igt@fbdev@pan - fail - Failed assertion: ret == expected_ret, No space left on device, ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0

https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_939/fi-tgl-1115g4/igt@fbdev@pan.html

Starting subtest: pan
(fbdev:7376) CRITICAL: Test assertion failure function pan_test, file ../tests/fbdev.c:65:
(fbdev:7376) CRITICAL: Failed assertion: ret == expected_ret
(fbdev:7376) CRITICAL: Last errno: 28, No space left on device
(fbdev:7376) CRITICAL: ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0
Subtest pan failed.

added feature: display/Other platform: GEN9 platform: ICL platform: ILK platform: SNB platform: TGL labels

The CI Bug Log issue associated to this bug has been updated by Lakshmi Vudum.

New filters associated

• ILK SNB GEN9 ICL TGL: igt@fbdev@pan - fail - Failed assertion: ret == expected_ret, No space left on device, ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_938/fi-ilk-650/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_938/fi-skl-guc/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_938/fi-snb-2600/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_938/fi-tgl-1115g4/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_939/fi-icl-u2/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_939/fi-ilk-650/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_939/fi-skl-guc/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_939/fi-snb-2600/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_939/fi-tgl-1115g4/igt@fbdev@pan.html

@vsyrjala you have rb'd. Thoughts?

Sadly no 0x10 in drm.debug so couldn't immediately see what's going on. But I managed to reproduce locally by having two different sized displays plugged in.

It's a bug in drm_fb_helper. xres/yres will be based on the smallest display whereas fb/xres_virtual/yres_virtual will be based on the largest display. res < res_virtual is telling the pan test that it can actually pan, which is not true since the largest display already eats up the whole framebuffer. The fix would involve making drm_fb_helper not lie about the virtual resolution we can actually use.

mentioned in commit igt-gpu-tools@6d172a5c

added platform: ADL_P label

The CI Bug Log issue associated to this bug has been updated by Lakshmi Vudum.

New filters associated

• ADL_P: igt@fbdev@pan - fail - Failed assertion: ret == expected_ret, Last errno: 22, Invalid argument, ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0
  • https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10852/re-adlp-5/igt@fbdev@pan.html
  • https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10853/re-adlp-5/igt@fbdev@pan.html

@adrinael , do you know if Thomas working on fix based on Ville's input?

I don't.

@jani.saarinen I didn't work on it. Maybe just drop a patch? You have the bug, the HW and apparently a possible fix.

Right, only issue is that originally caused by your changes ;(.

My patches didn't cause the problem, they just exposed it.

right, sorry using wrong word.But no problem we have person taking this up.

A CI Bug Log filter associated to this bug has been updated by Lakshmi Vudum:

Description: ADL_P: igt@fbdev@pan - fail - Failed assertion: ret == expected_ret, Last errno: 22, Invalid argument, ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0

Equivalent query: runconfig_tag IS IN ["DRM-TIP"] AND (machine_name IS IN ["re-adlp-pub3", "re-adlp-pub2", "re-adlp-5", "re-adlp-11", "bat-adlp-4", "bat-adlp-6", "re-adlp-pub1"] OR machine_tag IS IN ["ADL-P"]) AND ((testsuite_name = "IGT" AND test_name IS IN ["igt@fbdev@pan"])) AND ((testsuite_name = "IGT" AND status_name IS IN ["fail"])) AND stderr ~= 'Failed assertion: ret == expected_ret\n.*Last errno: 22, Invalid argument\n.*ioctl\(FBIOPAN_DISPLAY\) returned ret=-1, expected 0'

New failures caught by the filter:

• https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10903/re-adlp-11/igt@fbdev@pan.html

A CI Bug Log filter associated to this bug has been updated by :

Description: ADL_P ADL_M: igt@fbdev@pan - fail - Failed assertion: ret == expected_ret, ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0

Equivalent query: runconfig_tag IS IN ["DRM-TIP"] AND (machine_name IS IN ["re-adlp-pub3", "re-adlp-pub2", "re-adlp-5", "re-adlm", "re-adlp-11", "bat-adlp-4", "bat-adlp-6", "re-adlp-pub1"] OR machine_tag IS IN ["ADL-M", "ADL-P"]) AND ((testsuite_name = "IGT" AND test_name IS IN ["igt@fbdev@pan"])) AND ((testsuite_name = "IGT" AND status_name IS IN ["fail"])) AND stderr ~= 'Failed assertion: ret == expected_ret\n.*\n.*ioctl\(FBIOPAN_DISPLAY\) returned ret=-1, expected 0'

New failures caught by the filter:

added platform: ADL_M label

On ADL_M, issue is observed with low resolution edp panel but testcase is passing with higher resolution 3840x2160 edp panel

Fail: ADL_M_DRM_10878_IGT_6280_igt_fbdev_pan_-_Details.html

Pass: ADL_M_DRM_10935_IGT_6293_igt_fbdev_pan_-_Details.html

added platform: RPL_S label

Issue is observed on RPL-S in manual resume run with DRM_10991 and IGT_6305.

• RPL_S_10991_igt_fbdev_pan_-_Details.html

The CI Bug Log issue associated to this bug has been updated by .

New filters associated

• RPL-S : igt@fbdev@pan - fail - Failed assertion: ret == expected_ret, No space left on device, ioctl(FBIOPAN_DISPLAY) returned ret=-1, expected 0