[FFmpeg-devel] [PATCH v2] riscv: Tweak names of cpu flags, print flags in libavutil/tests/cpu

[Martin Storsjö]

On Sun, 17 Dec 2023, Rémi Denis-Courmont wrote:

> Le perjantaina 15. joulukuuta 2023, 22.52.51 EET Martin Storsjö a écrit :
>> The names of the cpu flags, when parsed from a string with
>> av_parse_cpu_caps, are parsed by the libavutil eval functions. These
>> interpret dashes as subtractions. Therefore, these previous cpu flag
>> names haven't been possible to set.
>> 
>> Use the official names for these extensions, as the previous ad-hoc
>> names wasn't parseable.
>> 
>> libavutil/tests/cpu tests that the cpu flags can be set, and prints
>> the detected flags.
>> ---
>>  libavutil/cpu.c       | 12 ++++++------
>>  libavutil/tests/cpu.c | 10 ++++++++++
>>  2 files changed, 16 insertions(+), 6 deletions(-)
>> 
>> diff --git a/libavutil/cpu.c b/libavutil/cpu.c
>> index 1e0607d581..f04068acda 100644
>> --- a/libavutil/cpu.c
>> +++ b/libavutil/cpu.c
>> @@ -186,12 +186,12 @@ int av_parse_cpu_caps(unsigned *flags, const char *s)
>>          { "rvi",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVI
>>     },    .unit = "flags" }, { "rvf",      NULL, 0, AV_OPT_TYPE_CONST, {
>> .i64 = AV_CPU_FLAG_RVF      },    .unit = "flags" }, { "rvd",      NULL, 0,
>> AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVD      },    .unit = "flags" }, -
>>        { "rvv-i32",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 =
>> AV_CPU_FLAG_RVV_I32 },     .unit = "flags" }, -        { "rvv-f32",  NULL,
>> 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F32 },     .unit = "flags"
>> }, -        { "rvv-i64",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 =
>> AV_CPU_FLAG_RVV_I64 },     .unit = "flags" }, -        { "rvv",      NULL,
>> 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F64 },     .unit = "flags"
>> }, -        { "rvb-addr",NULL, 0, AV_OPT_TYPE_CONST, { .i64 =
>> AV_CPU_FLAG_RVB_ADDR },   .unit = "flags" }, -        { "rvb-basic",NULL,
>> 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVB_BASIC },   .unit = "flags"
>> }, +        { "zve32x",   NULL, 0, AV_OPT_TYPE_CONST, { .i64 =
>> AV_CPU_FLAG_RVV_I32  },    .unit = "flags" }, +        { "zve32f",   NULL,
>> 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F32  },    .unit = "flags"
>> }, +        { "zve64x",   NULL, 0, AV_OPT_TYPE_CONST, { .i64 =
>> AV_CPU_FLAG_RVV_I64  },    .unit = "flags" },
>> +        { "zve64f",   NULL,
>> 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F64  },    .unit = "flags"
>
> That's Zve64d. Or V though strictly speaking V also implies a vector length of 
> at least 128 bits, while Zve64d only implies 64 bits.

Oh, right. But we'd use it lowercased here, as "zve64d", as that's what we 
use with the function macros and with .option arch, +<foo>, right?

Using the single-letter forms here for cpu flags would probably feel a bit 
obscure... Is there some similar names like these, that would be used for 
.option arch, for what we call rvi/rvf/rvd above?

// Martin