public inbox for igt-dev@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Louis Chauvet <louis.chauvet@bootlin.com>
To: Kamil Konieczny <kamil.konieczny@linux.intel.com>,
	igt-dev@lists.freedesktop.org, thomas.petazzoni@bootlin.com,
	luca.ceresoli@bootlin.com, kory.maincent@bootlin.com,
	markyacoub@google.com, khaled.almahallawy@intel.com
Subject: Re: [PATCH i-g-t v9 03/49] lib/igt_kms: Add function to list connected connectors
Date: Tue, 24 Mar 2026 14:56:36 +0100	[thread overview]
Message-ID: <ac19d33d-f1ec-435c-bcf4-37cf048795be@bootlin.com> (raw)
In-Reply-To: <20260324103402.oslsmbm6zkglw2kg@kamilkon-DESK.igk.intel.com>



On 3/24/26 11:34, Kamil Konieczny wrote:
> Hi Louis,
> On 2026-03-16 at 17:17:24 +0100, Louis Chauvet wrote:
>> Introduce the igt_get_connected_connectors() function, which returns a
>> list of connector IDs that are currently connected according to DRM.
>>
>> This function includes a timeout mechanism because connectors can be
>> unplugged at any time. Also, especially with MST, the connector can be
>> listed by drmModeGetResources() but not yet accessible with
>> drmModeGetConnector().
>>
>> Reviewed-by: Kory Maincent <kory.maincent@bootlin.com>
>> Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
>> Signed-off-by: Louis Chauvet <louis.chauvet@bootlin.com>
>> ---
>>   lib/igt_kms.c | 80 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   lib/igt_kms.h |  1 +
>>   2 files changed, 81 insertions(+)
>>
>> diff --git a/lib/igt_kms.c b/lib/igt_kms.c
>> index 4a496ab256f0..72a38334f921 100644
>> --- a/lib/igt_kms.c
>> +++ b/lib/igt_kms.c
>> @@ -8094,6 +8094,31 @@ igt_crtc_t *igt_random_crtc(igt_display_t *display)
>>   	return crtcs[rand() % n];
>>   }
>>   
>> +static drmModeConnectorPtr igt_wait_for_connector(int drm_fd, unsigned int connector_id,
>> +						  double timeout)
>> +{
>> +	drmModeConnectorPtr connector = NULL;
>> +	struct timespec start, end;
>> +
>> +	connector = drmModeGetConnector(drm_fd, connector_id);
>> +	/*
>> +	 * This time is required as sometimes some id in the connector list are not totally
>> +	 * ready or can disappear
>> +	 */
>> +	igt_assert_eq(igt_gettime(&start), 0);
>> +	end = start;
>> +
>> +	while (!connector &&
>> +	       igt_time_elapsed(&start, &end) <= timeout) {
>> +		drmModeFreeConnector(connector);
>> +		connector = drmModeGetConnector(drm_fd, connector_id);
>> +		usleep(10000);
> 
> Same here, use defined constant.
> 
>> +		igt_assert_eq(igt_gettime(&end), 0);
>> +	}
>> +
>> +	return connector;
>> +}
>> +
>>   /**
>>    * igt_wait_for_connector_status:
>>    * @drm_fd: drm file descriptor
>> @@ -8134,3 +8159,58 @@ bool igt_wait_for_connector_status(int drm_fd, unsigned int connector_id, double
>>   
>>   	return false;
>>   }
>> +
>> +/**
>> + * igt_get_connected_connectors:
>> + *
>> + * @drm_fd: DRM file descriptor
>> + * @connector_ids: Out pointer for the array of connector ids connected. The function
>> + * will allocate and store the pointer in this out pointer. The caller is
>> + * responsible to free the inner pointer using free(*connector_ids). caller.
> 
> s/ caller.//
> 
>> + *
>> + * This will probe all the connectors and return the list of all
>> + * connected connectors.
> 
> Document that this could assert. Also add newline here.
> 
>> + * Returns: The number of connectors in @connector_ids
>> + */
>> +int igt_get_connected_connectors(int drm_fd, uint32_t **connector_ids)
>> +{
>> +	uint32_t *tmp_alloc;
>> +	int connected_count = 0;
>> +	double timeout = igt_default_display_detect_timeout();
>> +	drmModeResPtr resources;
>> +
>> +	igt_assert(drm_fd);
> 
> -1 is invalid but will not assert.
> 
>> +	*connector_ids = NULL;
>> +
>> +	resources = drmModeGetResources(drm_fd);
>> +	igt_assert(resources);
> 
> Cannot you just return 0 instead of asserting?
> Or at least use igt_assert_f() with explanation message.

