + return &f->base; +} + +static struct dma_fence *mock_array(unsigned int num_fences, ...) +{ + struct dma_fence_array *array; + struct dma_fence **fences; + va_list valist; + int i; + + fences = kcalloc(num_fences, sizeof(*fences), GFP_KERNEL); + if (!fences) + return NULL; + + va_start(valist, num_fences); + for (i = 0; i < num_fences; ++i) + fences[i] = va_arg(valist, typeof(*fences)); + va_end(valist); + + array = dma_fence_array_create(num_fences, fences, + dma_fence_context_alloc(1), + 1, false); + if (!array) + goto cleanup; + return &array->base; + +cleanup: + for (i = 0; i < num_fences; ++i) + dma_fence_put(fences[i]); + kfree(fences); + return NULL; +} + +static struct dma_fence *mock_chain(struct dma_fence *prev, + struct dma_fence *fence) +{ + struct dma_fence_chain *f; + + f = dma_fence_chain_alloc(); + if (!f) { + dma_fence_put(prev); + dma_fence_put(fence); + return NULL; + } + + dma_fence_chain_init(f, prev, fence, 1); + return &f->base; +} + +static int sanitycheck(void *arg) +{ + struct dma_fence *f, *chain, *array; + int err = 0; + + f = mock_fence(); + if (!f) + return -ENOMEM; + + array = mock_array(1, f); + if (!array) + return -ENOMEM; + + chain = mock_chain(NULL, array); + if (!chain) + return -ENOMEM; + + dma_fence_signal(f); + dma_fence_put(chain); + return err; +} + +static int unwrap_array(void *arg) +{ + struct dma_fence *fence, *f1, *f2, *array; + struct dma_fence_unwrap iter; + int err = 0; + + f1 = mock_fence(); + if (!f1) + return -ENOMEM; + + f2 = mock_fence(); + if (!f2) { + dma_fence_put(f1); + return -ENOMEM; + } + + array = mock_array(2, f1, f2); + if (!array) + return -ENOMEM; + + dma_fence_unwrap_for_each(fence, &iter, array) { + if (fence == f1) { + f1 = NULL; + } else if (fence == f2) { + f2 = NULL; + } else { + pr_err("Unexpected fence!\n"); + err = -EINVAL; + } + } + + if (f1 || f2) { + pr_err("Not all fences seen!\n"); + err = -EINVAL; + } + + dma_fence_signal(f1); + dma_fence_signal(f2); + dma_fence_put(array); + return 0; +} + +static int unwrap_chain(void *arg) +{ + struct dma_fence *fence, *f1, *f2, *chain; + struct dma_fence_unwrap iter; + int err = 0; + + f1 = mock_fence(); + if (!f1) + return -ENOMEM; + + f2 = mock_fence(); + if (!f2) { + dma_fence_put(f1); + return -ENOMEM; + } + + chain = mock_chain(f1, f2); + if (!chain) + return -ENOMEM; + + dma_fence_unwrap_for_each(fence, &iter, chain) { + if (fence == f1) { + f1 = NULL; + } else if (fence == f2) { + f2 = NULL; + } else { + pr_err("Unexpected fence!\n"); + err = -EINVAL; + } + } + + if (f1 || f2) { + pr_err("Not all fences seen!\n"); + err = -EINVAL; + } + + dma_fence_signal(f1); + dma_fence_signal(f2); + dma_fence_put(chain); + return 0; +} + +static int unwrap_chain_array(void *arg) +{ + struct dma_fence *fence, *f1, *f2, *array, *chain; + struct dma_fence_unwrap iter; + int err = 0; + + f1 = mock_fence(); + if (!f1) + return -ENOMEM; + + f2 = mock_fence(); + if (!f2) { + dma_fence_put(f1); + return -ENOMEM; + } + + array = mock_array(2, f1, f2); + if (!array) + return -ENOMEM; + + chain = mock_chain(NULL, array); + if (!chain) + return -ENOMEM; + + dma_fence_unwrap_for_each(fence, &iter, chain) { + if (fence == f1) { + f1 = NULL; + } else if (fence == f2) { + f2 = NULL; + } else { + pr_err("Unexpected fence!\n"); + err = -EINVAL; + } + } + + if (f1 || f2) { + pr_err("Not all fences seen!