Re: no unlock in bio/b_sock.c
Goetz Babin-Ebell <[EMAIL PROTECTED]>:
>>I think in b_sock.c I've stumbled into some error cases in which some locking
>>isn't undone:
> - if (ret == NULL) return(NULL);
> + if (ret == NULL)
> + goto err;
> - if (ret == NULL) return(NULL);
> - if (j > 128) return(ret); /* too big to cache */
> + if (ret == NULL || j > 128) /* not found or too big to
> cache */
> + {
> + CRYPTO_w_unlock(CRYPTO_LOCK_BIO_GETHOSTBYNAME);
> + return(ret);
> + }
Your diff is relative to b_sock.c version 1.12 or older (in the
OpenSSL CVS repository), which corresponds to OpenSSL 0.9.2b.
OpenSSL 0.9.3 has more sophisticated bugs :-) (their presence is
indicated by a comment, although that comment does not note
that the function interface makes it impossible to make
BIO_gethostbyname threadsafe).
__
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
Re: no unlock in bio/b_sock.c
On Mon, Jun 07, 1999 at 05:52:24PM +0200, Goetz Babin-Ebell wrote: > I think in b_sock.c I've stumbled into some error cases in which some locking > isn't undone: It's wrong anyway, I haven't come around to change it ... That caching gethostbyname function is only suitable for single-threaded use -- it's an API problem, not an implementation problem; for multi-threaded use we need something with a gethostbyname_r-like interface (BIO_gethostbyname_r). All locking will be done there: lock, call BIO_gethostbyname, copy result, unlock. __ OpenSSL Project http://www.openssl.org Development Mailing List [EMAIL PROTECTED] Automated List Manager [EMAIL PROTECTED]
Re: no unlock in bio/b_sock.c
At 17:52 07.06.99 +0200, you wrote:
>Hallo,
>
>I think in b_sock.c I've stumbled into some error cases in which some locking
>isn't undone:
sorry, wrong diff...
*** OpenSSL/crypto/bio/b_sock.c Mon Apr 26 19:00:12 1999
--- OpenSSL/crypto/bio/b_sock.fixed.c Mon Jun 7 17:46:45 1999
***
*** 247,253
MemCheck_off();
ret=(struct hostent *)Malloc(sizeof(struct hostent));
! if (ret == NULL) return(NULL);
memset(ret,0,sizeof(struct hostent));
for (i=0; a->h_aliases[i] != NULL; i++)
--- 247,254
MemCheck_off();
ret=(struct hostent *)Malloc(sizeof(struct hostent));
! if (ret == NULL)
! goto err;
memset(ret,0,sizeof(struct hostent));
for (i=0; a->h_aliases[i] != NULL; i++)
***
*** 349,356
BIO_ghbn_miss++;
ret=gethostbyname(name);
! if (ret == NULL) return(NULL);
! if (j > 128) return(ret); /* too big to cache */
/* else add to cache */
if (ghbn_cache[lowi].ent != NULL)
--- 350,360
BIO_ghbn_miss++;
ret=gethostbyname(name);
! if (ret == NULL || j > 128) /* not found or too big to
cache */
! {
! CRYPTO_w_unlock(CRYPTO_LOCK_BIO_GETHOSTBYNAME);
! return(ret);
! }
/* else add to cache */
if (ghbn_cache[lowi].ent != NULL)
By
Goetz
--
Goetz Babin-Ebell mailto:[EMAIL PROTECTED]
TC Trust Center for Security http://www.trustcenter.de
in Data Networks GmbH Tel.: +49-40-766 29 3301
Am Werder 1 / 21073 Hamburg / Germany Fax.: +49-40-766 29 577
__
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
no unlock in bio/b_sock.c
Hallo,
I think in b_sock.c I've stumbled into some error cases in which some locking
isn't undone:
> diff b_sock.c b_sock.fixed.c
250c250,251
< if (ret == NULL) return(NULL);
---
> if (ret == NULL)
> goto err;
352,353c353,357
< if (ret == NULL) return(NULL);
< if (j > 128) return(ret); /* too big to cache */
---
> if (ret == NULL || j > 128) /* not found or too big to
cache */
> {
> CRYPTO_w_unlock(CRYPTO_LOCK_BIO_GETHOSTBYNAME);
> return(ret);
> }
By
Goetz
--
Goetz Babin-Ebell mailto:[EMAIL PROTECTED]
TC Trust Center for Security http://www.trustcenter.de
in Data Networks GmbH Tel.: +49-40-766 29 3301
Am Werder 1 / 21073 Hamburg / Germany Fax.: +49-40-766 29 577
__
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
