From: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
To: Zhanjun Dong <zhanjun.dong@intel.com>, <intel-xe@lists.freedesktop.org>
Subject: Re: [PATCH v2 1/1] drm/xe/gsc: Fix GSC proxy cleanup on early initialization failure
Date: Tue, 27 Jan 2026 15:32:26 -0800 [thread overview]
Message-ID: <b6736da7-a430-43b5-bf48-d09a2af79af0@intel.com> (raw)
In-Reply-To: <20260127024315.359176-1-zhanjun.dong@intel.com>
On 1/26/2026 6:43 PM, Zhanjun Dong wrote:
> xe_gsc_proxy_remove undoes what is done in both xe_gsc_proxy_init and
> xe_gsc_proxy_start; however, if we fail between those 2 calls, it is
> possible that the HW forcewake access hasn't been initialized yet and so
> we hit errors when the cleanup code tries to write GSC register. To
> avoid that, split the cleanup in 2 functions so that the HW cleanup is
> only called if the HW setup was completed successfully.
>
> Additionally, fix error handling in xe_gsc_proxy_start to properly
> disable interrupts on failure paths before returning, ensuring cleanup
> is performed correctly when xe_gsc_proxy_request_handler() or
> xe_gsc_proxy_init_done() fails.
This second one is not a fix. In the current behavior, if
xe_gsc_proxy_start fails then xe_gsc_proxy_remove is called and the
cleanup is performed correctly. Since you're removing that part of the
cleanup from xe_gsc_proxy_remove then this new change to
xe_gsc_proxy_start becomes necessary.
>
> Fixes: ff6cd29b690b ("drm/xe: Cleanup unwind of gt initialization")
> Signed-off-by: Zhanjun Dong <zhanjun.dong@intel.com>
> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
> ---
> v2:
> - Split cleanup into two functions: xe_gsc_proxy_remove() for SW cleanup
> and xe_gsc_proxy_stop() for HW cleanup that requires forcewake access.
> - Add error handling in xe_gsc_proxy_start to disable interrupts on
> early error exits.
> ---
> drivers/gpu/drm/xe/xe_gsc_proxy.c | 42 ++++++++++++++++++++-----------
> 1 file changed, 28 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/xe_gsc_proxy.c b/drivers/gpu/drm/xe/xe_gsc_proxy.c
> index 42438b21f235..0f62ee7dab4a 100644
> --- a/drivers/gpu/drm/xe/xe_gsc_proxy.c
> +++ b/drivers/gpu/drm/xe/xe_gsc_proxy.c
> @@ -444,16 +444,6 @@ static void xe_gsc_proxy_remove(void *arg)
> if (!gsc->proxy.component_added)
> return;
>
> - /* disable HECI2 IRQs */
> - scoped_guard(xe_pm_runtime, xe) {
> - CLASS(xe_force_wake, fw_ref)(gt_to_fw(gt), XE_FW_GSC);
> - if (!fw_ref.domains)
> - xe_gt_err(gt, "failed to get forcewake to disable GSC interrupts\n");
> -
> - /* try do disable irq even if forcewake failed */
> - gsc_proxy_irq_toggle(gsc, false);
> - }
> -
> xe_gsc_wait_for_worker_completion(gsc);
in v1 you had xe_gsc_wait_for_worker_completion moved to
xe_gsc_proxy_stop(). I think that works better, since the worker
shouldn't be queued anymore after we disable interrupts.
Daniele
>
> component_del(xe->drm.dev, &xe_gsc_proxy_component_ops);
> @@ -502,6 +492,23 @@ int xe_gsc_proxy_init(struct xe_gsc *gsc)
> return devm_add_action_or_reset(xe->drm.dev, xe_gsc_proxy_remove, gsc);
> }
>
> +static void xe_gsc_proxy_stop(void *arg)
> +{
> + struct xe_gsc *gsc = arg;
> + struct xe_gt *gt = gsc_to_gt(gsc);
> + struct xe_device *xe = gt_to_xe(gt);
> +
> + /* disable HECI2 IRQs */
> + scoped_guard(xe_pm_runtime, xe) {
> + CLASS(xe_force_wake, fw_ref)(gt_to_fw(gt), XE_FW_GSC);
> + if (!fw_ref.domains)
> + xe_gt_err(gt, "failed to get forcewake to disable GSC interrupts\n");
> +
> + /* try do disable irq even if forcewake failed */
> + gsc_proxy_irq_toggle(gsc, false);
> + }
> +}
> +
> /**
> * xe_gsc_proxy_start() - start the proxy by submitting the first request
> * @gsc: the GSC uC
> @@ -510,6 +517,8 @@ int xe_gsc_proxy_init(struct xe_gsc *gsc)
> */
> int xe_gsc_proxy_start(struct xe_gsc *gsc)
> {
> + struct xe_gt *gt = gsc_to_gt(gsc);
> + struct xe_device *xe = gt_to_xe(gt);
> int err;
>
> /* enable the proxy interrupt in the GSC shim layer */
> @@ -521,12 +530,17 @@ int xe_gsc_proxy_start(struct xe_gsc *gsc)
> */
> err = xe_gsc_proxy_request_handler(gsc);
> if (err)
> - return err;
> + goto err_irq_disable;
>
> if (!xe_gsc_proxy_init_done(gsc)) {
> - xe_gt_err(gsc_to_gt(gsc), "GSC FW reports proxy init not completed\n");
> - return -EIO;
> + xe_gt_err(gt, "GSC FW reports proxy init not completed\n");
> + err = -EIO;
> + goto err_irq_disable;
> }
>
> - return 0;
> + return devm_add_action_or_reset(xe->drm.dev, xe_gsc_proxy_stop, gsc);
> +
> +err_irq_disable:
> + gsc_proxy_irq_toggle(gsc, false);
> + return err;
> }
next prev parent reply other threads:[~2026-01-27 23:32 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-01-27 2:43 [PATCH v2 1/1] drm/xe/gsc: Fix GSC proxy cleanup on early initialization failure Zhanjun Dong
2026-01-27 2:50 ` ✓ CI.KUnit: success for series starting with [v2,1/1] " Patchwork
2026-01-27 5:14 ` ✗ Xe.CI.Full: failure " Patchwork
2026-01-27 23:32 ` Daniele Ceraolo Spurio [this message]
2026-01-28 21:52 ` [PATCH v2 1/1] " Dong, Zhanjun
2026-01-28 8:05 ` ✗ Xe.CI.Full: failure for series starting with [v2,1/1] " Patchwork
2026-01-28 13:01 ` ✓ CI.KUnit: success for series starting with [v2,1/1] drm/xe/gsc: Fix GSC proxy cleanup on early initialization failure (rev2) Patchwork
2026-01-28 13:52 ` ✓ Xe.CI.BAT: " 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=b6736da7-a430-43b5-bf48-d09a2af79af0@intel.com \
--to=daniele.ceraolospurio@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=zhanjun.dong@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