E4500 with 24GB RAM

[Hiroki Sato]

Pyun YongHyeon <yongari at rndsoft.co.kr> wrote
  in <20050611072632.GB19976 at rndsoft.co.kr>:

yo> On Sat, Jun 11, 2005 at 03:40:28PM +0900, Hiroki Sato wrote:
yo>  >  I do not know why but the freeze occurs only when displaying "invalid
yo>  >  packet size xxx; dropping".  When I tried "vmstat 1" on the serial
yo>  >  console and fetching a large file via ftp at the same time
yo>  >  with 12GB RAM configuration, the freeze did not occur.
yo>  >  Once "hme0: too may errors; not reporting any more" is displayed,
yo>  >  the box seems to work fine and I can check out the ports tree via NFS
yo>  >  without problems.
yo>  > 
yo> 
yo> Normally the "invalid packet size" message comes from link mismatch.
yo> If your HME's PHY is DP83840 there are known issues on link neogotiation.
yo> AFAIK the issue has nothing to do with panic as I always see that on my
yo> Ultra2 which has DP83840 PHY too.

 I see.  E4500 has a DP83840 PHY, too.

yo> I wonder how you can use NFS reliably on sparc64. Due to failure of
yo> alignment(both server and client) it's really easy to get panic on sparc64.

 IIRC there are few NFS-related panics.  I did not try heavy load, but
 massive read-only access to /home/ncvs works fine at least.

yo>  >  I tried to comment out the HME_WHINE line of hme_read() in if_hme.c,
yo>  >  and it seems to make the box work fine so far.
yo> 
yo> HME_WHINE() just prints a message. I can't think removing the function
yo> can cure your problem.

 I also think so, but the freeze surely happens at that moment.
 Very strange...

 Since doing "vmstat 1" on console when the message is displayed
 prevents the freeze, there may be a kind of timing-dependent problem.

-- 
| Hiroki SATO
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-sparc64/attachments/20050611/836c9e63/attachment.bin