Description of problem: Output of command 'dmesg' exhibits error multiple blocks reading: [..] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [..] swiotlb: coherent allocation failed for device 0000:01:00.0 size=2097152 [..] CPU: 0 PID: 1914 Comm: gnome-shell Not tainted 4.15.3-300.fc27.x86_64 #1 [..] Hardware name: LENOVO 2768W9J/2768W9J, BIOS 7UET94WW (3.24 ) 10/17/2012 [..] Call Trace: [..] dump_stack+0x5c/0x85 [..] swiotlb_alloc_coherent+0xe0/0x150 [..] ttm_dma_pool_get_pages+0x20e/0x5e0 [ttm] [..] ttm_dma_populate+0x24d/0x340 [ttm] [..] ttm_tt_bind+0x29/0x60 [ttm] [..] ttm_bo_handle_move_mem+0x5da/0x610 [ttm] [..] ? verify_cpu+0xdd/0x100 [..] ttm_bo_evict+0x14d/0x330 [ttm] [..] ? ttm_bo_man_get_node+0x3c/0xc0 [ttm] [..] ? update_load_avg+0x558/0x6c0 [..] ? update_curr+0x175/0x1d0 [..] ttm_mem_evict_first+0x161/0x1d0 [ttm] [..] ttm_bo_mem_space+0x344/0x4c0 [ttm] [..] ttm_bo_validate+0xce/0x150 [ttm] [..] ttm_bo_init_reserved+0x385/0x430 [ttm] [..] ttm_bo_init+0x2f/0x90 [ttm] [..] ? radeon_update_memory_usage.isra.3+0x50/0x50 [radeon] [..] radeon_bo_create+0x180/0x230 [radeon] [..] ? radeon_update_memory_usage.isra.3+0x50/0x50 [radeon] [..] radeon_gem_object_create+0xa7/0x1d0 [radeon] [..] ? radeon_gem_pwrite_ioctl+0x30/0x30 [radeon] [..] radeon_gem_create_ioctl+0x66/0x100 [radeon] [..] ? radeon_gem_pwrite_ioctl+0x30/0x30 [radeon] [..] drm_ioctl_kernel+0x5b/0xb0 [drm] [..] drm_ioctl+0x2d5/0x370 [drm] [..] ? radeon_gem_pwrite_ioctl+0x30/0x30 [radeon] [..] ? wake_up_q+0x70/0x70 [..] radeon_drm_ioctl+0x49/0x80 [radeon] [..] do_vfs_ioctl+0xa4/0x620 [..] SyS_ioctl+0x74/0x80 [..] ? SyS_clock_gettime+0x81/0xa0 [..] do_syscall_64+0x75/0x180 [..] entry_SYSCALL_64_after_hwframe+0x21/0x86 [..] RIP: 0033:0x7f13b6b748e7 [..] RSP: 002b:00007ffc83c37418 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 [..] RAX: ffffffffffffffda RBX: 0000000000000004 RCX: 00007f13b6b748e7 [..] RDX: 00007ffc83c37490 RSI: 00000000c020645d RDI: 000000000000000c [..] RBP: 00007ffc83c37490 R08: 0000000000000004 R09: 0000000000000002 [..] R10: 00007ffc83c37460 R11: 0000000000000246 R12: 00000000c020645d [..] R13: 000000000000000c R14: 0000564f3c007960 R15: 0000564f3c007960 Version-Release number of selected component (if applicable): kernel-4.15.3-300.fc27 How reproducible: Frequently Steps to Reproduce: 1. Boot current Fedora 27 on system with AMD Radeon RV620 video device. 2. Run GNOME session. Actual results: Swiotlb buffer is reported to be full all the time. Expected results: Swiotlb buffer is not running out of memory. Additional info: None
I'm seeing this on a Radeon HD 5450. It's new in the 4.15.3-300.fc27.x86_64 kernel; it didn't happen in 4.14.18-300.fc27 or any previous kernel.
