Re: [Xquartz-dev] XQuartz 2.6.3_rc1
Jeremy Huddleston <jeremyhu@apple.com> wrote:
If you hit a crash in 2.6.3_rc1 or later, I want to know about it, but anything before that will just result in me asking "does it happen on 2.6.3?" ... ;)
Actually I am experiencing a crash in 2.6.3_rc1 that I did not have before. I built a window manager (e16) against the libs and headers of 2.6.3_rc1, and X11.bin was crashing whenever I exit the window manager. Normally, I exit the window manager and X11.bin quits cleanly. I reverted to 2.6.2 and rebuilt the same code, and no crash. Crash report is here: http://pastebin.com/2Wztpkrq -Dave
I believe that code has remained pretty much unchanged for quite some time... it's also not XQuartz specific. I don't see anything between 2.6.2 and 2.6.3 which should be related to this. Are you able to reproduce it 100% of the time? Would you mind helping debug this? 1) Install 2.6.3. 2) Edit /Applications/Utilities/X11.app/Contents/MacOS/X11 ... add the following to the second line (below the #! line): export DYLD_INSERT_LIBRARIES=/usr/lib/libgmalloc.dylib This will cause X11 to run with guardmalloc which will slow it down a bit but will cause it to crash at a more helpful location (hopefully the location where we had a buffer overrun or similar). 3) Run X11 and try triggering the bug (by quitting Enlightenment?). 4) Install the 1.10.1 server over 2.6.3*. $ install_x11_version_sl 1.10.1 5) Run X11 and try triggering the bug (by quitting Enlightenment?). 6) Install the 1.10.1.901 server over 2.6.3. $ install_x11_version_sl 1.10.1.901 7) Run X11 and try triggering the bug (by quitting Enlightenment?). 8) Install the 1.10.1.902 server over 2.6.3. $ install_x11_version_sl 1.10.1.902 9) Run X11 and try triggering the bug (by quitting Enlightenment?). What is the first version that exhibits the bug (note that the version in 2.6.3 is slightly newer than 1.10.1.902). Please attach the associated crash log as well. Could you provide me your built binaries (you can do 'sudo make install DESTDIR=/tmp/for_jeremy' then just tar up /tmp/for_jeremy)? Thanks, Jeremy *: You can use this bash function. More info is at http://xquartz.macosforge.org/trac/wiki/Releases install_x11_version_sl () { local version=$1 if [[ ! -f X11.bin-${version} ]] ; then if [[ ! -f X11.bin-${version}.bz2 ]] ; then curl -LO http://static.macosforge.org/xquartz/downloads/SL/X11.bin-${version}.bz2 || die "Failed to fetch version ${version}" fi bunzip2 X11.bin-${version}.bz2 || die "Failed to decompress version ${version}" fi chmod 755 X11.bin-${version} || die "Failed to chmod version ${version}" sudo cp X11.bin-${version} /Applications/Utilities/XQuartz.app/Contents/MacOS/X11.bin } On May 28, 2011, at 10:44 AM, Dave Ray wrote:
Jeremy Huddleston <jeremyhu@apple.com> wrote:
If you hit a crash in 2.6.3_rc1 or later, I want to know about it, but anything before that will just result in me asking "does it happen on 2.6.3?" ... ;)
Actually I am experiencing a crash in 2.6.3_rc1 that I did not have before. I built a window manager (e16) against the libs and headers of 2.6.3_rc1, and X11.bin was crashing whenever I exit the window manager. Normally, I exit the window manager and X11.bin quits cleanly. I reverted to 2.6.2 and rebuilt the same code, and no crash.
Crash report is here: http://pastebin.com/2Wztpkrq
-Dave
For #2, please edit: /Applications/Utilities/XQuartz.app/Contents/MacOS/X11 that is XQuartz.app, not X11.app ... sorry for the confusion. On May 28, 2011, at 12:34 PM, Jeremy Huddleston wrote:
I believe that code has remained pretty much unchanged for quite some time... it's also not XQuartz specific.
I don't see anything between 2.6.2 and 2.6.3 which should be related to this.
Are you able to reproduce it 100% of the time? Would you mind helping debug this?
1) Install 2.6.3.
