https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
Andrew Pinski changed:
What|Removed |Added
Target Milestone|--- |14.0
Status|ASSIGNED
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
--- Comment #7 from CVS Commits ---
The trunk branch has been updated by Andrew Pinski :
https://gcc.gnu.org/g:64ca6aa74b6b5a9737f3808bf4a947dd5c122d47
commit r14-1508-g64ca6aa74b6b5a9737f3808bf4a947dd5c122d47
Author: Andrew Pinski
Date: Th
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
--- Comment #6 from Andrew Pinski ---
Created attachment 55238
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=55238&action=edit
Patch which I will be testing
This adds a check for the address space in DSE.
Even tested:
```
void test_null
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
Andrew Pinski changed:
What|Removed |Added
Status|NEW |ASSIGNED
Assignee|unassigned
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
--- Comment #4 from Andrew Pinski ---
(In reply to Martin Sebor from comment #3)
> A couple of data points: I see the same behavior for any constant address
> (not just null).
Right because DSE does not compare MEM_ADDR_SPACE. It is literally
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
--- Comment #3 from Martin Sebor ---
A couple of data points: I see the same behavior for any constant address (not
just null). Clang 12 behaves the same as GCC, but Clang 13 emits both stores,
suggesting they fixed it as a bug:
https://godb
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
--- Comment #2 from Andrew Pinski ---
(In reply to Andrew Pinski from comment #1)
> RTL level DSE removes the first store. I suspect dse.c does not check
> ADDR_SPACE_GENERIC_P.
I mean MEM_ADDR_SPACE. And yes it looks like it does not check M
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102733
Andrew Pinski changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever confirmed|0