Moving this to the graphics team for tracking
(In reply to Joachim Frieben from comment #0) > Steps to Reproduce: > 1. Boot current Fedora 27 on system with AMD Radeon RV620 video device. > 2. Run GNOME session. > > Actual results: > Swiotlb buffer is reported to be full all the time. I can't repoduce. 01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] RV620/M82 [Mobility Radeon HD 3450/3470] 4.15.3-300.fc27.x86_64 Gnome (wayland)
I got the same on 03:00.0 VGA compatible controller: NVIDIA Corporation C79 [GeForce 9300 / nForce 730i] (rev b1) [ 2492.016898] nouveau 0000:03:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 2492.016900] swiotlb: coherent allocation failed for device 0000:03:00.0 size=2097152 [ 2492.016904] CPU: 0 PID: 1652 Comm: Xwayland Not tainted 4.15.4-300.vanilla.knurd.1.fc27.x86_64 #1 [ 2492.016904] Hardware name: NVIDIA MCP7A/MCP7A, BIOS 6.00 PG 04/22/2009 [ 2492.016905] Call Trace: [ 2492.016916] dump_stack+0x5c/0x85 [ 2492.016920] swiotlb_alloc_coherent+0xe0/0x150 [ 2492.016933] ttm_dma_pool_get_pages+0x20e/0x5e0 [ttm] [ 2492.016939] ttm_dma_populate+0x24d/0x340 [ttm] [ 2492.016943] ttm_tt_bind+0x29/0x60 [ttm] [ 2492.016947] ttm_bo_handle_move_mem+0x5da/0x610 [ttm] [ 2492.016952] ttm_bo_validate+0x135/0x150 [ttm] [ 2492.016956] ttm_bo_init_reserved+0x385/0x430 [ttm] [ 2492.016961] ttm_bo_init+0x2f/0x90 [ttm] [ 2492.017029] ? nouveau_bo_invalidate_caches+0x10/0x10 [nouveau] [ 2492.017031] ? _cond_resched+0x15/0x40 [ 2492.017066] nouveau_bo_new+0x416/0x590 [nouveau] [ 2492.017102] ? nouveau_bo_invalidate_caches+0x10/0x10 [nouveau] [ 2492.017136] ? nouveau_gem_new+0x120/0x120 [nouveau] [ 2492.017170] nouveau_gem_new+0x5d/0x120 [nouveau] [ 2492.017205] nouveau_gem_ioctl_new+0x51/0xd0 [nouveau] [ 2492.017234] drm_ioctl_kernel+0x5b/0xb0 [drm] [ 2492.017246] drm_ioctl+0x2d5/0x370 [drm] [ 2492.017280] ? nouveau_gem_new+0x120/0x120 [nouveau] [ 2492.017283] ? update_load_avg+0x558/0x6b0 [ 2492.017285] ? update_load_avg+0x558/0x6b0 [ 2492.017319] nouveau_drm_ioctl+0x64/0xc0 [nouveau] [ 2492.017323] do_vfs_ioctl+0xa4/0x620 [ 2492.017325] SyS_ioctl+0x74/0x80 [ 2492.017328] do_syscall_64+0x75/0x180 [ 2492.017330] entry_SYSCALL_64_after_hwframe+0x21/0x86 [ 2492.017333] RIP: 0033:0x7f6eaf4b38e7 [ 2492.017334] RSP: 002b:00007ffc55176d28 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 [ 2492.017336] RAX: ffffffffffffffda RBX: 0000000001e80b20 RCX: 00007f6eaf4b38e7 [ 2492.017337] RDX: 00007ffc55176d80 RSI: 00000000c0306480 RDI: 000000000000000b [ 2492.017338] RBP: 00007ffc55176d80 R08: 0000000000000004 R09: 00007f6eaf784c80 [ 2492.017339] R10: 0000000000000000 R11: 0000000000000246 R12: 00000000c0306480 [ 2492.017340] R13: 000000000000000b R14: 00000000020498a8 R15: 00000000019c66b0
For me this seems to only happen during periods of high disk IO load, possibly only for disk IO load on HDs. Both times I've seen this were probably write IO load to an ext3/4 filesystem on LVM on a software RAID mirror (on two HDs).
