From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andrei Petru Mura <mapandrei@gmail.com>
cc: pgsql-interfaces@postgresql.org
Subject: Re: PQconnectdbParams returns exit value 01 (from gdb).
In-reply-to: 
 <CAK18DJUsbdMYS3m-7az4Zc32S-AurcMZG7MkoQ5G3SCub_1-fg@mail.gmail.com>
References: 
 <CAK18DJUsbdMYS3m-7az4Zc32S-AurcMZG7MkoQ5G3SCub_1-fg@mail.gmail.com>
Comments: In-reply-to Andrei Petru Mura <mapandrei@gmail.com>
	message dated "Mon, 25 Feb 2013 12:42:10 +0200"
Date: Mon, 25 Feb 2013 06:13:24 -0500
Message-ID: <19857.1361790804@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-interfaces-owner@postgresql.org

Andrei Petru Mura <mapandrei@gmail.com> writes:
> I'm trying to create a connection to my database with libpq. Bellow is my
> code:

>         keywords = malloc(6 * sizeof(char *));
>         values = malloc(6 * sizeof(char *));

That looks less than safe ... what happens if you have more than five
lines in the creds file?

>         while (fgets(line, LINE_SIZE, creds) != NULL) {
>                 if (line[strlen(line) - 1] == '\n')
>                         line[strlen(line) - 1] = '\0';
>                 prop = line;
>                 while(*(prop++) != '=') {
>                         i++;
>                 }
>                 tmp = prop;
>                 prop = malloc(i + 1);
>                 strncpy(prop, line, i);
>                 prop[i] = '\0';
>                 keywords[j++] = prop;
>                 val = malloc(strlen(line) - strlen(prop) + 1);
>                 strcpy(val, tmp);
>                 values[k++] = val;
>                 i = 0;
>         }

This has got a few issues, like it'll die badly if there's no '='
in a line, and not behave too well if a line overruns the fixed
buffer length.

>         keywords[++j] = NULL;
>         values[++k] = NULL;

But your real problem is here, where you're leaving undefined holes
in the arrays.  These should be j++ and k++, not ++j and ++k.
If malloc doesn't give you back all-zeroes storage, PQconnectdbParams
will be led to try to dereference garbage pointer values, resulting
in your SIGSEGVs.

			regards, tom lane


-- 
Sent via pgsql-interfaces mailing list (pgsql-interfaces@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-interfaces