stress-ng pty stressor is reporting an EINTR as an error

[Colin Ian King]

I received the following failure messages from stress-ng-pty:

stress-ng: 19:27:57.21 fail: [3323] stress-ng-pty: ioctl TCSETAF on slave pty failed, errno=4 (Interrupted system call)
 2019-08-07 19:27:57.239
stress-ng: 19:27:57.21 fail: [3322] stress-ng-pty: ioctl TCSETSW on slave pty failed, errno=4 (Interrupted system call)

I believe system calls may in principle be interrupted at any time; is this really a failure?

Using stress-ng-0.10

[Colin Ian King (colin-king)]

Do you mind explaining how you are running stress-ng to get this error?

[Colin Ian King (colin-king)]

Never mind, I can see the issue. Fix pushed to the repository.

https://kernel.ubuntu.com/git/cking/stress-ng.git/commit/?id=6698d6439ccb922e85adc0e7a36969fe051cb7e7

This will be land in the next stress-ng release.

[Gábor Lehel (glaebhoerl)]

Thanks!

[Launchpad Janitor (janitor)]

This bug was fixed in the package stress-ng - 0.10.02-1

---------------
stress-ng (0.10.02-1) unstable; urgency=medium

  * Makefile: bump version
  * stress-pthread: protect counter with spinlock not mutex lock
  * Manual: update date
  * stress-pthread: don't block forever on pthread_cond_wait() (LP: #1840867)
  * stress-pty: remove some white spaces in source
  * core-cache: ensure read buffer has trailing '\0'
  * core-cache: update comments
  * core-cache: use a fixed size buffer for reading in /sys info
  * stress-cpu: remove a debugging message
  * stress-pty: don't report a failure on EINTR (LP: #1840512)
  * stress-cpu: add an x86 cpuid instruction exerciser
  * stress-cpu: add a 64 bit division stressor
  * snapcraft: add automatic versioning
  * snapcraft: bump version
  * stress-pthread: don't tkill pthreads, can lead to futex blocking

 -- Colin King <email address hidden> Wed, 21 Aug 2019 10:00:01 +0100