How to debug Xorg problems: Difference between revisions

Revision as of 09:44, 3 March 2015

Foreword

If you are experiencing a problem with Xorg, please see the common bugs document before filing a bug. Some easy configuration tweaks that fix a wide range of issues are listed there. If the problem you are seeing is not listed there or none of the workarounds seem to help, please consider filing a bug to help us make Fedora run better on your hardware.

Be prepared to include some information (logs) about your system as well. These should be complete (no snippets please), not in an archive, uncompressed, with MIME type set as text/plain.

Identifying your problem area

What driver am I using?

If you do not know already, try to find out what video driver you are using. Examine the output of journalctl -e _COMM=gdm-x-session (Fedora 22+), journalctl -e _COMM=Xorg.bin (Fedora 21) or journalctl -e _COMM=Xorg (Fedora 20 and earlier). Quite early on, you will see some lines like this:

(II) VESA(0): initializing int10
(II) VESA(0): Bad V_BIOS checksum

The word in capital letters after (II) is the name of the driver in use (so, in this case, the word is VESA, indicating the vesa driver is in use). Drivers are packaged with the name xorg-x11-drv-(name), so the vesa driver is in the package xorg-x11-drv-vesa.

If you are using a driver with the name nvidia (not nv or nouveau) or fglrx, you are using a proprietary third-party video driver (respectively, the proprietary drivers provided by NVIDIA and AMD/ATI). Please do not report any bugs in these drivers to Fedora, as we do not provide or support these drivers. Report bugs either to the place where you got these drivers, or to NVIDIA or AMD.

What area might the problem be in?

If the problem in question occurs when using 3D-accelerated applications - for instance, Blender, or 3D-accelerated games - the problem is a 3D acceleration issue, and you should include the information outlined in the appropriate section further down this page.

If you experience a "lockup", please distinguish what kind of lockup you are seeing. There are three major categories:
- Mouse moves, and cursor still changes when moving over window borders. This is usually a bug in the window manager's state machine where it doesn't let go of a grab in the X server.
- Mouse moves, but cursor does not change. This is usually the X server being stuck far away from the dispatch loop (position updates happen asynchronously, but glyph updates do not), either waiting on the kernel or in an infinite loop. The X server will probably print a message in the log about the event queue overflowing when this happens; this is _not_ the bug, it is merely the symptom.
- Mouse doesn't move. This is the X server being stuck, usually in the kernel (which will not show a backtrace in the X log, but will probably show something in dmesg) but occasionally in an internal deadlock with the cursor update code disabled (which will often show up as a backtrace in the X log).

Information to include in your report

All bug reports

In all cases, the following should be attached to your bug report:

All of the X server log messages: journalctl -e _COMM=gdm-x-session (Fedora 22+), journalctl -e _COMM=Xorg.bin (Fedora 21) or journalctl -e _COMM=Xorg (Fedora 20 and earlier). If you cannot use the system in the affected configuration, but you can use it some other way (e.g. with nomodeset or a different kernel), first boot the affected configuration and reproduce the bug, then boot the unaffected configuration and run journalctl -e _COMM=gdm-x-session (Fedora 22+), journalctl -b-1 -e _COMM=Xorg.bin (Fedora 21) or journalctl -b-1 -e _COMM=Xorg (Fedora 20 and earlier). The -b-1 asks for the output from the previous boot; you can adjust it as desired if you need logs from more than one boot ago, -b-2 etc.
The output of the command su -c 'lspci -nn'
If you use a xorg.conf, please include it in the bug report, otherwise, please specify in the bug report that it does not exist. Usually this would be located at /etc/X11/xorg.conf, but see the xorg.conf manpage - man xorg.conf - for other standard locations. Also include any files in the directory /etc/X11/xorg.conf.d
X log messages from a trial run where you move your xorg.conf aside and let Xorg autodetect your hardware (if you have such a file)
output of the dmesg command after adding the boot parameters drm.debug=14 log_buf_len=16M and rebooting, especially in case of crashes. If you cannot run dmesg in the affected configuration, but you can boot an unaffected configuration, boot the affected configuration with the parameters, then boot the unaffected configuration, run journalctl -b-1 and attach that output
Details of your monitor configuration: what monitors you have connected to what ports (e.g. VGA, DVI, HDMI...). Please note if you are using a display switcher

Rendering problems (unreadable text, corrupted display...)

As well as the information from the 'all bug reports' section, include the following information:

A screenshot showing the problem if at all possible.

