Bug 88522

Summary:	unreal effectscave demo locks up gpu
Product:	Mesa	Reporter:	almos <aaalmosss>
Component:	Drivers/Gallium/r600	Assignee:	Default DRI bug account <dri-devel>
Status:	RESOLVED MOVED	QA Contact:
Severity:	normal
Priority:	medium	CC:	marek.zdunek, vmerlet
Version:	git
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:
Attachments:	messages1 log1

Description almos 2015-01-17 01:12:50 UTC

I tried the Effects Cave demo from here https://wiki.unrealengine.com/Linux_Demos and it doesn't render anything, just locks up the GPU, the kernel calls a softreset, then the next frame locks up again and so on and so forth, until I kill it from a text console. This is printed to dmesg each time (nothing on the console):

[760734.139721] radeon 0000:01:00.0: ring 0 stalled for more than 10379msec
[760734.139727] radeon 0000:01:00.0: GPU lockup (waiting for 0x0000000006cb14fc last fence id 0x0000000006cb14fb on ring 0)
[760734.139734] radeon 0000:01:00.0: failed to get a new IB (-35)
[760734.139736] [drm:radeon_cs_ib_fill] *ERROR* Failed to get ib !
[760734.276346] radeon 0000:01:00.0: Saved 663 dwords of commands on ring 0.
[760734.276370] radeon 0000:01:00.0: GPU softreset: 0x00000009
[760734.276372] radeon 0000:01:00.0:   GRBM_STATUS               = 0xF0001828
[760734.276375] radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x80000003
[760734.276377] radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x80000003
[760734.276379] radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
[760734.276381] radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
[760734.276383] radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
[760734.276385] radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x40040000
[760734.276388] radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00048006
[760734.276390] radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x80268647
[760734.276392] radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
[760734.277882] radeon 0000:01:00.0: GRBM_SOFT_RESET=0x00007F6B
[760734.277936] radeon 0000:01:00.0: SRBM_SOFT_RESET=0x00000100
[760734.279096] radeon 0000:01:00.0:   GRBM_STATUS               = 0x00003828
[760734.279098] radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x00000007
[760734.279100] radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x00000007
[760734.279102] radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
[760734.279104] radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
[760734.279107] radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
[760734.279109] radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x00000000
[760734.279111] radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00000000
[760734.279113] radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x00000000
[760734.279116] radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
[760734.279147] radeon 0000:01:00.0: GPU reset succeeded, trying to resume
[760734.305227] [drm] enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0
[760734.306448] [drm] PCIE GART of 1024M enabled (table at 0x0000000000273000).
[760734.306543] radeon 0000:01:00.0: WB enabled
[760734.306546] radeon 0000:01:00.0: fence driver on ring 0 use gpu addr 0x0000000040000c00 and cpu addr 0xffff88041e6dbc00
[760734.306548] radeon 0000:01:00.0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0xffff88041e6dbc0c
[760734.307995] radeon 0000:01:00.0: fence driver on ring 5 use gpu addr 0x0000000000072118 and cpu addr 0xffffc900120b2118
[760734.324270] [drm] ring test on 0 succeeded in 2 usecs
[760734.324281] [drm] ring test on 3 succeeded in 7 usecs
[760734.501676] [drm] ring test on 5 succeeded in 2 usecs
[760734.501685] [drm] UVD initialized successfully.
[760744.513661] radeon 0000:01:00.0: ring 0 stalled for more than 10000msec
[760744.513666] radeon 0000:01:00.0: GPU lockup (waiting for 0x0000000006cb1511 last fence id 0x0000000006cb1502 on ring 0)
[760744.513672] [drm:r600_ib_test] *ERROR* radeon: fence wait failed (-35).
[760744.513677] [drm:radeon_ib_ring_tests] *ERROR* radeon: failed testing IB on GFX ring (-35).
[760744.513678] radeon 0000:01:00.0: ib ring test failed (-35).
[760744.522064] radeon 0000:01:00.0: GPU softreset: 0x00000009
[760744.522067] radeon 0000:01:00.0:   GRBM_STATUS               = 0xF7731828
[760744.522069] radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0xFC000003
[760744.522072] radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0xFC000003
[760744.522074] radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
[760744.522076] radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
[760744.522078] radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
[760744.522080] radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x400C0000
[760744.522083] radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00048006
[760744.522085] radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x80268647
[760744.522087] radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
[760744.522255] radeon 0000:01:00.0: GRBM_SOFT_RESET=0x00007F6B
[760744.522310] radeon 0000:01:00.0: SRBM_SOFT_RESET=0x00000100
[760744.523469] radeon 0000:01:00.0:   GRBM_STATUS               = 0x00003828
[760744.523471] radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0x00000007
[760744.523475] radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0x00000007
[760744.523477] radeon 0000:01:00.0:   SRBM_STATUS               = 0x200000C0
[760744.523479] radeon 0000:01:00.0:   SRBM_STATUS2              = 0x00000000
[760744.523481] radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0x00000000
[760744.523483] radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0x00000000
[760744.523486] radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0x00000000
[760744.523488] radeon 0000:01:00.0:   R_008680_CP_STAT          = 0x00000000
[760744.523490] radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
[760744.523525] radeon 0000:01:00.0: GPU reset succeeded, trying to resume
[760744.535154] [drm] enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0
[760744.538833] [drm] PCIE GART of 1024M enabled (table at 0x0000000000273000).
[760744.538926] radeon 0000:01:00.0: WB enabled
[760744.538929] radeon 0000:01:00.0: fence driver on ring 0 use gpu addr 0x0000000040000c00 and cpu addr 0xffff88041e6dbc00
[760744.538931] radeon 0000:01:00.0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0xffff88041e6dbc0c
[760744.540384] radeon 0000:01:00.0: fence driver on ring 5 use gpu addr 0x0000000000072118 and cpu addr 0xffffc900120b2118
[760744.556776] [drm] ring test on 0 succeeded in 2 usecs
[760744.556787] [drm] ring test on 3 succeeded in 7 usecs
[760744.734161] [drm] ring test on 5 succeeded in 2 usecs
[760744.734170] [drm] UVD initialized successfully.
[760744.734257] [drm] ib test on ring 0 succeeded in 0 usecs
[760744.734340] [drm] ib test on ring 3 succeeded in 0 usecs
[760744.885694] [drm:uvd_v1_0_ib_test] *ERROR* radeon: failed to get create msg (-22).
[760744.885701] [drm:radeon_ib_ring_tests] *ERROR* radeon: failed testing IB on ring 5 (-22).
[760744.885793] switching from power state:
[760744.885797]         ui class: none
[760744.885798]         internal class: boot 
[760744.885799]         caps: 
[760744.885801]         uvd    vclk: 0 dclk: 0
[760744.885802]                 power level 0    sclk: 10000 mclk: 30000 vddc: 950 vddci: 950
[760744.885803]                 power level 1    sclk: 10000 mclk: 30000 vddc: 950 vddci: 950
[760744.885804]                 power level 2    sclk: 10000 mclk: 30000 vddc: 950 vddci: 950
[760744.885805]         status: c b 
[760744.885806] switching to power state:
[760744.885807]         ui class: performance
[760744.885808]         internal class: none
[760744.885808]         caps: 
[760744.885809]         uvd    vclk: 0 dclk: 0
[760744.885810]                 power level 0    sclk: 10000 mclk: 15000 vddc: 950 vddci: 950
[760744.885811]                 power level 1    sclk: 60000 mclk: 100000 vddc: 1100 vddci: 1100
[760744.885812]                 power level 2    sclk: 86000 mclk: 110000 vddc: 1150 vddci: 1100
[760744.885813]         status: r 

