Re: cvs commit: httpd-test/flood flood_net_ssl.c

[Aaron Bannert <aa...@clove.org>]

So I may be missing something in the haze that is our beloved OpenSSL,
but why wasn't this detected before? Are these locks only used under
certain conditions (like not on any of my boxen)? I've tested this on
Linux 2.4, Solaris 8/intel, and FreeBSD-CURRENT.

-aaron


On Thu, Dec 06, 2001 at 06:03:32PM -0000, aaron@apache.org wrote:
> aaron       01/12/06 10:03:32
> 
>   Modified:    flood    flood_net_ssl.c
>   Log:
>   Whoops! Fix backwards CPP logic that would prevent rwlocks from
>   being initialized.
>   
>   Revision  Changes    Path
>   1.16      +2 -2      httpd-test/flood/flood_net_ssl.c
>   
>   Index: flood_net_ssl.c
>   ===================================================================
>   RCS file: /home/cvs/httpd-test/flood/flood_net_ssl.c,v
>   retrieving revision 1.15
>   retrieving revision 1.16
>   diff -u -r1.15 -r1.16
>   --- flood_net_ssl.c	2001/12/04 10:08:30	1.15
>   +++ flood_net_ssl.c	2001/12/06 18:03:32	1.16
>   @@ -93,9 +93,9 @@
>    
>        l = apr_palloc(ssl_pool, sizeof(CRYPTO_dynlock_value));
>    #ifdef USE_RW_LOCK_FOR_SSL 
>   -    apr_lock_create(&l->lock, APR_MUTEX, APR_INTRAPROCESS, NULL, ssl_pool);
>   -#else
>        apr_lock_create(&l->lock, APR_READWRITE, APR_INTRAPROCESS, NULL, ssl_pool);
>   +#else
>   +    apr_lock_create(&l->lock, APR_MUTEX, APR_INTRAPROCESS, NULL, ssl_pool);
>    #endif
>        return l;
>    }
>   
>   
>   

Re: cvs commit: httpd-test/flood flood_net_ssl.c

[Justin Erenkrantz <je...@ebuilt.com>]

On Thu, Dec 06, 2001 at 10:25:16AM -0800, Aaron Bannert wrote:
> So I may be missing something in the haze that is our beloved OpenSSL,
> but why wasn't this detected before? Are these locks only used under
> certain conditions (like not on any of my boxen)? I've tested this on
> Linux 2.4, Solaris 8/intel, and FreeBSD-CURRENT.

I have no idea.  The whole locking API with OpenSSL is just odd.
I think I was never able to trigger that locking code.  I wonder
why Solaris 2.6/Sparc triggers it?  -- justin