Igt-dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Jani Nikula <jani.nikula@intel.com>
To: Arun R Murthy <arun.r.murthy@intel.com>, igt-dev@lists.freedesktop.org
Cc: karthik.b.s@intel.com, kunal1.joshi@intel.com,
	Arun R Murthy <arun.r.murthy@intel.com>
Subject: Re: [PATCH i-g-t 1/2] tests/intel/kms_joiner_helper: Add igt_get_free_pipes_mask()
Date: Wed, 07 Oct 2026 16:36:19 +0300	[thread overview]
Message-ID: <45b38494fa17817bbcbae211010c3161706f364a@intel.com> (raw)
In-Reply-To: <20261007-joiner-v1-1-b40826411fa8@intel.com>

On Wed, 07 Oct 2026, Arun R Murthy <arun.r.murthy@intel.com> wrote:
> The joiner tests derive the set of usable pipes from the pipe count
> alone, which assumes every pipe is free. A pipe already driven by
> another client is then handed to a modeset, making the commit fail.
>
> Add igt_get_free_pipes_mask(), which gets the current CRTC usage via
> drmModeGetCrtcList() and translates it into a bitmask of hardware
> pipes.
>
> drmModeGetCrtcList() is a recent libdrm addition, so probe for it in
> meson and fall back to plain CRTC enumeration when it is missing.
>
> Assisted-by: LLM
> Signed-off-by: Arun R Murthy <arun.r.murthy@intel.com>
> ---
>  meson.build                     |  7 +++++
>  tests/intel/kms_joiner_helper.c | 66 +++++++++++++++++++++++++++++++++++++++++
>  tests/intel/kms_joiner_helper.h |  1 +
>  3 files changed, 74 insertions(+)
>
> diff --git a/meson.build b/meson.build
> index d7bbfd9698e5616ab777ac58128192ba1e0bec23..f41588c9d73361dbfd405de5ea8f764c6768a9c9 100644
> --- a/meson.build
> +++ b/meson.build
> @@ -129,6 +129,13 @@ endif
>  
>  build_info += 'With libdrm: ' + ','.join(libdrm_info)
>  
> +# drmModeGetCrtcList() is a recent libdrm addition; keep building against
> +# older libdrm by falling back to plain resource enumeration.
> +if cc.has_function('drmModeGetCrtcList', dependencies : libdrm,
> +		   prefix : '#include <xf86drmMode.h>')
> +	config.set('HAVE_DRMMODEGETCRTCLIST', 1)
> +endif
> +
>  pciaccess = dependency('pciaccess', version : '>=0.10')
>  libkmod = dependency('libkmod')
>  # igt_procps.c reads /proc directly; libprocps and libproc2 are not needed.
> diff --git a/tests/intel/kms_joiner_helper.c b/tests/intel/kms_joiner_helper.c
> index 36168030475c2725130e6c8ad924dcfb971a07f6..235e227fe9d82f598d0bd7df57ff65d73d717ad0 100644
> --- a/tests/intel/kms_joiner_helper.c
> +++ b/tests/intel/kms_joiner_helper.c
> @@ -113,6 +113,72 @@ void igt_set_all_master_pipes_for_platform(igt_display_t *display, uint32_t *mas
>  	}
>  }
>  
> +/**
> + * igt_get_free_pipes_mask:
> + * @display: The display structure containing pipe information.
> + *
> + * Get the CRTCs that are currently usable and translate that into a
> + * bitmask of hardware pipes. A CRTC is reported free only when it is neither
> + * enabled through KMS nor active in hardware, so pipes claimed by another
> + * client (fbcon, a compositor, joiner) are never handed out to a modeset.
> + *
> + * The returned mask reflects the state at call time, so query it right before
> + * a modeset and after releasing the pipes a previous commit was using.
> + *
> + * Returns: bitmask of hardware pipes whose CRTC is free to be assigned. When
> + * libdrm lacks drmModeGetCrtcList() every pipe of @display is reported free.
> + */
> +uint32_t igt_get_free_pipes_mask(igt_display_t *display)
> +{
> +	uint32_t free_mask = 0;
> +#ifdef HAVE_DRMMODEGETCRTCLIST

#ifdefs belong outside of functions, not inside.

