keyctl tests failing with EDQUOT
Hi!
I've found another problem while doing pre-release testing.
The keyctl03 and sometimes keyctl05 fails with EDQUOT when executed right after keyctl02 which creates and revokes keys in the loop and also adjusts the quota temporarily. It fails for me on kernel 3.16 inside of a qemu virtual machine. Looks like the keys are not deleted after the keyctl02 test exits. Have anybody seen a failure like this one?
Signature · 2 lines
Cyril Hrubis [email protected]
Hi!
Quoted 9 lines · Cyril Hrubis
Hi!
I've found another problem while doing pre-release testing.The keyctl03 and sometimes keyctl05 fails with EDQUOT when executed right after keyctl02 which creates and revokes keys in the loop and also adjusts the quota temporarily. It fails for me on kernel 3.16 inside of a qemu virtual machine. Looks like the keys are not deleted after the keyctl02 test exits. Have anybody seen a failure like this one?
I tested on RHEL7.4GA(Kernel 3.10.0-693.el7.x86_64) in my virtual machine and physical machine, keyctl03 right after keyctl02 or keyctl05 right after keyctl02, both worked well.
On mainline kernel 3.16, keyctl02 and keyctl03 triggered the bug and crashed the system.
Best Regards,
Guangwen Feng
Hi!
Quoted 10 lines
Quoted 6 lines
The keyctl03 and sometimes keyctl05 fails with EDQUOT when executed right after keyctl02 which creates and revokes keys in the loop and also adjusts the quota temporarily. It fails for me on kernel 3.16 inside of a qemu virtual machine. Looks like the keys are not deleted after the keyctl02 test exits. Have anybody seen a failure like this one?
I tested on RHEL7.4GA(Kernel 3.10.0-693.el7.x86_64) in my virtual machine and physical machine, keyctl03 right after keyctl02 or keyctl05 right after keyctl02, both worked well.
I've managed to reproduce it on 3.12 as well. The problem is visible on SLES12 SP1 as well as Debian 8.8.
Quoted 2 lines
On mainline kernel 3.16, keyctl02 and keyctl03 triggered the bug and crashed the system.
You have to have the patch that fixes the kernel crash backported obviously, which happens mostly for distribution maintained kernels.
The question is if we care enough about these failures to do something about them before the release. I guess that we do not.
Signature · 2 lines
Cyril Hrubis [email protected]