From: Matthew Auld <matthew.auld@intel.com>
To: Badal Nilawar <badal.nilawar@intel.com>,
intel-xe@lists.freedesktop.org, linux-acpi@vger.kernel.org,
linux-pci@vger.kernel.org
Cc: anshuman.gupta@intel.com, rafael@kernel.org, lenb@kernel.org,
bhelgaas@google.com, ilpo.jarvinen@linux.intel.com,
lucas.demarchi@intel.com, rodrigo.vivi@intel.com,
varun.gupta@intel.com, ville.syrjala@linux.intel.com,
uma.shankar@intel.com, karthik.poosa@intel.com
Subject: Re: [PATCH v5 11/12] drm/xe/pm/s2idle: Don't evict user BOs for D3hot and D3cold-VRSR state
Date: Fri, 17 Oct 2025 12:00:46 +0100 [thread overview]
Message-ID: <749a5f42-f1eb-488b-826a-ca9b6981b048@intel.com> (raw)
In-Reply-To: <20251015080710.1468409-12-badal.nilawar@intel.com>
On 15/10/2025 09:07, Badal Nilawar wrote:
> In D3hot and D3cold-VRSR, VRAM remains active, so it is not necessary
> to evict user BOs.
>
> Cc: Matthew Auld <matthew.auld@intel.com>
> Co-developed-by: Sk Anirban <sk.anirban@intel.com>
> Signed-off-by: Sk Anirban <sk.anirban@intel.com>
> Signed-off-by: Badal Nilawar <badal.nilawar@intel.com>
> ---
> drivers/gpu/drm/xe/xe_pm.c | 15 ++++++++++-----
> 1 file changed, 10 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/xe_pm.c b/drivers/gpu/drm/xe/xe_pm.c
> index 84cd9e84f548..51ff9b90fbfd 100644
> --- a/drivers/gpu/drm/xe/xe_pm.c
> +++ b/drivers/gpu/drm/xe/xe_pm.c
> @@ -169,7 +169,7 @@ static void xe_rpm_lockmap_release(const struct xe_device *xe)
>
> static void xe_pm_suspend_prepare(struct xe_device *xe)
> {
> - if (pm_suspend_target_state == PM_SUSPEND_TO_IDLE)
> + if (pm_suspend_default_s2idle())
> xe_pm_d3cold_target_state_toggle(xe);
> else
> xe->d3cold.target_state = XE_D3COLD_OFF;
> @@ -191,8 +191,6 @@ int xe_pm_suspend(struct xe_device *xe)
> xe_pm_block_begin_signalling();
> trace_xe_pm_suspend(xe, __builtin_return_address(0));
>
> - xe_pm_suspend_prepare(xe);
> -
> err = xe_pxp_pm_suspend(xe->pxp);
> if (err)
> goto err;
> @@ -543,8 +541,12 @@ static int xe_pm_notifier_callback(struct notifier_block *nb,
> int err = 0;
>
> switch (action) {
> - case PM_HIBERNATION_PREPARE:
> case PM_SUSPEND_PREPARE:
> + xe_pm_suspend_prepare(xe);
> + if (xe->d3cold.target_state != XE_D3COLD_OFF)
> + break;
I think this will break hibernation on igpu platforms with flat-CSS? We
need evict_all() to move CCS from stolen memory back to normal RAM such
that it is saved and then restored correctly on resume from hibernation.
What is the value of target_state for a hibernation event? Does the
evict_all() still get called later? AFAICT it is not called any more.
