PR #36576 Simplify llvmcall

Simplify llvmcall #36576

maleadt merged 7 commits into master from tb/llvmcall

maleadt added compiler:codegen

maleadt added gpu

maleadt requested a review from

vtjnash 5 years ago

maleadt marked this pull request as draft 5 years ago

maleadt force pushed from e74e01fe to 157b0b9f 5 years ago

maleadt force pushed from 157b0b9f to 0729ccee 5 years ago

maleadt marked this pull request as ready for review 5 years ago

maleadt force pushed from 0729ccee to 9829772b 5 years ago

maleadt force pushed from 9829772b to 3fa501e2 5 years ago

yuyichao added breaking

maleadt force pushed from 3fa501e2 to cfe250f7 5 years ago

vtjnash commented on 2020-07-08

maleadt commented on 2020-07-09

maleadt force pushed from 983e59ef to b27d8b9c 5 years ago

maleadt commented on 2020-07-09

vchuravy approved these changes on 2020-07-17

Simplify llvmcall, only supporting literal function or module IR inputs.

96d36737

Make it possible to specify module bitcode as a byte array.

0fb70158

Link the llvmcall IR modules after codegen.

3601351a

Update the llvmcall documentation.

b94f920e

Fix off-by-one in do_invoke.

6ee43712

Remove support for late evaluation of llvmcall arguments.

82ab1458

Make some more llvmcall errors dynamic.

04712bf0

maleadt force pushed from 41b4a9d0 to 04712bf0 5 years ago

maleadt merged e58fb286 into master 5 years ago

maleadt deleted the tb/llvmcall branch 5 years ago

Reviewers

vchuravy

vtjnash

Assignees

No one assigned

Labels

breaking compiler:codegen gpu

Milestone

No milestone