httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mathihalli, Madhusudan" <mad...@hp.com>
Subject Bug 27751 (was RE: Error messages from ssl_scache_shmcb)
Date Thu, 25 Mar 2004 22:27:03 GMT
>-----Original Message-----
>From: Geoff Thorpe [mailto:geoff@geoffthorpe.net]
[SNIP]
>It's been a long time since I last looked at shmcb in any 
>detail, and it's possible that there were still bugs waiting 
>to turn up eventually just as it's also possible that some
>of the porting to apache2 might have introduced some niggles
>along the way. As it happens, I'm going to be delving in the
>next few days anyway because of bug 27751[1] - perhaps this
>is related? If you want to get some context from the mement
>this error occurs, we could try and dig in.

I don't see a SEGV as reported in the bugzilla - but the size keeps increasing, and I'm pretty
sure it'll either cause a hang or die wanting more memory :)

Here's my observation :

- process size remains constant under the following conditions:
  - SSLVerifyClient is set to 'none'
  - SSLVerifyClient is set to 'require' and SSLSessionCache is set to 'none'

OTOH, when both of them are set to valid values, the size of the process keeps increasing
like it has no end !.. I narrowed down the cause to ssl_engine_init.c(line 490)

   490          cache_mode = SSL_SESS_CACHE_SERVER|SSL_SESS_CACHE_NO_INTERNAL

If the flag 'SSL_SESS_CACHE_SERVER' is NOT used, then everything seems normal - Ofcourse !
it should be that way because the session cache is not used at all (doh!)

I'm getting a feeling that OpenSSL just ignores the SSL_SESS_CACHE_NO_INTERNAL flag and continues
happily allocating memory for the cache.

Any comments ?

-Madhu

Mime
View raw message