2) Edit /Applications/Utilities/X11.app/Contents/MacOS/X11 ... add the following to the second line (below the #! line):
export DYLD_INSERT_LIBRARIES=/usr/lib/libgmalloc.dylib
This will cause X11 to run with guardmalloc which will slow it down a bit but will cause it to crash at a more helpful location (hopefully the location where we had a buffer overrun or similar).
3) Run X11 and try triggering the bug (by quitting Enlightenment?).
4) Install the 1.10.1 server over 2.6.3*. $ install_x11_version_sl 1.10.1
5) Run X11 and try triggering the bug (by quitting Enlightenment?).
6) Install the 1.10.1.901 server over 2.6.3. $ install_x11_version_sl 1.10.1.901
7) Run X11 and try triggering the bug (by quitting Enlightenment?).
8) Install the 1.10.1.902 server over 2.6.3. $ install_x11_version_sl 1.10.1.902
9) Run X11 and try triggering the bug (by quitting Enlightenment?).
What is the first version that exhibits the bug (note that the version in 2.6.3 is slightly newer than 1.10.1.902). Please attach the associated crash log as well.
Could you provide me your built binaries (you can do 'sudo make install DESTDIR=/tmp/for_jeremy' then just tar up /tmp/for_jeremy)?
Thanks, Jeremy
*: You can use this bash function. More info is at http://xquartz.macosforge.org/trac/wiki/Releases
install_x11_version_sl () { local version=$1 if [[ ! -f X11.bin-${version} ]] ; then if [[ ! -f X11.bin-${version}.bz2 ]] ; then curl -LO http://static.macosforge.org/xquartz/downloads/SL/X11.bin-${version}.bz2 || die "Failed to fetch version ${version}" fi bunzip2 X11.bin-${version}.bz2 || die "Failed to decompress version ${version}" fi chmod 755 X11.bin-${version} || die "Failed to chmod version ${version}" sudo cp X11.bin-${version} /Applications/Utilities/XQuartz.app/Contents/MacOS/X11.bin }
On May 28, 2011, at 10:44 AM, Dave Ray wrote:
Jeremy Huddleston <jeremyhu@apple.com> wrote:
If you hit a crash in 2.6.3_rc1 or later, I want to know about it, but anything before that will just result in me asking "does it happen on 2.6.3?" ... ;)
Actually I am experiencing a crash in 2.6.3_rc1 that I did not have before. I built a window manager (e16) against the libs and headers of 2.6.3_rc1, and X11.bin was crashing whenever I exit the window manager. Normally, I exit the window manager and X11.bin quits cleanly. I reverted to 2.6.2 and rebuilt the same code, and no crash.
Crash report is here: http://pastebin.com/2Wztpkrq
-Dave
_______________________________________________ Xquartz-dev mailing list Xquartz-dev@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/xquartz-dev
I've installed enlightenment from MacPorts, set it as my WM, and launched X11 with guardmalloc enabled 3 times. X11 quit cleanly each time that I quit enlightenment. Can you please file a bug report and attach a tarball of your ~/.e16 directory in addition to the information I've requested below? Please do this in the next day or so because I'm holding up the 1.10.2 release of xorg-server pending this investigation. If I don't hear back soon, I will release 1.10.2 as is since I cannot reproduce the crash. Thanks, Jeremy On May 28, 2011, at 12:52 PM, Jeremy Huddleston wrote:
For #2, please edit: /Applications/Utilities/XQuartz.app/Contents/MacOS/X11
that is XQuartz.app, not X11.app ... sorry for the confusion.
On May 28, 2011, at 12:34 PM, Jeremy Huddleston wrote:
I believe that code has remained pretty much unchanged for quite some time... it's also not XQuartz specific.
I don't see anything between 2.6.2 and 2.6.3 which should be related to this.
Are you able to reproduce it 100% of the time? Would you mind helping debug this?
1) Install 2.6.3.
2) Edit /Applications/Utilities/X11.app/Contents/MacOS/X11 ... add the following to the second line (below the #! line):
export DYLD_INSERT_LIBRARIES=/usr/lib/libgmalloc.dylib
This will cause X11 to run with guardmalloc which will slow it down a bit but will cause it to crash at a more helpful location (hopefully the location where we had a buffer overrun or similar).
3) Run X11 and try triggering the bug (by quitting Enlightenment?).
