Re: Re [2]: [PATCH] Mark CONFIG_UFS_FS_WRITE as BROKEN

[Alexey Dobriyan <adobriyan@gmail.com>]

Thanks, these two patches makes things better but not much better.

1.

        inode->i_blocks = sb->s_blocksize / UFS_SECTOR_SIZE;
+       inode->i_size = sb->s_blocksize;
        de = (struct ufs_dir_entry *) dir_block->b_data;

This creates directories which are 2048 bytes in size. Native ones are
512 bytes.

	inode->i_size = 512;

makes mkdir and rm reliable for me both on linux and OpenBSD.

2. Second patch indeed makes hangs disaapear. However, data corruption
   is still in place:

Try

	for i in $(seq 1 10000); do echo $i; done >10000-linux.txt
	cp 10000-linux.txt >/mnt/openbsd/10000-openbsd.txt

Now corruption structure is following:

00000000  39 0a 33 30 39 30 0a 33  30 39 31 0a 33 30 39 32  |9.3090.3091.3092|
	[snip]
000007f0  33 34 39 36 0a 33 34 39  37 0a 33 34 39 38 0a 33  |3496.3497.3498.3|
00000800  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
00004000  36 36 0a 36 33 36 37 0a  36 33 36 38 0a 36 33 36  |66.6367.6368.636|
	[snip]
000047f0  0a 36 37 37 33 0a 36 37  37 34 0a 36 37 37 35 0a  |.6773.6774.6775.|
00004800  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
00008000  36 34 33 0a 39 36 34 34  0a 39 36 34 35 0a 39 36  |643.9644.9645.96|
	[snip]
000086f0  39 38 0a 39 39 39 39 0a  31 30 30 30 30 0a 00 00  |98.9999.10000...|
00008700  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
0000bef0  00 00 00 00 00 00 00 00  00 00 00 00 00 00        |..............|
0000befe

There are all zeros in 0800-4000 and 4800-8000 range. 0000-3800 from
original file is dropped (that 9 it's from 3089).

-
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/