mirrors/gcc
0
0
Fork 0
mirror of https://gcc.gnu.org/git/gcc.git synced 2024-12-27 04:55:01 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

re PR target/17692 (ICE splitting sse conditional move)

Browse Source 
PR target/17692
        * config/i386/i386.c (ix86_split_sse_movcc): Emit DELETED note
        when expanding to nothing.

From-SVN: r102509
This commit is contained in:
Richard Henderson 2005-07-28 13:51:18 -07:00 committed by Richard Henderson
parent baa923061b
commit b188ebb19e
2 changed files with 27 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
gcc/ChangeLog
View File

@ -1,3 +1,9 @@
2005-07-28 Richard Henderson <rth@redhat.com>
PR target/17692
* config/i386/i386.c (ix86_split_sse_movcc): Emit DELETED note
when expanding to nothing.
2005-07-28 Josh Conner <jconner@apple.com>
* ipa-inline.c (update_caller_keys): Fix estimated_growth caching.

21
gcc/testsuite/gcc.target/i386/pr17692.c Normal file
View File

@ -0,0 +1,21 @@
/* { dg-do compile } */
/* { dg-options "-O -mfpmath=sse -msse2" } */
/* The fact that t1 and t2 are uninitialized is critical. With them
uninitialized, the register allocator is free to put them in the same
hard register, which results in
xmm0 = xmm0 >= xmm0 ? xmm0 : xmm0
Which is of course a nop, but one for which we would ICE splitting the
pattern. */
double out;
static void foo(void)
{
double t1, t2, t3, t4;
t4 = t1 >= t2 ? t1 : t2;
t4 = t4 >= t3 ? t4 : t3;
out = t4;
}