5 Commits

Author SHA1 Message Date
Nikita Popov
a96f691985 [Hexagon] Convert some tests to opaque pointers (NFC) 2022-12-19 12:53:32 +01:00
Nikita Popov
884b919f2e Reapply [Hexagon] Use default attributes for intrinsics
The issue that caused the revert has been fixed in:
44bd80751274a81c870882968ecd478b03af292a

-----

This switches Hexagon intrinsics to use the default attributes
(nosync, nofree, nocallback and willreturn). Especially willreturn
is needed to prevent optimization regressions in the future.

The only intrinsics I've excluded here are the load/store locked
intrinsics, which presumably aren't nosync.

Differential Revision: https://reviews.llvm.org/D137623
2022-11-15 16:01:14 +01:00
Alina Sbirlea
57dbca20ea Revert "[Hexagon] Use default attributes for intrinsics"
This reverts commit 8a8983b279dd5e4dceabe1fadbb8980b6adb88f9.

Uncovers existing regalloc issue in Hexagon backend - blocking for Halide
Hexagon users. Reverting to unblock, to be recommitted when underlying issue is resolved.
Reproducer available shortly.
2022-11-11 15:10:05 -08:00
Nikita Popov
8a8983b279 [Hexagon] Use default attributes for intrinsics
This switches Hexagon intrinsics to use the default attributes
(nosync, nofree, nocallback and willreturn). Especially willreturn
is needed to prevent optimization regressions in the future.

The only intrinsics I've excluded here are the load/store locked
intrinsics, which presumably aren't nosync.

Differential Revision: https://reviews.llvm.org/D137623
2022-11-11 10:59:00 +01:00
Krzysztof Parzyszek
0f791f44c7 [Hexagon] Remove dead nodes from SelectionDAG to avoid cycles
Recent changes to the instruction selection code exposed a problem where
a dead node was not removed on time. This node had both input and output
chains, which lead to an apparent cycle.

llvm-svn: 269458
2016-05-13 18:48:15 +00:00