Re: [PATCH v2] target/loongarch/gdbstub: Add vector registers support

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 11/7/24 04:44, Song Gao wrote:
> GDB already support LoongArch vector extension[1], QEMU gdb adds
> LoongArch vector registers support, so that users can use 'info all-registers'
> to get all vector registers values.
> 
> [1]: https://sourceware.org/git/?p=binutils-gdb.git;a=commitdiff;h=1e9569f383a3d5a88ee07d0c2401bd95613c222e
> 
> Signed-off-by: Song Gao <gaosong@loongson.cn>
> ---
> based-on:
>   https://patchew.org/QEMU/20240607035016.2975799-1-maobibo@loongson.cn/
> 
> v2:
> - fix tab line wrapper issue.
> - Link to v1: https://patchew.org/QEMU/20240621065406.864232-1-gaosong@loongson.cn/
> 
>   configs/targets/loongarch64-linux-user.mak |  2 +-
>   configs/targets/loongarch64-softmmu.mak    |  2 +-
>   target/loongarch/gdbstub.c                 | 70 +++++++++++++++++++++-
>   gdb-xml/loongarch-lasx.xml                 | 60 +++++++++++++++++++
>   gdb-xml/loongarch-lsx.xml                  | 59 ++++++++++++++++++
>   5 files changed, 189 insertions(+), 4 deletions(-)
>   create mode 100644 gdb-xml/loongarch-lasx.xml
>   create mode 100644 gdb-xml/loongarch-lsx.xml


> +static int loongarch_gdb_get_vec(CPUState *cs, GByteArray *mem_buf, int n, int vl)
> +{
> +    LoongArchCPU *cpu = LOONGARCH_CPU(cs);
> +    CPULoongArchState *env = &cpu->env;
> +    int i, length = 0;
> +
> +    if (0 <= n && n < 32) {
> +        for (i = 0; i < vl / 64; i++) {

Preferably using FOO_PER_BAR definitions for 32 & 64 magic values,

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

> +            length += gdb_get_reg64(mem_buf, env->fpr[n].vreg.D(i));
> +        }
> +    }
> +
> +    return length;
> +}
> +
> +static int loongarch_gdb_set_vec(CPUState *cs, uint8_t *mem_buf, int n, int vl)
> +{
> +    LoongArchCPU *cpu = LOONGARCH_CPU(cs);
> +    CPULoongArchState *env = &cpu->env;
> +    int i, length = 0;
> +
> +    if (0 <= n && n < 32) {
> +        for (i = 0; i < vl / 64; i++) {
> +            env->fpr[n].vreg.D(i) = ldq_le_p(mem_buf + 8 * i);
> +            length += 8;
> +        }
> +    }
> +
> +    return length;
> +}