>structure of your code (not the source, but the notion of adding
>preemption points at locations that were identifiable problems in
IMHO adding preemption point in middle of read(2) and write(2) isn't an
option, but it should be done someway in 2.2.x too (IIRC also DaveM
pointed this out a few months ago), because it's a fix (it can be achieved
by adding checks in copy-user, file_read_actor, making the copy-user
preemtable or making all the kernel preemptable or whatever way you
prefer). Adding the preemption point in copy-user covers all cases and
fses without risk. If the data that you're going to read is in cache you
will never get preempted until it completes (unless your user buffer was
in the swap space only). That's not very visible because of I/O, because
memory is fast, and because we usually read/write only a few kbytes per
syscall but I guess with 2giga of cache and 2giga of user buffer the hang
should become quite visible.
Andrea
-
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/