Intel-specific

For intel display setup issues (outputs shaky or not lighting up):

collect and attach the video ROM for the chip: su -c 'dd if=/dev/mem of=/tmp/rom bs=64k skip=12 count=1'
collect and attach the opregion data: su -c 'mount -t debugfs debugfs /sys/kernel/debug' ; cp /sys/kernel/debug/dri/*/i915_opregion /tmp/opregion
Collect and attach a register snapshot from the broken configuration: su -c 'intel_reg_snapshot > /tmp/snapshot'

The command intel_reg_snapshot is provided by the intel-gpu-tools package. Note that you may need to ssh into the machine in order to collect this information. In some cases intel_reg_snapshot might fail to run, then please provide output of intel_reg_dumper (if possible, both before and after you see the issue).

For errors indicating a GPU hang:

collect and attach the GPU error state: su -c 'mount -t debugfs debugfs /sys/kernel/debug' ; cp /sys/kernel/debug/dri/0/i915_error_state /tmp

You may find the upstream intel driver bug filing instructions useful - much of the information from them is contained here, but there may be additional ideas there.

3D acceleration issues

As well as the information from the 'all bug reports' section, include the following information:

Output of the command glxinfo (if this is not installed, install the package glx-utils)
A screenshot, if possible (if the system has crashed but the display on screen is something other than just blank, take a picture with a digital camera and attach that)
Information as to whether or not other OpenGL applications are able to run without problems.
Console output from running the application with LIBGL_DEBUG=verbose. For instance, if the bug is in the application glxgears, you would run LIBGL_DEBUG=verbose glxgears from a console, reproduce the problem, and attach any output from the console

Creating a xorg.conf

If you need to make manual changes to X configuration, you will need to create a xorg.conf file if it doesn't already exist.

Input devices

Debugging of input devices is covered by a special wiki page.

Stack traces

You will need a stack trace if your X server crashes. You should be able to submit one to bugzilla by running Automatic Bug Reporting Tool (gnome-abrt) and selecting the relevant crash from System crashes (if abrtd was running and noticed the crash). If not, see the documentation on the upstream wiki.

