[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 --- Comment #7 from CVS Commits --- The releases/gcc-8 branch has been updated by Jakub Jelinek : https://gcc.gnu.org/g:fbb4d21086a20cd996049309d6c24d7fb3099123 commit r8-10019-gfbb4d21086a20cd996049309d6c24d7fb3099123 Author: Jakub Jelinek Date: Fri Feb 14 15:49:32 2020 +0100 i386: Fix k*shift* intrinsics [PR93673] As mentioned in the PR, the intrinsics allow counts from 0 to 255, but we actually reject values from 128 to 255. That is because QImode CONST_INTs can be only -128 to 127. Fixed by using const_0_to_255_operand and dropping the modes for the operands with those predicates (the IL actually contains the CONST_INT which has VOIDmode). 2020-02-13 Jakub Jelinek PR target/93673 * config/i386/sse.md (k): Drop mode from last operand and use const_0_to_255_operand predicate instead of immediate_operand. (avx512dq_fpclass, avx512dq_vmfpclass, vgf2p8affineinvqb_, vgf2p8affineqb_): Drop mode from const_0_to_255_operand predicated operands. * gcc.target/i386/avx512f-pr93673.c: New test. * gcc.target/i386/avx512dq-pr93673.c: New test. * gcc.target/i386/avx512bw-pr93673.c: New test.
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 --- Comment #6 from CVS Commits --- The releases/gcc-9 branch has been updated by Jakub Jelinek : https://gcc.gnu.org/g:488a947b2ddd57a6f44a6aecc32862f8cbf4ec77 commit r9-8225-g488a947b2ddd57a6f44a6aecc32862f8cbf4ec77 Author: Jakub Jelinek Date: Thu Feb 13 08:17:07 2020 +0100 i386: Fix k*shift* intrinsics [PR93673] As mentioned in the PR, the intrinsics allow counts from 0 to 255, but we actually reject values from 128 to 255. That is because QImode CONST_INTs can be only -128 to 127. Fixed by using const_0_to_255_operand and dropping the modes for the operands with those predicates (the IL actually contains the CONST_INT which has VOIDmode). 2020-02-13 Jakub Jelinek PR target/93673 * config/i386/sse.md (k): Drop mode from last operand and use const_0_to_255_operand predicate instead of immediate_operand. (avx512dq_fpclass, avx512dq_vmfpclass, vgf2p8affineinvqb_, vgf2p8affineqb_): Drop mode from const_0_to_255_operand predicated operands. * gcc.target/i386/avx512f-pr93673.c: New test. * gcc.target/i386/avx512dq-pr93673.c: New test. * gcc.target/i386/avx512bw-pr93673.c: New test.
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 Hongtao.liu changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution|--- |FIXED --- Comment #5 from Hongtao.liu --- Fixed in GCC10
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 --- Comment #4 from CVS Commits --- The master branch has been updated by Jakub Jelinek : https://gcc.gnu.org/g:dc6d0f89d4be3ed7fde73417606a78c73d954cdf commit r10-6612-gdc6d0f89d4be3ed7fde73417606a78c73d954cdf Author: Jakub Jelinek Date: Thu Feb 13 08:17:07 2020 +0100 i386: Fix k*shift* intrinsics [PR93673] As mentioned in the PR, the intrinsics allow counts from 0 to 255, but we actually reject values from 128 to 255. That is because QImode CONST_INTs can be only -128 to 127. Fixed by using const_0_to_255_operand and dropping the modes for the operands with those predicates (the IL actually contains the CONST_INT which has VOIDmode). 2020-02-13 Jakub Jelinek PR target/93673 * config/i386/sse.md (k): Drop mode from last operand and use const_0_to_255_operand predicate instead of immediate_operand. (avx512dq_fpclass, avx512dq_vmfpclass, vgf2p8affineinvqb_, vgf2p8affineqb_): Drop mode from const_0_to_255_operand predicated operands. * gcc.target/i386/avx512f-pr93673.c: New test. * gcc.target/i386/avx512dq-pr93673.c: New test. * gcc.target/i386/avx512bw-pr93673.c: New test.
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 --- Comment #3 from Jakub Jelinek --- Created attachment 47816 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=47816=edit gcc10-pr93673.patch I meant this actually. QImode for const_0_to_255_operand is wrong, because QImode CONST_INTs are -128 to 127 rather than 0 to 255.
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673
--- Comment #2 from H.J. Lu ---
Something like this:
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 902ea318999..b3b6552e13b 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -1650,7 +1650,7 @@ (define_insn "k"
[(set (match_operand:SWI1248_AVX512BWDQ 0 "register_operand" "=k")
(any_lshift:SWI1248_AVX512BWDQ
(match_operand:SWI1248_AVX512BWDQ 1 "register_operand" "k")
-(match_operand:QI 2 "immediate_operand" "n")))
+(match_operand:QI 2 "const_0_to_255_operand")))
(unspec [(const_int 0)] UNSPEC_MASKOP)]
"TARGET_AVX512F"
"k\t{%2, %1, %0|%0, %1, %2}"
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 Jakub Jelinek changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2020-02-11 CC||jakub at gcc dot gnu.org Assignee|unassigned at gcc dot gnu.org |jakub at gcc dot gnu.org Ever confirmed|0 |1
[Bug target/93673] Fake error given by gcc when compiling for _kshift intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93673 --- Comment #1 from Hongtao.liu --- Affected instrinsics _kshiftli_mask16 _kshiftri_mask16