> +	drmModeCrtcListPtr list;
> +	int i;
> +
> +	list = drmModeGetCrtcList(display->drm_fd);
> +	igt_assert_f(list, "drmModeGetCrtcList() failed\n");
> +
> +	for (i = 0; i < list->count; i++) {
> +		drmModeCrtcInfoPtr info = &list->crtcs[i];
> +		igt_crtc_t *crtc = igt_crtc_for_crtc_id(display, info->crtc_id);
> +
> +		/* A CRTC the display never enumerated is not ours to use. */
> +		if (!crtc)
> +			continue;
> +
> +		if (info->in_use) {
> +			igt_debug("CRTC %s in use: %ux%u, fb %u\n",
> +				  igt_crtc_name(crtc), info->width,
> +				  info->height, info->buffer_id);
> +			continue;
> +		}
> +
> +		if (info->driver_active) {
> +			igt_debug("CRTC %s active in hardware, not enabled via KMS\n",
> +				  igt_crtc_name(crtc));
> +			continue;
> +		}
> +
> +		free_mask |= BIT(crtc->hardware_pipe);
> +	}
> +
> +	igt_info("%d of %d CRTCs free, free pipe mask 0x%x\n",
> +		 list->count_available, list->count, free_mask);
> +
> +	drmModeFreeCrtcList(list);
> +#else
> +	igt_crtc_t *crtc;
> +
> +	for_each_crtc(display, crtc)
> +		free_mask |= BIT(crtc->hardware_pipe);
> +
> +	igt_debug("libdrm has no drmModeGetCrtcList(), assuming pipe mask 0x%x free\n",
> +		  free_mask);
> +#endif
> +
> +	return free_mask;
> +}
> +
>  /*
>   * @drm_fd: DRM file descriptor
>   * @outputs: array of pointers to igt_output_t
> diff --git a/tests/intel/kms_joiner_helper.h b/tests/intel/kms_joiner_helper.h
> index a0ab75b2c165084b643ead3f3ffc05cf5364e90b..14083d25c6f09e0577f123130697c34c2b6a15e4 100644
> --- a/tests/intel/kms_joiner_helper.h
> +++ b/tests/intel/kms_joiner_helper.h
> @@ -11,6 +11,7 @@
>  
>  void igt_set_all_master_pipes_for_platform(igt_display_t *display,
>  					   uint32_t *master_pipes);
> +uint32_t igt_get_free_pipes_mask(igt_display_t *display);
>  bool igt_assign_pipes_for_outputs(int drm_fd,
>  				  igt_output_t **outputs,
>  				  int num_outputs,

-- 
Jani Nikula, Intel

  parent reply	other threads:[~2026-10-07 13:37 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-07  9:40 [PATCH i-g-t 0/2] EDITME: Update pipe joiner test to get free crtc Arun R Murthy
2026-10-07  9:40 ` [PATCH i-g-t 1/2] tests/intel/kms_joiner_helper: Add igt_get_free_pipes_mask() Arun R Murthy
2026-10-07 13:34   ` Jani Nikula
2026-10-07 15:39     ` Murthy, Arun R
2026-10-07 13:36   ` Jani Nikula [this message]
2026-10-07 15:43     ` Murthy, Arun R
2026-10-07 16:19       ` Jani Nikula
2026-10-08  4:13         ` Murthy, Arun R
2026-10-07  9:40 ` [PATCH i-g-t 2/2] tests/intel/kms_joiner: Use the CRTCs reported free before a modeset Arun R Murthy
2026-10-07 10:33 ` ✓ i915.CI.BAT: success for EDITME: Update pipe joiner test to get free crtc Patchwork
2026-10-07 10:49 ` ✓ Xe.CI.BAT: " Patchwork
2026-10-07 12:06 ` ✓ Xe.CI.FULL: " Patchwork
2026-10-07 12:49 ` ✗ i915.CI.Full: failure " 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=45b38494fa17817bbcbae211010c3161706f364a@intel.com \
    --to=jani.nikula@intel.com \
    --cc=arun.r.murthy@intel.com \
    --cc=igt-dev@lists.freedesktop.org \
    --cc=karthik.b.s@intel.com \
    --cc=kunal1.joshi@intel.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