@@ Line 1: / Line 1: @@
-==General bug reports==
+'''Foreword'''
-If you are experiencing a problem with Xorg, the following would be useful to attach to your bug report (individually, uncompressed):
+If you are experiencing a problem with Xorg, please see the [[Bugs/Common|common bugs]] document before filing a bug. Some easy configuration tweaks that fix a wide range of issues are listed there. If the problem you are seeing is not listed there or none of the workarounds seem to help, please consider filing a bug to help us make Fedora run better on your hardware.
-* {{filename|/etc/X11/xorg.conf}} (or, if this file does not exist, specify in the bug report that it does not)
+Be prepared to include some information (logs) about your system as well. These should be complete (no snippets please), not in an archive, uncompressed, with MIME type set as text/plain.
-* X server log file(s) {{filename|/var/log/Xorg.*.log}}
-* {{filename|/var/log/Xorg.0.log}} from a trial run where you move /etc/X11/xorg.conf aside and let Xorg autodetect your hardware (if you have such a file).
-==Stack traces==
+== Identifying your problem area ==
-Did you know, that all of the modular X packages now come with full debuginfo packages, including the X server and driver modules?  The X server now uses dlopen, so it can be debugged using the stock gdb that comes with Fedora.
+=== What driver am I using? ===
-In order to debug the running X server, install the debuginfo packages for it, and any drivers you are loading.  You then need to have 2 computers[1]  networked together, or using a serial port or some other remote access method.
+If you do not know already, try to find out what video driver you are using. Examine the output of {{command|<nowiki>journalctl -e _COMM=gdm-x-session</nowiki>}} (Fedora 22+), {{command|<nowiki>journalctl -e _COMM=Xorg.bin</nowiki>}} (Fedora 21) or {{command|<nowiki>journalctl -e _COMM=Xorg</nowiki>}} (Fedora 20 and earlier). Quite early on, you will see some lines like this:
+<pre>
+(II) VESA(0): initializing int10
+(II) VESA(0): Bad V_BIOS checksum
+</pre>
+The word in capital letters after (II) is the name of the driver in use (so, in this case, the word is VESA, indicating the ''vesa'' driver is in use). Drivers are packaged with the name ''xorg-x11-drv-(name)'', so the ''vesa'' driver is in the package {{package|xorg-x11-drv-vesa}}.
-Edit the xorg.conf file, and enable the option:
+If you are using a driver with the name ''nvidia'' (not ''nv'' or ''nouveau'') or ''fglrx'', you are using a proprietary third-party video driver (respectively, the proprietary drivers provided by NVIDIA and AMD/ATI). Please do not report any bugs in these drivers to Fedora, as we do not provide or support these drivers. Report bugs either to the place where you got these drivers, or to NVIDIA or AMD.
-Option "NoTrapSignals"
+=== What area might the problem be in? ===
-Then, enable coredumps on the system, if you have not done so already, by using "ulimit -c unlimited" in a script dropped into /etc/profile.d/enable-coredumps.sh  or whatever.  Be sure to run the script or log out and back in.
+* If the problem in question occurs when using 3D-accelerated applications - for instance, Blender, or 3D-accelerated games - the problem is a ''3D acceleration issue'', and you should include the information outlined in the [[Xorg/Debugging#3DInfo|appropriate section]] further down this page.
-As the X server is a setuid root process, and the kernel will not allow coredumps of SUID processes by default, you have 2 choices.
+* If you experience a "lockup", please distinguish what kind of lockup you are seeing.  There are three major categories:
+** Mouse moves, and cursor still changes when moving over window borders.  This is usually a bug in the window manager's state machine where it doesn't let go of a grab in the X server.
+** Mouse moves, but cursor does not change.  This is usually the X server being stuck far away from the dispatch loop (position updates happen asynchronously, but glyph updates do not), either waiting on the kernel or in an infinite loop.  The X server will probably print a message in the log about the event queue overflowing when this happens; this is _not_ the bug, it is merely the symptom.
+** Mouse doesn't move.  This is the X server being stuck, usually in the kernel (which will not show a backtrace in the X log, but will probably show something in dmesg) but occasionally in an internal deadlock with the cursor update code disabled (which will often show up as a backtrace in the X log).
-You can tweak the kernel to permit SUID coredumps by doing:
+== Information to include in your report ==
-echo 1 > /proc/sys/kernel/core_setuid_ok
+{{Anchor|AllInfo}}
+=== All bug reports ===
-or you can run the X server as root.  If you enable the kernel to allow SUID coredumps as per above, just be sure to disable it afterwards by echoing '0' to the same file, to avoid potential security issues.
+In all cases, the following should be attached to your bug report:
-Now you are ready to break things and debug them.  I prefer to start the X server with "startx" all the time, but particularly when debugging.  I recommend doing the same, unless the problem you are debugging only occurs when using gdm/kdm/xdm, as it tends to be much simpler to debug with startx.  YMMV however.
+* All of the X server log messages: {{command|<nowiki>journalctl -e _COMM=gdm-x-session</nowiki>}} (Fedora 22+), {{command|<nowiki>journalctl -e _COMM=Xorg.bin</nowiki>}} (Fedora 21) or {{command|<nowiki>journalctl -e _COMM=Xorg</nowiki>}} (Fedora 20 and earlier). If you cannot use the system in the affected configuration, but you can use it some other way (e.g. with {{code|nomodeset}} or a different kernel), first boot the affected configuration and reproduce the bug, then boot the unaffected configuration and run {{command|<nowiki>journalctl -e _COMM=gdm-x-session</nowiki>}} (Fedora 22+), {{command|<nowiki>journalctl -b-1 -e _COMM=Xorg.bin</nowiki>}} (Fedora 21) or {{command|<nowiki>journalctl -b-1 -e _COMM=Xorg</nowiki>}} (Fedora 20 and earlier). The {{code|-b-1}} asks for the output from the previous boot; you can adjust it as desired if you need logs from more than one boot ago, {{code|-b-2}} etc.
