On Tue, Nov 27, 2018 at 3:51 PM Jeffrey Walton wrote:
>
> I've got some self tests failing on PowerPC with GCC 8.2. The program
> was compiled with -O1 and -fsanitize=undefined. The program uses
> Altivec. A sample finding is shown below.
Oh, sorry about this noise. This was supposed to go to GCC
On Tue, Nov 27, 2018 at 3:51 PM Jeffrey Walton wrote:
>
> I've got some self tests failing on PowerPC with GCC 8.2. The program
> was compiled with -O1 and -fsanitize=undefined. The program uses
> Altivec. A sample finding is shown below.
I managed to get it reduced to a test case:
https://gcc.gn
On Tue, Nov 27, 2018 at 3:51 PM Jeffrey Walton wrote:
>
> I've got some self tests failing on PowerPC with GCC 8.2. The program
> was compiled with -O1 and -fsanitize=undefined. The program uses
> Altivec. A sample finding is shown below.
>
> ppc_simd.h:965 is a vec_add, and it happens on a unsign
On Tue, Nov 27, 2018 at 4:17 PM Jerry James wrote:
>
> On Tue, Nov 27, 2018 at 1:53 PM Jeffrey Walton via cfarm-users
> wrote:
> > ppc_simd.h:965 is a vec_add, and it happens on a unsigned int vector.
> > We don't use signed vector types. I don't believe overflow can happen
> > and the algorithm
On Tue, Nov 27, 2018 at 3:51 PM Jeffrey Walton wrote:
>
> I've got some self tests failing on PowerPC with GCC 8.2. The program
> was compiled with -O1 and -fsanitize=undefined. The program uses
> Altivec. A sample finding is shown below.
>
> ppc_simd.h:965 is a vec_add, and it happens on a unsign
On Tue, Nov 27, 2018 at 1:53 PM Jeffrey Walton via cfarm-users
wrote:
> ppc_simd.h:965 is a vec_add, and it happens on a unsigned int vector.
> We don't use signed vector types. I don't believe overflow can happen
> and the algorithm depends on unsigned wrap.
Does building with -fwrapv help?
--
