Joe Eli McIlvain
Joe Eli McIlvain
Segmentation fault caused by likely LLVM bug that results in incorrect exception handling on aarch64
@SeanTAllen and I think this is related to us needing to write some ARM64-specific code in `posix_except.c`. We currently have ARM32-specific code, and then catch-all code for everything else. Here's...
Segmentation fault caused by likely LLVM bug that results in incorrect exception handling on aarch64
That PDF links to this doc, which seems to have the more relevant information: https://itanium-cxx-abi.github.io/cxx-abi/abi-eh.html
Segmentation fault caused by likely LLVM bug that results in incorrect exception handling on aarch64
@SeanTAllen - try this diff: ```diff diff --git a/src/libponyc/codegen/codegen.c b/src/libponyc/codegen/codegen.c index 6a972f88..147435b5 100644 --- a/src/libponyc/codegen/codegen.c +++ b/src/libponyc/codegen/codegen.c @@ -265,6 +265,7 @@ static void init_runtime(compile_t* c) unsigned int align_value = target_is_ilp32(c->opt->triple)...
Segmentation fault caused by likely LLVM bug that results in incorrect exception handling on aarch64
In case anything here is helpful: https://llvm.org/docs/ExceptionHandling.html#exception-handling-support-on-the-target
In the sync call, I agreed with Jason that conceptually the function with default arguments should be a subtype of the function without arguments. Sean does not agree. Making it...
We agreed in the sync call that extending the special case error that @ergl mentioned is the right fix.
Yeah, we don't have a generic "unparser" within ponyc. Perhaps an easier (but still significant effort) approach would be to update our parser to track not only the start position...
I see no objections to the above diff.
Discussed on the sync call today. The fastest fix seems to be removing the assert and changing the error string at https://github.com/ponylang/ponyc/blob/deb1494a4ba423b5c46743062b00ccf5f4bf7942/src/libponyc/expr/reference.c#L173-L185 Jason also says it may be possible to...
Okay, I've taken a first stab at auditing the possible error types for `accept4` on Linux (see the comment after this one). I'm looking for someone like @SeanTAllen to check...