[PATCH] Fix rtx_equal_p and similar predicates on ASM_OPERANDS/ASM_INPUT (PR rtl-optimization/46865)

[Chung-Lin Tang]

On 2010/12/10 06:06, Jakub Jelinek wrote:
> The last operand of ASM_OPERANDS and ASM_INPUT is a RTL locator,
> for which we unfortunately don't guarantee uniqueness.  One has
> to call locator_eq to actually compare them.
> On the testcases below, without -save-temps the locators are equal
> while without them some other location_t got a locator in between.

Actually, I think the issue is not just about comparing locators, but 
rather: why do you need locator equality for ASM_INPUT/ASM_OPERANDS to 
be equivalent? The checking of the other fields should be sufficient. 
This way for the testcase, the non-macro version would then also be 
combined into a single asm.

I was about to submit this patch, which simply skips the ending 'i' 
field for ASM_INPUT/ASM_OPERANDS when checking equivalence. Bootstrapped 
and tested on i686 and x86_64; tested on ARM-Linux with QEMU.

Thanks,
Chung-Lin

2010-12-10  Chung-Lin Tang  <cltang@codesourcery.com>

	PR rtl-optimization/46865
	* rtl.c (rtx_equal_p_cb): Skip location operand when comparing
	for equivalence.
	(rtx_equal_p): Same.
	* jump.c (rtx_renumbered_equal_p): Same.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: asm.diff
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20101210/e8faa2ab/attachment.ksh>