Re: [PATCH] selftests/powerpc: Fix earlyclobber in tm-vmxcopy

2019-06-30 Thread Michael Ellerman
On Mon, 2019-06-17 at 21:24:58 UTC, Gustavo Romero wrote: > In some cases, compiler can allocate the same register for operand 'res' > and 'vecoutptr', resulting in segfault at 'stxvd2x 40,0,%[vecoutptr]' > because base register will contain 1, yielding a false-positive. > > This is because

Re: [PATCH] selftests/powerpc: Fix earlyclobber in tm-vmxcopy

2019-06-18 Thread Segher Boessenkool
On Mon, Jun 17, 2019 at 05:24:58PM -0400, Gustavo Romero wrote: > In some cases, compiler can allocate the same register for operand 'res' > and 'vecoutptr', resulting in segfault at 'stxvd2x 40,0,%[vecoutptr]' > because base register will contain 1, yielding a false-positive. > > This is because

[PATCH] selftests/powerpc: Fix earlyclobber in tm-vmxcopy

2019-06-17 Thread Gustavo Romero
In some cases, compiler can allocate the same register for operand 'res' and 'vecoutptr', resulting in segfault at 'stxvd2x 40,0,%[vecoutptr]' because base register will contain 1, yielding a false-positive. This is because output 'res' must be marked as an earlyclobber operand so it may not