Обсуждение: Wierd MD5-authentication crash on Solaris 8

Severity: Core dump
Frequency: Unusual
PostgreSQL Version: 7.4.0 release
Platform:=20=20
    Solaris 8
    gmake 3.80
    gcc 3.23
    440mhz UltraSparc III
    Netra T1/105
Reproducability:  100% on this machine and an identical one.=20
     Not yet tested on other machines running Solaris 8.
Summary: attempting to connect via MD5 authentication as a user=20
    who has no password triggers a core dump of Postmaster.
Steps to Reproduce:=20=20
    1) set localhost authentication to MD5
    2) set up a new user without setting a password.
    3) attempt to connect as that user.

Core dump file is available.    Strace can also be done if desired.

--=20
-Josh Berkus
 Aglio Database Solutions
 San Francisco

Josh Berkus <josh@agliodbs.com> writes:
> Summary: attempting to connect via MD5 authentication as a user
>     who has no password triggers a core dump of Postmaster.
[...]
> Core dump file is available.

Can you post a stacktrace? (Building the postmaster with debugging
symbols first would be nice.)

-Neil

Neil,

> Can you post a stacktrace? (Building the postmaster with debugging
> symbols first would be nice.)

I'll see what I can do.   A regular strace should be easy.

--
Josh Berkus
Aglio Database Solutions
San Francisco

Josh Berkus <josh@agliodbs.com> writes:
>     Solaris 8
> Summary: attempting to connect via MD5 authentication as a user
>     who has no password triggers a core dump of Postmaster.

Is this the bsearch-of-no-elements problem recently discussed?
(If you have other users who do have passwords, then it's not...)

            regards, tom lane

Tom,

> Is this the bsearch-of-no-elements problem recently discussed?
> (If you have other users who do have passwords, then it's not...)

Actually, it could be.  Is it patched in the current source code?

--
Josh Berkus
Aglio Database Solutions
San Francisco

Josh Berkus <josh@agliodbs.com> writes:
>> Is this the bsearch-of-no-elements problem recently discussed?

> Actually, it could be.  Is it patched in the current source code?

Yeah, see
http://developer.postgresql.org/cvsweb.cgi/pgsql-server/src/backend/libpq/hba.c

            regards, tom lane