> + fallthrough;
> + case PM_HIBERNATION_PREPARE:
> {
> struct xe_validation_ctx ctx;
>
> @@ -569,8 +571,11 @@ static int xe_pm_notifier_callback(struct notifier_block *nb,
> xe_pm_block_end_signalling();
> break;
> }
> - case PM_POST_HIBERNATION:
> case PM_POST_SUSPEND:
> + if (xe->d3cold.target_state != XE_D3COLD_OFF)
> + break;
> + fallthrough;
> + case PM_POST_HIBERNATION:
> complete_all(&xe->pm_block);
> xe_pm_wake_rebind_workers(xe);
> xe_bo_notifier_unprepare_all_pinned(xe);
next prev parent reply other threads:[~2025-10-17 11:00 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-15 8:06 [PATCH v5 00/12] VRAM Self Refresh Badal Nilawar
2025-10-15 8:06 ` [PATCH v5 01/12] PCI/ACPI: Add D3cold Aux Power Limit_DSM method Badal Nilawar
2025-10-27 19:03 ` Rafael J. Wysocki
2025-10-15 8:07 ` [PATCH v5 02/12] PCI/ACPI: Per Root/Switch Downstream Port allow one aux power limit request Badal Nilawar
2025-10-27 19:11 ` Rafael J. Wysocki
2025-12-17 14:45 ` Rafael J. Wysocki
2025-10-15 8:07 ` [PATCH v5 03/12] PCI/ACPI: Add PERST# Assertion Delay _DSM method Badal Nilawar
2025-10-27 19:18 ` Rafael J. Wysocki
2025-10-15 8:07 ` [PATCH v5 04/12] drm/xe/vrsr: Introduce flag has_vrsr Badal Nilawar
2025-10-15 8:07 ` [PATCH v5 05/12] drm/xe/vrsr: Detect VRSR Capability Badal Nilawar
2025-10-15 8:07 ` [PATCH v5 06/12] drm/xe/vrsr: Initialize VRSR feature Badal Nilawar
2025-10-15 8:07 ` [PATCH v5 07/12] drm/xe/vrsr: Enable VRSR on default VGA boot device Badal Nilawar
2025-10-15 8:07 ` [PATCH v5 08/12] drm/xe/vrsr: Refactor d3cold.allowed to a enum Badal Nilawar
2025-10-15 8:39 ` Raag Jadav
2025-10-15 9:04 ` Nilawar, Badal
2025-10-15 9:08 ` Nilawar, Badal
2025-10-15 9:08 ` Gupta, Anshuman
2025-10-15 8:07 ` [PATCH v5 09/12] drm/xe/pm: D3cold target state Badal Nilawar
2025-10-15 8:07 ` [PATCH v5 10/12] drm/xe/vrsr: Enable VRSR Badal Nilawar
2025-10-15 8:07 ` [PATCH v5 11/12] drm/xe/pm/s2idle: Don't evict user BOs for D3hot and D3cold-VRSR state Badal Nilawar
2025-10-17 11:00 ` Matthew Auld [this message]
2025-10-15 8:07 ` [PATCH v5 12/12] drm/xe/vrsr: Introduce a debugfs node named vrsr_capable Badal Nilawar
2025-10-15 8:43 ` Poosa, Karthik
2025-10-15 11:10 ` ✓ CI.KUnit: success for VRAM Self Refresh (rev5) Patchwork
2025-10-15 11:25 ` ✗ CI.checksparse: warning " Patchwork
2025-10-15 12:05 ` ✓ Xe.CI.BAT: success " Patchwork
2025-10-15 21:29 ` ✗ Xe.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=749a5f42-f1eb-488b-826a-ca9b6981b048@intel.com \
--to=matthew.auld@intel.com \
--cc=anshuman.gupta@intel.com \
--cc=badal.nilawar@intel.com \
--cc=bhelgaas@google.com \
--cc=ilpo.jarvinen@linux.intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=karthik.poosa@intel.com \
--cc=lenb@kernel.org \
--cc=linux-acpi@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=lucas.demarchi@intel.com \
--cc=rafael@kernel.org \
--cc=rodrigo.vivi@intel.com \
--cc=uma.shankar@intel.com \
--cc=varun.gupta@intel.com \
--cc=ville.syrjala@linux.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