-
Notifications
You must be signed in to change notification settings - Fork 4.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix some flag uses #34969
Fix some flag uses #34969
Conversation
These were accidental overlaps in flag usage between Windows x86 and x64 had zero diffs. The others were: arm32: -180 (-0.00% of base) (4 improved, 1 regressed) |
@@ -3794,7 +3794,8 @@ void Compiler::lvaMarkLclRefs(GenTree* tree, BasicBlock* block, Statement* stmt, | |||
allowStructs || genActualType(varDsc->TypeGet()) == genActualType(tree->gtType) || | |||
(tree->gtType == TYP_BYREF && varDsc->TypeGet() == TYP_I_IMPL) || | |||
(tree->gtType == TYP_I_IMPL && varDsc->TypeGet() == TYP_BYREF) || (tree->gtFlags & GTF_VAR_CAST) || | |||
varTypeIsFloating(varDsc->TypeGet()) && varTypeIsFloating(tree->gtType)); | |||
(varTypeIsFloating(varDsc) && varTypeIsFloating(tree)) || | |||
(varTypeIsStruct(varDsc) == varTypeIsStruct(tree))); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There are cases where we have a SIMD that we don't normalize when the lclVar is created because the basetype is "Canon". These were getting by before because of the bug in the importer where we were setting GTF_IND_TGT_ANYWHERE
on the lclVar, which for a lclVar is GTF_VAR_CAST
, so it passed the above check.
Note that prior to the introduction of SIMD types, struct types would always "match" here.
(Also, I think the allowStructs
is just wrong and should be eliminated - we can have structs on all targets.)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How do we ever see SIMD base types being __Canon? I though these had to be value types.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I will say that this is a bit of a mystery to me - it only happens during crossgen, and I've only observed it on a GT_RET_EXPR
(which is the importer case that was incorrectly setting the flag).
@@ -9443,7 +9443,7 @@ void LinearScan::lsraGetOperandString(GenTree* tree, | |||
unsigned operandStringLength) | |||
{ | |||
const char* lastUseChar = ""; | |||
if ((tree->gtFlags & GTF_VAR_DEATH) != 0) | |||
if (tree->OperIsScalarLocal() && ((tree->gtFlags & GTF_VAR_DEATH) != 0)) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is just a fix to the dump - we shouldn't be checking last use on anything but lclVars.
@dotnet/jit-contrib PTAL |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
These issues were found while working on rearranging the flags for #34822.