\n"); + err = -EINVAL; + } + + dma_fence_signal(f1); + dma_fence_signal(f2); + dma_fence_put(chain); + return 0; +} + +int dma_fence_unwrap(void) +{ + static const struct subtest tests[] = { + SUBTEST(sanitycheck), + SUBTEST(unwrap_array), + SUBTEST(unwrap_chain), + SUBTEST(unwrap_chain_array), + }; + int ret; + + slab_fences = KMEM_CACHE(mock_fence, + SLAB_TYPESAFE_BY_RCU | + SLAB_HWCACHE_ALIGN); + if (!slab_fences) + return -ENOMEM; + + ret = subtests(tests, NULL); + + kmem_cache_destroy(slab_fences); + return ret; +} diff --git a/include/linux/dma-fence-unwrap.h b/include/linux/dma-fence-unwrap.h new file mode 100644 index 000000000000..54963df00c98 --- /dev/null +++ b/include/linux/dma-fence-unwrap.h @@ -0,0 +1,99 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * fence-chain: chain fences together in a timeline + * + * Copyright (C) 2022 Advanced Micro Devices, Inc. + * Authors: + * Christian König <christian.koenig(a)amd.com> + */ + +#ifndef __LINUX_DMA_FENCE_UNWRAP_H +#define __LINUX_DMA_FENCE_UNWRAP_H + +#include <linux/dma-fence-chain.h> +#include <linux/dma-fence-array.h> + +/** + * struct dma_fence_unwrap - cursor into the container structure + */ +struct dma_fence_unwrap { + /** + * @chain: potential dma_fence_chain, but can be other fence as well + */ + struct dma_fence *chain; + /** + * @array: potential dma_fence_array, but can be other fence as well + */ + struct dma_fence *array; + /** + * @index: last returned index if @array is really a dma_fence_array + */ + unsigned int index; +}; + +/** + * dma_fence_unwrap_array - helper to unwrap dma_fence_arrays + * @cursor: cursor to initialize + * + * Helper function to unwrap dma_fence_array containers, don't touch directly. + * Use dma_fence_unwrap_first/next instead. + */ +static inline struct dma_fence * +dma_fence_unwrap_array(struct dma_fence_unwrap * cursor) +{ + cursor->array = dma_fence_chain_contained(cursor->chain); + cursor->index = 0; + return dma_fence_array_first(cursor->array); +} + +/** + * dma_fence_unwrap_first - return the first fence from fence containers + * @head: the entrypoint into the containers + * @cursor: current position inside the containers + * + * Unwraps potential dma_fence_chain/dma_fence_array containers and return the + * first fence. + */ +static inline struct dma_fence * +dma_fence_unwrap_first(struct dma_fence *head, struct dma_fence_unwrap *cursor) +{ + cursor->chain = dma_fence_get(head); + return dma_fence_unwrap_array(cursor); +} + +/** + * dma_fence_unwrap_next - return the next fence from a fence containers + * @cursor: current position inside the containers + * + * Continue unwrapping the dma_fence_chain/dma_fence_array containers and return + * the next fence from them. + */ +static inline struct dma_fence * +dma_fence_unwrap_next(struct dma_fence_unwrap *cursor) +{ + struct dma_fence *tmp; + + ++cursor->index; + tmp = dma_fence_array_next(cursor->array, cursor->index); + if (tmp) + return tmp; + + cursor->chain = dma_fence_chain_walk(cursor->chain); + return dma_fence_unwrap_array(cursor); +} + +/** + * dma_fence_unwrap_for_each - iterate over all fences in containers + * @fence: current fence + * @cursor: current position inside the containers + * @head: starting point for the iterator + * + * Unwrap dma_fence_chain and dma_fence_array containers and deep dive into all + * potential fences in them. If @head is just a normal fence only that one is + * returned. + */ +#define dma_fence_unwrap_for_each(fence, cursor, head) \ + for (fence = dma_fence_unwrap_first(head, cursor); fence; \ + fence = dma_fence_unwrap_next(cursor)) + +#endif -- 2.25.1

