2
0
mirror of https://git.postgresql.org/git/postgresql.git synced 2025-01-18 18:44:06 +08:00

Cover heap_page_prune_opt()'s cleanup lock tactic in README.

Jeff Janes, reviewed by Jim Nasby.
This commit is contained in:
Noah Misch 2016-01-01 21:52:22 -05:00
parent c7e27becd2
commit dfcd9cb302

View File

@ -81,18 +81,20 @@ it won't be able to actually examine the page until it acquires shared
or exclusive content lock. or exclusive content lock.
Rule #5 only affects VACUUM operations. Obtaining the Obtaining the lock needed under rule #5 is done by the bufmgr routines
necessary lock is done by the bufmgr routine LockBufferForCleanup(). LockBufferForCleanup() or ConditionalLockBufferForCleanup(). They first get
It first gets an exclusive lock and then checks to see if the shared pin an exclusive lock and then check to see if the shared pin count is currently
count is currently 1. If not, it releases the exclusive lock (but not the 1. If not, ConditionalLockBufferForCleanup() releases the exclusive lock and
caller's pin) and waits until signaled by another backend, whereupon it then returns false, while LockBufferForCleanup() releases the exclusive lock
tries again. The signal will occur when UnpinBuffer decrements the shared (but not the caller's pin) and waits until signaled by another backend,
pin count to 1. As indicated above, this operation might have to wait a whereupon it tries again. The signal will occur when UnpinBuffer decrements
good while before it acquires lock, but that shouldn't matter much for the shared pin count to 1. As indicated above, this operation might have to
concurrent VACUUM. The current implementation only supports a single wait a good while before it acquires lock, but that shouldn't matter much for
waiter for pin-count-1 on any particular shared buffer. This is enough concurrent VACUUM. The current implementation only supports a single waiter
for VACUUM's use, since we don't allow multiple VACUUMs concurrently on a for pin-count-1 on any particular shared buffer. This is enough for VACUUM's
single relation anyway. use, since we don't allow multiple VACUUMs concurrently on a single relation
anyway. Anyone wishing to obtain a cleanup lock outside of recovery or a
VACUUM must use the conditional variant of the function.
Buffer Manager's Internal Locking Buffer Manager's Internal Locking