> I noticed strange output that I can't understand.

I can't understand it either.
It doesn't make any sense to me.
It might be interesting to run

cpu -h io
iostats venti/checkarenas -v /dev/sdC0/arenas arena.30

and see if the problem happens then.

The only thing I can think of is that
when run with something proxying the
9p connection, some reads return
less than the amount of data requested,
and venti is not correctly reissuing
the read for the remainder of the block.
That's a very scary thought.
The relevant function is /sys/src/cmd/venti/srv/part.c:/^prwb
but it looks okay.  Maybe there is some
bit of code that's not calling it.

Russ

Reply via email to