clang in-process cc1 causes ASan failures in interface stubs generation

[zmodem]

Bugzilla Link	44865
Version	10.0
OS	Linux
Blocks	#43900
CC	@plotfi

Extended Description

First reported here: https://reviews.llvm.org/D69825#1831258

Worked around with:
https://reviews.llvm.org/rGc38e42527b21acee8d01a016d5bfa2fb83202e29
https://reviews.llvm.org/rGe174da447c180b586719cb28f7bd556e30625762

Later on in D69825 it's suggested that if we can't fix this properly, maybe -emit-interface-stubs should imply -fno-integrated-cc1

Filing this to keep track of the issue, especially for the 10.0.0 release.

[zmodem]

Puyan, have you had any chance to look into this some more?

[zmodem]

Puyan, have you had any chance to look into this some more?

Ping?

[zmodem]

Puyan replied by email that they're looking into suppressing in-process cc1 when combined with -emit-interface-stubs.

[llvmbot]

@llvm/issue-subscribers-clang-driver

Author: Hans (zmodem)

| | | | --- | --- | | Bugzilla Link | [44865](https://llvm.org/bz44865) | | Version | 10.0 | | OS | Linux | | Blocks | llvm/llvm-project#43900 | | CC | @plotfi |

Extended Description

First reported here: https://reviews.llvm.org/D69825#1831258

Worked around with:
https://reviews.llvm.org/rGc38e42527b21acee8d01a016d5bfa2fb83202e29
https://reviews.llvm.org/rGe174da447c180b586719cb28f7bd556e30625762

Later on in D69825 it's suggested that if we can't fix this properly, maybe -emit-interface-stubs should imply -fno-integrated-cc1

Filing this to keep track of the issue, especially for the 10.0.0 release.