From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Simon Riggs <simon(at)2ndquadrant(dot)com> |
Cc: | pgsql-committers(at)postgresql(dot)org |
Subject: | Re: pgsql: Fix (some of the) breakage introduced into query-cancel |
Date: | 2010-01-08 02:51:24 |
Message-ID: | 24830.1262919084@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-committers |
Simon Riggs <simon(at)2ndquadrant(dot)com> writes:
> On Thu, 2010-01-07 at 16:29 +0000, Tom Lane wrote:
>> Instead of that, undo the original optimizations that threw away
>> QueryCancelPending anytime we were starting or finishing a command
>> read, and instead discard the cancel request within ProcessInterrupts
>> if we find that there is no HS reason for forcing a cancel and we are
>> DoingCommandRead.
> Is there a reason why we are calling DisableNotifyInterrupt() and
> DisableCatchupInterrupt() before we call ProcessInterrupts() and again
> within it?
Different purposes. ProcessInterrupts is called from a lot of places
and should not assume the status of those flags --- in particular, if it
is called directly from the signal handler while we are waiting for a
command, those disables are needed because the interrupts are active.
The call site you are probably looking at is where we are exiting the
active-command-read state, and we want to disable the interrupts before
beginning active processing, whether or not there is any cancel interrupt
waiting.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | User Kasahara | 2010-01-08 08:30:28 | pgstatsinfo - pg_statsinfo: New Directory |
Previous Message | Tom Lane | 2010-01-08 02:44:00 | pgsql: Fix oversight in EvalPlanQualFetch: after failing to lock a tuple |