Stream: git-wasmtime

Topic: wasmtime / PR #4851 Cranelift AArch64: Fix the get_return...

Wasmtime GitHub notifications bot (Sep 02 2022 at 11:48):

akirilov-arm opened PR #4851 from get_return_address to main:

The previous implementation assumed that nothing had clobbered the LR register since the current function had started executing, so it would be incorrect for a non-leaf function, for example, that contains the get_return_address operation right after a call. The operation is valid only if the preserve_frame_pointers flag is enabled, which implies that the presence of a frame record on the stack is guaranteed.

Wasmtime GitHub notifications bot (Sep 02 2022 at 16:42):

cfallin created PR review comment:

Can we add a comment here describing the reload? I think this is so that we know that LR is in a consistent state (the original signed return address), even if another get_return_address op had already been executed; is that right? And then it's safe to clobber below with xpaclri because preserve_frame_pointers implies that LR will be reloaded before return?

Wasmtime GitHub notifications bot (Sep 02 2022 at 16:42):

cfallin submitted PR review.

Wasmtime GitHub notifications bot (Sep 02 2022 at 16:42):

cfallin submitted PR review.

Wasmtime GitHub notifications bot (Sep 07 2022 at 14:04):

akirilov-arm updated PR #4851 from get_return_address to main.

Wasmtime GitHub notifications bot (Sep 07 2022 at 14:05):

akirilov-arm created PR review comment:

Actually the main issue is not other get_return_address operations, but rather function calls - I added the documentation.

Wasmtime GitHub notifications bot (Sep 07 2022 at 14:05):

akirilov-arm submitted PR review.

Wasmtime GitHub notifications bot (Sep 07 2022 at 18:09):

cfallin submitted PR review.

Wasmtime GitHub notifications bot (Sep 07 2022 at 18:09):

cfallin merged PR #4851.

Last updated: Dec 23 2024 at 12:05 UTC