onstat -g spi Spin Locks with Long Spins

The onstat -g spi command displays a list of the resources in the system where a wait was required to obtain a latch on a resource. During the wait, the thread will spin (or loop) trying to acquire the resource. The number of times a wait was required for the resource and the number of total loops performed are given.

Spin locks with waits:                                                    
                                                                          
Num Waits   Num Loops   Avg Loop/Wait    Name                             
11          14          1.27             vproc vp_lock, id = 1            
1           1           1.00             vproc vp_lock, id = 7            
1           1           1.00             tcb lock, tid = 5                

</div>

Output Description

Heading	Description	Format	See Also
Num Waits	The number of times the resource had to be waited on due to another thread holding a latch on that resource.	Dec
Num Loops	The number of times a loop was performed to test for the release of the latch on the resource.	Dec	onstat -g spi
Avg Loop/Wait	The average number of loops that were performed for each wait.	Flt
Name	The name of the resource upon which a spin locking condition has existed. lockfr The lock free list. The number after lockfr is the index into the lock free list array. lockhash[] The lock hash bucket. The field inside the brackets is the index into the lock hash bucket array. :bhash [] The buffer hash bucket. The field before the colon is the buffer pool index; the field inside the brackets after bhash is the index into the buffer hash bucket array. :lru- The LRU latch. The field before the colon is the buffer pool index; the field after lru- identifies the buffer chain pairs that are being used. :bf[] The buffer latch. The field before the colon is the buffer pool index; the field inside the brackets after bf is the position of buffer in the buffer array. The next two fields are the partition number and the page header address in memory for the buffer in hex form. </div>	Str	onstat -g wmx

Notes

A long spin is one where the number of loops for a single wait has exceeded 10,000.

Only resources that have had at least one wait are listed.

Monitoring and Tuning

Spin locks occur as part of the mechanism for acquiring control of a resource. When a thread needs access to a resource in shared memory, it must first acquire a latch on the resource. If the latch is currently held by another thread, the current thread will spin waiting for the latch to be released.

When the number of loops is zero, it indicates that the lock was acquired immediately. This can occur when the latch is released between the time when the request is first made by the thread and the request is remade in the spin lock function call.