[PATCH] D31530: [ARM] Use new assembler diags for ARM (WIP)

[Oliver Stannard via Phabricator via llvm-commits]

olista01 added a comment.

> I imagine a generic engine which calls a target specific for the messages.

Do you think it's ok to develop this as ARM-specific code first then refactor it once we have stabilised it, or would it be better to start with the target-independent core?

> Also, can you give an example of a more general error that should silence the more specific ones?

There are plenty of examples of this in Thumb, where we have 16-bit and 32-bit versions of an instruction, where the 32-bit version accepts a wider range of immediates or registers. Note that this case won't be hit often at all with just this patch, because we have diagnostic messages for very few operand classes (just those listed in getOperandMatchFailDiag). Expanding that is on my TODO list.


Repository:
  rL LLVM

https://reviews.llvm.org/D31530