Graphical Issues on Skylake with 14.2

atelszewski · 04-19-2016, 04:27 PM

Hi,

Quote:

Originally Posted by ppencho

Code:

Tue Apr 19 18:59:10 UTC 2016
x/xf86-video-intel-git_20160418_562ae1f-x86_64-1.txz: Upgraded.
       Bumped to latest git, dropped glamor fallback patch.

No issues here (using conf with sna and TearFree).

Try removing SNA, it should not be necessary any more.
On the other side, you can keep it, if you wish

--
Best regards,
Andrzej Telszewski

D1ver · 04-19-2016, 04:49 PM

Reporting in:
Removed /etc/X11/xorg.conf.d/20-intel.conf as we now default to SNA.

- April 18th upgrade broke x11 completely
- April 19th upgrade - Everything works again. No artifacts in the usual places. Libreoffice seems ok. Scrolling in firefox is smooth, no desktop performance issues. Seems good to me!

lazardo · 04-19-2016, 09:10 PM

20160418_562ae1f works with no 20-intel.conf and with SNA, TearFree, DRI 2

Cheers,

GazL · 04-20-2016, 11:25 AM

Not skylake, but it's working fine on a Broadwell 5157U with Iris 6100, though I do have to enable "TearFree" to stop video tearing. VSync on it's own doesn't seem enough to do the job.

lazardo · 04-21-2016, 03:25 AM

https://bugzilla.kernel.org/show_bug.cgi?id=109081 points at major skylake boot/ACPI/cstate fixes, so for 4.6.0-rc4 special boot params are no longer needed.

They have also been queued for upstream so maybe one more kernel bump cycle should be considered, if my opinion counts

https://bugzilla.kernel.org/show_bug.cgi?id=109081#c92

Cheers,

atelszewski · 04-21-2016, 09:38 AM

Hi,

I'm now running git_20160420_46caee8.

One commit before, might be something of interest to Skylake users.

--
Best regards,
Andrzej Telszewski

ReaperX7 · 04-21-2016, 09:47 AM

Quote:

Originally Posted by atelszewski

Hi,

It looks like SNA is the way to go.

You (Skylake users) might also try:

Code:

Section "Device"
  Identifier  "card0"
  Driver      "modesetting"
  Option      "AccelMethod" "glamor"
EndSection

but my guess is that this is what the Pat's patch for git_20160417_81029be was supposed to do and it failed for @lazardo.

--
Best regards,
Andrzej Telszewski

This seems to work fairly well for other cards too, not just Intel. I'm working on a broader aspect of this on another topic http://www.linuxquestions.org/questi...ng-4175577805/ but perhaps the issue is more with the Intel DDX driver than anything. I have a few log readouts from this against an Nvidia card if anyone is interested. Seems to work well as long as your card has support in Mesa.

atelszewski · 04-21-2016, 03:53 PM

Hi,

That's just an update on my git getter script. I hope I don't pollute the thread too much

I didn't like the intermediate copy of the source directory that had to be performed in the previous version of git-get-mksrctarball.sh. I did a bit of searching and found some interesting options supported by tar in Slackware.

Here they come (refer to man for explanation):

Code:

--transform=
--show-transformed-names
--exclude-vcs
--exclude-vcs-ignores

Using those options, it's possible to make the source tarball, without the intermediate copy.

The updated script follows:

Code:

$ cat git-get-mksrctarball.sh 
#!/bin/sh

REPODIR=/home/software/src/01-git
PACKDIR=/home/software/src

set -e

ADDRESS=$1
DESTDIR=$2
BRANCH=${3:-master}

if [ -z "${ADDRESS}" ] || [ -z "${DESTDIR}" ]; then
  echo "usage: $0 ADDRESS DESTDIR [branch]"
  exit 1
fi

UMASK=$(umask)
umask 022

cd "${REPODIR}"

if [ -e "${DESTDIR}" ]; then
  cd "${DESTDIR}"
    git checkout master
    git pull -f
  cd ..
else
  git clone "${ADDRESS}" "${DESTDIR}"
fi

cd "${DESTDIR}"
  git checkout "${BRANCH}"
  VERSION="git_$( git log -1 --format=%ad_%h --date=format:%Y%m%d )"
cd ..

umask ${UMASK}
tar -Jvcf "${DESTDIR}-${VERSION}.tar.xz" "${DESTDIR}" \
  --transform="s/${DESTDIR}/${DESTDIR}-${VERSION}/" \
  --show-transformed-names \
  --exclude-vcs \
  --exclude-vcs-ignores
md5sum "${DESTDIR}-${VERSION}.tar.xz" > "${DESTDIR}-${VERSION}.tar.xz.md5"

mv "${DESTDIR}-${VERSION}.tar.xz"     "${PACKDIR}"
mv "${DESTDIR}-${VERSION}.tar.xz.md5" "${PACKDIR}"

echo
echo "${DESTDIR} packaged as ${PACKDIR}/${DESTDIR}-${VERSION}.tar.xz"
echo
echo "PACKAGE=${DESTDIR}"
echo "VERSION=${VERSION}"
echo "TARBALL=${PACKDIR}/${DESTDIR}-${VERSION}.tar.xz"
echo

Hope this helps.

