https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124336
--- Comment #5 from Zdenek Sojka <zsojka at seznam dot cz> --- (In reply to Uroš Bizjak from comment #4) > (In reply to Zdenek Sojka from comment #3) > > I think, from the performance point of view, generally the size of the > > memory access should be the same as other accesses to the same memory > > location [rsp+0], in order to allow store-to-load forwarding. (if there are > > such accesses) > This is only a probe insn that accesses the stack at the end of the page: > > c: 48 81 ec 00 10 00 00 sub $0x1000,%rsp > 13: 83 0c 24 00 orl $0x0,(%rsp) > > There is "nothing" at this location, so any kind of STL is not effective. Thank you for the explanation!
