From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Thomas Munro <thomas(dot)munro(at)enterprisedb(dot)com> |
Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, Andres Freund <andres(at)anarazel(dot)de>, Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Condition variable live lock |
Date: | 2018-01-08 02:34:11 |
Message-ID: | 31228.1515378851@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Thomas Munro <thomas(dot)munro(at)enterprisedb(dot)com> writes:
> On Mon, Jan 8, 2018 at 12:38 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Concretely, as per attached.
> +1 for the idea. Haven't looked at the code yet but I'll review this
> and the proclist patch shortly.
Thanks. BTW, I realized that there is a second (and perhaps more
important) reason why we can only prepare one CV sleep at a time:
we only have one cvWaitLink in our PGPROC. So I'm now inclined
to word the revised comment in ConditionVariablePrepareToSleep as
/*
* If some other sleep is already prepared, cancel it; this is necessary
* because we have just one static variable tracking the prepared sleep,
* and also only one cvWaitLink in our PGPROC. It's okay to do this
* because whenever control does return to the other test-and-sleep loop,
* its ConditionVariableSleep call will just re-establish that sleep as
* the prepared one.
*/
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Thomas Munro | 2018-01-08 04:25:53 | Re: Condition variable live lock |
Previous Message | Thomas Munro | 2018-01-08 02:21:30 | Re: Condition variable live lock |