2 years, 9 months

[PATCH] dma-buf: WIP dma_fence_array_first fix

by Christian König

Looks like somebody is creating an empty dma_fence_array. Signed-off-by: Christian König <christian.koenig(a)amd.com> --- drivers/dma-buf/dma-fence-array.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/dma-buf/dma-fence-array.c b/drivers/dma-buf/dma-fence-array.c index 52b85d292383..5c8a7084577b 100644 --- a/drivers/dma-buf/dma-fence-array.c +++ b/drivers/dma-buf/dma-fence-array.c @@ -159,6 +159,8 @@ struct dma_fence_array *dma_fence_array_create(int num_fences, struct dma_fence_array *array; size_t size = sizeof(*array); + WARN_ON(!num_fences || !fences); + /* Allocate the callback structures behind the array. */ size += num_fences * sizeof(struct dma_fence_array_cb); array = kzalloc(size, GFP_KERNEL); @@ -231,6 +233,9 @@ struct dma_fence *dma_fence_array_first(struct dma_fence *head) if (!array) return head; + if (!array->num_fences) + return NULL; + return array->fences[0]; } EXPORT_SYMBOL(dma_fence_array_first); -- 2.25.1

2 years, 9 months

Re: [RFC v3 5/8] dmabuf: Add gpu cgroup charge transfer function

by Michal Koutný

On Mon, Mar 21, 2022 at 04:54:26PM -0700, "T.J. Mercier" <tjmercier(a)google.com> wrote: > Since the charge is duplicated in two cgroups for a short period > before it is uncharged from the source cgroup I guess the situation > you're thinking about is a global (or common ancestor) limit? The common ancestor was on my mind (after the self-shortcut). > I can see how that would be a problem for transfers done this way and > an alternative would be to swap the order of the charge operations: > first uncharge, then try_charge. To be certain the uncharge is > reversible if the try_charge fails, I think I'd need either a mutex > used at all gpucg_*charge call sites or access to the gpucg_mutex, Yes, that'd provide safe conditions for such operations, although I'm not sure these special types of memory can afford global lock on their fast paths. > which implies adding transfer support to gpu.c as part of the gpucg_* > API itself and calling it here. Am I following correctly here? My idea was to provide a special API (apart from gpucp_{try_charge,uncharge}) to facilitate transfers... > This series doesn't actually add limit support just accounting, but > I'd like to get it right here. ...which could be implemented (or changed) depending on how the charging is realized internally. Michal

2 years, 9 months

[PATCH 04/23] RDMA: use dma_resv_wait() instead of extracting the fence

by Christian König

Use dma_resv_wait() instead of extracting the exclusive fence and waiting on it manually. Signed-off-by: Christian König <christian.koenig(a)amd.com> Reviewed-by: Daniel Vetter <daniel.vetter(a)ffwll.ch> Cc: Jason Gunthorpe <jgg(a)ziepe.ca> Cc: Leon Romanovsky <leon(a)kernel.org> Cc: Maor Gottlieb <maorg(a)nvidia.com> Cc: Gal Pressman <galpress(a)amazon.com> Cc: linux-media(a)vger.kernel.org Cc: linaro-mm-sig(a)lists.linaro.org --- drivers/infiniband/core/umem_dmabuf.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) diff --git a/drivers/infiniband/core/umem_dmabuf.c b/drivers/infiniband/core/umem_dmabuf.c index f0760741f281..d32cd7538835 100644 --- a/drivers/infiniband/core/umem_dmabuf.c +++ b/drivers/infiniband/core/umem_dmabuf.c @@ -16,7 +16,6 @@ int ib_umem_dmabuf_map_pages(struct ib_umem_dmabuf *umem_dmabuf) { struct sg_table *sgt; struct scatterlist *sg; - struct dma_fence *fence; unsigned long start, end, cur = 0; unsigned int nmap = 0; int i; @@ -68,11 +67,8 @@ int ib_umem_dmabuf_map_pages(struct ib_umem_dmabuf *umem_dmabuf) * may be not up-to-date. Wait for the exporter to finish * the migration. */ - fence = dma_resv_excl_fence(umem_dmabuf->attach->dmabuf->resv); - if (fence) - return dma_fence_wait(fence, false); - - return 0; + return dma_resv_wait_timeout(umem_dmabuf->attach->dmabuf->resv, false, + false, MAX_SCHEDULE_TIMEOUT); } EXPORT_SYMBOL(ib_umem_dmabuf_map_pages); -- 2.25.1

