Is this true?
I was looking through the source last night (2.1.106) and while
kflushd/swapd were started by the kernel, bdflush still appeared to be
the only thing which wrote out some meta data.
The bdflush source forks twice:-
o The first fork calls the obsolete syscall to start kflushd (which
just returns).
o The second call appears to periodically call the syscall with func==1
(I may have the number wrong, I don't have the source here). That
appeared to call a very similar function to the kflushd one, BUT it
included calls to two extra functions to write out some extra meta
data (which the source comments has no age and therefore is always
written out). These calls do not appear to exist in kflushd.
So is this periodic syscall REQUIRED to ensure all meta-data
eventually gets written? (could this be why some people are having
ext2 corruption?)
Cheers
Mark
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu