Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
To: Nareshkumar Gollakoti <naresh.kumar.g@intel.com>,
	<intel-xe@lists.freedesktop.org>
Cc: <stuart.summers@intel.com>
Subject: Re: [PATCH V4] drm/xe/: Mutual Exclusivity b/w Multi CCS Mode & SRIOV VF Provisioning
Date: Tue, 7 Oct 2025 13:29:37 +0200	[thread overview]
Message-ID: <449707fb-8701-4806-a301-c511d1894957@intel.com> (raw)
In-Reply-To: <20251006115013.1305805-2-naresh.kumar.g@intel.com>



On 10/6/2025 1:50 PM, Nareshkumar Gollakoti wrote:
> Due to SLA agreement between PF and VFs, multi CCS mode can't
> be enabled when VFs are already enabled.
> Similarly, enabling VFs is disabled when multi CCS mode enabled.

s/is disabled/must be blocked
> 
> v2:
> - function xe_device_is_vf_enabled has been refactored to
>   xe_sriov_pf_has_vfs_enabled and moved to xe_sriov_pf_helper.h.
> - The code now distinctly checks for SR-IOV VF mode and
>   SR-IOV PF with VFs enabled.
> - Log messages have been updated to explicitly state the current mode.
> - The function xe_multi_ccs_mode_enabled is moved to xe_device.h
> 
> v3: Described missed arg documentation for xe_sriov_pf_has_vfs_enabled
> 
> v4:
> - sysfs interface for CCS mode is not initialized
>   when operating in SRIOV VF Mode.
> - xe_sriov_pf_has_vfs_enabled() check is sufficient while CCS mode
>   enablement.
> - remove unnecessary comments as flow is self explanatory.
> 
> Signed-off-by: Nareshkumar Gollakoti <naresh.kumar.g@intel.com>
> ---
>  drivers/gpu/drm/xe/xe_device.h           |  8 ++++++++
>  drivers/gpu/drm/xe/xe_gt_ccs_mode.c      | 10 ++++++----
>  drivers/gpu/drm/xe/xe_pci_sriov.c        |  5 +++++
>  drivers/gpu/drm/xe/xe_sriov_pf_helpers.h | 13 +++++++++++++
>  4 files changed, 32 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_device.h b/drivers/gpu/drm/xe/xe_device.h
> index 32cc6323b7f6..986f9cabb897 100644
> --- a/drivers/gpu/drm/xe/xe_device.h
> +++ b/drivers/gpu/drm/xe/xe_device.h
> @@ -172,6 +172,14 @@ static inline bool xe_device_has_lmtt(struct xe_device *xe)
>  	return IS_DGFX(xe);
>  }
>  
> +static inline bool xe_multi_ccs_mode_enabled(struct xe_device *xe)
> +{
> +	/* Multi CCS mode supported exclusively on GT0 */
> +	struct xe_gt *gt = xe_device_get_gt(xe, 0);
> +
> +	return gt->ccs_mode > 1;
> +}
> +
>  u32 xe_device_ccs_bytes(struct xe_device *xe, u64 size);
>  
>  void xe_device_snapshot_print(struct xe_device *xe, struct drm_printer *p);
> diff --git a/drivers/gpu/drm/xe/xe_gt_ccs_mode.c b/drivers/gpu/drm/xe/xe_gt_ccs_mode.c
> index 50fffc9ebf62..91908faf3768 100644
> --- a/drivers/gpu/drm/xe/xe_gt_ccs_mode.c
> +++ b/drivers/gpu/drm/xe/xe_gt_ccs_mode.c
> @@ -13,6 +13,7 @@
>  #include "xe_gt_sysfs.h"
>  #include "xe_mmio.h"
>  #include "xe_sriov.h"
> +#include "xe_sriov_pf_helpers.h"

beware that most of the helpers in this header are for PF-only, so including them in non-PF code must be done with care

>  
>  static void __xe_gt_apply_ccs_mode(struct xe_gt *gt, u32 num_engines)
>  {
> @@ -117,9 +118,8 @@ ccs_mode_store(struct device *kdev, struct device_attribute *attr,
>  	u32 num_engines, num_slices;
>  	int ret;
>  
> -	if (IS_SRIOV(xe)) {
> -		xe_gt_dbg(gt, "Can't change compute mode when running as %s\n",
> -			  xe_sriov_mode_to_string(xe_device_sriov_mode(xe)));
> +	if (xe_sriov_pf_has_vfs_enabled(xe)) {
> +		xe_gt_dbg(gt, "Can't change CCS mode in SR-IOV PF Mode with VFs Enabled\n");
>  		return -EOPNOTSUPP;

this is still potentially racy, can you consider similar approach as it was done in eudebug [1] which has the same problem but is trying to close the race by using a flag protected by a mutex ?

[1] https://patchwork.freedesktop.org/patch/678636/?series=155452&rev=1

>  	}
>  
> @@ -184,6 +184,8 @@ static void xe_gt_ccs_mode_sysfs_fini(void *arg)
>   * The number of available compute slices is exposed to user through a per-gt
>   * 'num_cslices' sysfs interface.
>   *
> + * The sysfs interface for CCS mode is not set up in SRIOV VF Mode.
> + *
>   * Returns: Returns error value for failure and 0 for success.
>   */
>  int xe_gt_ccs_mode_sysfs_init(struct xe_gt *gt)
> @@ -191,7 +193,7 @@ int xe_gt_ccs_mode_sysfs_init(struct xe_gt *gt)
>  	struct xe_device *xe = gt_to_xe(gt);
>  	int err;
>  
> -	if (!xe_gt_ccs_mode_enabled(gt))
> +	if (!xe_gt_ccs_mode_enabled(gt) || IS_SRIOV_VF(xe))
>  		return 0;
>  
>  	err = sysfs_create_files(gt->sysfs, gt_ccs_mode_attrs);
> diff --git a/drivers/gpu/drm/xe/xe_pci_sriov.c b/drivers/gpu/drm/xe/xe_pci_sriov.c
> index 9c1c9e669b04..0715f87a5ecc 100644
> --- a/drivers/gpu/drm/xe/xe_pci_sriov.c
> +++ b/drivers/gpu/drm/xe/xe_pci_sriov.c
> @@ -153,6 +153,11 @@ static int pf_enable_vfs(struct xe_device *xe, int num_vfs)
>  	xe_assert(xe, num_vfs <= total_vfs);
>  	xe_sriov_dbg(xe, "enabling %u VF%s\n", num_vfs, str_plural(num_vfs));
>  
> +	if (xe_multi_ccs_mode_enabled(xe)) {
> +		xe_sriov_info(xe, "Disable multi-CCS mode before enabling VF's\n");
> +		return -ECANCELED;
> +	}
> +
>  	err = xe_sriov_pf_wait_ready(xe);
>  	if (err)
>  		goto out;
> diff --git a/drivers/gpu/drm/xe/xe_sriov_pf_helpers.h b/drivers/gpu/drm/xe/xe_sriov_pf_helpers.h
> index dd1df950b021..e26837091375 100644
> --- a/drivers/gpu/drm/xe/xe_sriov_pf_helpers.h
> +++ b/drivers/gpu/drm/xe/xe_sriov_pf_helpers.h
> @@ -43,4 +43,17 @@ static inline struct mutex *xe_sriov_pf_master_mutex(struct xe_device *xe)
>  	return &xe->sriov.pf.master_lock;
>  }
>  
> +/**
> + * xe_sriov_pf_has_vfs_enabled() - Determines if the PF has any VFs enabled
> + * @xe: ptr to xe_device
> + *
> + * Return: true if one or more VFs are enabled on the PF, false otherwise.
> + */
> +static inline bool xe_sriov_pf_has_vfs_enabled(const struct xe_device *xe)
> +{
> +	struct pci_dev *pdev = to_pci_dev(xe->drm.dev);
> +
> +	return pci_num_vf(pdev) > 0;
> +}
> +
>  #endif


  parent reply	other threads:[~2025-10-07 11:29 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-06 11:50 [PATCH V4] drm/xe/: Mutual Exclusivity b/w Multi CCS Mode & SRIOV VF Provisioning Nareshkumar Gollakoti
2025-10-06 13:20 ` ✓ CI.KUnit: success for drm/xe/: Mutual Exclusivity b/w Multi CCS Mode & SRIOV VF Provisioning (rev5) Patchwork
2025-10-06 13:56 ` ✓ Xe.CI.BAT: " Patchwork
2025-10-06 17:26 ` ✗ Xe.CI.Full: failure " Patchwork
2025-10-07 11:29 ` Michal Wajdeczko [this message]
  -- strict thread matches above, loose matches on Subject: below --
2025-10-06  8:45 [PATCH V4] drm/xe/: Mutual Exclusivity b/w Multi CCS Mode & SRIOV VF Provisioning Nareshkumar Gollakoti

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=449707fb-8701-4806-a301-c511d1894957@intel.com \
    --to=michal.wajdeczko@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=naresh.kumar.g@intel.com \
    --cc=stuart.summers@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