BUG #14859: Config file parsing is allergic to spaces on last line of file

boatcoder@gmail.com writes:

Edit your pg_hba.conf file, add a space (without a new line) at the end of
the file.
Attempt to start postgres. It will fail with messages like this.

Hmph. I can reproduce this in 9.6 and earlier, but not in v10 or HEAD.
Apparently somebody fixed it by accident. Looking ...

regards, tom lane

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

On Tue, Oct 17, 2017 at 7:44 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

boatcoder@gmail.com writes:

Edit your pg_hba.conf file, add a space (without a new line) at the end

of

the file.
Attempt to start postgres. It will fail with messages like this.

Hmph. I can reproduce this in 9.6 and earlier, but not in v10 or HEAD.
Apparently somebody fixed it by accident. Looking ...

​Probably as a side-effect of:

"​Add view pg_hba_file_rules to display the contents of pg_hba.conf"

to v10 (you can probably find the commit hash faster than I).

David J.

"David G. Johnston" <david.g.johnston@gmail.com> writes:

On Tue, Oct 17, 2017 at 7:44 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

Hmph. I can reproduce this in 9.6 and earlier, but not in v10 or HEAD.
Apparently somebody fixed it by accident. Looking ...

​Probably as a side-effect of:
"​Add view pg_hba_file_rules to display the contents of pg_hba.conf"
to v10 (you can probably find the commit hash faster than I).

Close, but apparently it was the adjacent commit 1e5a5d03d,
"Simplify some long-obsolete code in hba.c's next_token()."
in which I wrote

Also, ensure that we don't return with *lineptr set to someplace past the
terminating '\0'; that would be catastrophic if a caller were to ask for
another token from the same line. This is just latent since no callers
actually do call again after a "false" return; but considering that it was
actually costing us extra code to do it wrong, we might as well make it
bulletproof.

I was wrong to think that the problem was only latent, because
tokenize_file() decides whether there's more on the line using

while (strlen(lineptr) > 0)

so that having advanced lineptr past the current line's '\0' allows
it to see (and process) whatever had been in the buffer beyond that
from preceding line(s).

This seems to have been broken since 9.3; kinda surprising nobody
noticed earlier. I shall go back-patch 1e5a5d03d --- it does a bit
more than the minimum needed to fix the bug, but the other changes
seem harmless enough.

regards, tom lane

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