(In reply to Chris Siebenmann from comment #5) Yes. This happens when I copy files from hdd (xfs) to usb stick (fat32).
(In reply to Chris Siebenmann from comment #1) > I'm seeing this on a Radeon HD 5450. It's new in the 4.15.3-300.fc27.x86_64 > kernel; it didn't happen in 4.14.18-300.fc27 or any previous kernel. HD6450 Have the same recurring issue at least since Fedora 26 kernels. Suddendly runs out of memory, real slow then everything freezes, disk activity led full. I've limited tmpfs to 2GB, and it allowed to recover from one freeze. Happened last when loading a page in chrome, but the log shows swiotlb_alloc_coherent started few hours before
Seeing this on a Dell Latitude: 01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GF119M [NVS 4200M] [10de:1056] (rev a1) [568281.598393] nouveau 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [568281.598395] swiotlb: coherent allocation failed for device 0000:01:00.0 size=2097152 [568281.598398] CPU: 6 PID: 1231 Comm: Xorg Not tainted 4.15.3-300.fc27.x86_64 #1 [568281.598399] Hardware name: Dell Inc. Latitude E6420/032T9K, BIOS A24 05/12/2017 [568281.598400] Call Trace: [568281.598409] dump_stack+0x5c/0x85 [568281.598413] swiotlb_alloc_coherent+0xe0/0x150 I don't know what is causing it, I just notice that the X session is locked up. This seems to be a known kernel problem, with a fix in the queue for 4.16? * https://www.spinics.net/lists/dri-devel/msg164108.html * https://lkml.org/lkml/2018/1/16/106
4.16.3-300.fc28.x86_64 freezes with dmesg output [29977.647331] nouveau 0000:03:00.0: swiotlb buffer is full (sz: 2097152 bytes) [29977.647336] nouveau 0000:03:00.0: swiotlb: coherent allocation failed, size=2097152 [29977.647339] CPU: 0 PID: 2614 Comm: Xwayland Tainted: G OE 4.16.3-300.fc28.x86_64 #1 [29977.647341] Hardware name: NVIDIA MCP7A/MCP7A, BIOS 6.00 PG 04/22/2009 [29977.647342] Call Trace: [29977.647354] dump_stack+0x5c/0x85 [29977.647359] swiotlb_alloc_coherent+0x1c3/0x1e0 [29977.647375] ttm_dma_pool_get_pages+0x21a/0x620 [ttm] [29977.647382] ttm_dma_populate+0xdd/0x390 [ttm] [29977.647389] ttm_tt_bind+0x2e/0x60 [ttm] [29977.647395] ttm_bo_handle_move_mem+0x4cd/0x530 [ttm] [29977.647402] ttm_bo_validate+0x119/0x130 [ttm] [29977.647431] ? drm_mode_convert_umode+0x16/0x100 [drm] [29977.647437] ttm_bo_init_reserved+0x334/0x380 [ttm] [29977.647444] ? ttm_bo_init+0x62/0xd0 [ttm] [29977.647517] ? nouveau_bo_invalidate_caches+0x10/0x10 [nouveau] [29977.647563] ? nouveau_bo_new+0x401/0x580 [nouveau] [29977.647610] ? nouveau_bo_invalidate_caches+0x10/0x10 [nouveau] [29977.647656] ? nouveau_gem_new+0x120/0x120 [nouveau] [29977.647702] ? nouveau_gem_new+0x5d/0x120 [nouveau] [29977.647748] ? nouveau_gem_ioctl_new+0x53/0xe0 [nouveau] [29977.647763] ? drm_ioctl_kernel+0x5b/0xb0 [drm] [29977.647778] ? drm_ioctl+0x1c0/0x380 [drm] [29977.647824] ? nouveau_gem_new+0x120/0x120 [nouveau] [29977.647829] ? __handle_mm_fault+0xdb3/0x12f0 [29977.647875] ? nouveau_drm_ioctl+0x65/0xc0 [nouveau] [29977.647878] ? do_vfs_ioctl+0xa4/0x610 [29977.647881] ? SyS_ioctl+0x74/0x80 [29977.647884] ? do_syscall_64+0x74/0x180 [29977.647887] ? entry_SYSCALL_64_after_hwframe+0x3d/0xa2
