[Bug target/61062] vzip and vuzp execution tests fail on armeb

[ktkachov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

ktkachov at gcc dot gnu.org changed:

   What|Removed |Added

 Target||armeb*
 Status|NEW |RESOLVED
 CC||ktkachov at gcc dot gnu.org
 Resolution|--- |FIXED

--- Comment #7 from ktkachov at gcc dot gnu.org ---
Fixed then

[Bug target/61062] vzip and vuzp execution tests fail on armeb

[christophe.lyon at st dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

--- Comment #6 from christophe.lyon at st dot com ---
It looks OK now, thanks.

[Bug target/61062] vzip and vuzp execution tests fail on armeb

[alan.lawrence at arm dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

--- Comment #5 from Alan Lawrence alan.lawrence at arm dot com ---
Should be fixed by rev 211369.

[Bug target/61062] vzip and vuzp execution tests fail on armeb

[rearnsha at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

Richard Earnshaw rearnsha at gcc dot gnu.org changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2014-05-07
 Ever confirmed|0   |1

--- Comment #3 from Richard Earnshaw rearnsha at gcc dot gnu.org ---
Better to fix the compiler.  arm_neon.h should work both big and little-endian.

[Bug target/61062] vzip and vuzp execution tests fail on armeb

[rearnsha at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

--- Comment #4 from Richard Earnshaw rearnsha at gcc dot gnu.org ---
Looks to me as though this is probably a 4.8 and later regression.  Prior to
that we had arm-specific builtins to deal with vzip and friends.

[Bug target/61062] vzip and vuzp execution tests fail on armeb

[alan.lawrence at arm dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

Alan Lawrence alan.lawrence at arm dot com changed:

   What|Removed |Added

 CC||alan.lawrence at arm dot com

--- Comment #2 from Alan Lawrence alan.lawrence at arm dot com ---
This'll be because the ARM arm_neon.h is wrong for bigendian - the shuffle
masks used to implement ZIP/UZP/etc. need to be different for each endianness.
(See gcc/config/aarch64/arm_neon.h vzip{1,2}_* where each contains a #ifdef
__AARCH64EB__).

(In short, arm_neon.h was already broken but the new tests have just exposed
this.)

[Bug target/61062] vzip and vuzp execution tests fail on armeb

[christophe.lyon at st dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61062

--- Comment #1 from christophe.lyon at st dot com ---
Similarly, vzup tests committed as SVN rev 209947 fail on armeb targets:
  gcc.target/arm/simd/vuzpf32_1.c execution test
  gcc.target/arm/simd/vuzpp16_1.c execution test
  gcc.target/arm/simd/vuzpp8_1.c execution test
  gcc.target/arm/simd/vuzpqf32_1.c execution test
  gcc.target/arm/simd/vuzpqp16_1.c execution test
  gcc.target/arm/simd/vuzpqp8_1.c execution test
  gcc.target/arm/simd/vuzpqs16_1.c execution test
  gcc.target/arm/simd/vuzpqs32_1.c execution test
  gcc.target/arm/simd/vuzpqs8_1.c execution test
  gcc.target/arm/simd/vuzpqu16_1.c execution test
  gcc.target/arm/simd/vuzpqu32_1.c execution test
  gcc.target/arm/simd/vuzpqu8_1.c execution test
  gcc.target/arm/simd/vuzps16_1.c execution test
  gcc.target/arm/simd/vuzps32_1.c execution test
  gcc.target/arm/simd/vuzps8_1.c execution test
  gcc.target/arm/simd/vuzpu16_1.c execution test
  gcc.target/arm/simd/vuzpu32_1.c execution test
  gcc.target/arm/simd/vuzpu8_1.c execution test