[Touch-packages] [Bug 1429756] Re: FTBFS: upstart test_job_process fails in majority of cases / Kernel returning unexpected EIO at near of file

2015-03-13 Thread Andy Whitcroft
Ok here are some test kernels with this commit reverted (the was a small conflict). Could you test these and confirm they fix the issue for you as well: http://people.canonical.com/~apw/lp1429756-vivid/ -- You received this bug notification because you are a member of Ubuntu Touch seeded

[Touch-packages] [Bug 1429756] Re: FTBFS: upstart test_job_process fails in majority of cases / Kernel returning unexpected EIO at near of file

2015-03-13 Thread Andy Whitcroft
For clarity this is triggered by the kernel sometimes exposing an EIO where we are expecting EOF. This tickles an issue with libnih which accumulates input in a read until EAGAIN loop, and dumps (arguably erroneously) its entire pending buffer when it sees an unexpected error. -- You received

[Touch-packages] [Bug 1429756] Re: FTBFS: upstart test_job_process fails in majority of cases / Kernel returning unexpected EIO at near of file

2015-03-13 Thread James Hunt
Running with a 3.19.0-8-generic kernel I see: - The Upstart test_job_process test fail within a few seconds when run in a loop looking for the errror. - My test program in #13 also finds the bug within ~30 seconds on my test system. Using the kernel in #11, I have run both tests simultaneously

[Touch-packages] [Bug 1429756] Re: FTBFS: upstart test_job_process fails in majority of cases / Kernel returning unexpected EIO at near of file

2015-03-13 Thread James Hunt
I've written a basic program that surfaces this bug. To recreate: $ cd /tmp $ tar xvf bug-1429756.tgz $ cd bug-1429756 $ make $ ./bug-1429756.sh loop 1 loop 2 ERROR: got failure scenario after 2 loops (got=16, count=17, log=/tmp/tmp.uwZHkHh0mc) $ ** Attachment added: bug-1429756.tgz