Comment # 4 on bug 1223152 from Martin Jambor

Strange developments happened yesterday.

First and foremost, the I noticed the issue reported here went away
with revision 24 of openssl-3
(https://build.opensuse.org/package/rdiff/openSUSE:Factory/openssl-3?linkrev=base&rev=24).
Both locally and in the Gcc7 staging project
(https://build.opensuse.org/package/live_build_log/openSUSE:Factory:Staging:Gcc7/openssl-3/standard/x86_64
)

We have nevertheless looked at the issue a bit deeper and found out
that ICF is not doing anything wrong on its own, rather unifying
functions i2d_ECDSA_SIG and i2d_DSA_SIG in fips.so leads to inlining
of function BUF_MEM_new because it is now the only call (the function
is also split making the situation even more blurry).  However, moving
the function into a header file and marking it always_inline leads to
test failures even when ICF is switched off.

I'm assigning the issue to myself in order for further analysis of
what is going on.