+* The output of the command {{command|su -c 'lspci -nn'}}
+* If you use a {{filename|xorg.conf}}, please include it in the bug report, otherwise, please specify in the bug report that it does not exist. Usually this would be located at {{filename|/etc/X11/xorg.conf}}, but see the xorg.conf manpage - {{command|man xorg.conf}} - for other standard locations. Also include any files in the directory {{filename|/etc/X11/xorg.conf.d}}
+* X log messages from a trial run where you move your {{filename|xorg.conf}} aside and let Xorg autodetect your hardware (if you have such a file)
+* output of the {{command|dmesg}} command after adding the boot parameters {{code|<nowiki>drm.debug=14 log_buf_len=16M</nowiki>}} and rebooting, especially in case of crashes. If you cannot run {{command|dmesg}} in the affected configuration, but you can boot an unaffected configuration, boot the affected configuration with the parameters, then boot the unaffected configuration, run {{command|journalctl -b-1}} and attach that output
+* Details of your monitor configuration: what monitors you have connected to what ports (e.g. VGA, DVI, HDMI...). Please note if you are using a display switcher
-If the server crashes, you can gdb the coredump now.  Assuming you installed debuginfo packages for the server (make sure the debuginfo package is the same version-release as the server that is installed), you should get useful backtraces, etc.
+=== Rendering problems (unreadable text, corrupted display...) ===
-If you want to debug the running server, single step it, etc., you need the second computer[1]  mentioned above.  Do all the same steps as above, but this time, ssh in from the remote computer (or use minicom on serial port or whatever), and get the PID
+As well as the information from [[Xorg/Debugging#AllInfo|the 'all bug reports' section]], include the following information:
-of the X server with 'pidof X'.  Then attach gdb to the X server:
-gdb --pid=$(pidof X)
+* A screenshot showing the problem if at all possible.
-Now go to town!  Find the bug, install the X server/driver sources, fix the bug, generate a patch with gendiff, and attach it to a bug report.  It's really that simple!
+==== Intel-specific ====
-[1]  The reason that 2 computers are required, is that if you start the X server, open an xterm and run gdb on the running X server, gdb will stop the X server, and try to print to stdout.  Since stdout is your xterm, and xterm is displaying to the X server which is now no longer responding because gdb stopped it, you are screwed. It's kindof like trying to stand on a carpet, and then pull it out from under yourself.  You can't really do it.  This is why you need[2]  2 computers.
+For intel display setup issues (outputs shaky or not lighting up):
+* collect and attach the video ROM for the chip: {{command|<nowiki>su -c 'dd if=/dev/mem of=/tmp/rom bs=64k skip=12 count=1'</nowiki>}}
+* collect and attach the opregion data: {{command|su -c 'mount -t debugfs debugfs /sys/kernel/debug' ; cp /sys/kernel/debug/dri/*/i915_opregion /tmp/opregion}}
+* Collect and attach a register snapshot from the broken configuration: {{command|su -c 'intel_reg_snapshot > /tmp/snapshot'}}
-[2]  Ok, it is technically possible to do it with one computer too, but it is complex enough to not really bother trying to do, so don't even waste time thinking of it.  If you don't have 2 computers, go dumpster diving at 3am in the garbage dumpster of a computer store.  They throw out good stuff I'm told, and you can have a "X server debugging terminal" in no time, like everyone else on the block.  Alternatively, surf ebay.com for a bargain, or hang out at a computer recycling shop.
+The command {{command|intel_reg_snapshot}} is provided by the {{package|intel-gpu-tools}} package.  Note that you may need to {{command|ssh}} into the machine in order to collect this information. In some cases {{command|intel_reg_snapshot}} might fail to run, then please provide output of {{filename|intel_reg_dumper}} (if possible, both before and after you see the issue).
+For errors indicating a GPU hang:
+* collect and attach the GPU error state: {{command|su -c 'mount -t debugfs debugfs /sys/kernel/debug' ; cp /sys/kernel/debug/dri/0/i915_error_state /tmp}}
+You may find [https://01.org/linuxgraphics/documentation/how-report-bugs-0 the upstream intel driver bug filing instructions] useful - much of the information from them is contained here, but there may be additional ideas there.
+{{Anchor|3DInfo}}
+=== 3D acceleration issues ===
+As well as the information from [[Xorg/Debugging#AllInfo|the 'all bug reports' section]], include the following information:
+* Output of the command {{command|glxinfo}} (if this is not installed, install the package {{package|glx-utils}})
+* A screenshot, if possible (if the system has crashed but the display on screen is something other than just blank, take a picture with a digital camera and attach that)
+* Information as to whether or not other OpenGL applications are able to run without problems.
+* Console output from running the application with LIBGL_DEBUG=verbose. For instance, if the bug is in the application ''glxgears'', you would run {{command|LIBGL_DEBUG<nowiki>=</nowiki>verbose glxgears}} from a console, reproduce the problem, and attach any output from the console
+== Creating a xorg.conf ==
+If you need to make manual changes to X configuration, you will need to [[How_to_create_xorg.conf|create a xorg.conf]] file if it doesn't already exist.
+== Input devices ==
+Debugging of input devices is covered by [[Xorg/Input_Triage_Algorithm|a special wiki page]].
+== Stack traces ==
+You will need a stack trace if your X server crashes. You should be able to submit one to bugzilla by running '''Automatic Bug Reporting Tool''' ({{command|gnome-abrt}}) and selecting the relevant crash from '''System''' crashes (if abrtd was running and noticed the crash). If not, see the [http://wiki.x.org/wiki/Development/Documentation/ServerDebugging documentation] on the upstream wiki.
+[[Category:Debugging|X]] [[Category:How to]]

Search