Bug 26779

Summary: make test fails on Snow Leopard
Product: cairo Reporter: Charles Doutriaux <doutriaux1>
Component: generalAssignee: Carl Worth <cworth>
Status: RESOLVED FIXED QA Contact: cairo-bugs mailing list <cairo-bugs>
Severity: blocker    
Priority: medium CC: ranma42, schlegel
Version: 1.9.6   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments: Logfile from make test
logs with 64bit gs

Description Charles Doutriaux 2010-02-26 11:21:05 UTC
Created attachment 33597 [details]
Logfile from make test

even though I try to build in pure 64Bit mode (but even w/o this) cairo builds but fails almost all the tests in the test suite. It seems to be linked to some 32bit vs 64bit issue.
Comment 1 Andrea Canciani 2010-02-27 01:40:41 UTC
(In reply to comment #0)
> Created an attachment (id=33597) [details]
> Logfile from make test
> 
> even though I try to build in pure 64Bit mode (but even w/o this) cairo builds
> but fails almost all the tests in the test suite. It seems to be linked to some
> 32bit vs 64bit issue.
It looks like your gs is linked to a cairo shared library with a different architecture from the one you're now using.
Does "make test TARGETS=image,quartz" complete?

NB: Quartz will probably fail a lot of tests when built 64-bits, see http://bugs.freedesktop.org/show_bug.cgi?id=25814 for the main reason.
Comment 2 Charles Doutriaux 2010-03-01 12:26:04 UTC
Created attachment 33667 [details]
logs with 64bit gs

recompiled gs to be 64bit built. New logs
Comment 3 Charles Doutriaux 2010-03-01 12:27:28 UTC
You were partly right I had an old 32bit gs. Recompiled all in 64bit.

It still faisl actually the test you asked me to run hangs.

I attached the new logs. 

Thanks for helping!

C.
Comment 4 Andrea Canciani 2010-03-05 15:07:14 UTC
(In reply to comment #3)
> You were partly right I had an old 32bit gs. Recompiled all in 64bit.
> 
> It still faisl actually the test you asked me to run hangs.
You should just wait for the test suite to complete. Some tests (dash-infinite-loop, for example) take a long time to complete/fail (expecially on old cairo versions).
> 
> I attached the new logs. 
It looks like you manually killed the testsuite. What happens if you let it run?
Andrea Canciani
Comment 5 Ryan Schlegel 2010-03-16 15:28:18 UTC
I'm having the same issue. I've built cairo 1.8.8 in 64 bit and it fails many tests or simply hangs. The test your discussing, dash-zero-length.quartz.argb32, I've had particular trouble with. I'm running it now with the intention of letting it run over night to see if it finishes. If it does, I will post the results.
Comment 6 Andrea Canciani 2010-03-21 01:29:23 UTC
(In reply to comment #5)
> I'm having the same issue. I've built cairo 1.8.8 in 64 bit and it fails many
> tests or simply hangs. The test your discussing,
> dash-zero-length.quartz.argb32, I've had particular trouble with. I'm running
> it now with the intention of letting it run over night to see if it finishes.
> If it does, I will post the results.
The fix for that test is not present in 1.8.8. Nonetheless the test should fail (with a "CRASHED" status) bacause of the timeout. Since you didn't post your results, I assume it didn't finish.
This is probably a bug in the test suite code.
Comment 7 Ryan Schlegel 2010-03-22 13:14:41 UTC
> The fix for that test is not present in 1.8.8. Nonetheless the test should fail
> (with a "CRASHED" status) bacause of the timeout. Since you didn't post your
> results, I assume it didn't finish.
> This is probably a bug in the test suite code.
> 

You are correct, the test never finished. 
Comment 8 Charles Doutriaux 2010-03-23 08:30:59 UTC
I doubt it's just a bug in the test suite. My code core dumps in cairo when i link it to the new build. Works great on every other platform. it COULD be a 64/32bit conflict with Mac OS X X11 libraries I have no idea if they've been compiled as 32 or 64bit...
Comment 9 Andrea Canciani 2010-03-23 08:44:03 UTC
(In reply to comment #8)
> I doubt it's just a bug in the test suite. My code core dumps in cairo when i
The test running forever instead of being terminated after the timeout is a bug in the test suite.
> link it to the new build. Works great on every other platform. it COULD be a
> 64/32bit conflict with Mac OS X X11 libraries I have no idea if they've been
> compiled as 32 or 64bit...
That should not really matter.
Quartz should "PASS" dash-infinite-loop on master (and 1.9.6) and "CRASH" on 1.8.8
There are other (unrelated) problems with Snow Leopard on x86_64 as I already pointed out.
Can you confirm that on 1.9.6 you see dash-infinite-loop run forever?
Comment 10 Andrea Canciani 2010-08-04 08:50:02 UTC
Is this still an issue?
Comment 11 Andrea Canciani 2010-10-30 07:59:19 UTC
Closing because quartz received a lot of fixes since the bug was reported and cairo 1.10 was released.
There are some failures but no crashes/hangs should happen with cairo 1.10 or master.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.