On Friday, October 21 2022, Steve Langasek wrote: > Hi Sergio,
Hello, > On Fri, Oct 21, 2022 at 06:47:09PM -0400, Sergio Durigan Junior wrote: >> * cryfs >> - FTBFS on ppc64el due to RLIMIT_MEMLOCK being too low. >> - After spending some time playing with setrlimit et al, it's clear to >> me that the problem cannot be easily fixed in the source code. I >> believe we will need to increase RLIMIT_MEMLOCK in the ppc64el >> builders. >> - Debian isn't affected by this problem. >> - There's also an s390x dep8 failure, which is also affecting Debian. >> I opened an upstream issue. >> - https://bugs.launchpad.net/ubuntu/+source/cryfs/+bug/1993578 > > Is there any reason not to instead change the package to skip these tests > when RLIMIT_MEMLOCK is too low (which seems detectable)? My initial reason would be that the package is in universe and in sync with Debian. On top of that, after reading #ubuntu-devel's log, it seems like we've encountered similar situations where RLIMIT_MEMLOCK got in the way. As far as I have investigated, these other cases were solved (correctly) by using setrlimit to properly extend RLIMIT_MEMLOCK when needed, but the packages in question had the necessary capabilities to do that. cryfs doesn't. Since cryfs in Debian doesn't suffer from this problem, and given that, based on local experiments, the RLIMIT_MEMLOCK value necessary to trigger this problem seems rather low, I think it's worth at least considering increasing it in our ppc64el builders. Please take the above with more than a grain of salt, since I obviously don't have access to our builders and didn't try to do a deep-dive into how their resource limits are configured. Thanks, -- Sergio GPG key ID: E92F D0B3 6B14 F1F4 D8E0 EB2F 106D A1C8 C3CB BF14 -- ubuntu-devel mailing list ubuntu-devel@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel