Date: Tue, 2 Jul 1996 09:58:04 -0500
   From: Miguel de Icaza <miguel@roxanne.nuclecu.unam.mx>

   I think one of the problems was related to not having the proper
   cable attached to the sparc at boot time, I'm sure you will find
   those in the archive.

Another issue was the usage of the delayed LEDMA ecache invalidations,
this is _critical_ as it is the only way to cause the infamous SS10
Lance problems not show up at all.  If you don't use delayed ecache
invalidations, the Lance is constantly pressuring the ledma to run the
data to memory before it will assert the interrupt, this is what trips
up all the SS10 motherboard buglets.  Believe me, I did everything
except for putting a scope on the fucker to validate these claims.

Later,
David S. Miller
dm@sgi.com