> For me, the main risk is to forget to remove the workaround once a new GCC version will be released (in N months).

This is why I suggested using the reproducer in configure.ac, setting something
like HAVE_GCC_MEMCPY_ROUNDING_BUG and then either a) wrap the *entire* PACK_SINGLE()
macro in an #ifdef or b) *preferably* add a global flag like -fno-inline or any
other flag that prevents the issue to CFLAGS.

b) under the condition that any such flag exists.


I may do a) in the weekend, which also addresses Florian Weimer's observation
that it is not known whether the issue is limited to POWER.