But why does this fail specifically on i686 and how do you know it is i686-specific?
It ran to completion on x86_64. And glibc default thread stack size should be 8MB, so it makes sense that they can't fit 1000 of those in the 4GB (or 3GB, I never remember how the split goes exactly) of address space available.
I did this
I ran the testsuite in a i686 chroot and test 3026 fails with the following:
I think making thread creation failures in the test non fatal might be enough, but I don't know if that's desired.
I expected the following
The testsuite should complete without failure.
curl/libcurl version
7.84.0
operating system
Linux $hostname 5.16.20_1 #1 SMP PREEMPT Thu Apr 14 17:43:53 UTC 2022 x86_64 GNU/Linux
(this was run in an i686 chroot)The text was updated successfully, but these errors were encountered: