[PATCH v6 04/25] drm/xe: Modify xe_force_wake_put to handle _get returned mask

Nilawar, Badal badal.nilawar at intel.com
Fri Oct 4 03:21:54 UTC 2024



On 01-10-2024 10:41, Ghimiray, Himal Prasad wrote:
> 
> 
> On 01-10-2024 01:45, Michal Wajdeczko wrote:
>>
>>
>> On 30.09.2024 22:13, Michal Wajdeczko wrote:
>>>
>>>
>>> On 30.09.2024 07:31, Himal Prasad Ghimiray wrote:
>>>> Instead of calling xe_force_wake_put on all domains that were input to
>>>> xe_force_wake_get, call _put only on the domains whose reference counts
>>>> were successfully incremented by the _get call. Since the return value
>>>> of _get can be a mask that does not match any specific value in the 
>>>> enum
>>>> xe_force_wake_domains, change the input parameter of _put to 
>>>> unsigned int.
>>>
>>> at this point in the series xe_force_wake_get() still returns errno
>>
>> oops, scratch that, somehow missed patch 03/25
>>
>>>
>>> maybe rephrase this to something like:
>>>
>>> "Prepare xe_force_wake_put() to accept force_wake reference with
>>> refcounted domains"
>>>
>>> and also maybe it is a good time to change this function to void (or
>>> maybe even do it before this path?
> 
> Changing this to void here or before this patch will break the
> compilation, since the cleanup is in subsequent patches. Therefore,
> patch 25/25 addresses changing it to void after all the cleanups.
> 
>>>
>>>>
>>>> v3
>>>> - Move WARN to this patch (Badal)
>>>> - use xe_gt_WARN instead of XE_WARN (Michal)
>>>> - Stop using xe_force_wake_domains for non enum values.
>>>> - Remove kernel-doc from this patch (Badal)
>>>>
>>>> -v5
>>>> - Fix global awake_domain
>>>>
>>>> -v6
>>>> - put all initialized domains in case of FORCEWAKE_ALL.
>>>> - Modify ret variable name (Michal)
>>>> - Modify input var name (Michal)
>>>> - Modify commit message and warn (Badal)
>>>>
>>>> Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
>>>> Cc: Badal Nilawar <badal.nilawar at intel.com>
>>>> Cc: Rodrigo Vivi <rodrigo.vivi at intel.com>
>>>> Cc: Lucas De Marchi <lucas.demarchi at intel.com>
>>>> Cc: Nirmoy Das <nirmoy.das at intel.com>
>>>> Reviewed-by: Badal Nilawar <badal.nilawar at intel.com>
>>>
>>> probably this applies to some earlier #rev, so please mark it as such
> 
> Sure. Miss at my end. Will do in this patch and others.

Changes in this patch looks good to me. Please maintain my RB.

Regards,
Badal

> 
>>>
>>>> Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray at intel.com>
>>>> ---
>>>>   drivers/gpu/drm/xe/xe_force_wake.c | 28 +++++++++++++++++++++-------
>>>>   drivers/gpu/drm/xe/xe_force_wake.h |  2 +-
>>>>   2 files changed, 22 insertions(+), 8 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/xe/xe_force_wake.c b/drivers/gpu/drm/ 
>>>> xe/xe_force_wake.c
>>>> index 7f358e42c5d4..372ea43b0d06 100644
>>>> --- a/drivers/gpu/drm/xe/xe_force_wake.c
>>>> +++ b/drivers/gpu/drm/xe/xe_force_wake.c
>>>> @@ -211,26 +211,40 @@ unsigned int xe_force_wake_get(struct 
>>>> xe_force_wake *fw,
>>>>   }
>>>>   int xe_force_wake_put(struct xe_force_wake *fw,
>>>> -              enum xe_force_wake_domains domains)
>>>> +              unsigned int fw_ref)
>>>>   {
>>>>       struct xe_gt *gt = fw->gt;
>>>>       struct xe_force_wake_domain *domain;
>>>> -    enum xe_force_wake_domains tmp, sleep = 0;
>>>> +    unsigned int tmp, sleep = 0;
>>>>       unsigned long flags;
>>>> -    int ret = 0;
>>>> +    int ack_fail = 0;
>>>> +
>>>> +    /*
>>>> +     * Avoid unnecessary lock and unlock when the function is called
>>>> +     * in error path of individual domains.
>>>> +     */
>>>> +    if (!fw_ref)
>>>> +        return 0;
>>>> +
>>>> +    if (fw_ref == XE_FORCEWAKE_ALL)
>>>> +        fw_ref = fw->initialized_domains;
>>>>       spin_lock_irqsave(&fw->lock, flags);
>>>> -    for_each_fw_domain_masked(domain, domains, fw, tmp) {
>>>> +    for_each_fw_domain_masked(domain, fw_ref, fw, tmp) {
>>>>           if (!--domain->ref) {
>>>>               sleep |= BIT(domain->id);
>>>>               domain_sleep(gt, domain);
>>>>           }
>>>>       }
>>>>       for_each_fw_domain_masked(domain, sleep, fw, tmp) {
>>>> -        ret |= domain_sleep_wait(gt, domain);
>>>> +        if (domain_sleep_wait(gt, domain) == 0)
>>>> +            fw->awake_domains &= ~BIT(domain->id);
>>>> +        else
>>>> +            ack_fail |= BIT(domain->id);
>>>>       }
>>>> -    fw->awake_domains &= ~sleep;
>>>>       spin_unlock_irqrestore(&fw->lock, flags);
>>>> -    return ret;
>>>> +    xe_gt_WARN(gt, ack_fail, "Forcewake domain%s %#x failed to 
>>>> acknowledge sleep request\n",
>>>> +           str_plural(hweight_long(ack_fail)), ack_fail);
>>>> +    return ack_fail;
>>>>   }
>>>> diff --git a/drivers/gpu/drm/xe/xe_force_wake.h b/drivers/gpu/drm/ 
>>>> xe/xe_force_wake.h
>>>> index eb638128952d..b5a75544d86a 100644
>>>> --- a/drivers/gpu/drm/xe/xe_force_wake.h
>>>> +++ b/drivers/gpu/drm/xe/xe_force_wake.h
>>>> @@ -18,7 +18,7 @@ void xe_force_wake_init_engines(struct xe_gt *gt,
>>>>   unsigned int xe_force_wake_get(struct xe_force_wake *fw,
>>>>                      enum xe_force_wake_domains domains);
>>>>   int xe_force_wake_put(struct xe_force_wake *fw,
>>>> -              enum xe_force_wake_domains domains);
>>>> +              unsigned int fw_ref);
>>>>   static inline int
>>>>   xe_force_wake_ref(struct xe_force_wake *fw,
>>>
>>
> 



More information about the Intel-xe mailing list