On Wed, 1 Jul 2026, Christophe Lyon wrote:

> This patch applies cleanly to gcc-15, OK to backport?

OK

> Thanks,
> 
> Christophe
> 
> 
> ---
> 
> Depending on how GCC was configured, default -mtune parameter can load
> the floating-point constant using either:
>     movw    r3, #52429
>     movt    r3, 16268
>     or
>     ldr     r3, .L4
> 
> Update the expected code to accept both versions.
> 
> Tested on:
> - arm-linux-gnueabihf --with-tune=cortex-a9 --with-arch=armv7-a
> - arm-none-eabi --with-cpu=cortex-m0
> - arm-none-eabi --with-cpu=cortex-m55
> 
> 2026-01-26  Christophe Lyon  <[email protected]>
> 
>       gcc/testsuite/
>       * gcc.target/arm/mve/intrinsics/vdupq_n_f32.c: Update expected
>       code.
> 
> (cherry picked from commit 858392b691d0f02bcb5b094191ea454dbaf49a33)
> ---
>  gcc/testsuite/gcc.target/arm/mve/intrinsics/vdupq_n_f32.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vdupq_n_f32.c 
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vdupq_n_f32.c
> index 81c2e29f5e5..6a01ce47b6d 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vdupq_n_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vdupq_n_f32.c
> @@ -24,7 +24,12 @@ foo (float32_t a)
>  /*
>  **foo1:
>  **   ...
> +** (
>  **   ldr     r[0-9]+, .L.*
> +** |
> +**   movw    (r[0-9]+), #52429
> +**   movt    \1, 16268
> +** )
>  **   vdup.32 q[0-9]+, r[0-9]+
>  **   ...
>  */
> 

-- 
Richard Biener <[email protected]>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Jochen Jaser, Andrew McDonald; (HRB 36809, AG Nuernberg)

Reply via email to