Bug 92792

At a light reading of the code, the edid override is done only at fill_modes (get connector), but by that time we have already setup the codec using eld from the edid loaded from the sink. It's a valid bug.

hi kaufmann,

Is this still a issue to you? if yes, please try the latest drm-tip branch(url = git://anongit.freedesktop.org/drm-tip). if this issue is already be fixed, please help to close this bug. thanks~

I am following all the time the change log (archiv) and saw no solution until now. 
So long 'intel_hdmi_set_edid' is calling 'drm_get_edid' it will always ignore the firmware EDID which is loaded during drm helper on boot cmd line.

intel_hdmi_set_edid needs to use the firmware file if available. drm-tip is also using drm_get_edid and i am using linux kernel 4.9.8. So yes the bug is still valid.

i have tried the patch commited in december: https://lists.freedesktop.org/archives/intel-gfx/2016-December/115248.html

but even with this patch the edid is once loaded from TV and not from the firmware.

Thanks for your info.
please retest latest drm-tips, with which GFX driver use SW API to sync ELD with audio driver on hot plug, then provide the kernel logs to us.

Please try [1] on top of current drm-tip. With that, the forced EDID loading gets moved at a lower level, and should work transparently for everything. The module parameter gets changed to drm.edid_firmware.

For future reference, please attach all files to bugzilla instead of external sites. (And please attach logs etc. individually as text/plain.) If the logs get dropped from the external sites, our job gets much harder. Thanks.

[1] http://patchwork.freedesktop.org/patch/msgid/1487344854-18777-5-git-send-email-jani.nikula@intel.com

The pathc indicated works well for me.
I am still doing some test, but forcing an edid to kernel command line I have now /proc/asound/card0/eld* popultaed although I have the monitor disconnected

I have build Linux Kernel 4.11 with patchset: https://patchwork.freedesktop.org/series/19764/
which includes given patch.

And when all devices are off resolution and eld from EDID are used. Looking good so far.

[1] http://patchwork.freedesktop.org/patch/msgid/1487344854-18777-5-git-send-email-jani.nikula@intel.com

According to the linked page, [1] still isn't upstream.

Please try current drm-tip, with:

commit ac6c35a4d8c77083525044a192cb1a8711381e94
Author: Jani Nikula <jani.nikula@intel.com>
Date:   Mon Sep 18 21:20:03 2017 +0300

    drm: add backwards compatibility support for drm_kms_helper.edid_firmware

commit 53fd40a90f3c0bdad86ec266ee5df833f54ace39
Author: Jani Nikula <jani.nikula@intel.com>
Date:   Tue Sep 12 11:19:26 2017 +0300

    drm: handle override and firmware EDID at drm_do_get_edid() level

Actually, based on the comments above and testing internally, closing as fixed. I'm afraid the change has such a wide impact on all drm drivers that it'll only hit Linus' tree at v4.15-rc1 timeframe. Can't backport to stable kernels.

Please file new bugs if there are related issues, and please reopen if the commits don't do what they claim. Thanks for the report.

For me it is okay to have the fix "drm: handle override and firmware EDID at drm_do_get_edid() Level" included in 4.15. Until then i will continue to compile the kernel with the patch by my own. 

I am now using the patch for a longer time and had never any issue regarding missing sound or resolution. Edid override fully works. Thanks for fixing this!

(In reply to kaufmann.mobil from comment #11)
> For me it is okay to have the fix "drm: handle override and firmware EDID at
> drm_do_get_edid() Level" included in 4.15. Until then i will continue to
> compile the kernel with the patch by my own. 
> 
> I am now using the patch for a longer time and had never any issue regarding
> missing sound or resolution. Edid override fully works. Thanks for fixing
> this!

Thanks for the report and testing!