RE: Bogus LBD patch

From: Peter Chubb (peter@chubb.wattle.id.au)
Date: Tue Jun 25 2002 - 01:59:02 EST


>>>>> "AmitR" == Amit Agrawal <Amit> writes:

AmitR> Hi Peter I've been found in your todo list ,in url u have
AmitR> given, that "large disc simulator" is planned, there is one
AmitR> solution developed by me regarding this,which ignores filedata
AmitR> and writes only metadata on the actual device so a large disc
AmitR> space is seen especially for big files. This produces a real

That sounds like a good idea.

AmitR> "large disc simulator" in you todo list.Please tell me your
AmitR> strategy for "large disc simulator" , if you have planned it

What I did was create a fake scsi device based on the HP simulator for
IA64, backed by a sparse file. So it's really a loop device that
obeys SCSI commands. It worked back on 2.5.9, but the changes since
then mean it deadlocks immediately now --- I intend to fix it, and
release it as soon as I can make some time.

The purpose was to test the SCSI layer's response to large discs.

What I'm trying to test now is what happens with real discs.
I found that ext2 has too much metadata for the amount of disc space I
have for the sparse file approach to work. JFS worked very well.
Reiserfs doesn't work at present, but I sent the bug report to them,
and they're working on the problems.

The testing I'm doing is:
    1. Is the size of the drive reported correctly by the kernel?
        -- at bootup
        -- via ioctl(GETBLKSIZE)
        -- via ioctl(GETBLKSIZE64)
        -- via lseek(,0, SEEK_END)
    2. Can the drive be partitioned?
        I expect Gnu parted with GDT to work properly; I expect the
        others to fail gracefully
    3. Can a file system be written to the drive?
        I've been testing ext2, jfs and reiserfs;
        I tested XFS on an earlier patch, but at present it's too much
        of a pain to integrate.
    4. If a filesystem is created, does fsck work?
    5. Can the filesystem be mounted?
    6. Then test how large a file I can create on the filesystem,
        both sparse and full.

--
Dr Peter Chubb				    peterc@gelato.unsw.edu.au
You are lost in a maze of BitKeeper repositories, all almost the same.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sun Jun 30 2002 - 22:00:09 EST