[PATCH 2/7] staging/android: display sync_pt name on debugfs

Maarten Lankhorst maarten.lankhorst at linux.intel.com
Tue Aug 9 10:04:49 UTC 2016


Op 08-08-16 om 21:59 schreef Gustavo Padovan:
> 2016-08-08 Maarten Lankhorst <maarten.lankhorst at linux.intel.com>:
>
>> Op 20-06-16 om 17:53 schreef Gustavo Padovan:
>>> From: Gustavo Padovan <gustavo.padovan at collabora.co.uk>
>>>
>>> When creating a sync_pt the name received wasn't used anywhere.
>>> Now we add it to the sync info debug output to make it easier to indetify
>>> the userspace name of that sync pt.
>>>
>>> Signed-off-by: Gustavo Padovan <gustavo.padovan at collabora.co.uk>
>>> ---
>>>  drivers/staging/android/sw_sync.c    | 16 ++++------------
>>>  drivers/staging/android/sync_debug.c |  5 +++--
>>>  drivers/staging/android/sync_debug.h |  9 +++++++++
>>>  3 files changed, 16 insertions(+), 14 deletions(-)
>>>
>>> diff --git a/drivers/staging/android/sw_sync.c b/drivers/staging/android/sw_sync.c
>>> index b4ae092..ea27512 100644
>>> --- a/drivers/staging/android/sw_sync.c
>>> +++ b/drivers/staging/android/sw_sync.c
>>> @@ -37,15 +37,6 @@ struct sw_sync_create_fence_data {
>>>  		struct sw_sync_create_fence_data)
>>>  #define SW_SYNC_IOC_INC			_IOW(SW_SYNC_IOC_MAGIC, 1, __u32)
>>>  
>>> -static const struct fence_ops timeline_fence_ops;
>>> -
>>> -static inline struct sync_pt *fence_to_sync_pt(struct fence *fence)
>>> -{
>>> -	if (fence->ops != &timeline_fence_ops)
>>> -		return NULL;
>>> -	return container_of(fence, struct sync_pt, base);
>>> -}
>>> -
>>>  struct sync_timeline *sync_timeline_create(const char *name)
>>>  {
>>>  	struct sync_timeline *obj;
>>> @@ -108,7 +99,7 @@ static void sync_timeline_signal(struct sync_timeline *obj, unsigned int inc)
>>>  }
>>>  
>>>  static struct sync_pt *sync_pt_create(struct sync_timeline *obj, int size,
>>> -			     unsigned int value)
>>> +			     unsigned int value, char *name)
>>>  {
>>>  	unsigned long flags;
>>>  	struct sync_pt *pt;
>>> @@ -120,6 +111,7 @@ static struct sync_pt *sync_pt_create(struct sync_timeline *obj, int size,
>>>  	if (!pt)
>>>  		return NULL;
>>>  
>>> +	strlcpy(pt->name, name, sizeof(pt->name));
>>>  	spin_lock_irqsave(&obj->child_list_lock, flags);
>>>  	sync_timeline_get(obj);
>>>  	fence_init(&pt->base, &timeline_fence_ops, &obj->child_list_lock,
>>> @@ -191,7 +183,7 @@ static void timeline_fence_timeline_value_str(struct fence *fence,
>>>  	snprintf(str, size, "%d", parent->value);
>>>  }
>>>  
>>> -static const struct fence_ops timeline_fence_ops = {
>>> +const struct fence_ops timeline_fence_ops = {
>>>  	.get_driver_name = timeline_fence_get_driver_name,
>>>  	.get_timeline_name = timeline_fence_get_timeline_name,
>>>  	.enable_signaling = timeline_fence_enable_signaling,
>>> @@ -252,7 +244,7 @@ static long sw_sync_ioctl_create_fence(struct sync_timeline *obj,
>>>  		goto err;
>>>  	}
>>>  
>>> -	pt = sync_pt_create(obj, sizeof(*pt), data.value);
>>> +	pt = sync_pt_create(obj, sizeof(*pt), data.value, data.name);
>>>  	if (!pt) {
>>>  		err = -ENOMEM;
>>>  		goto err;
>>> diff --git a/drivers/staging/android/sync_debug.c b/drivers/staging/android/sync_debug.c
>>> index 4c5a855..b732ea3 100644
>>> --- a/drivers/staging/android/sync_debug.c
>>> +++ b/drivers/staging/android/sync_debug.c
>>> @@ -75,13 +75,14 @@ static void sync_print_fence(struct seq_file *s, struct fence *fence, bool show)
>>>  {
>>>  	int status = 1;
>>>  	struct sync_timeline *parent = fence_parent(fence);
>>> +	struct sync_pt *pt = fence_to_sync_pt(fence);
>>>  
>>>  	if (fence_is_signaled_locked(fence))
>>>  		status = fence->status;
>>>  
>>> -	seq_printf(s, "  %s%sfence %s",
>>> +	seq_printf(s, "  %s%sfence %s %s",
>>>  		   show ? parent->name : "",
>>> -		   show ? "_" : "",
>>> +		   show ? "_" : "", pt->name,
>>>  		   sync_status_str(status));
>>>  
>> NAK,
>> A fence in sync_print_fence can be of any type. If you want to print the name, use the fence_value_str callback.
> Indeed. But fence_value_str doesn't return the sync_pt name, but the
> seqno of that fence. I'll keep this change out for the de-staging and
> then try to come with something that works better.
>
> 	Gustavo

This will probably cause a kernel panic if you keep it like it is and you look at
debugfs for non sync_pt fences, it should definitely be fixed before destaging.
Ignoring sync_pt name would be better than crashing.

~Maarten



More information about the devel mailing list