Add a "me too". 4.16.5-200.fc27.x86_64 xorg-x11-drv-amdgpu-18.0.1-1.fc27.x86_64 F27 x64 fully updated, Amd RX 560 2GB, Cpu amd FX 8350, 32GB ram. dmesg shows many of those messages, but the machine and video runs fine. [295774.677184] amdgpu 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [295774.677186] amdgpu 0000:01:00.0: swiotlb: coherent allocation failed, size=2097152 [295774.677189] CPU: 4 PID: 5913 Comm: Xorg Not tainted 4.16.5-200.fc27.x86_64 #1 [295774.677190] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./990FX Extreme4, BIOS P2.70 06/05/2014 [295774.677190] Call Trace: [295774.677197] dump_stack+0x5c/0x85 [295774.677200] swiotlb_alloc_coherent+0x1be/0x1d0 [295774.677208] ttm_dma_pool_get_pages+0x235/0x620 [ttm] [295774.677213] ttm_dma_populate+0x25e/0x350 [ttm] [295774.677216] ttm_tt_bind+0x2c/0x60 [ttm] [295774.677220] ttm_bo_handle_move_mem+0x577/0x5b0 [ttm] [295774.677224] ttm_bo_validate+0x120/0x130 [ttm] [295774.677243] ? drm_pcie_get_speed_cap_mask+0x8e/0xe0 [drm] [295774.677247] ttm_bo_init_reserved+0x378/0x420 [ttm] [295774.677312] amdgpu_bo_do_create+0x22e/0x4b0 [amdgpu] [295774.677342] ? amdgpu_fill_buffer+0x300/0x300 [amdgpu] [295774.677370] amdgpu_bo_create+0x4f/0x2c0 [amdgpu] [295774.677374] ? finish_wait+0x80/0x80 [295774.677403] amdgpu_gem_object_create+0x7f/0x110 [amdgpu] [295774.677432] ? amdgpu_gem_object_close+0x210/0x210 [amdgpu] [295774.677461] amdgpu_gem_create_ioctl+0x1d4/0x260 [amdgpu] [295774.677490] ? amdgpu_gem_object_close+0x210/0x210 [amdgpu] [295774.677499] drm_ioctl_kernel+0x5b/0xb0 [drm] [295774.677507] drm_ioctl+0x2d5/0x370 [drm] [295774.677536] ? amdgpu_gem_object_close+0x210/0x210 [amdgpu] [295774.677538] ? vfs_writev+0xb9/0x110 [295774.677566] amdgpu_drm_ioctl+0x49/0x80 [amdgpu] [295774.677568] do_vfs_ioctl+0xa4/0x620 [295774.677569] SyS_ioctl+0x74/0x80 [295774.677571] do_syscall_64+0x74/0x180 [295774.677574] entry_SYSCALL_64_after_hwframe+0x3d/0xa2 [295774.677576] RIP: 0033:0x7f8c4f84c0f7 [295774.677580] RSP: 002b:00007ffd9d76c498 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 [295774.677582] RAX: ffffffffffffffda RBX: 000000000319c1a0 RCX: 00007f8c4f84c0f7 [295774.677583] RDX: 00007ffd9d76c4e0 RSI: 00000000c0206440 RDI: 000000000000000d [295774.677583] RBP: 00007ffd9d76c4e0 R08: 000000000319c1a0 R09: 0000000000000004 [295774.677584] R10: 0000000001d60010 R11: 0000000000000246 R12: 00000000c0206440 [295774.677585] R13: 000000000000000d R14: 00007ffd9d76c568 R15: 000000000388cde0
We probably shoud merge the bugreports https://bugzilla.redhat.com/show_bug.cgi?id=1571454 https://bugzilla.redhat.com/show_bug.cgi?id=1556797 https://bugzilla.redhat.com/show_bug.cgi?id=1546449