This is with AMD Barts (HD6850), kernel is 3.17.7, mesa 10.3.2 and 10.5-dev.

I also tried the Realistic Rendering demo, and it works fine (with mesa 10.1 it was completely blue, but it has been fixed).

Comment 1 Marek Olšák 2015-02-07 13:49:37 UTC

Do any of these fix the bug?

R600_DEBUG=nosb
R600_DEBUG=nocpdma
https://bugs.freedesktop.org/attachment.cgi?id=113251
https://bugs.freedesktop.org/attachment.cgi?id=113246
https://bugs.freedesktop.org/attachment.cgi?id=109669

Comment 2 Marek Olšák 2015-02-07 13:50:20 UTC

Also:

R600_DEBUG=nohyperz

Comment 3 Marek Zdunek 2015-02-28 01:00:39 UTC

same here on HD6850, instant gpu lockup when run ./Effects

I tested:
./Effects -> gpu lockup
R600_DEBUG=nosb ./Effects -> gpu lockup
R600_DEBUG=nocpdma ./Effects -> gpu lockup
R600_DEBUG=nohyperz ./Effects -> gpu lockup

kernel.x86_64 3.18.3-201.fc21
mesa-libGL.x86_64 10.4.3-1.20150124.fc21

Comment 4 Marek Zdunek 2015-02-28 01:01:41 UTC

Created attachment 113880 [details]
messages1

messages log

Comment 5 Marek Zdunek 2015-02-28 01:03:33 UTC

Created attachment 113881 [details]
log1

program output log

Comment 6 almos 2016-01-21 13:43:05 UTC

I no longer have the HD6850 card, unfortunately. With my new card (R9 270x) and recent software stack (kernel 4.3.3, mesa 11.1.1, llvm 3.7.1) this demo runs perfectly.

Comment 7 GitLab Migration User 2019-09-18 19:18:57 UTC

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/540.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.