Re: [cfe-commits] [PATCH] Inline asm mult-alt constraints changes and tests, Clang-side

Tue, 02 Nov 2010 13:46:50 -0700

+    case 'z':  // First SSE register.

This one isn't in gcc-4.2, it was added later.   OK unless we think it's covered by GPL3 somehow?  I don't see how, but IANAL.

+      Name++;

This doesn't  currently accomplish anything.  From the tests I see handling "Y?" is a work in progress, so OK if you have a plan in mind for how to use it.

+  case 'I': // Integer constant in the range 0..31 for 32-bit shifts.
+  case 'J': // Integer constant in the range 0..63 for 64-bit shifts.
+  case 'K': // 8-bit signed integer constant.
+  case 'L': // 0xff or 0xffff, for andsi as a zero-extending move.
+  case 'M': // 0, 1, 2, 3 (shifts for the lea instruction).
+  case 'N': // Unsigned 8-bit integer constant (for in and out instructions).

These are recognized by the generic code in validateInputConstraint, we shouldn't duplicate that here.

Testcases are OK (assuming they pass)

On Nov 2, 2010, at 11:41 AMPDT, John Thompson wrote:

May I check in the changes in the enclosed patch?
 
I added a few constraint characters for x86 from the gcc docs, and two new test files for validating the .ll output with respect to mult-alt constraints.
 
Thanks.
 
-John

--
John Thompson
[email protected]

Attachment: clangmultalt10.patch
Description: Binary data

_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to