However, it looks like with an SMP machine that two processes can't
concurrently read, say, a regular file and a fast (non-sleeping)
device, since only one process can own the global kernel lock. I was a
little surprised at this, since I thought we'd pretty much gotten rid
of the global kernel lock in 2.1.x. Although, I have to admit I
haven't been paying close attention.
Anything which does anything referencing the vfs at all will need to
have the global kernel lock the entire time still.
Most inode only operations are protected with spinlocks, but not all.
Same for some dentry stuff, and again not all.
Don't expect things to "loosen up" until 2.3.x :-)
Later,
David S. Miller
davem@dm.cobaltmicro.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/