[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618 --- Comment #7 from Kewen Lin --- Sorry that I didn't catch the bugzilla mails timely, thanks for fixing this!
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618 Jakub Jelinek changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution|--- |FIXED --- Comment #6 from Jakub Jelinek --- Fixed.
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618 --- Comment #5 from Jakub Jelinek --- Author: jakub Date: Fri Nov 22 18:17:04 2019 New Revision: 278626 URL: https://gcc.gnu.org/viewcvs?rev=278626=gcc=rev Log: PR tree-optimization/92618 * tree-ssa-reassoc.c (v_info): Change from auto_vec to a struct containing the auto_vec and a tree. (undistribute_bitref_for_vector): Handle the case when element type of vec is not the same as type of the BIT_FIELD_REF. Formatting fixes. * gcc.c-torture/compile/pr92618.c: New test. * gcc.c-torture/execute/pr92618.c: New test. Added: trunk/gcc/testsuite/gcc.c-torture/compile/pr92618.c trunk/gcc/testsuite/gcc.c-torture/execute/pr92618.c Modified: trunk/gcc/ChangeLog trunk/gcc/testsuite/ChangeLog trunk/gcc/tree-ssa-reassoc.c
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618 --- Comment #4 from Jakub Jelinek --- Created attachment 47332 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=47332=edit gcc10-pr92618.patch Untested fix.
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618 Jakub Jelinek changed: What|Removed |Added Status|NEW |ASSIGNED Assignee|unassigned at gcc dot gnu.org |jakub at gcc dot gnu.org
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618
--- Comment #3 from Jakub Jelinek ---
That optimization seems to ignore completely the involved types.
Consider following testcase instead, where in foo the addition is performed
originally in unsigned long long type and in baz in double type, that reassoc
change happily performs the addition in long long type in both cases.
#if __SIZEOF_LONG_LONG__ == __SIZEOF_DOUBLE__
typedef long long __m128i __attribute__((__may_alias__, __vector_size__(2 *
sizeof (long long;
unsigned long long b[4];
__m128i bar (void);
double e[4];
void
foo (unsigned long long *x)
{
__m128i c = bar ();
__m128i d = bar ();
*(__m128i *) [0] = c;
*(__m128i *) [2] = d;
*x = b[0] + b[1] + b[2] + b[3];
}
void
baz (double *x)
{
__m128i c = bar ();
__m128i d = bar ();
*(__m128i *) [0] = c;
*(__m128i *) [2] = d;
*x = e[0] + e[1] + e[2] + e[3];
}
#endif
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618
Jakub Jelinek changed:
What|Removed |Added
CC||jakub at gcc dot gnu.org
--- Comment #2 from Jakub Jelinek ---
Testcase in C and with fewer casts, plus portable between -m32/-m64:
typedef long long __m128i __attribute__((__may_alias__, __vector_size__(2 *
sizeof (long long;
void
foo (__m128i *x)
{
__m128i a;
*x = a;
}
void
bar (__m128i *x)
{
foo (x);
foo (x + 1);
}
unsigned long long b[4];
void
baz (long long *x)
{
bar ((__m128i *) b);
*x = b[0] + b[1] + b[2] + b[3];
}
[Bug tree-optimization/92618] [10 Regression] error: type mismatch in binary expression in reassoc since r273490
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92618 Martin Liška changed: What|Removed |Added Priority|P3 |P1 Status|UNCONFIRMED |NEW Last reconfirmed||2019-11-21 Known to work||9.2.0 Target Milestone|--- |10.0 Ever confirmed|0 |1 Known to fail||10.0 --- Comment #1 from Martin Liška --- Seen in gdal2 package.
