staging-panda and tracking-panda do not come up with the right resolution

Bug #892837 reported by Zach Pfeffer
20
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Linaro Android
Fix Released
Low
vishal
linaro-landing-team-ti
Fix Released
Medium
Unassigned

Bug Description

On https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=101 hardware graphics fail to come up on 4430. Log attached.

I think this may be the race case Michael Edwards has been seeing.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :
Changed in linaro-android:
status: New → Confirmed
importance: Undecided → High
assignee: nobody → vishal (vishalbhoj)
Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Booted 3 times with the same effect.

It was last tested to work on:

https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=97

I replaced just the 101 kernel with the 97 kernel and graphics came up okay. Log of a boot of 101 with just the kenrel replaced attached.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Looking at the attached diff I think this error may point to the problem:

-omap_device: omapdss_dss.-1: new worst case deactivate latency 0: 2532958
 omap_device: omapdss_dss.-1: new worst case deactivate latency 0: 2563476
+omapdss CORE error: driver probe failed: -22
+generic_dpi_panel: probe of display1 failed with error -22
+omap_device: omapdss_dispc.-1: new worst case deactivate latency 0: 2593994

Changed in linaro-android:
importance: High → Critical
importance: Critical → High
Revision history for this message
Zach Pfeffer (pfefferz) wrote :
Revision history for this message
warmcat (andy-warmcat) wrote :

I just tested tracking on 4430 in case that's the issue, but it is coming up fine here in 1080p accelerated too.

The error from the logs about omapdss failing probe looks horrible alright and would lead to no video. But I can't reproduce it at all on 4460, 4430, tracking, tilt-android-3.1 basis.

Are you plugged in the HDMI socket?

Revision history for this message
warmcat (andy-warmcat) wrote :

Can't reproduce that log error with cable in DVI (no video is coming then though, it's a known issue on tracking) or no video cable in anywhere.

Revision history for this message
Jassi Brar (jassisinghbrar) wrote :

The error
     omapdss CORE error: driver probe failed: -22
     generic_dpi_panel: probe of display1 failed with error -22
tells only issue with DVI (no panel config available to the driver and even rudimentary
doesn't work, I checked)

From pinned-manifest.xml, the build 101 is using TILT's tilt-android-3.1
I tested and it works fine over HDMI port.

Revision history for this message
warmcat (andy-warmcat) wrote :

Thanks Jassi. DVI has never been workable with our SGX IIRC.

Can you confirm you are plugged in HDMI socket Zach?

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Yeah, I'm plugged into DVI.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

I mean I'm plugged into HDMI, and it doesn't work. When I plug into DVI I see no output on the screen.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Jassi, the logs were taken with me plugged into the HDMI port.

Revision history for this message
Alexander Sack (asac) wrote :

zach, can you submit a fresh HDMI plugged log?

Jessi/Andy: can you also submit a fresh HDMI plugged log since you seem to not see it?

seems jessi couldnt reproruce the 101 build, which seems to be broken for zach, so using that build would be good.

Revision history for this message
Alexander Sack (asac) wrote :

s/Jessi/Jassi/

-> sorry for name typo.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Just to clarify. All of the logs I've posted were from me plugged into the HDMI 1080P port including the first log. When I plug into the DVI port I see no output on the screen which is a known issue, so I never take logs of that build.

Revision history for this message
Alexander Sack (asac) wrote :

zach confirmed that logs in #1 and #2 are HDMI logs.

Revision history for this message
Alexander Sack (asac) wrote :

subscribed release team; eliminating this regression is very important if we want to continue to have an accelerated android LEB this month.

Changed in linaro-landing-team-ti:
status: New → Confirmed
Revision history for this message
Zach Pfeffer (pfefferz) wrote :

https://bugs.launchpad.net/linaro-android/+bug/892837/+attachment/2602986/+files/minicom.panda.txt is plugged into HDMI and is from https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=101.

https://bugs.launchpad.net/linaro-android/+bug/892837/+attachment/2603004/+files/minicom.panda.97.txt is plugged into HDMI and is from https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=97.

I think the issue may be that its not actually selecting the right port, so even though I'm plugged into HDMI the kernel thinks its DVI. WOuld you guys post your logs and the user space your using so we can see other differences, like u-boot or bootargs.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

I wonder if this has something to do with changes in EDID parsing. I'm plugged a different HDMI cable and now I see it come up in 800x600 DVI (even though I'm plugged into the HDMI port). Andy did you remove the EDID patch that we've had for a while?

Revision history for this message
Jassi Brar (jassisinghbrar) wrote :

> seems jessi couldnt reproruce the 101 build, which seems to be broken for zach, so using that build would be good.
http://android-build.linaro.org/builds/~linaro-android/staging-panda/101/pinned-manifest.xml tells uImage is built
from '39857a6923986aa', which is 'tilt-android-3.1' as of now.

Also I used http://android-build.linaro.org/builds/~linaro-android/staging-panda/101/kernel_config
Though I see one difference from my .config
@@ -24,7 +24,7 @@
 CONFIG_GENERIC_CALIBRATE_DELAY=y
 CONFIG_NEED_DMA_MAP_STATE=y
 CONFIG_VECTORS_BASE=0xffff0000
-# CONFIG_ARM_PATCH_PHYS_VIRT is not set
+CONFIG_ARM_PATCH_PHYS_VIRT=y
 CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
 CONFIG_HAVE_IRQ_WORK=y

I have attached my uImage and boot log. Btw I don't see any problem with original build-101 uImage either.

Revision history for this message
Jassi Brar (jassisinghbrar) wrote :

uImage here....

