[Bug target/101845] wrong code at -O1 on aarch64 with __builtin_shufflevector()

2021-08-11 Thread pinskia at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101845

Andrew Pinski  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |DUPLICATE

--- Comment #2 from Andrew Pinski  ---
(In reply to Tejas Belagod from comment #1)
> Confirmed.
> 
> (insn 11 10 12 2 (set (reg:V8QI 92 [ _1 ])
> (mult:V8QI (reg:V8QI 100)
> (reg/v:V8QI 97 [ v ]))) "shuf.c":11:35 -1
>  (nil))
> (insn 12 11 13 2 (set (reg:V8QI 102)
> (const_vector:V8QI [
> (const_int -1 [0x]) repeated x8
> ])) -1
>  (nil))
> 
> // Overwriting 92 [_1] looks suspicious?!

Yes so it is the same problem as PR 101529.

*** This bug has been marked as a duplicate of bug 101529 ***

[Bug target/101845] wrong code at -O1 on aarch64 with __builtin_shufflevector()

2021-08-11 Thread belagod at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101845

Tejas Belagod  changed:

   What|Removed |Added

   Last reconfirmed||2021-08-11
 Ever confirmed|0   |1
 Status|UNCONFIRMED |NEW
 CC||belagod at gcc dot gnu.org

--- Comment #1 from Tejas Belagod  ---
Confirmed.

(insn 11 10 12 2 (set (reg:V8QI 92 [ _1 ])
(mult:V8QI (reg:V8QI 100)
(reg/v:V8QI 97 [ v ]))) "shuf.c":11:35 -1
 (nil))
(insn 12 11 13 2 (set (reg:V8QI 102)
(const_vector:V8QI [
(const_int -1 [0x]) repeated x8
])) -1
 (nil))

// Overwriting 92 [_1] looks suspicious?!
(insn 13 12 14 2 (set (subreg:V4HI (reg:V8QI 92 [ _1 ]) 0)
(subreg:V4HI (reg:V8QI 102) 0)) -1
 (nil))

// Seems like merging 92 into itself leads to the wrong code...
(insn 14 13 15 2 (set (subreg:V4HI (reg:V8QI 92 [ _1 ]) 0)
(vec_merge:V4HI (vec_duplicate:V4HI (vec_select:HI (subreg:V4HI
(reg:V8QI 92 [ _1 ]) 0)
(parallel [
(const_int 0 [0])
])))
(subreg:V4HI (reg:V8QI 92 [ _1 ]) 0)
(const_int 1 [0x1]))) -1
 (nil))