02:01 <vmeson> tgamblin: builds fine without your patch and with your patch I see the same error from list thread, strace is indeed huge but this helps reduce the scope: rg -B 22222 "returned -11" /tmp/full.log | tee /tmp/return-11.log

02:03 <ak77> ok. still not there with producing wic disk image with squashfs rootfs. having IMAGE_FSTYPES = "squashfs wic" and IMAGE_TYPEDEP:wic = "squashfs", fails as IMAGE_NAME (or IMAGE_LINK_NAME) mentioned in .wks.in cannot be found. no squashfs file produced. if i use IMAGE_FSTYPES = "squashfs", squashfs gets built and file is present.

02:03 <ak77> seems like IMAGE_TYPEDEP is not respected.

02:05 <vmeson> tgamblin: but that's too much even for me. I've wrapped dnf-3 to narrow things down to a 6 MB, 35K line files but really I think its just python core dumping.

02:06 <tgamblin> vmeson: I think so too, just need to get it to produce a relevant stack trace

02:11 <vmeson> tgamblin: did you check for dnf updates, python3 issues?

02:11 <vmeson> tgamblin: that's enough for me for today. I'll sleep on it and may disable stripping tomorrow and see what I can figure out.

02:13 <vmeson> oh and as you likely know, dnf execve'd several (6?) times before it goes boom so we may have a bug that others haven't tripped over yet.

02:13 <tgamblin> vmeson: yeah, looked in libdnf, dnf, and python3. Didn't spot anything obvious

02:14 <tgamblin> vmeson: interesting, thanks for looking

02:14 <tgamblin> I'll take another look either tomorrow or on Monday

02:15 <tgamblin> the systemd-coredump service is handy...

04:03 <khem> tgamblin: check the log.do_compile of python3-native and see if -g is reflected