Revision history for this message
Jassi Brar (jassisinghbrar) wrote :

Ok, here is log of build-101 run. The HDMI comes up fine.
Though I do see the error message
    omapdss CORE error: driver probe failed: -22
    generic_dpi_panel: probe of display1 failed with error -22
but as I said, that is harmless to HDMI. And I do confirm DVI doesn't come up.

Anmar Oueja (anmar)
Changed in linaro-landing-team-ti:
importance: Undecided → Critical
Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Hmmm...

So in every recent build up and including to https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=97 I saw SGX come up on my Dynex DX-24L230A12...though I think it was always in DVI mode, even though I was connected to the HDMI port.

Now with https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=101 I don't see hardware accelerated graphics, though I do see output.

I'll try my HP Compaq LA2405wg. WOuld other people mind listed the monitors they're using?

Revision history for this message
Alexander Sack (asac) wrote :

thanks jassi.

@TILT: would be really great if you could use the tip images that we produce from your kernel branch to verify whether the latest kernel works or not. That would remove lots of hazzles and inefficiencies on both sides imo.

If we can make those builds more useful or provide other builds that would be more helpful for you guys let us know.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Jassi, rsalveti reminded me of passing omapdss.debug=1 on the boot lin, would you rerun 101 with that addition to the bootargs and I'll do the same. I think we're narrowing in on an EDID parsing issue.

Revision history for this message
Jassi Brar (jassisinghbrar) wrote :

Zach, I use ASUS-MS246H monitor and am currently staring at 0xBenchmark running smoothly at 1080p.
Is it exactly your way of testing too ?

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Jassi, I tried https://bugs.launchpad.net/linaro-android/+bug/892837/+attachment/2603940/+files/uImage and the image failed to boot.. I've attached the log.

All I did was program the build with:

https://android-build.linaro.org/builds/~linaro-android/staging-panda/#build=101

bzr branch lp:linaro-image-tools
./linaro-image-tools/linaro-android-media-create --mmc /dev/sdc --dev panda --system system.tar.bz2 --userdata userdata.tar.bz2 --boot boot.tar.bz2

then mount the first partition and copy your uImage over the one that got programmed.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

Log from 101 connected to my HP Compaq LA2405wg.

Revision history for this message
Jassi Brar (jassisinghbrar) wrote :

Linux version 2.6.38.6+ (jassi@Jassi-Linaro) (gcc version 4.5.1 (Sourcery G++ Lite 2010.09-50) ) #4 SMP PREEMPT Sat May 28 00:33:42 IST 2011
  Oops, that is wrong uImage. Here's the proper one.

Revision history for this message
warmcat (andy-warmcat) wrote :

Alexander you're right it makes sense to use recent image. However our situation with multiple kernels, themselves on someone else's head and being worked on constantly, multiple kinds of board under test means we need some stability in our testing somewhere. If we swap out the test rootfs every time we lose ability to infer where the problem came from easily by adding a new dimension. However it doesn't make sense to use ancient rootfs either.

warmcat (andy-warmcat)
Changed in linaro-landing-team-ti:
status: Confirmed → New
Revision history for this message
warmcat (andy-warmcat) wrote :

There should be no differences in result coming from plugging in a different hdmi cable, it suggests cable quality / connectivity issue.

There are now different hdmi implementations in 3.1 based and tracking based kernels. That's due to hdmi patches landing upstream and my performing a major adaptation to keep all that working. That you're reporting both 3.1 and tracking both act in the same broken way suggests what you need to get this symptom is specific to your side somehow and is independent of the kernel.

Revision history for this message
Zach Pfeffer (pfefferz) wrote :

I retested 101 and ran logcat I see the hardware GL libs getting loaded:

D/libEGL ( 1736): loaded /vendor/lib/egl/libGLESv1_CM_POWERVR_SGX540_120.so
D/libEGL ( 1736): loaded /vendor/lib/egl/libGLESv2_POWERVR_SGX540_120.so

and glmark2 runs and 0xbenchmark runs at speed. SO we're good.

So the only remaining issue is the build not coming up in 1080p. Are other people seeing this?

Changed in linaro-android:
importance: High → Low
milestone: none → 11.11
Revision history for this message
Zach Pfeffer (pfefferz) wrote :

I tested the 101 build.

Revision history for this message
vishal (vishalbhoj) wrote :

I tried the #101 build and and I observed that on first boot it booted with VGA resolution and on second boot it came up with 1080p resolution. Attached are the logs for VGA res case

Revision history for this message
vishal (vishalbhoj) wrote :

Attaching the logs when it came up at 1080p. I am trying it with Dell ST2320L monitor

Zach Pfeffer (pfefferz)
summary: - Accelerated graphics fails to come up on staging-panda
+ staging-panda and tracking-panda do not come up with the right
+ resolution
Zach Pfeffer (pfefferz)
Changed in linaro-android:
milestone: 11.11 → 11.12
Revision history for this message
warmcat (andy-warmcat) wrote :

Guys this isn't really "critical" now it's downgraded to issue about resolution sometimes.

On vishal log from #33, there's a WARN() but in tracking -- and we backported those changes to tilt-3.1 at request of TI -- that whole function has gone and the architecture has changed radically.

Tomorrow I will try Androidized versions of current tilt-3.1 and 3.2-rc3 tracking and push / post on this bug for a retry if they're workable here.

Changed in linaro-landing-team-ti:
importance: Critical → Medium
Zach Pfeffer (pfefferz)
Changed in linaro-android:
milestone: 11.12 → 12.01
Revision history for this message
warmcat (andy-warmcat) wrote :

Can't reproduce, always getting 1080p on ICS with tilt-android-tracking... sgx-1.8 stuff has new EDID processing

Changed in linaro-landing-team-ti:
status: New → Fix Released
Revision history for this message
vishal (vishalbhoj) wrote :
Changed in linaro-android:
status: Confirmed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.