[ubuntu-x] State of X for quantal

[Chase Douglas]

On Sat, Oct 6, 2012 at 6:27 PM, Bryce Harrington <bryce at canonical.com> wrote:
> On Sat, Oct 06, 2012 at 08:51:33AM -0700, Chase Douglas wrote:
>> To be honest, the stack corruption issue has always been a "potential"
>> issue. It's hard to say whether the issue was the real cause of bugs
>> or not, but it could be. I don't have any proof either way.
>>
>> The current X server in quantal should tell us if a message has been
>> printed unsafely. It will say that the message printing call needs to
>> be converted to a safe version. The message printing should always
>> make it to the log, I think, so we will definitely know if an unsafe
>> log occurred.
>>
>> Thus, if we see a bug and we do *not* see an unsafe log warning, then
>> the bug must be something else. Perhaps that will help narrow down the
>> possibilities.
>
> Well, bug #1059994 as a typical example.  It's hitting a FatalError:
>
> #12 0x00007fa94a96b3fd in FatalError (f=f at entry=0x7fa94a97e438 "%s: VT_ACTIVATE failed: %s\n") at ../../os/log.c:793
>
> However there are no "VT_ACTIVATE failed" messages in the attached
> Xorg.0.log files, and no messages about "unsafe log" in Xorg.0.log,
> dmesg, or lightdm logs.

The logs that are in the bug are not from the crash, unfortunately.
Both logs are clean. We would see lots of messages at the end if it
was.

> What is the text of the unsafe log message that we should scan for?  And
> does it get printed to Xorg.0.log or elsewhere?

This would be printed:

"""
Warning: attempting to log data in a signal unsafe manner while in
signal context.
Please update to check inSignalContext and/or use LogMessageVerbSigSafe().
The offending header and log message formats are...
"""

The two places in the server code that this is printed from are shown
well in this commit:

http://cgit.freedesktop.org/xorg/xserver/commit/os/log.c?id=1ebba43052d68d874148e63c9ae38489ddfc5ec1

The messages are only slightly different at the very end.

-- Chase