4) Install the 1.10.1 server over 2.6.3*. $ install_x11_version_sl 1.10.1
5) Run X11 and try triggering the bug (by quitting Enlightenment?).
6) Install the 1.10.1.901 server over 2.6.3. $ install_x11_version_sl 1.10.1.901
7) Run X11 and try triggering the bug (by quitting Enlightenment?).
8) Install the 1.10.1.902 server over 2.6.3. $ install_x11_version_sl 1.10.1.902
9) Run X11 and try triggering the bug (by quitting Enlightenment?).
What is the first version that exhibits the bug (note that the version in 2.6.3 is slightly newer than 1.10.1.902). Please attach the associated crash log as well.
Could you provide me your built binaries (you can do 'sudo make install DESTDIR=/tmp/for_jeremy' then just tar up /tmp/for_jeremy)?
Thanks, Jeremy
*: You can use this bash function. More info is at http://xquartz.macosforge.org/trac/wiki/Releases
install_x11_version_sl () { local version=$1 if [[ ! -f X11.bin-${version} ]] ; then if [[ ! -f X11.bin-${version}.bz2 ]] ; then curl -LO http://static.macosforge.org/xquartz/downloads/SL/X11.bin-${version}.bz2 || die "Failed to fetch version ${version}" fi bunzip2 X11.bin-${version}.bz2 || die "Failed to decompress version ${version}" fi chmod 755 X11.bin-${version} || die "Failed to chmod version ${version}" sudo cp X11.bin-${version} /Applications/Utilities/XQuartz.app/Contents/MacOS/X11.bin }
On May 28, 2011, at 10:44 AM, Dave Ray wrote:
Jeremy Huddleston <jeremyhu@apple.com> wrote:
If you hit a crash in 2.6.3_rc1 or later, I want to know about it, but anything before that will just result in me asking "does it happen on 2.6.3?" ... ;)
Actually I am experiencing a crash in 2.6.3_rc1 that I did not have before. I built a window manager (e16) against the libs and headers of 2.6.3_rc1, and X11.bin was crashing whenever I exit the window manager. Normally, I exit the window manager and X11.bin quits cleanly. I reverted to 2.6.2 and rebuilt the same code, and no crash.
Crash report is here: http://pastebin.com/2Wztpkrq
-Dave
_______________________________________________ Xquartz-dev mailing list Xquartz-dev@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/xquartz-dev
_______________________________________________ Xquartz-dev mailing list Xquartz-dev@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/xquartz-dev
I re-installed 2.6.3rc1 and again recompiled e16 against, and tried to repeat all my steps, following your debug instructions. I cannot repeat the crash that I was seeing before! It was previously happening every time. I rebooted and re-compiled e16 again just to be sure the new libs were being used, in case something had been cached. Still no crash. I'll leave 2.6.3rc1 installed so that I can report anything unexpected. It was definitely crashing before. Sorry I wasn't able to reproduce. FYI I'm installing e16-1.0.7 directly into /usr/local, thought that wouldn't make a difference. -Dave On May 28, 2011, at 12:34 PM, Jeremy Huddleston wrote:
I believe that code has remained pretty much unchanged for quite some time... it's also not XQuartz specific.
I don't see anything between 2.6.2 and 2.6.3 which should be related to this.
Are you able to reproduce it 100% of the time? Would you mind helping debug this?
1) Install 2.6.3.
2) Edit /Applications/Utilities/X11.app/Contents/MacOS/X11 ... add the following to the second line (below the #! line):
export DYLD_INSERT_LIBRARIES=/usr/lib/libgmalloc.dylib
This will cause X11 to run with guardmalloc which will slow it down a bit but will cause it to crash at a more helpful location (hopefully the location where we had a buffer overrun or similar).
3) Run X11 and try triggering the bug (by quitting Enlightenment?).
4) Install the 1.10.1 server over 2.6.3*. $ install_x11_version_sl 1.10.1
5) Run X11 and try triggering the bug (by quitting Enlightenment?).
6) Install the 1.10.1.901 server over 2.6.3. $ install_x11_version_sl 1.10.1.901
7) Run X11 and try triggering the bug (by quitting Enlightenment?).
