12 Commits

Author SHA1 Message Date
Matt Arsenault
d873a14e93 ValueTracking: Implement computeKnownFPClass for frexp
Work around the lack of proper multiple return values by looking
at the extractvalue.

https://reviews.llvm.org/D150982
2023-07-21 16:04:13 -04:00
Matt Arsenault
e1ac984a10 ValueTracking: Implement computeKnownFPClass for ldexp
https://reviews.llvm.org/D149590
2023-07-11 09:26:41 -04:00
Arthur Eubanks
f4f826bcd4 Revert "Revert "ValueTracking: Fix nan result handling for fmul""
This reverts commit 464dcab8a6c823c9cb462bf4107797b8173de088.

Going to fix forward size regression instead due to more dependent patches needing to be reverted otherwise.
2023-06-16 13:53:32 -07:00
Arthur Eubanks
3e39cfe5b4 Revert "Revert "InstSimplify: Require instruction be parented""
This reverts commit 0c03f48480f69b854f86d31235425b5cb71ac921.

Going to fix forward size regression instead due to more dependent patches needing to be reverted otherwise.
2023-06-16 13:53:31 -07:00
Arthur Eubanks
0c03f48480 Revert "InstSimplify: Require instruction be parented"
This reverts commit 1536e299e63d7788f38117b0212ca50eb76d7a3b.

Causes large binary size regressions, see comments on https://reviews.llvm.org/rG1536e299e63d7788f38117b0212ca50eb76d7a3b.
2023-06-16 11:24:29 -07:00
Arthur Eubanks
464dcab8a6 Revert "ValueTracking: Fix nan result handling for fmul"
This reverts commit a632ca4b00279baf18e72a171ec0ce526e9d80aa.

Dependent commit to be reverted
2023-06-16 11:24:28 -07:00
Alan Zhao
d6b4f6786b Revert "Revert "InstSimplify: Require instruction be parented""
This reverts commit 00264eac4d0938ae8a0826da38e4777be269124c.

Reason: caused a bunch of bots to break
2023-06-16 10:58:54 -07:00
Alan Zhao
00264eac4d Revert "InstSimplify: Require instruction be parented"
This reverts commit 1536e299e63d7788f38117b0212ca50eb76d7a3b.

Reason: causes a regression in the inliner (see https://crbug.com/1454531 and https://reviews.llvm.org/rG1536e299e63d7788f38117b0212ca50eb76d7a3b#1217141)
2023-06-16 10:36:49 -07:00
Matt Arsenault
a632ca4b00 ValueTracking: Fix nan result handling for fmul
This was mishandling maybe 0 * inf.

Fixes issue #63316
2023-06-15 09:35:12 -04:00
Matt Arsenault
1536e299e6 InstSimplify: Require instruction be parented
Unlike every other analysis and transform, simplifyInstruction
permitted operating on instructions which are not inserted
into a function. This created an edge case no other code needs
to really worry about, and limited transforms in cases that
can make use of the context function. Only the inliner and a handful
of other utilities were making use of this, so just fix up these
edge cases. Results in some IR ordering differences since
cloned blocks are inserted eagerly now. Plus some additional
simplifications trigger (e.g. some add 0s now folded out that
previously didn't).
2023-06-02 18:14:28 -04:00
Matt Arsenault
4130ccc8be ValueTracking: Check context instruction is in a function 2023-05-18 14:40:13 +01:00
Matt Arsenault
f42136d4d6 ValueTracking: Check instruction is in a parent in computeKnownFPClass
For some reason the inliner calls simplifyInstruction with disembodied
instructions. I consider this to be an API defect. Either the instruction
should always be inserted prior to simplification, or we at least
should pass in the new function for the context.
2023-05-18 12:21:47 +01:00