Created attachment 736342 [details] build.log [...] make check-TESTS make[2]: Entering directory '/var/tmp/portage/app-arch/libarchive-3.5.2/work/libarchive-3.5.2-.ppc64' make[3]: Entering directory '/var/tmp/portage/app-arch/libarchive-3.5.2/work/libarchive-3.5.2-.ppc64' PASS: bsdcat_test FAIL: bsdcpio_test PASS: bsdtar_test PASS: libarchive_test ============================================================================ Testsuite summary for libarchive 3.5.2 ============================================================================ # TOTAL: 4 # PASS: 3 # SKIP: 0 # XFAIL: 0 # FAIL: 1 # XPASS: 0 # ERROR: 0
Created attachment 736345 [details] test-suite.log
Created attachment 736366 [details] bsdcpio_test test_basic.log
Created attachment 736369 [details] emerge --info
I'm afraid I'm going to need your help figuring this out. I can't reproduce it on ppc64 or ppc64le. I don't see anything relevant in configure output. And worst of all, this doesn't seem really possible given the code: https://github.com/libarchive/libarchive/blob/master/libarchive/archive_write_set_format_by_name.c#L43 This seems to be the function causing this error. Since the function just iterates over the table and strcmp()s it with predefined format list, the only way I see this could have happened is if either name or the list were damaged. For a start, could you try running: "/var/tmp/portage/app-arch/libarchive-3.5.2/work/libarchive-3.5.2-.ppc64/bsdcpio" -o -H bin and seeing if it outputs the same error? ( bsdcpio: No such format 'bin'. ) If it does, then either attach a debugger and trace archive_write_set_format_by_name() or add some printfs to verify what it's doing.
Created attachment 736648 [details] build.log (3.5.2 with tests passing) (In reply to Michał Górny from comment #4) > I'm afraid I'm going to need your help figuring this out. > > I can't reproduce it on ppc64 or ppc64le. I don't see anything relevant in > configure output. I was just about to check this out but can't reproduce it any longer... Tried several times but tests passed. Did not do much besides rebooting the machine so don't know what it was. I'll flag the bug as invalid and if I encounter it again throw gdb in immediately.
