TransactionIdGetCommitTsData and its dereferenced pointers - Mailing list pgsql-hackers

From Michael Paquier
Subject TransactionIdGetCommitTsData and its dereferenced pointers
Date
Msg-id CAB7nPqS_kF7vJaQA6qNbcuQnr5B6JBDyxAz5MuXPsB3KtmoBbA@mail.gmail.com
Whole thread Raw
Responses Re: TransactionIdGetCommitTsData and its dereferenced pointers  (Alvaro Herrera <alvherre@2ndquadrant.com>)
List pgsql-hackers
Hi all,

TransactionIdGetCommitTsData@commit_ts.c does the following:
    if (ts)
        *ts = entry.time;
    [...]
    return *ts != 0;
This is a bad idea, because if TransactionIdGetCommitTsData is called
with ts == NULL this would simply crash. It seems to me that it makes
little sense to have ts == NULL either way because this function is
intended to return a timestamp in any case, hence I think that we
should simply Assert(ts == NULL) and remove those checks.

Petr, Alvaro, perhaps you intended something like the patch attached,
or perhaps something like that? This would be useful to not ERROR
should an OOM happen when allocating a timestamp pointer, though I
doubt that this is the first intention:
return ts != NULL ? *ts != 0 : false;
Regards,
--
Michael

Attachment

pgsql-hackers by date:

Previous
From: Michael Paquier
Date:
Subject: Re: WIP: SCRAM authentication
Next
From: Andres Freund
Date:
Subject: Re: Raising our compiler requirements for 9.6