From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from majordomo by infradead.org with local (Exim 3.20 #2) id 14SLMf-0003l2-00 for mtd-list@infradead.org; Mon, 12 Feb 2001 15:58:09 +0000 Message-ID: <3A880E37.E5D3DFC8@danielind.com> Date: Mon, 12 Feb 2001 10:24:23 -0600 From: Vipin Malik MIME-Version: 1.0 To: David Woodhouse CC: mendel@mobach.nl, "mtd@infradead.org" Subject: Re: Errors with M-Systems' DOC v MD2200 D16 and ext2 References: <3A86F72B.A1971413@mobach.nl> <19636.981968145@redhat.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-mtd@infradead.org List-ID: David, I got something similar. I'm (I think) using the latest versions from CVS (nftl.c is ver. 1.66 and nftlmount.c is ver. 1.12). I just did a cvs update and no new versions came down. Here is my situation: I have a DOC2000 32MB w/ 0x4000 (16k) erase size. This causes nftl_format to bail out as it explicitely checks for an erase size of 0x2000 (8k) before it does anything. Since not having much to loose, I removed that check and went ahead and formatted the disk anyway. It did not complain. I could then modprobe in docprobe (After removing the module after the format). This time, it detected AND installed the NFTL device under /dev/nftla. I could then fdisk it and create a partition on it, create a fs on (ext2 with mke2fs) and mount it. My question is: Why was the explicit check being done? Are there any traps "lurking" for me now that I have removed the check and formatted my disk with nftl_format? While copying data to the disk (cp -a /bin /mnt/doc2000 etc.) I would get the data to copy, but after a few seconds (I guess when the cache is flushed to disk) I would get a bunch of ext2-fs errors like: end_request: I/O error, dev 5d:01 (unknown) sector 28 EXT2-fs error (device nftl(93,1)): ext2_write_inode: unable to read inode block - inode=73, block=14 .... ... repeat above for inodes 74...80 rest of error same (same sector (28), block =14). I then unmounted the fs and did an e2fsck on /dev/nftl1 I then got error messages regarding short reads reading block 14 (on sector 28). Same problem as above. Is this related to the format kludge that I did or something else? Thanks Vipin Vipin Malik Sr. Design Engr. Daniel Industries Houston, TX David Woodhouse wrote: > mendel@mobach.nl said: > > With the 2800 I've got no problems but with the 2200 I've got > > problems and get some very strange kernel messages > > Running out of free blocks to write to - again. This has happened to a few > people. I'm confused. I put a change into the CVS tree to attempt to fix > this - can you try it? Just drop nftl.c and nftlmount.c from CVS on top of > the versions in your kernel tree. > > -- > dwmw2 > > To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org