8) Install the 1.10.1.902 server over 2.6.3. $ install_x11_version_sl 1.10.1.902
9) Run X11 and try triggering the bug (by quitting Enlightenment?).
What is the first version that exhibits the bug (note that the version in 2.6.3 is slightly newer than 1.10.1.902). Please attach the associated crash log as well.
Could you provide me your built binaries (you can do 'sudo make install DESTDIR=/tmp/for_jeremy' then just tar up /tmp/for_jeremy)?
Thanks, Jeremy
*: You can use this bash function. More info is at http://xquartz.macosforge.org/trac/wiki/Releases
install_x11_version_sl () { local version=$1 if [[ ! -f X11.bin-${version} ]] ; then if [[ ! -f X11.bin-${version}.bz2 ]] ; then curl -LO http://static.macosforge.org/xquartz/downloads/SL/X11.bin-${version}.bz2 || die "Failed to fetch version ${version}" fi bunzip2 X11.bin-${version}.bz2 || die "Failed to decompress version ${version}" fi chmod 755 X11.bin-${version} || die "Failed to chmod version ${version}" sudo cp X11.bin-${version} /Applications/Utilities/XQuartz.app/Contents/MacOS/X11.bin }
On May 28, 2011, at 10:44 AM, Dave Ray wrote:
Jeremy Huddleston <jeremyhu@apple.com> wrote:
If you hit a crash in 2.6.3_rc1 or later, I want to know about it, but anything before that will just result in me asking "does it happen on 2.6.3?" ... ;)
Actually I am experiencing a crash in 2.6.3_rc1 that I did not have before. I built a window manager (e16) against the libs and headers of 2.6.3_rc1, and X11.bin was crashing whenever I exit the window manager. Normally, I exit the window manager and X11.bin quits cleanly. I reverted to 2.6.2 and rebuilt the same code, and no crash.
Crash report is here: http://pastebin.com/2Wztpkrq
-Dave
On May 28, 2011, at 16:14, Dave Ray wrote:
I re-installed 2.6.3rc1 and again recompiled e16 against, and tried to repeat all my steps, following your debug instructions. I cannot repeat the crash that I was seeing before! It was previously happening every time.
I rebooted and re-compiled e16 again just to be sure the new libs were being used, in case something had been cached. Still no crash.
I'll leave 2.6.3rc1 installed so that I can report anything unexpected.
It was definitely crashing before. Sorry I wasn't able to reproduce.
Dang. That is bizarre. If you are able to reproduce it at some point, please try to capture it with guard malloc as that might provide a more useful backtrace. At this point, I'm going to assume that it isn't a regression since it's not reproducible and there's nothing obvious that could be causing it between the versions.
FYI I'm installing e16-1.0.7 directly into /usr/local, thought that wouldn't make a difference.
I was using e16-0.16.8 from MacPorts ... I'll look into updating that version later. Thanks, Jeremy
Based on your stack trace, my guess is that you're hitting this: http://lists.x.org/archives/xorg-devel/2011-May/022281.html Would you mind filing a bug report, so I can track the fix getting in to XQuartz? Thanks, Jeremy On May 28, 2011, at 16:57, Jeremy Huddleston wrote:
On May 28, 2011, at 16:14, Dave Ray wrote:
I re-installed 2.6.3rc1 and again recompiled e16 against, and tried to repeat all my steps, following your debug instructions. I cannot repeat the crash that I was seeing before! It was previously happening every time.
I rebooted and re-compiled e16 again just to be sure the new libs were being used, in case something had been cached. Still no crash.
I'll leave 2.6.3rc1 installed so that I can report anything unexpected.
It was definitely crashing before. Sorry I wasn't able to reproduce.
Dang. That is bizarre. If you are able to reproduce it at some point, please try to capture it with guard malloc as that might provide a more useful backtrace. At this point, I'm going to assume that it isn't a regression since it's not reproducible and there's nothing obvious that could be causing it between the versions.
FYI I'm installing e16-1.0.7 directly into /usr/local, thought that wouldn't make a difference.
I was using e16-0.16.8 from MacPorts ... I'll look into updating that version later.
Thanks, Jeremy
_______________________________________________ Xquartz-dev mailing list Xquartz-dev@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/xquartz-dev
participants (2)
-
Dave Ray
-
Jeremy Huddleston