[PING] Improved TableGen support for PowerPC pre-inc operand matching
Ulrich Weigand
Ulrich.Weigand at de.ibm.com
Mon Feb 25 07:57:54 PST 2013
Jakub,
I'm still working on asm parser support for PowerPC, and it turns out a
couple of TableGen changes would make the back-end implementation
significantly cleaner. In particular, a couple of weeks ago I posted a
patch to extend TableGen code to allow writing "Pat" patterns that produce
an output instruction having a single operand (with multiple sub-operands)
by providing the sub-operands as distinct operands at the pattern level:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130204/164592.html
[ Note that I've continued to investigate the X86 TLS pattern I've refered
to in that mail. As far as I can see, this pattern is really simply
broken, but then again it never gets used, since code matching (load (i64
(X86Wrapper tglobaltlsaddr :$dst))) can never be generated by the X86
back-end. So I'd now indeed suggest to simply remove the pattern. ]
I'd really appreciate a review of those TableGen changes. Do you think
this is the correct approach, or do you have any other suggestions?
Thanks for your help!
Hal,
asssuming the TableGen changes are OK, would you be OK with the PowerPC
changes to rewrite the pre-inc patterns?
Updated patch against current sources is attached. Re-tested with no
regressions.
(See attached file: diff-llvm-asm-preinc)
Mit freundlichen Gruessen / Best Regards
Ulrich Weigand
--
Dr. Ulrich Weigand | Phone: +49-7031/16-3727
STSM, GNU compiler and toolchain for Linux on System z and Cell/B.E.
IBM Deutschland Research & Development GmbH
Vorsitzende des Aufsichtsrats: Martina Koederitz | Geschäftsführung: Dirk
Wittkopp
Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht
Stuttgart, HRB 243294
-------------- next part --------------
A non-text attachment was scrubbed...
Name: diff-llvm-asm-preinc
Type: application/octet-stream
Size: 31561 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130225/b0dbc2c0/attachment.obj>
More information about the llvm-commits
mailing list