Webkit Rendering Issues

Hello all,
             Although i hark from the gumstix land, I wanted to know
if anyone else has been having issues with Webkit rendering on the
BeagleBoard - since i can't seem to attach images here, I'm attaching
the link to my original post on this issue found here:
http://old.nabble.com/Webkit-Rendering-Issues-td27714291.html

If anyone else has been seeing this, or has a suggested fix, please do
help me out as I can't seem to move ahead past this.

Thank you kindly,

Jerry

I've not seen your first problem (wrong colors), but will check
it when I get my system running again (currently rebuilding...)

However, I've also been plagued by the second issue. I think it's
a VRAM refresh issue. I've found that just refreshing the screen
will make it go away.

n.b. I'm running Angstrom & Poky on my own 3530 based board.

I don't know it is the same issue, but you may want to verify that you are not
affected by VFP state corruption in signal handlers:
https://bugs.launchpad.net/ubuntu/+source/linux-fsl-imx51/+bug/507503

The latest kernel patches with a fix are available here unless I'm mistaken:
http://kernel.ubuntu.com/git?p=roc/ubuntu-lucid.git;a=shortlog;h=refs/heads/vfp_sig

Thanks Siarhei! The quirky Xorg symptoms defn. match that addressed by
the VFP patch - will try the patch out on my 2.6.32 kernel.

However, I think the WebKit color rendering issues might be something
else. I've done my own latest build of the xserver-xorg, and webkit gtk
+ and the color issue is defn. persistent there as well.

This should be more widespread..unless of course it is specific to my
setup (nothing peculiar there really..just an Overo system)

Jerry

Yup, those two patches courtesy of Imre Deak (found here at:
http://lists.arm.linux.org.uk/lurker/message/20100204.213712.acad59f5.en.html)
applied to my git tree defn fixed the Xorg noise issue - tested with a
few Gnome applications including the browser and no noise is to be
seen :slight_smile:

Had a slim hope that it would fix my color table issue, but it didn't
- guess it'll have to wait for another day

Jerry

It's good that at least some of your graphics issues got fixed.

This VFP patch still has some issues to be fixed as can be seen from the
discussion in the mailing list. But from the practical point of view, IMHO
even in its current shape it is better to have this fix applied than not.

I also strongly recommend to apply the following patch to u-boot:
http://git.denx.de/?p=u-boot/u-boot-arm.git;a=commit;h=42d97d08838d4e010d80c92ad1300c426320dd18
It is on the way into upstream u-boot, but the sooner everyone gets it, the
better. Of course that is unless somebody enjoys a rather small, but nonzero
probability of getting some nice CPU deadlocks so that your board hangs and
needs to be rebooted.

Has anyone been able to replicate the color table issue? I'm sure
others must be using Webkit-based browsers on the Beagle board as well

Jerry

2010/3/2 Jerry Johns <jerry.johns@gmail.com>

Has anyone been able to replicate the color table issue? I’m sure
others must be using Webkit-based browsers on the Beagle board as well

Sounds like you are hitting https://bugs.webkit.org/show_bug.cgi?id=20253

It does look like it, but we're building r53071 which should have the check for a proper flex version....

regards,

Koen

I just tried this on an up to date angstrom build ('unstable' in narcissus) and I get this:

http://scap.linuxtogo.org/files/dd479dc2b06bfab0ce122f21a9ea69b2.png

So it seems the colour issue has been resolved :slight_smile:

regards,

Koen

It looks like my Flex version on the gumstix branch of OE did not have
the .35 version, but rather the .31 version.
I noticed the latest OE and gumstix branches have flex updated to .35
- this is interesting, considering i just updated my Gumstix OE branch
a week ago, so the flex recipe for .35 must have come in last week

will test with the latest flex recipe and see if my webkit-gtk+ works
fine

Jerry