KMS/radeon regression: failure to set modes on CEDAR & TAHITI; v3.11 vs 3.12-rc5/rc7

KMS/radeon regression: failure to set modes on CEDAR & TAHITI; v3.11 vs 3.12-rc5/rc7

[Robin H. Johnson]

[-- Attachment #1.1: Type: text/plain, Size: 2712 bytes --]

(Please CC, not sub'd to LKML; Was originally posted to
xorg-driver-ati@lists.x.org, dri-devel@lists.freedesktop.org, but I
think the dmesg attachments got it eaten as spam).

I apologize for not sending this sooner, but I wanted to be really sure that I
wasn't introducing the bug elsewhere, and I was lacking time to even narrow it
down this far.

TL;DR:
3.11 kernel does proper modeset for dual 1920x1200 outputs
3.12-rc5/rc7 kernel fails to do proper modeset, blank screen after radeon loaded

Symptoms:
- With implicit video= modes, after the radeon module loads, both monitors just go blank.
  Kernel message: radeon 0000:01:00.0: No connectors reported connected with modes
  X does not visible start, screen stays black.
- With explicit video= modes, after the radeon module loads, both monitors during the
  kernel boot display very garbled output; after X starts, they are in
  1024x768, with no higher modes listed.

Attached are four dmesg [1] runs, on v3.11 vs v3.12-rc7, with both explicit
video parameters, and implicit video parameters; as well as drm.debug=0x06 in
ALL cases; as well as two xorg logs (both from the implicit variants).

testcase 1 (implicit): modeset=1 fbcon=map:0 drm.debug=0x06
testcase 2 (explicit): video=HDMI-A-1:1920x1200e video=DVI-I-1:1920x1200e modeset=1 fbcon=map:0 drm.debug=0x06

I have produced this bug with three different cards:
CEDAR 0x1002:0x68F9 0x1043:0x03CA (used for all these bug reports)
CEDAR 0x1002:0x68F9 0x174B:0xE164
TAHITI 0x1002:0x6798 0x1458:0x2261 (7970, borrowed before)

This happened as far back as 3.12-rc5 for me, but I didn't debug more at the
time.

Monitors:
(Left) Dell U2410 via DVI
(Center) Dell U2413 via HDMI
(Right is another U2410, not connected at all)

Userspace:
xorg-server-1.14.3
mesa-9.2.1
xf86-video-ati-7.2.0

I think the error might be and off by one in the connector probing, based on
this:

3.11:
[drm:drm_helper_probe_single_connector_modes], [CONNECTOR:15:HDMI-A-1]
[drm:radeon_atombios_connected_scratch_regs], DFP1 connected
[drm:drm_helper_probe_single_connector_modes], [CONNECTOR:15:HDMI-A-1] probed modes :

3.12-rc7:
[drm:drm_helper_probe_single_connector_modes], [CONNECTOR:16:HDMI-A-1]
[drm:radeon_process_i2c_ch], hw_i2c error
[drm:radeon_atombios_connected_scratch_regs], DFP1 disconnected
[drm:drm_helper_probe_single_connector_modes], [CONNECTOR:16:HDMI-A-1] disconnected

[1] yes I know about the bonding error, that's a separate issue I need
to mail to LKML/netdev.

-- 
Robin Hugh Johnson
Gentoo Linux: Developer, Trustee & Infrastructure Lead
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85

[-- Attachment #1.2: 3.11.0-working-explicit.gz --]
[-- Type: application/octet-stream, Size: 26611 bytes --]

[-- Attachment #1.3: 3.11.0-working-implicit.gz --]
[-- Type: application/octet-stream, Size: 26378 bytes --]

[-- Attachment #1.4: 3.11.0-working-implicit-xorg.gz --]
[-- Type: application/octet-stream, Size: 9330 bytes --]

[-- Attachment #1.5: 3.12.0-rc7-broken-explicit.gz --]
[-- Type: application/octet-stream, Size: 27341 bytes --]

[-- Attachment #1.6: 3.12.0-rc7-broken-implicit.gz --]
[-- Type: application/octet-stream, Size: 25820 bytes --]

