[Bug middle-end/88010] noinline function alias unexpectedly inlined
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88010 --- Comment #5 from Martin Sebor --- (In reply to Jan Hubicka from comment #3) > Is there some critical use of this that we would want to have for GCC 9? I don't think making it work as I suggested is critical for GCC 9 in comment #0. I noticed it while testing the new -Wattribute-alias/-Wmissing-attributes options with Glibc, but (AFAICT) Glibc doesn't actually depend on the inlining attributes for aliases working this way. It would make sense to me to issue a warning for this use case until it is implemented (if it ever is).
[Bug middle-end/88010] noinline function alias unexpectedly inlined
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88010 --- Comment #4 from Jan Hubicka --- > Yep, GCC considers attributes to be part of the definition of a function for > IPA passes. We are not consitent here (i.e. warning attributes on aliases > counts), so it makes sense to support this (and is not too difficult to do in > this case). Actually siplicity was bit overrated - currently all function redirection code eliminates aliases, so we will need to carefully update those so attributes are not lost. Still implementable.
[Bug middle-end/88010] noinline function alias unexpectedly inlined
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88010 Jan Hubicka changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2018-11-14 Ever confirmed|0 |1 --- Comment #3 from Jan Hubicka --- Yep, GCC considers attributes to be part of the definition of a function for IPA passes. We are not consitent here (i.e. warning attributes on aliases counts), so it makes sense to support this (and is not too difficult to do in this case). Is there some critical use of this that we would want to have for GCC 9? Honza
[Bug middle-end/88010] noinline function alias unexpectedly inlined
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88010 Richard Biener changed: What|Removed |Added CC||hubicka at gcc dot gnu.org --- Comment #2 from Richard Biener --- Aliases are generally "transparent" for most (late) attribute lookups in IPA since IPA looks through aliases.
[Bug middle-end/88010] noinline function alias unexpectedly inlined
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88010 Martin Sebor changed: What|Removed |Added Known to fail||4.1.0, 8.2.0, 9.0 --- Comment #1 from Martin Sebor --- All GCC versions appear to behave this way.