Re: Dead code or buggy code? - Mailing list pgsql-hackers

From Greg Stark
Subject Re: Dead code or buggy code?
Date
Msg-id CAM-w4HOprJ68g3BW6R4i6idewKZNniqfShPpjb7et4dLHztJxA@mail.gmail.com
Whole thread Raw
In response to Re: Dead code or buggy code?  (Robert Haas <robertmhaas@gmail.com>)
Responses Re: Dead code or buggy code?
List pgsql-hackers
<p dir="ltr">So I'm just going to make the code defensive and assume NULL is possible when if maybe it isn't. <p
dir="ltr">Incase it's not clear, this is one of the thing's Xi Wang's took picked up. There not to many but it turns
outthey are indeed not all in the adt code so I might wait until after the commit fest to commit it to avoid causing
bitchurn.<p dir="ltr">-- <br /> greg<div class="gmail_quote">On 19 Sep 2013 12:52, "Robert Haas" <<a
href="mailto:robertmhaas@gmail.com">robertmhaas@gmail.com</a>>wrote:<br type="attribution" /><blockquote
class="gmail_quote"style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> On Wed, Sep 18, 2013 at 6:20
PM,Greg Stark <<a href="mailto:stark@mit.edu">stark@mit.edu</a>> wrote:<br /> > The following code is in the
ProcSleepat proc.c:1138.<br /> > GetBlockingAutoVacuumPgproc() should presumably always return a vacuum<br /> >
pgprocentry since the deadlock state says it's blocked by autovacuum.<br /> > But I'm not really familiar enough
withthis codepath to know whether<br /> > there's not a race condition here where it can sometimes return null.<br
/>> The following code checks autovac != NULL but the PGXACT initializer<br /> > would have seg faulted if it
returnedNULL if that's possible.<br /> ><br /> >         if (deadlock_state == DS_BLOCKED_BY_AUTOVACUUM
&&<br/> > allow_autovacuum_cancel)<br /> >         {<br /> >             PGPROC       *autovac =
GetBlockingAutoVacuumPgproc();<br/> >             PGXACT       *autovac_pgxact =<br /> >
&ProcGlobal->allPgXact[autovac->pgprocno];<br/> ><br /> >             LWLockAcquire(ProcArrayLock,
LW_EXCLUSIVE);<br/> ><br /> >             /*<br /> >              * Only do it if the worker is not working to
protectagainst Xid<br /> >              * wraparound.<br /> >              */<br /> >             if ((autovac
!=NULL) &&<br /> >                 (autovac_pgxact->vacuumFlags & PROC_IS_AUTOVACUUM) &&<br
/>>                 !(autovac_pgxact->vacuumFlags & PROC_VACUUM_FOR_WRAPAROUND))<br /> >             {<br
/><br/> Hmm, yeah.  I remember noticing this some time ago but never got<br /> around to fixing it.  +1 for rearranging
thingsthere somehow.<br /><br /> --<br /> Robert Haas<br /> EnterpriseDB: <a href="http://www.enterprisedb.com"
target="_blank">http://www.enterprisedb.com</a><br/> The Enterprise PostgreSQL Company<br /></blockquote></div> 

pgsql-hackers by date:

Previous
From: Andres Freund
Date:
Subject: Re: [PERFORM] encouraging index-only scans
Next
From: Tatsuo Ishii
Date:
Subject: Looking for information on our elephant