[Backtracing] Fix ARM64e build failure. #69859
Conversation
Accessing the thread context structures is complicated because their member variables change name depending on various macros. This caused a build failure for ARM64e. rdar://118402226
|
@swift-ci Please smoke test |
| #elif defined(__arm64__) | ||
| pc = (void *)(ctx->uc_mcontext->CTX_MEMBER(ss).CTX_MEMBER(pc)); | ||
| pc = (void *)(ctx->uc_mcontext->CTX_MEMBER(ss).THREAD_STATE_MEMBER(pc)); |
There was a problem hiding this comment.
Would be good to strip this, since the signature bits won't be useful to us.
There was a problem hiding this comment.
It's really only a "we're doing a backtrace, here's the program counter in case it goes wrong" message, so I'm not too worried about it — we'll normally present it stripped very shortly afterwards anyway.
There was a problem hiding this comment.
In fact, thinking about it, it might be nice to keep the ptrauth bits here, because if we're interrupted while attempting to generate the backtrace, we aren't going to show a register dump (so we won't have any way to see what they were if we strip them, and if the failure happened for ptrauth reasons, we'd have no clue at that point).
There was a problem hiding this comment.
Gotcha, that all makes sense to me.
|
@swift-ci Please smoke test macOS platform |
Accessing the thread context structures is complicated because their member variables change name depending on various macros. This caused a build failure for ARM64e.
rdar://118402226