Re: [PATCH] target/riscv: don't enable Zfa by default

[Daniel Henrique Barboza]

On 11/6/23 08:14, Jerry Zhang Jian wrote:
> - Zfa requires F, we should not assume all CPUs have F extension
>    support.

We do not have a case where this happen, do we? The default CPUs have F
enabled (see misa_ext_cfgs[] in target/riscv/tcg/tcg-cpu.c), so zfa being
enable isn't a problem for them. Vendor CPUs might not have F enabled, but
they don't use the default values for extensions, so they're not affected.
Having zfa enabled by default does not hurt the default CPU setups we have.

I am not a fan of these defaults for rv64 and so on, but once we set them to
'true' people can complain if we set them to 'false' because it might break
existing configs in the wild. We need a strong case (i.e. a bug) to do so.


Thanks,

Daniel


> Signed-off-by: Jerry Zhang Jian <jerry.zhangjian@sifive.com>
> ---
>   target/riscv/cpu.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
> index ac4a6c7eec..c9f11509c8 100644
> --- a/target/riscv/cpu.c
> +++ b/target/riscv/cpu.c
> @@ -1247,7 +1247,7 @@ const RISCVCPUMultiExtConfig riscv_cpu_extensions[] = {
>       MULTI_EXT_CFG_BOOL("zihintntl", ext_zihintntl, true),
>       MULTI_EXT_CFG_BOOL("zihintpause", ext_zihintpause, true),
>       MULTI_EXT_CFG_BOOL("zawrs", ext_zawrs, true),
> -    MULTI_EXT_CFG_BOOL("zfa", ext_zfa, true),
> +    MULTI_EXT_CFG_BOOL("zfa", ext_zfa, false),
>       MULTI_EXT_CFG_BOOL("zfh", ext_zfh, false),
>       MULTI_EXT_CFG_BOOL("zfhmin", ext_zfhmin, false),
>       MULTI_EXT_CFG_BOOL("zve32f", ext_zve32f, false),