This is the mail archive of the
ecos-discuss@sourceware.org
mailing list for the eCos project.
Re: AW: FreeBSD not always releasing ACK MBUFs
- From: Laurie Gellatly <laurie dot gellatly at netic dot com>
- To: Richard Rauch <rrauch at itrgmbh dot de>
- Cc: "<ecos-discuss at sourceware dot org>" <ecos-discuss at sourceware dot org>
- Date: Sun, 28 Apr 2013 21:55:40 +1000
- Subject: Re: AW: FreeBSD not always releasing ACK MBUFs
- References: <001201ce4404$f89d4f20$e9d7ed60$ at netic dot com> <001501ce4406$9f7a0670$de6e1350$ at itrgmbh dot de>
FreeBSD. How did I miss that important fact???
...Laurie:{)
+61 416 114419
On 28/04/2013, at 9:50 PM, "Richard Rauch" <rrauch@itrgmbh.de> wrote:
> Hello,
>
> which TCP/IP Stack are you using? FreeBSD or lwIP?
>
> Richard
>
> ITR GmbH
> web: http://www.itrgmbh.com
> email: info@itrgmbh.com
>
>
>> -----UrsprÃngliche Nachricht-----
>> Von: ecos-discuss-owner@ecos.sourceware.org [mailto:ecos-discuss-
>> owner@ecos.sourceware.org] Im Auftrag von Laurie Gellatly
>> Gesendet: Sonntag, 28. April 2013 13:39
>> An: ecos-discuss@sourceware.org
>> Betreff: [ECOS] FreeBSD not always releasing ACK MBUFs
>>
>> Hi All,
>> I'm using eCosPro from pre eCos 3.0 to serve web pages with embedded
>> images, CSS and javascripts.
>> Some pages are set to refresh their content and eventually the application runs
>> out of MBUFs and restarts.
>> I can use my own web server or the CYGPKG_HTTPD server with the same
>> result.
>>
>> I used the CYGPKG_HTTPD server on port 81 to monitor network stats.
>> What I find is that the âHEADERâ type MBUFs grow and never shrink, eventually
>> running out of MBUFs completely.
>> The amount of growth in each iteration varies.
>> When I dump the data in the HEADER MBUFs (modified support.c to print these
>> short MBUFs) and compared them to the network traffic seen by wireshark I
>> have confirmed that the HEADERs are correct ACK packets sent at the close of
>> various HTTP get sequences.
>> For some reason they are not being released.
>>
>> I also found that the number of timeout entries (NTIMEOUTS) in timeout.c
>> (default 8) is WAY below what is needed (about 70 being used).
>>
>> I'm thinking it's something to do with TIME_WAIT not closing the tcpcb when it
>> should.
>>
>> Has anyone seen this before or can suggest a way to troubleshoot this?
>>
>> Thanks ...Laurie:{)
>>
>>
>>
>>
>> --
>> Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
>> and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss
>
>
>
--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss