Date: Mon, 17 Jan 2000 09:35:57 -0800 (PST)
From: Linus Torvalds <torvalds@transmeta.com>
The thing is, that you can do that as easily in user space by having a
"ext2fastcheck" thing or similar, something that CAN make sense.
[ Sorry for the possible resend, but I accidentally sent this directly
while dialuped up from my Compuserve (since I'm currently travelling
on business), and I know from bitter experience that Linux-kernel
blocks mail sent from there. So I have to manually transfer the mail
to MIT, and then resend the mail from there. -- Ted ]
Agreed, and I was about to send you patches to take out the checks
altogether.
Stephen has asked me to keep the checks, though, since it's a useful
double-check after running the journal for ext3 to makes sure things are
sane. And, the current ext3 design calls for the journal to be run at
mount time from inside the kernel so that there's no need to run an
external program. (This is especially useful when mounting the root
filesystem.) If that's the case, you'd want to run the checks after
running the journal, which implies leaving the checks in the kernel.
What I intend to do is to #ifdef out the checks (so if necessary they
can be re-enabled for ext3), and re-arranging some of the current
STRICT_CHECK checks so that they happen all the time --- they're super
cheap, so there was no reason to not do they by default, and in that
case it provides almost just as much protection as doing the mount-time
checks. I also want to change some ext2_warning() calls to ext2_error(),
since if ext2_free_block finds out that a block it was freeing had its
bit already cleared in the bitmap, there's almost certainly some other
corruption on the filesystem, so it should be treated as an ext2_error
--- in which case the kernel can either ignore the error, panic and
reboot, or remount the filesystem read-only, depending on how the
filesystem is configured using tune2fs.
Fair enough? If so, you can expect to see patches soon....
- Ted
-
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/
This archive was generated by hypermail 2b29 : Sun Jan 23 2000 - 21:00:17 EST