Horton
Menu

Re: Don't pass NULL pointer to strcmp().

Aleksander Alekseev <aleksander@timescale.com>

View thread
From: Aleksander Alekseev <aleksander@timescale.com>
To: pgsql-hackers <pgsql-hackers@postgresql.org>
Cc: Junwang Zhao <zhjwpku@gmail.com>, Xing Guo <higuoxing@gmail.com>
Date: 2023-11-01T11:44:07Z
Lists: pgsql-hackers 
Hi,

> > I found that there's a nullable pointer being passed to strcmp() and
> > can make the server crash. It can be reproduced on the latest master
> > branch by crafting an extension[1]. Patch for fixing it is attatched.
> >
> > [1] https://github.com/higuoxing/guc_crash/tree/pg

Thanks for reporting. I can confirm that the issue reproduces on the
`master` branch and the proposed patch fixes it.

> Can we set a string guc to NULL? If not, `*lconf->variable == NULL` would
> be unnecessary.

Judging by the rest of the code we better keep it, at least for consistenc.

I see one more place with a similar code in guc.c around line 1472.
Although I don't have exact steps to trigger a crash I suggest adding
a similar check there.

-- 
Best regards,
Aleksander Alekseev

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Make GetConfigOption/GetConfigOptionResetString return "" for NULL.

  2. Be more wary about NULL values for GUC string variables.

  3. doc: Replace reference to ERRCODE_RAISE_EXCEPTION by "raise_exception"