Re: [BUGS] Possible password authentication bug in 6.3.1
От | Fernando Carello |
---|---|
Тема | Re: [BUGS] Possible password authentication bug in 6.3.1 |
Дата | |
Msg-id | 199804041303.PAA09515@shadow.srd.it обсуждение исходный текст |
Список | pgsql-hackers |
> Try adding another host line to the end of the file, and let me know if > that fixes it. Added: host users 192.168.0.1 255.255.255.255 password at the end of pg_hba.conf, but the error is still there. Please note that I don't make use of Unix sockets for the connection, I use TCP/IP instead (" -i "). I've also commented out the (original) last two lines that allowed restrictless connections from the localhost. I'm not very familiar with Postgres internals, but it *seems* to me that the variable "areq" is not getting the right value: it should be "3" ( = AUTH_REQ_PASSWORD) for plain-password authentication, while it gets "13824". ---- Now I'm at home, and I'm playing a little with libpq sources: here I've got Postgres 6.3 (not 6.3.1) and I get a value of areq = 14336 (and the same error, of course). So I printed out areq value in "fe-connect.c", just after the pqGetInt call: I get areq = "14336d", that is quite strange; of course, shortly after, the call to fe_sendauth fails. Then I tried to force areq=3 just before calling fe_sendauth (we are near the middle of fe-connect.c), and it happens that the error becomes: FATAL 1: Socket command option. Don't know if that helps in some way ! :-) Please let me know if I can do something useful (btw, I'm in trouble with that authentication stuff: for now I'm not able to protect my data, so I shutted down the SQL server), and as always thanks to all you people. Fernando Carello
В списке pgsql-hackers по дате отправления: