llvm-gcc inline asm regression wrt 2.7

[llvmbot]

Bugzilla Link	8332
Resolution	WONTFIX
Resolved on	Apr 08, 2011 13:28
Version	unspecified
OS	Linux
Reporter	LLVM Bugzilla Contributor

Extended Description

Both gcc and llvm-gcc from LLVM 2.7 accept this, however latest llvm-gcc says:

$ llvm-gcc -S asm-4.c
error: invalid operand in inline asm: 'test2 X${0:a}Y${0:a}Z'

This is on x86-64 linux. Testcase:

int main()
{
int z;
asm volatile ("test2 X%a0Y%a[arg]Z" : : [arg] "p" (&z));
}

[llvmbot]

Getting the "p" constraint right is hard; it's not documented in a way that's understandable, and it behaves differently on different targets. The case here works fine if you use "r" instead, and the assembly code is obvious gibberish. Is it based on something real?

[llvmbot]

This is gcc.dg/asm-4.c from the gcc testsuite. It was added in this commit:

---

r48299 | rth | 2001-12-24 07:15:31 +0100 (Mon, 24 Dec 2001) | 4 lines

    * stmt.c (resolve_operand_names): Handle operand modifiers.

    * gcc.dg/asm-4.c: Test operand modifiers.

---

The content of the commit was:

Index: gcc/stmt.c

--- gcc/stmt.c (revision 48298)
+++ gcc/stmt.c (revision 48299)
@@ -2080,8 +2080,16 @@
p = buffer;
while ((p = strchr (p, '%')) != NULL)
{

•  if (*++p != '[')
  

•   continue;
  

•  if (p[1] == '[')
  

•   p += 1;
  

•  else if (ISALPHA (p[1]) && p[2] == '[')
  

•   p += 2;
  

•  else
  

•   {
  

•     p += 1;
  

•     continue;
  

•   }
  

•  p = resolve_operand_name_1 (p, outputs, inputs);
  

  }

[lattner]

LLVM-GCC has reached its end of life as far as llvm.org releases go. Please migrate to Clang or DragonEgg when possible. The llvm-gcc SVN repo will still continue to exist, so additional patches can go into it if you so desire, but since we have no more releases of it coming, it doesn't make sense to track bugs for it anymore.

If this is a problem that still exists with clang or DragonEgg, please file a new bug to track it. Thanks!

-Chris

[llvmbot]

mentioned in issue llvm/llvm-bugzilla-archive#9688