[PATCH xts] tet: Ignore SIGWINCH (terminal window size change)

Aaron Plattner aplattner at nvidia.com
Mon Aug 15 07:06:25 PDT 2011


On Sun, Aug 14, 2011 at 05:55:19PM -0700, Peter Hutterer wrote:
> On Fri, Aug 12, 2011 at 03:51:24PM -0700, Aaron Plattner wrote:
> > This signal is asserted when the controlling terminal changes size,
> > for example if you're running xts from an xterm and you resize the
> > window.  This causes test tests to fail:
> > 
> >   70||"VSW5TESTSUITE CASE FillPoly 3"
> >   10|34 /Xproto/pFillPoly 15:28:48|TC Start, scenario ref 71-0
> >   15|34 3.3-lite 3|TCM Start
> >   510|34|system 0: Abandoning testset: caught unexpected signal 28 (unknown signal)
> >   80|34 1 15:28:49|TC End, scenario ref 71-0
> > 
> > These signals can't affect the test environment, so just ignore them
> > all the time.
> > 
> > Signed-off-by: Aaron Plattner <aplattner at nvidia.com>
> > ---
> > Tracking down test intermittency only to find that it's caused by me
> > unconsciously rearranging my windows all the time gets really old really
> > quickly.
> 
> Acked-by: Peter Hutterer <peter.hutterer at who-t.net>

Thanks!

> do you actually run the test suite on your main box? I found it gives me a
> lot of false positives and negatives alike - the WM screws up the
> expectations.

No, but I often run it from an SSH session on a different computer, or
against Xvfb from a terminal.

One of the most annoying things with the old version of XTS was that it
would abort the whole test suite run rather than just an individual test,
so I'd connect to a system after letting it run for a long time, run
"screen -r" to check on its progress, and then kick myself when the
resulting SIGWINCH killed the whole thing.

-- Aaron


More information about the xorg-devel mailing list