https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
--- Comment #7 from Andrew Pinski ---
There was a scheduler change I thought which prevented moving prefetches across
load/stores but I can't find it for some reason. Maybe it didn't make it in.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
--- Comment #6 from Oleg Endo ---
(In reply to pietro from comment #5)
> Created attachment 59210 [details]
> Add a blockage instruction before the prefetch
>
> I did a few tests on sh4-elf and adding a blockage before the prefetch keeps
> the
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
--- Comment #5 from pietro ---
Created attachment 59210
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=59210&action=edit
Add a blockage instruction before the prefetch
I did a few tests on sh4-elf and adding a blockage before the prefetch
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
--- Comment #4 from Oleg Endo ---
(In reply to pietro from comment #3)
> It looks like it's a more general GCC issue. The prefetch gets moved on both
> x86_64 and aarch64 on GCC, but not on clang: https://godbolt.org/z/Ycjr7Tq8b
>
> > It looks
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
--- Comment #3 from pietro ---
It looks like it's a more general GCC issue. The prefetch gets moved on both
x86_64 and aarch64 on GCC, but not on clang: https://godbolt.org/z/Ycjr7Tq8b
> It looks like the problem can be "fixed" by inserting a
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
Oleg Endo changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last reconfirmed|
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116713
pietro changed:
What|Removed |Added
CC||pietro.gcc at sociotechnical
dot x