If the drmModeGetResources fails, it means that there is a big kernel 
issue or that the drm_fd is not a DRM file descriptor. In either case I 
think it is useless to continue or try to handle the error in the test case.

>> +	for (int j = 0; j < resources->count_connectors; j++) {
>> +		int status = DRM_MODE_UNKNOWNCONNECTION;
>> +
>> +		drmModeConnectorPtr connector = igt_wait_for_connector(drm_fd,
>> +								       resources->connectors[j],
>> +								       timeout);
>> +
>> +		if (connector)
>> +			status = connector->connection;
>> +		drmModeFreeConnector(connector);
>> +
>> +		if (status == DRM_MODE_CONNECTED) {
>> +			tmp_alloc = reallocarray(*connector_ids,
>> +						 connected_count
>> +						 + 1, sizeof(**connector_ids));
> 
> Why this line breaks? imho better:
> 			tmp_alloc = reallocarray(*connector_ids,
> 						 connected_count + 1,
> 						 sizeof(**connector_ids));

I agree, it was probably the autowrap of my editor.

>> +			if (tmp_alloc) {
>> +				*connector_ids = tmp_alloc;
>> +			} else {
>> +				free(*connector_ids);
>> +				drmModeFreeResources(resources);
>> +				igt_assert(false);
> 
> Use igt_assert_f() with explanation message.
> 
>> +			}
> 
> Btw better flow is with error check first:
> 			if (!tmp_alloc) {
> 				...handle error here with assert
> 			}
> 
> 			*connector_ids = tmp_alloc;

Right, done for the next iteration.

> Regards,
> Kamil
> 
>> +			(*connector_ids)[connected_count] = resources->connectors[j];
>> +			connected_count++;
>> +		}
>> +	}
>> +	drmModeFreeResources(resources);
>> +
>> +	return connected_count;
>> +}
>> diff --git a/lib/igt_kms.h b/lib/igt_kms.h
>> index 934f24d5915b..4efd40a3ad08 100644
>> --- a/lib/igt_kms.h
>> +++ b/lib/igt_kms.h
>> @@ -1303,5 +1303,6 @@ igt_colorop_t *igt_find_colorop(igt_display_t *display, uint32_t id);
>>   
>>   bool igt_wait_for_connector_status(int drm_fd, unsigned int connector_id, double timeout,
>>   				   int drm_mode);
>> +int igt_get_connected_connectors(int drm_fd, uint32_t **connector_ids);
>>   
>>   #endif /* __IGT_KMS_H__ */
>>
>> -- 
>> 2.52.0
>>


  reply	other threads:[~2026-03-24 13:56 UTC|newest]

