A question regarding LWLock in ProcSleep

From: Kenan Yao <kyao(at)pivotal(dot)io>
To: pgsql-hackers(at)postgresql(dot)org
Subject: A question regarding LWLock in ProcSleep
Date: 2015-12-17 08:21:30
Message-ID: CAO1BoPwHnPsmaG+ifg=zOYt0xSigVW==Ye=Bo2NP_scAHGwAsA@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi there,

In function ProcSleep, after the process has been waken up, either with
lock granted or deadlock detected, it would re-acquire the lock table's
partition LWLock.

The code episode is here:

/*
* Re-acquire the lock table's partition lock. We have to do this to hold
* off cancel/die interrupts before we can mess with lockAwaited (else we
* might have a missed or duplicated locallock update).
*/
LWLockAcquire(partitionLock, LW_EXCLUSIVE);

/*
* We no longer want LockErrorCleanup to do anything.
*/
lockAwaited = NULL;

/*
* If we got the lock, be sure to remember it in the locallock table.
*/
if (MyProc->waitStatus == STATUS_OK)
GrantAwaitedLock();

/*
* We don't have to do anything else, because the awaker did all the
* necessary update of the lock table and MyProc.
*/
return MyProc->waitStatus;


Questions are:

(1) The comment says that "we might have a missed or duplicated locallock
update", in what cases would we hit this if without holding the LWLock?

(2) The comment says "we have to do this to hold off cancel/die
interrupts", then:

- why using LWLockAcquire instead of HOLD_INTERRUPTS directly?
- From the handler of SIGINT and SIGTERM, seems nothing serious would be
processed here, since no CHECK_FOR_INTERRUPS is called before releasing
this LWLock. Why we should hold off cancel/die interrupts here?

(3) Before releasing this LWLock, the only share memory access is
MyProc->waitStatus; since the process has been granted the lock or removed
from the lock's waiting list because of deadlock, is it possible some other
processes would access this field? if not, then why we need LWLock here?
what does this lock protect?

Apologize if I missed anything here, and appreciate if someone can help me
on this question. Thanks

Cheers,
Kenan

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Ashutosh Bapat 2015-12-17 08:32:17 Re: Getting sorted data from foreign server for merge join
Previous Message Peter Geoghegan 2015-12-17 07:55:22 Re: Refactoring speculative insertion with unique indexes a little