2 years, 9 months

[RFC v3 0/8] Proposal for a GPU cgroup controller

by T.J. Mercier

This patch series revisits the proposal for a GPU cgroup controller to track and limit memory allocations by various device/allocator subsystems. The patch series also contains a simple prototype to illustrate how Android intends to implement DMA-BUF allocator attribution using the GPU cgroup controller. The prototype does not include resource limit enforcements. Changelog: v3: Remove Upstreaming Plan from gpu-cgroup.rst per John Stultz Use more common dual author commit message format per John Stultz Remove android from binder changes title per Todd Kjos Add a kselftest for this new behavior per Greg Kroah-Hartman Include details on behavior for all combinations of kernel/userspace versions in changelog (thanks Suren Baghdasaryan) per Greg Kroah-Hartman. Fix pid and uid types in binder UAPI header v2: See the previous revision of this change submitted by Hridya Valsaraju at: https://lore.kernel.org/all/20220115010622.3185921-1-hridya@google.com/ Move dma-buf cgroup charge transfer from a dma_buf_op defined by every heap to a single dma-buf function for all heaps per Daniel Vetter and Christian König. Pointers to struct gpucg and struct gpucg_device tracking the current associations were added to the dma_buf struct to achieve this. Fix incorrect Kconfig help section indentation per Randy Dunlap. History of the GPU cgroup controller ==================================== The GPU/DRM cgroup controller came into being when a consensus[1] was reached that the resources it tracked were unsuitable to be integrated into memcg. Originally, the proposed controller was specific to the DRM subsystem and was intended to track GEM buffers and GPU-specific resources[2]. In order to help establish a unified memory accounting model for all GPU and all related subsystems, Daniel Vetter put forth a suggestion to move it out of the DRM subsystem so that it can be used by other DMA-BUF exporters as well[3]. This RFC proposes an interface that does the same. [1]: https://patchwork.kernel.org/project/dri-devel/cover/20190501140438.9506-1-… [2]: https://lore.kernel.org/amd-gfx/20210126214626.16260-1-brian.welty@intel.co… [3]: https://lore.kernel.org/amd-gfx/YCVOl8%2F87bqRSQei@phenom.ffwll.local/ Hridya Valsaraju (5): gpu: rfc: Proposal for a GPU cgroup controller cgroup: gpu: Add a cgroup controller for allocator attribution of GPU memory dmabuf: heaps: export system_heap buffers with GPU cgroup charging dmabuf: Add gpu cgroup charge transfer function binder: Add a buffer flag to relinquish ownership of fds T.J. Mercier (3): dmabuf: Use the GPU cgroup charge/uncharge APIs binder: use __kernel_pid_t and __kernel_uid_t for userspace selftests: Add binder cgroup gpu memory transfer test Documentation/gpu/rfc/gpu-cgroup.rst | 183 +++++++ Documentation/gpu/rfc/index.rst | 4 + drivers/android/binder.c | 26 + drivers/dma-buf/dma-buf.c | 100 ++++ drivers/dma-buf/dma-heap.c | 27 + drivers/dma-buf/heaps/system_heap.c | 3 + include/linux/cgroup_gpu.h | 127 +++++ include/linux/cgroup_subsys.h | 4 + include/linux/dma-buf.h | 22 +- include/linux/dma-heap.h | 11 + include/uapi/linux/android/binder.h | 5 +- init/Kconfig | 7 + kernel/cgroup/Makefile | 1 + kernel/cgroup/gpu.c | 304 +++++++++++ .../selftests/drivers/android/binder/Makefile | 8 + .../drivers/android/binder/binder_util.c | 254 +++++++++ .../drivers/android/binder/binder_util.h | 32 ++ .../selftests/drivers/android/binder/config | 4 + .../binder/test_dmabuf_cgroup_transfer.c | 480 ++++++++++++++++++ 19 files changed, 1598 insertions(+), 4 deletions(-) create mode 100644 Documentation/gpu/rfc/gpu-cgroup.rst create mode 100644 include/linux/cgroup_gpu.h create mode 100644 kernel/cgroup/gpu.c create mode 100644 tools/testing/selftests/drivers/android/binder/Makefile create mode 100644 tools/testing/selftests/drivers/android/binder/binder_util.c create mode 100644 tools/testing/selftests/drivers/android/binder/binder_util.h create mode 100644 tools/testing/selftests/drivers/android/binder/config create mode 100644 tools/testing/selftests/drivers/android/binder/test_dmabuf_cgroup_transfer.c -- 2.35.1.616.g0bdcbb4464-goog

2 years, 9 months

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

Linaro-mm-sig March 2022