--
Best regards,
Andrzej Telszewski

lazardo · 04-21-2016, 04:01 PM

Need to setup a cron job to do a daily git pull/build on video-intel, apparently.

I'm starting to think video-intel s/b split in to architecture focused pieces.

atelszewski · 04-23-2016, 05:18 AM

Hi,

I had my first X crash since very long time:

Code:

$ grep EE /var/log/Xorg.0.log.old
[  2164.110] (EE) Backtrace:
[  2164.110] (EE) 0: /usr/libexec/Xorg (xorg_backtrace+0x41) [0x580ba1]
[  2164.110] (EE) 1: /usr/libexec/Xorg (0x400000+0x184ac9) [0x584ac9]
[  2164.110] (EE) 2: /lib64/libpthread.so.0 (0x7f942f803000+0x113b0) [0x7f942f8143b0]
[  2164.110] (EE) 3: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f942adc5000+0x42312) [0x7f942ae07312]
[  2164.110] (EE) 4: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f942adc5000+0x4ccba) [0x7f942ae11cba]
[  2164.111] (EE) 5: /usr/libexec/Xorg (0x400000+0x10d4e8) [0x50d4e8]
[  2164.111] (EE) 6: /usr/libexec/Xorg (0x400000+0x31b79) [0x431b79]
[  2164.111] (EE) 7: /usr/libexec/Xorg (0x400000+0x34d2f) [0x434d2f]
[  2164.111] (EE) 8: /usr/libexec/Xorg (0x400000+0x38c13) [0x438c13]
[  2164.111] (EE) 9: /lib64/libc.so.6 (__libc_start_main+0xf0) [0x7f942de497d0]
[  2164.111] (EE) 10: /usr/libexec/Xorg (_start+0x29) [0x4242a9]
[  2164.111] (EE) 
[  2164.111] (EE) Segmentation fault at address 0x47048
[  2164.111] (EE) 
[  2164.111] (EE) Caught signal 11 (Segmentation fault). Server aborting
[  2164.111] (EE) 
[  2164.111] (EE) 
[  2164.111] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[  2164.111] (EE) 
[  2164.245] (EE) Server terminated with error (1). Closing log file.

This happened with driver git_20160420_46caee8.

My xorg.conf:

Code:

$ cat /etc/X11/xorg.conf
Section "Device"
  Identifier  "card0"
  Driver      "intel"
  Option      "DRI"       "3"
  Option      "TearFree"  "true"
  Option      "VSync"     "false"
EndSection

Before the crash, I was playing with KWin, --replace(ing) it multiple times. Then terminal emulator did some weird scrolling. Then I fired up Scilab and X crashed before Scilab even loaded.

I cannot reproduce it for the moment.

--
Best regards,
Andrzej Telszewski

schmatzler · 04-24-2016, 04:00 AM

After the latest updates I now have tearing on Intel Ironlake (5th gen).

It's not excessive, but noticeable in Chromium and VLC. Get's a little better after enabling desktop effects in KDE and setting VSync to Automatic in the settings.

If anyone has an idea to get completely rid of it like before, please tell.

atelszewski · 04-24-2016, 05:14 AM

Hi,

Quote:

Originally Posted by schmatzler

After the latest updates I now have tearing on Intel Ironlake (5th gen).

It's not excessive, but noticeable in Chromium and VLC. Get's a little better after enabling desktop effects in KDE and setting VSync to Automatic in the settings.

If anyone has an idea to get completely rid of it like before, please tell.

What do you have in xorg.conf?
It looks like "TearFree" does not get always enabled and maybe enabling it could help?
Or maybe before you were running on UXA and now you are running SNA (which is default now)?

I am currently testing driver version git_20160423_bca4e0e.

--
Best regards,
Andrzej Telszewski

schmatzler · 04-24-2016, 05:27 AM

Quote:

Originally Posted by atelszewski

It looks like "TearFree" does not get always enabled and maybe enabling it could help?

Indeed. Why did they disable this option by default?

I didn't even need an xorg.conf before.

Now it works fine on Gen5 again. Tested it on this website.

archfan · 04-24-2016, 09:44 AM

I can't get DRI3 working.

atelszewski · 04-24-2016, 03:13 PM

Hi,

@archfan What do you have in xorg.conf?

Mine looks like that:

Code:

$ cat /etc/X11/xorg.conf
Section "Device"
  Identifier  "card0"
  Driver      "intel"
  Option      "DRI"       "3"
  Option      "TearFree"  "true"
  Option      "VSync"     "false"
EndSection

and that's what I have in the log:

Code:

$ grep DRI /var/log/Xorg.0.log
[     8.373] (**) intel(0): Option "DRI" "3"
[     8.378] (II) intel(0): [DRI2] Setup complete
[     8.378] (II) intel(0): [DRI2]   DRI driver: i965
[     8.378] (II) intel(0): [DRI2]   VDPAU driver: va_gl
[     8.378] (II) intel(0): direct rendering: DRI2 DRI3 enabled
[     8.410] (II) GLX: Initialized DRI2 GL provider for screen 0

I'm not sure if DRI3 is working for me or not, though.

--
Best regards,
Andrzej Telszewski