[-- Attachment #1.7: 3.12.0-rc7-broken-implicit-xorg.gz --]
[-- Type: application/octet-stream, Size: 7468 bytes --]

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 460 bytes --]

Re: KMS/radeon regression: failure to set modes on CEDAR & TAHITI; v3.11 vs 3.12-rc5/rc7

[Veaceslav Falico]

On Sun, Nov 3, 2013 at 7:48 PM, Robin H. Johnson <robbat2@gentoo.org> wrote:
> (Please CC, not sub'd to LKML; Was originally posted to
> xorg-driver-ati@lists.x.org, dri-devel@lists.freedesktop.org, but I
> think the dmesg attachments got it eaten as spam).
>
> I apologize for not sending this sooner, but I wanted to be really sure that I
> wasn't introducing the bug elsewhere, and I was lacking time to even narrow it
> down this far.
>
> TL;DR:
> 3.11 kernel does proper modeset for dual 1920x1200 outputs
> 3.12-rc5/rc7 kernel fails to do proper modeset, blank screen after radeon loaded
>
> Symptoms:
> - With implicit video= modes, after the radeon module loads, both monitors just go blank.
>   Kernel message: radeon 0000:01:00.0: No connectors reported connected with modes
>   X does not visible start, screen stays black.
> - With explicit video= modes, after the radeon module loads, both monitors during the
>   kernel boot display very garbled output; after X starts, they are in
>   1024x768, with no higher modes listed.
>
> Attached are four dmesg [1] runs, on v3.11 vs v3.12-rc7, with both explicit
> video parameters, and implicit video parameters; as well as drm.debug=0x06 in
> ALL cases; as well as two xorg logs (both from the implicit variants).
>
> testcase 1 (implicit): modeset=1 fbcon=map:0 drm.debug=0x06
> testcase 2 (explicit): video=HDMI-A-1:1920x1200e video=DVI-I-1:1920x1200e modeset=1 fbcon=map:0 drm.debug=0x06
>
> I have produced this bug with three different cards:
> CEDAR 0x1002:0x68F9 0x1043:0x03CA (used for all these bug reports)
> CEDAR 0x1002:0x68F9 0x174B:0xE164
> TAHITI 0x1002:0x6798 0x1458:0x2261 (7970, borrowed before)
>
> This happened as far back as 3.12-rc5 for me, but I didn't debug more at the
> time.
>
> Monitors:
> (Left) Dell U2410 via DVI
> (Center) Dell U2413 via HDMI
> (Right is another U2410, not connected at all)
>
> Userspace:
> xorg-server-1.14.3
> mesa-9.2.1
> xf86-video-ati-7.2.0
>
> I think the error might be and off by one in the connector probing, based on
> this:
>
> 3.11:
> [drm:drm_helper_probe_single_connector_modes], [CONNECTOR:15:HDMI-A-1]
> [drm:radeon_atombios_connected_scratch_regs], DFP1 connected
> [drm:drm_helper_probe_single_connector_modes], [CONNECTOR:15:HDMI-A-1] probed modes :
>
> 3.12-rc7:
> [drm:drm_helper_probe_single_connector_modes], [CONNECTOR:16:HDMI-A-1]
> [drm:radeon_process_i2c_ch], hw_i2c error
> [drm:radeon_atombios_connected_scratch_regs], DFP1 disconnected
> [drm:drm_helper_probe_single_connector_modes], [CONNECTOR:16:HDMI-A-1] disconnected
>
> [1] yes I know about the bonding error, that's a separate issue I need
> to mail to LKML/netdev.

Bonding issue should be fixed with
b32418705107265dfca5edfe2b547643e53a732e ("bonding: RCUify
bond_set_rx_mode()") in net-next.

>
> --
> Robin Hugh Johnson
> Gentoo Linux: Developer, Trustee & Infrastructure Lead
> E-Mail     : robbat2@gentoo.org
> GnuPG FP   : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85



-- 
Best regards,
Veaceslav Falico