Im not sure what the NFS server status is on tcp NFS. The problem is in the
NFS client. I fixed the tcp protocol layer bugs it had but some of the wakeup
code is definitely not doing the right thing. I got lost in a maze of twisty
little state machines all alike and never figured it out.
If anyone is curious then trace an NFS client in 2.1.125 doing NFS over tcp.
When you do a big I/O operation it will sooner or later do a partial write.
That used to break, what it does now is to back off until the protocol
stacks says "there is room". That then tries to wake up the current rpc
and continue. The wakeup is called but for some reason has no effect. THe
timer eventually goes off and completes the I/O.
Alan
-
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/