This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
RE: lib/timeout.c
- To: tb <terrencebarr at yahoo dot com>
- Subject: RE: [ECOS] lib/timeout.c
- From: Gary Thomas <gthomas at cambridge dot redhat dot com>
- Date: Fri, 08 Jun 2001 20:22:25 -0600 (MDT)
- Cc: ecos-discuss at sources dot redhat dot com
- Organization: Red Hat, Inc.
On 09-Jun-2001 tb wrote:
> Ah, ok.
>
> Now, can I upgrade to the new net stuff under CVS without updating
> all of my eCos sources? Seems like the net changes require new
> packages like CYGPKG_MEMALLOC and CYGPKG_ISOINFRA which are not
> present in the 1_3_1 release.
>
I personally wouldn't try - too much has changed and not always along
easily defined borders.
As I said below, use the CVS. There may be some initial pain, but the
benefits will far outweigh it.
> Thanks,
>
> -- Terrence
>
> --- Gary Thomas <gthomas@cambridge.redhat.com> wrote:
>>
>> On 08-Jun-2001 tb wrote:
>> > Hi,
>> >
>> > The TCP/IP stack comes with timeout.c code. In testing our ethernet driver
>> > and the TCP/IP stack we've noticed that timeouts sometimes get screwed up.
>> > It seems that when timeout() gets called while a previous timeout is still
>> > pending the computation of min_delta results in negative values and from
>> > that point on the timeout queue is broken. As a result the TCP/IP resources
>> > don't get released and opening new connections fails.
>> >
>> > I haven't found any clues in the archives. Any ideas? PS: We are using
>> > the released net package, not the CVS one.
>>
>> There's the problem; the timeout code has been completely restructured because
>> of just these sort of problems with the 1.3.1 network code.
>>
>> Use the CVS - it's up to date and it works.
>
>
> __________________________________________________
> Do You Yahoo!?
> Get personalized email addresses from Yahoo! Mail - only $35
> a year! http://personal.mail.yahoo.com/