Created attachment 569410 [details] glibc-2.28-r5-build.log.gz Looks like the majority are conform/*/linknamespace and ULPs-need-updating failures.
Actually, from spot checking some of the .out files it seems the math failures might be because the tests have started checking for more exceptional cases; alpha's implementations seem to fail to signal some exceptions. Sigh.
https://sourceware.org/glibc/wiki/Release/2.28#Alpha FAIL: conform/ISO/assert.h/linknamespace FAIL: conform/ISO/locale.h/linknamespace FAIL: conform/ISO/setjmp.h/linknamespace FAIL: conform/ISO/stdio.h/linknamespace ... ^ you can ignore these, this is bad interference between installed glibc <2.28 and tested glibc (bug in test suite) FAIL: crypt/badsalttest FAIL: elf/tst-prelink-cmp ^ known upstream FAIL: intl/tst-gettext ^ no idea FAIL: io/tst-fts-lfs ^ known upstream FAIL: math/test-double-finite-fma FAIL: math/test-double-finite-llrint FAIL: math/test-double-finite-llround ... ^ Known upstream ("Expected failures due to the default (non-)handling of inexact exceptions"). Likely nothing we can do, so let's ignore it. FAIL: resolv/tst-resolv-ai_idn FAIL: resolv/tst-resolv-ai_idn-latin1 ^ known upstream FAIL: stdlib/tst-makecontext ^ no idea
FAIL: intl/tst-gettext FAIL: stdlib/tst-makecontext ^ if you could add the .out file content of these two here, that would be great. On the whole, this doesnt look too bad to me compared to upstream report, so I'd go ahead. It's your call, but I would like to mask old glibc at some point because of sec bugs.
