check_recovery_target_lsn() does a PG_CATCH without a throw

From: Andres Freund <andres(at)anarazel(dot)de>
To: pgsql-hackers(at)postgresql(dot)org, Peter Eisentraut <peter_e(at)gmx(dot)net>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Subject: check_recovery_target_lsn() does a PG_CATCH without a throw
Date: 2019-06-11 06:11:15
Message-ID: 20190611061115.njjwkagvxp4qujhp@alap3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

While working on fixing [1] I noticed that 2dedf4d9a899 "Integrate
recovery.conf into postgresql.conf" added two non-rethrowing PG_CATCH
uses. That's not OK. See

https://www.postgresql.org/message-id/1676.1548726280%40sss.pgh.pa.us
https://postgr.es/m/20190206160958.GA22304%40alvherre.pgsql
etc.

static bool
check_recovery_target_time(char **newval, void **extra, GucSource source)
{
if (strcmp(*newval, "") != 0)
{
TimestampTz time;
TimestampTz *myextra;
MemoryContext oldcontext = CurrentMemoryContext;

/* reject some special values */
if (strcmp(*newval, "epoch") == 0 ||
strcmp(*newval, "infinity") == 0 ||
strcmp(*newval, "-infinity") == 0 ||
strcmp(*newval, "now") == 0 ||
strcmp(*newval, "today") == 0 ||
strcmp(*newval, "tomorrow") == 0 ||
strcmp(*newval, "yesterday") == 0)
{
return false;
}

PG_TRY();
{
time = DatumGetTimestampTz(DirectFunctionCall3(timestamptz_in,
CStringGetDatum(*newval),
ObjectIdGetDatum(InvalidOid),
Int32GetDatum(-1)));
}
PG_CATCH();
{
ErrorData *edata;

/* Save error info */
MemoryContextSwitchTo(oldcontext);
edata = CopyErrorData();
FlushErrorState();

/* Pass the error message */
GUC_check_errdetail("%s", edata->message);
FreeErrorData(edata);
return false;
}
PG_END_TRY();

same in check_recovery_target_lsn.

I'll add an open item.

Greetings,

Andres Freund

[1] CALDaNm1KXK9gbZfY-p_peRFm_XrBh1OwQO1Kk6Gig0c0fVZ2uw(at)mail(dot)gmail(dot)com

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message kuroda.hayato@fujitsu.com 2019-06-11 06:35:36 RE: [PATCH] memory leak in ecpglib
Previous Message John Naylor 2019-06-11 06:04:34 Re: Multivariate MCV stats can leak data to unprivileged users