On Sat, Jul 15, 2006 at 12:45:30AM -0400, Theodore Tso wrote: > On Sat, Jul 15, 2006 at 04:09:29AM +0200, Zoran Dzelajlija wrote: > > tar xfSj theres-a-sparse-file-in-this-archive.tar.bz2 > > > > will just uncompress the archive (about .5GB), and tar will seek over the > > holes when creating the file. > > mmm, yes, I had forgotten about the -S option. However, tar is still > exec'ing bunzip2 when you use the -j optoin, so 200GB is still going > to be sent across stdout.
I don't think so. > So it's no better than explicitly using bunzip2, but no worse, either. But, bzip2 is used like bzip2 -d foo.tar.bz2 tar xfS foo.tar ... and foo.tar is only 500MB, not 200GB. Compare, if you wish: wget http://ext3:[EMAIL PROTECTED]/~jelly/linux/hdb1.e2i.bz2 time bunzip2 < hda1.e2i.bz2 | make-sparse >hda1.e2i wget http://ext3:[EMAIL PROTECTED]/~jelly/linux/hdb1.e2i.tar.bz2 time tar xfSj hdb1.e2i.tar.bz2 What's more interesting is that the compression took a lot more time when it was done in the pipe. 3:50 for e2image|bzip2, and 1:05 for e2image; tar; bzip2 combined. Anyway, back to the bug report: running e2fsck without -n fixes the htrees for two directories, and a repeated e2fsck finds no errors. Accessing the directory which would previously make kernel oops now works too. # e2fsck.static -vfC0 /dev/hdb1 e2fsck 1.39 (29-May-2006) Pass 1: Checking inodes, blocks, and sizes HTREE directory inode 1060239 has an invalid root node. Clear HTree index<y>? yes HTREE directory inode 1630724 has an invalid root node. Clear HTree index<y>? yes Pass 2: Checking directory structure Pass 3: Checking directory connectivity Pass 3A: Optimizing directories Pass 4: Checking reference counts Pass 5: Checking group summary information storage: ***** FILE SYSTEM WAS MODIFIED ***** 1235327 inodes used (49%) 9141 non-contiguous inodes (0.7%) # of inodes with ind/dind/tind blocks: 77373/3656/4 45456181 blocks used (93%) 0 bad blocks 4 large files 1122207 regular files 83709 directories 3179 character device files 7777 block device files 51 fifos 1092 links 18312 symbolic links (17309 fast symbolic links) 83 sockets -------- 1236410 files Maybe I should ask the author of "Ext2 IFS For Windows" to disallow writing if -O dir_index is present. Regards, Zoran -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
