Re: Possible pointer dereference - Mailing list pgsql-hackers

From Haribabu Kommi
Subject Re: Possible pointer dereference
Date
Msg-id CAJrrPGdBnQdEUDGR+60D8h9thUD5DmhrrKUdVoU8ns+YM7iGiw@mail.gmail.com
Whole thread Raw
In response to Possible pointer dereference  (Gaetano Mendola <mendola@gmail.com>)
Responses Re: Possible pointer dereference  (Robert Haas <robertmhaas@gmail.com>)
List pgsql-hackers
On Thu, May 28, 2015 at 6:07 AM, Gaetano Mendola <mendola@gmail.com> wrote:
> I'm playing with a static analyzer and it's giving out some real error
> analyzing postgresql code base like the following one
>
> src/backend/access/transam/commit_ts.c
>    return *ts != 0  // line 321
> but a few line up (line 315) ts is checked for null, so either is not needed
> to check for null or *ts can lead to a null pointer dereference. Same
> happens a few line later lines 333 and 339

Thanks for providing detailed information.

The function "TransactionIdGetCommitTsData" is currently used only at
one place. The caller
always passes an valid pointer to this function. So there shouldn't be
a problem. But in future
if the same function is used at somewhere by passing the NULL pointer
then it leads to a crash.

By correcting the following way will solve the problem.

return ts ? (*ts != 0) : false; instead of retun *ts != 0;

Attached a patch for it.

Regards,
Hari Babu
Fujitsu Australia

Attachment

pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: fsync-pgdata-on-recovery tries to write to more files than previously
Next
From: "David G. Johnston"
Date:
Subject: Re: Can we add syntax for references auto create index or not.