Bug 151541 - AMD hybrid graphics: [drm:evergreen_resume [radeon]] *ERROR* evergreen startup failed on resume
Summary: AMD hybrid graphics: [drm:evergreen_resume [radeon]] *ERROR* evergreen startu...
Status: NEW
Alias: None
Product: Drivers
Classification: Unclassified
Component: Video(DRI - non Intel) (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: drivers_video-dri
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-08-05 09:51 UTC by Dainius Masiliūnas
Modified: 2016-08-05 19:32 UTC (History)
1 user (show)

See Also:
Kernel Version: 4.7.0
Subsystem:
Regression: No
Bisected commit-id:

Attachments
dmesg (63.41 KB, text/plain)
2016-08-05 09:51 UTC, Dainius Masiliūnas 		Details
4.7.0-next-20160803 dmesg (61.36 KB, text/plain)
2016-08-05 19:32 UTC, Dainius Masiliūnas 		Details
Add an attachment (proposed patch, testcase, etc.)

Description Dainius Masiliūnas 2016-08-05 09:51:15 UTC 
Created attachment 227711 [details]
dmesg

I am using an Acer Aspire 7560G laptop with muxless AMD-AMD hybrid graphics, officially 6520G integrated and 6720G dedicated:

00:01.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] BeaverCreek [Radeon HD 6520G]
01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Whistler [Radeon HD 6630M/6650M/6750M/7670M/7690M] (rev ff)

This configuration causes kernel errors whenever the dedicated card gets touched (usually by either running lspci - which takes a while to run - or turning the monitor off and back on again). Visually that shows up as flickering and parts of the screen not refreshing properly.

If in the BIOS I turn off the dedicated card completely, I don't have this problem any more.

I tested it with kernel 4.7 and 4.1, both had the same results. I'm running openSUSE Leap 42.1.

Here's the relevant part of dmesg (full dmesg is attached):

