I would be very surprised to hear that compiler choice matters.
Believe it or not we had to fix a problem exactly like this in the
2.0.x Sparc kernels, in exactly the ESP driver, and exactly because
egcs was being used as the compiler instead of gcc.
The issue back then, was that egcs would move some memory references
around (in a legal way, was not a compiler bug) that on slower
machines would cause a race condition which would confuse the ESP
driver and make it barf on disks that normally acted correctly.
I am not saying we are certainly hitting the same issue here, but when
I read your report, bells went off in my head and this is why I
instantly asked what compiler you were using.
I've also been told by a couple people that they were able to solve
this problem by either getting a new drive or playing around with
the cables and termination. Something somewhere in the new esp.c
is probing something that esp.c under 2.0.x didn't.
If you do resolve the issue in this way, please let me know, because I
still consider the bug open at this point.
Later,
David S. Miller
davem@redhat.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/