[PATCH] D48585: [X86] Use a std::vector for the memory unfolding vector.

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon accepted this revision.
RKSimon added a comment.
This revision is now accepted and ready to land.

LGTM - better error reporting would be great but not mandatory.



================
Comment at: lib/Target/X86/X86InstrInfo.cpp:5424
+           MemOp2RegOpTable.end() &&
+           "MemOp2RegOpTable is not unique!");
     assert(std::is_sorted(std::begin(MemoryFoldTable2Addr),
----------------
craig.topper wrote:
> greened wrote:
> > One thing I have wanted with the previous assert is to dump the duplicate entry so I know what to look for.  That's probably a separate change, but a useful one, I think.  I've been bitten by this before and it's a pain to figure out what happened.  Since this is under NDEBUG it seems like we can do a little more work here to inform about what's going on.
> What would that look like? Do you just want to print the enum values from the failing row or do you want to try to turn that back into a string that a human could read?
Anything that can be used to debug would be useful - enums are easy enough to work with as you can manually look them up in X86GenInstrInfo.inc - but an actual string would be even better.


Repository:
  rL LLVM

https://reviews.llvm.org/D48585