sparc64/68869: netcard: Unexpect packet size, drop packet.

Pyun YongHyeon yongari at kt-is.co.kr
Fri Jul 9 20:06:21 PDT 2004


On Fri, Jul 09, 2004 at 04:26:09PM +0000, Magicloud wrote:
...
 > 
 > >Number:         68869
 > >Category:       sparc64
 > >Synopsis:       netcard: Unexpect packet size, drop packet.
 > >Confidential:   no
 > >Severity:       non-critical
 > >Priority:       low
 > >Responsible:    freebsd-sparc64
 > >State:          open
 > >Quarter:        
 > >Keywords:       
 > >Date-Required:
 > >Class:          sw-bug
 > >Submitter-Id:   current-users
 > >Arrival-Date:   Fri Jul 09 16:30:28 GMT 2004
 > >Closed-Date:
 > >Last-Modified:
 > >Originator:     Magicloud
 > >Release:        5.2.1
 > >Organization:
 > >Environment:
 > FreeBSD Sun.teda.nankai.edu.cn 5.2.1-RELEASE-p8 FreeBSD 5.2.1-RELEASE-p8 #2: Fri Jul  9 16:14:19 CST 2004     root at Sun.teda.nankai.edu.cn:/usr/obj/usr/src/sys/GENERIC  sparc64
 > >Description:
 >       I have a ftp server (pure-ftpd) on the machine without speed limit, and the machine is in a 100M LAN. When file transfering, randomly, console reported that the netcard drop the packet for unexpect packet size.
 > >How-To-Repeat:
 >       Randomly. Sometime, even if a lot of file transfer, there is nothing happened. While sometime, when I check the log, a lot of error in it. And in daily mail, before the ftp was used heavily, there were tens of IErrs, while after, there are thousands of.
 > >Fix:
 >       I use current on another machine which has the same problem when use 5.2.1, and ... I think, the problem does not occur till now.

I had the same problem on my Ultra2 which runs -CURRENT.
The real cause of the problem seems to be the lack of
auto-negotiation capability of hme(4). According to the commit
logs from Marius, hme(4) with DP83840A PHY has the problem.

I fixed the problem by reviving commented code.
Activate "mii_mediachg(sc->sc_mii);" statement located
at line 822 of sys/dev/hme/if_hme.c(rev. 1.27) if your
system is NOT "Netra t1 100".

If your system is "Netra t1 100", your best bet would be 
half-duplex with some kind force mechanism.
Hope this helps.

Regards,
Pyun YongHyeon
-- 
Pyun YongHyeon <http://www.kr.freebsd.org/~yongari>


More information about the freebsd-sparc64 mailing list