-
Notifications
You must be signed in to change notification settings - Fork 10.8k
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
[GIsel][AArch64] Legalize <2 x i16> for G_INSERT_VECTOR_ELT #65830
Conversation
Do we need to add |
llvm/test/CodeGen/AArch64/GlobalISel/legalize-insert-vector-elt.mir
Outdated
Show resolved
Hide resolved
Now only do the minimal fix type v2s16. |
And why not do |
Thanks @aemerson. Apply your comment, refactor the widenVectorEltsToVectorMinSize to support the scalar type of Query.Types[1], then pass 1 as the type index to avoid add handling for the destination type in LegalizerHelper. |
Sorry, I hadn't understood your earlier comment about the operand indexes not including the destination. Your earlier change with the LegalizerHelper addition makes sense in that context. If you bring back that LegalizerHelper change, it seems you don't need the additional parameter to
What do you think? |
Widen the vector elements to 64 bits to make sure it legal instead by clamping the number of elements. refer to commit ccffc27. Fixes llvm#63826
I roll back to avoid additional parameter to widenVectorEltsToVectorMinSize(), thanks very much for detail suggestion. |
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.
Thanks, LGTM.
Widen the vector elements to 64 bits to make sure it legal instead by clamping the number of elements. Depend on D153394. Fixes llvm#63826
Widen the vector elements to 64 bits to make sure it legal instead by
clamping the number of elements. Depend on D153394.
Fixes #63826