Re: [PATCH v4 22/32] drm/xe/survivability: Report 'Boot Mode enabled' status using SIGID

On 13-08-2026 16:31, Michal Wajdeczko wrote:
>
> On 8/13/2026 12:52 PM, Mallesh, Koujalagi wrote:
>> On 13-08-2026 12:44 am, Michal Wajdeczko wrote:
>>> Report 'Boot Mode' status or failure using various xe_log() helpers.
>>>
>>> Signed-off-by: Michal Wajdeczko <[email protected]>
>>> Cc: Rodrigo Vivi <[email protected]>
>>> Cc: Riana Tauro <[email protected]>
>>> Cc: Aravind Iddamsetty <[email protected]>
>>> Cc: Mallesh Koujalagi <[email protected]>
>>> ---
>>>    drivers/gpu/drm/xe/xe_survivability_mode.c | 21 +++++++++++++--------
>>>    1 file changed, 13 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/xe/xe_survivability_mode.c b/drivers/gpu/drm/xe/xe_survivability_mode.c
>>> index 2d8c532157fd..ebd288986c11 100644
>>> --- a/drivers/gpu/drm/xe/xe_survivability_mode.c
>>> +++ b/drivers/gpu/drm/xe/xe_survivability_mode.c
>>> @@ -304,14 +304,13 @@ static int create_survivability_sysfs(struct pci_dev *pdev)
>>>      static int enable_boot_survivability_mode(struct pci_dev *pdev)
>>>    {
>>> -    struct device *dev = &pdev->dev;
>>>        struct xe_device *xe = pdev_to_xe_device(pdev);
>>>        struct xe_survivability *survivability = &xe->survivability;
>>> -    int ret = 0;
>>> +    int ret;
>>>          ret = create_survivability_sysfs(pdev);
>>>        if (ret)
>>> -        return ret;
>>> +        goto failed;
>>>          /* Make sure xe_heci_gsc_init() and xe_i2c_probe() are aware of survivability */
>>>        survivability->mode = true;
>>> @@ -323,19 +322,25 @@ static int enable_boot_survivability_mode(struct pci_dev *pdev)
>>>        if (survivability->fdo_mode) {
>>>            ret = xe_nvm_init(xe);
>>>            if (ret)
>>> -            goto err;
>>> +            goto failed;
>>>        }
>>>          ret = xe_i2c_probe(xe);
>>>        if (ret)
>>> -        goto err;
>>> +        goto failed;
>>>    -    dev_err(dev, "In Survivability Mode\n");
>>> +    if (check_boot_failure(xe))
>>> +        xe_log_comp_fatal(pdev, SURVIVABILITY,
>>> +                  &survivability->boot_status,
>>> +                  sizeof(survivability->boot_status),
>>> +                  "Boot Mode enabled!\n");
Do we need this check? This function is only called if it is a boot failure.

We can have a error log with the type here. We don't need if else.

How about something like this? and remove else


<3> xe 0000:03:00.0: [drm] *ERROR* SIGID=<N> FATAL (01000000) 
SURVIVABILITY: mode=Boot
>> In case of fatal, will make sense to return "0" OR we can add return as -ENXIO right? any thoughts?

Mallesh, you cannot change return codes here. This defeats the purpose 
of survivability mode

> you're a reviewer here ;)
>
> but seriously, enable_boot_survivability_mode() is called from
> xe_survivability_mode_boot_enable() which says:
>
>   * Return: 0 if boot survivability mode is enabled or not requested, negative error
>   * code otherwise.
>
> so returning 0 as success code in enabling boot mode is the correct one
> returning -ENXIO will be no different than failing to enter boot mode
>
> @Riana, this is your code, can you please confirm that
>
>> After handling fatal case
>>
>> Reviewed-by: Mallesh Koujalagi <[email protected]>
>>
>>> +    else
>>> +        xe_log_info(pdev, SURVIVABILITY, "Boot Mode enabled!\n");
>>>          return 0;
>>>    -err:
>>> -    dev_err(dev, "Failed to enable Survivability Mode\n");
>>> +failed:
>>> +    xe_log_err_fatal(pdev, SURVIVABILITY, ret, "Failed to enable Boot Mode!\n");

Can we retain the previous dmesg?

Thanks
Riana

>>>        survivability->mode = false;
>>>        return ret;
>>>    }