[  668.895509] radeon 0000:01:00.0: Refused to change power state, currently in D3
[  668.971201] radeon 0000:01:00.0: Refused to change power state, currently in D3
[  668.987293] radeon 0000:01:00.0: Refused to change power state, currently in D3
[  668.988397] radeon 0000:01:00.0: GPU softreset: 0x00000BDD
[  668.988405] radeon 0000:01:00.0:   GRBM_STATUS               = 0xFFFFFFFF
[  668.988412] radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0xFFFFFFFF
[  668.988417] radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0xFFFFFFFF
[  668.988423] radeon 0000:01:00.0:   SRBM_STATUS               = 0xFFFFFFFF
[  668.988428] radeon 0000:01:00.0:   SRBM_STATUS2              = 0xFFFFFFFF
[  668.988434] radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0xFFFFFFFF
[  668.988439] radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0xFFFFFFFF
[  668.988445] radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0xFFFFFFFF
[  668.988450] radeon 0000:01:00.0:   R_008680_CP_STAT          = 0xFFFFFFFF
[  668.988456] radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0xFFFFFFFF
[  670.587759] radeon 0000:01:00.0: Wait for MC idle timedout !
[  670.587762] radeon 0000:01:00.0: GRBM_SOFT_RESET=0xFFFFFFFF
[  670.587817] radeon 0000:01:00.0: SRBM_SOFT_RESET=0xFFFFFFFF
[  670.588971] radeon 0000:01:00.0:   GRBM_STATUS               = 0xFFFFFFFF
[  670.588973] radeon 0000:01:00.0:   GRBM_STATUS_SE0           = 0xFFFFFFFF
[  670.588976] radeon 0000:01:00.0:   GRBM_STATUS_SE1           = 0xFFFFFFFF
[  670.588978] radeon 0000:01:00.0:   SRBM_STATUS               = 0xFFFFFFFF
[  670.588981] radeon 0000:01:00.0:   SRBM_STATUS2              = 0xFFFFFFFF
[  670.588983] radeon 0000:01:00.0:   R_008674_CP_STALLED_STAT1 = 0xFFFFFFFF
[  670.588986] radeon 0000:01:00.0:   R_008678_CP_STALLED_STAT2 = 0xFFFFFFFF
[  670.588988] radeon 0000:01:00.0:   R_00867C_CP_BUSY_STAT     = 0xFFFFFFFF
[  670.588991] radeon 0000:01:00.0:   R_008680_CP_STAT          = 0xFFFFFFFF
[  670.588993] radeon 0000:01:00.0:   R_00D034_DMA_STATUS_REG   = 0xFFFFFFFF
[  675.594474] [drm:atom_op_jump [radeon]] *ERROR* atombios stuck in loop for more than 5secs aborting
[  675.594493] [drm:atom_execute_table_locked [radeon]] *ERROR* atombios stuck executing CE66 (len 62, WS 0, PS 0) @ 0xCE82
[  675.594511] [drm:atom_execute_table_locked [radeon]] *ERROR* atombios stuck executing BB66 (len 1036, WS 4, PS 0) @ 0xBC63
[  675.594529] [drm:atom_execute_table_locked [radeon]] *ERROR* atombios stuck executing BAFC (len 76, WS 0, PS 8) @ 0xBB04
[  675.594653] [drm:radeon_pm_resume [radeon]] *ERROR* radeon: dpm resume failed
[  677.087926] radeon 0000:01:00.0: Wait for MC idle timedout !
[  677.274849] radeon 0000:01:00.0: Wait for MC idle timedout !
[  677.292534] [drm] PCIE GART of 1024M enabled (table at 0x0000000000162000).
[  677.292646] radeon 0000:01:00.0: WB enabled
[  677.292651] radeon 0000:01:00.0: fence driver on ring 0 use gpu addr 0x0000000040000c00 and cpu addr 0xffff8800c6e69c00
[  677.292654] radeon 0000:01:00.0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0xffff8800c6e69c0c
[  677.297315] radeon 0000:01:00.0: fence driver on ring 5 use gpu addr 0x0000000000072118 and cpu addr 0xffffc90001c32118
[  677.510244] [drm:r600_ring_test [radeon]] *ERROR* radeon: ring 0 test failed (scratch(0x8504)=0xFFFFFFFF)
[  677.510268] [drm:evergreen_resume [radeon]] *ERROR* evergreen startup failed on resume
[  677.510433] [drm:radeon_pm_resume [radeon]] *ERROR* radeon: dpm resume failed
[  682.513984] [drm:atom_op_jump [radeon]] *ERROR* atombios stuck in loop for more than 5secs aborting
[  682.514002] [drm:atom_execute_table_locked [radeon]] *ERROR* atombios stuck executing C510 (len 1136, WS 0, PS 0) @ 0xC53A
Comment 1 Dainius Masiliūnas 2016-08-05 10:05:02 UTC 
Using radeon.dpm=0 doesn't change anything, only the "*ERROR* radeon: dpm resume failed" line is not there in that case.
Comment 2 Alex Deucher 2016-08-05 13:18:36 UTC 
Does setting radeon.runpm=0 help?
Comment 3 Alex Deucher 2016-08-05 13:18:49 UTC 
On the kernel command line in grub.
Comment 4 Dainius Masiliūnas 2016-08-05 16:08:38 UTC 
Yes, it does. With that setting, there are no errors either, and lspci runs quickly.
Comment 5 Alex Deucher 2016-08-05 18:15:29 UTC 
Does it work any better with kernel 4.8?
Comment 6 Dainius Masiliūnas 2016-08-05 19:32:31 UTC 
Created attachment 227741 [details]
4.7.0-next-20160803 dmesg

No. I tried the linux-next tree and it gave me the same errors even during boot itself. dmesg is attached.
Note You need to log in before you can comment on or make changes to this bug.