[igt-dev] [PATCH i-g-t 15/17] benchmarks/gem_wsim: for_each_ctx macro

[Bernatowicz, Marcin]

Hi,

On 10/6/2023 10:49 AM, Tvrtko Ursulin wrote:
> 
> On 05/10/2023 19:57, Marcin Bernatowicz wrote:
>> for_each_ctx_ctx_idx, for_each_ctx macros to easy traverse contexts.
>>
>> Signed-off-by: Marcin Bernatowicz <marcin.bernatowicz at linux.intel.com>
>> ---
>>   benchmarks/gem_wsim.c | 46 ++++++++++++++++++++++++-------------------
>>   1 file changed, 26 insertions(+), 20 deletions(-)
>>
>> diff --git a/benchmarks/gem_wsim.c b/benchmarks/gem_wsim.c
>> index 0c360d891..03a86b39c 100644
>> --- a/benchmarks/gem_wsim.c
>> +++ b/benchmarks/gem_wsim.c
>> @@ -231,6 +231,13 @@ struct workload {
>>       unsigned int nrequest[NUM_ENGINES];
>>   };
>> +#define for_each_ctx_ctx_idx(__ctx, __wrk, __ctx_idx) \
>> +    for (typeof((__wrk)->nr_ctxs) __ctx_idx = 0; __ctx_idx < 
>> (__wrk)->nr_ctxs && \
>> +         (__ctx = &(__wrk)->ctx_list[__ctx_idx]); ++__ctx_idx)
>> +
> 
> Is the macro name a typical naming convention for IGT stuff using 
> igt_unique? IMO it reads a bit odd and personally I think __for_each_ctx 
> + for_each_ctx would read better, but perhaps it is a personal preference.

igt_unique allows to nest the for_each_ctx loops without a warning on 
shadowed variable, I see it in macros like igt_subtest_group, 
igt_subtest_with_dynamic_f to name variables igt_unique(__tmpint), 
igt_unique(__tmpchar) .

I agree __for_each_ctx reads better.

> 
>> +#define for_each_ctx(__ctx, __wrk) \
>> +    for_each_ctx_ctx_idx(__ctx, __wrk, igt_unique(__ctx_idx))
>> +
>>   static unsigned int master_prng;
>>   static int verbose = 1;
>> @@ -1804,16 +1811,15 @@ static int prepare_contexts(unsigned int id, 
>> struct workload *wrk)
>>   {
>>       uint32_t share_vm = 0;
>>       struct w_step *w;
>> -    int i, j;
>> +    struct ctx *ctx, *ctx2;
>> +    unsigned int i, j;
>>       /*
>>        * Transfer over engine map configuration from the workload step.
>>        */
>> -    for (j = 0; j < wrk->nr_ctxs; j++) {
>> -        struct ctx *ctx = &wrk->ctx_list[j];
>> -
>> +    for_each_ctx_ctx_idx(ctx, wrk, ctx_idx) {
> 
> ctx ctx ctx ctx.. yeah it just reads wrong IMO. One ctx less would be 
> better. Maybe even as far as s/ctx_idx/idx/ for readability.
> 
> __for_each_ctx(ctx, wrk, ctx_idx)
> 
> I guess it is passable.
> 
>>           for (i = 0, w = wrk->steps; i < wrk->nr_steps; i++, w++) {
>> -            if (w->context != j)
>> +            if (w->context != ctx_idx)
>>                   continue;
>>               if (w->type == ENGINE_MAP) {
>> @@ -1850,32 +1856,32 @@ static int prepare_contexts(unsigned int id, 
>> struct workload *wrk)
>>       /*
>>        * Create and configure contexts.
>>        */
>> -    for (i = 0; i < wrk->nr_ctxs; i++) {
>> +    for_each_ctx(ctx, wrk) {
>>           struct drm_i915_gem_context_create_ext_setparam ext = {
>>               .base.name = I915_CONTEXT_CREATE_EXT_SETPARAM,
>>               .param.param = I915_CONTEXT_PARAM_VM,
>>           };
>>           struct drm_i915_gem_context_create_ext args = { };
>> -        struct ctx *ctx = &wrk->ctx_list[i];
>>           uint32_t ctx_id;
>>           igt_assert(!ctx->id);
>>           /* Find existing context to share ppgtt with. */
>> -        for (j = 0; !share_vm && j < wrk->nr_ctxs; j++) {
>> -            struct drm_i915_gem_context_param param = {
>> -                .param = I915_CONTEXT_PARAM_VM,
>> -                .ctx_id = wrk->ctx_list[j].id,
>> -            };
>> -
>> -            if (!param.ctx_id)
>> -                continue;
>> +        if (!share_vm)
>> +            for_each_ctx(ctx2, wrk) {
>> +                struct drm_i915_gem_context_param param = {
>> +                    .param = I915_CONTEXT_PARAM_VM,
>> +                    .ctx_id = ctx2->id,
>> +                };
>> +
>> +                if (!param.ctx_id)
>> +                    continue;
>> -            gem_context_get_param(fd, &param);
>> -            igt_assert(param.value);
>> -            share_vm = param.value;
>> -            break;
>> -        }
>> +                gem_context_get_param(fd, &param);
>> +                igt_assert(param.value);
>> +                share_vm = param.value;
>> +                break;
>> +            }
>>           if (share_vm) {
>>               ext.param.value = share_vm;
> 
> Conversion looks correct.
> 
> Hopefully you agree __for_each_ctx + for_each_ctx is more readable, in 
> which case:

Yes, proposed names look much better.

Thank You for review,
marcin
> 
> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
> 
> Regards,
> 
> Tvrtko