Re: [PATCH i-g-t v8 8/9] tests/intel/kms_ccs: limit #pipes tested during suspend

["Juha-Pekka Heikkilä" <juhapekka.heikkila@gmail.com>]

On 13/10/2026 19.10, Michał Grzelak wrote:
> Limit execution of suspend tests to first and last active pipe unless
> using `extended` flag. This reduces default execution time of the test.
> In turn this can lower chances of hitting CI timeout.
> 
> Alternatively test could be run on first and last CRTC; but due to pipe
> reordering, this would result in using arbitrary pipes. Thus set first
> and last pipe as the only available in suspend resume case.
> 
> v7->v8
> - use pipe mask (Jani)
> 
> v6->v7
> - use first and last hardware pipe instead of CRTCs (J-P, Jani)
> 
> v5->v6
> - describe how pipe reordering affects the test (Jani)
> 
> v3->v4
> - set `first' and `last' indices outside of crtc loop (Thasleem)
> 
> Signed-off-by: Michał Grzelak <michal.grzelak@linux.intel.com>
> ---
>   tests/intel/kms_ccs.c | 14 ++++++++++++--
>   1 file changed, 12 insertions(+), 2 deletions(-)
> 
> diff --git a/tests/intel/kms_ccs.c b/tests/intel/kms_ccs.c
> index 734dafd110..3af091fd87 100644
> --- a/tests/intel/kms_ccs.c
> +++ b/tests/intel/kms_ccs.c
> @@ -192,6 +192,7 @@ typedef struct {
>   	bool available_pipes[IGT_MAX_PIPES];
>   	bool do_hibernate;
>   	bool all_outputs;
> +	bool all_crtcs;
>   	struct {
>   		struct igt_fb fb;
>   		int width, height;
> @@ -1098,7 +1099,9 @@ static bool set_available_hw_pipes(data_t *data)
>   
>   	for_each_crtc(&data->display, crtc) {
>   		hw_pipe = crtc->hardware_pipe;
> -		data->available_pipes[hw_pipe] = true;
> +
> +		if (data->all_crtcs)
> +			data->available_pipes[hw_pipe] = true;
>   
>   		if (last_pipe < hw_pipe)
>   			last_pipe = hw_pipe;
> @@ -1110,6 +1113,9 @@ static bool set_available_hw_pipes(data_t *data)
>   	if (first_pipe == IGT_MAX_PIPES || last_pipe == PIPE_NONE)
>   		return false;
>   
> +	data->available_pipes[first_pipe] = true;
> +	data->available_pipes[last_pipe] = true;
> +
>   	return true;
>   }
>   
> @@ -1260,6 +1266,9 @@ static int opt_handler(int opt, int opt_index, void *opt_data)
>   	case 'n':
>   		data->all_outputs = true;
>   		break;
> +	case 'e':
> +		data->all_crtcs = true;
> +		break;
>   	default:
>   		return IGT_OPT_HANDLER_ERROR;
>   	}
> @@ -1274,9 +1283,10 @@ static const char *help_str =
>   "  -s <seed>\tSeed for random number generator\n"
>   "  -r\t\tOn suspend test do full hibernate with reboot\n"
>   "  -n\t\tRun on all connectors. (By default suspend subtests will run on 3 connectors)\n"
> +"  -e\t\tRun on all CRTCs. (By default subtests will run on two CRTCs)\n"
>   ;
>   
> -int igt_main_args("csrn:", NULL, help_str, opt_handler, &data)
> +int igt_main_args("csrne:", NULL, help_str, opt_handler, &data)

Here's needed "cs:rne", this probably is actually my sin that has gone 
unnoticed previously when I added that 'r' option. Eg. seed option need 
the parameter while your new 'e' should fail complaining about missing 
argument.

With that fixed

Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

>   {
>   	igt_fixture() {
>   		data.drm_fd = drm_open_driver_master(DRIVER_INTEL | DRIVER_XE);