Thread overview: 83+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-16 16:17 [PATCH i-g-t v9 00/49] Unigraf integration Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 01/49] lib/igt_kms: Add a detect timeout value Louis Chauvet
2026-03-16 18:50   ` Kamil Konieczny
2026-03-16 16:17 ` [PATCH i-g-t v9 02/49] lib/igt_kms: Add helper to wait for a specific status on a connector Louis Chauvet
2026-03-24 10:22   ` Kamil Konieczny
2026-03-16 16:17 ` [PATCH i-g-t v9 03/49] lib/igt_kms: Add function to list connected connectors Louis Chauvet
2026-03-24 10:34   ` Kamil Konieczny
2026-03-24 13:56     ` Louis Chauvet [this message]
2026-03-16 16:17 ` [PATCH i-g-t v9 04/49] lib/igt_kms: Add helper to obtain a connector by its name or MST path Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 05/49] lib/igt_kms: Add helper to wait for new connectors Louis Chauvet
2026-03-17 14:56   ` Kory Maincent
2026-03-23 14:40     ` Louis Chauvet
2026-03-24 10:11   ` Kamil Konieczny
2026-03-16 16:17 ` [PATCH i-g-t v9 06/49] lib/tests: Add tests for array manipulations Louis Chauvet
2026-03-17 15:00   ` Kory Maincent
2026-03-16 16:17 ` [PATCH i-g-t v9 07/49] lib/igt_kms: Add helper to get a pipe from a connector Louis Chauvet
2026-03-24  7:58   ` Jani Nikula
2026-03-24 10:02   ` Kamil Konieczny
2026-03-16 16:17 ` [PATCH i-g-t v9 08/49] lib/igt_kms: Expose dump_connector_attrs Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 09/49] lib/igt_kms: Expose reset_connectors_at_exit Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 10/49] lib/igt_kms: Expose connector_attr_set and igt_connector_attr_set Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 11/49] lib/igt_debugfs: Move debugfs helpers to the proper location Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 12/49] lib/igt_debugfs: Add const when make sense Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 13/49] lib/igt_amd: " Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 14/49] lib/igt_kms: " Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 15/49] lib/monitor_edids: Add helper functions for using monitor_edid objects Louis Chauvet
2026-03-17 15:09   ` Kory Maincent
2026-03-24 10:15   ` Kamil Konieczny
2026-03-24 14:09     ` Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 16/49] lib/monitor_edids: Add helper to get an EDID by its name Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 17/49] lib/monitor_edids: Add helper to print all available EDID names Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 18/49] lib/unigraf: Add used defines for TSI_Types Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 19/49] lib/unigraf: Add TSI.h Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 20/49] lib/unigraf: Initial Unigraf support Louis Chauvet
2026-03-24  8:19   ` Naladala, Ramanaidu
2026-03-24  9:37     ` Louis Chauvet
2026-03-26  9:59       ` Louis Chauvet
2026-03-24  9:58   ` Kamil Konieczny
2026-03-16 16:17 ` [PATCH i-g-t v9 21/49] lib/igt_kms: Automatically connect unigraf on display require Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 22/49] lib/unigraf: Introduce device configuration Louis Chauvet
2026-03-17 15:16   ` Kory Maincent
2026-03-16 16:17 ` [PATCH i-g-t v9 23/49] lib/unigraf: Introduce role configuration Louis Chauvet
2026-03-17 15:19   ` Kory Maincent
2026-03-16 16:17 ` [PATCH i-g-t v9 24/49] lib/unigraf: Introduce input configuration Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 25/49] lib/unigraf: Add reset function Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 26/49] lib/unigraf: Add unigraf assert and deassert helpers Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 27/49] lib/unigraf: Add plug/unplug helpers Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 28/49] lib/unigraf: Allows sst/mst configuration Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 29/49] lib/unigraf: Add helpers to read and write edid Louis Chauvet
2026-03-17 15:32   ` Kory Maincent
2026-03-23 14:46     ` Louis Chauvet
2026-03-23 16:26   ` Kamil Konieczny
2026-03-23 17:34     ` Louis Chauvet
2026-03-24  9:35       ` Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 30/49] lib/unigraf: Add connector and EDID configuration Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 31/49] tests/unigraf: Add basic unigraf tests Louis Chauvet
2026-03-17 15:46   ` Kory Maincent
2026-03-23 14:49     ` Louis Chauvet
2026-03-24 19:25   ` Naladala, Ramanaidu
2026-03-16 16:17 ` [PATCH i-g-t v9 32/49] lib/unigraf: Add unigraf CRC capture Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 33/49] lib/unigraf: Add configuration for CRC usage Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 34/49] lib/unigraf: add unigraf_get_connector_by_stream Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 35/49] lib/unigraf: Add helper to check timings received by unigraf Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 36/49] lib/igt_pipe_crc: Add unigraf crc calculation Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 37/49] lib/i915/dp: Move DP-related function for i915 to proper folder Louis Chauvet
2026-03-16 16:17 ` [PATCH i-g-t v9 38/49] lib/i915/dp: Rename functions to avoid confusion Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 39/49] lib/i915/dp: Add helper to get maximum supported rate Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 40/49] lib/i915/dp: Properly check sscanf results Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 41/49] lib/i915/dp: Use igt_output_name instead of private field Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 42/49] lib/igt_dp: Create generic helpers for DP information Louis Chauvet
2026-03-17 15:55   ` Kory Maincent
2026-03-23 14:52     ` Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 43/49] lib/igt_kms: Add asserts to avoid null pointer dereference Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 44/49] lib/igt_kms: Add helper to get a CRTC from an output Louis Chauvet
2026-03-17 15:56   ` Kory Maincent
2026-03-16 16:18 ` [PATCH i-g-t v9 45/49] lib/unigraf: Add lane count configuration Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 46/49] docs/unigraf: Add unigraf documentation Louis Chauvet
2026-03-17 15:58   ` Kory Maincent
2026-03-24  8:07   ` Naladala, Ramanaidu
2026-03-16 16:18 ` [PATCH i-g-t v9 47/49] lib/unigraf: Add helpers to set maximum link rate Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 48/49] lib/unigraf: Add helpers to get the current LT status Louis Chauvet
2026-03-16 16:18 ` [PATCH i-g-t v9 49/49] tests/unigraf/unigraf_lt: Add test for link training Louis Chauvet
2026-03-17 12:02 ` ✗ Fi.CI.BUILD: failure for Unigraf integration (rev8) Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ac19d33d-f1ec-435c-bcf4-37cf048795be@bootlin.com \
    --to=louis.chauvet@bootlin.com \
    --cc=igt-dev@lists.freedesktop.org \
    --cc=kamil.konieczny@linux.intel.com \
    --cc=khaled.almahallawy@intel.com \
    --cc=kory.maincent@bootlin.com \
    --cc=luca.ceresoli@bootlin.com \
    --cc=markyacoub@google.com \
    --cc=thomas.petazzoni@bootlin.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox