Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-02 01:35:21 -0700 (Sun, 02 Jun 2002)
Revision: 1586
Log message:

      
      Compiling version of file.c.
      
      However, the new inode function is not implemented in this commit, as I am
      still working on it. This is an older version of the file which passed the
      make command.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-03 11:56:36 -0700 (Mon, 03 Jun 2002)
Revision: 1587
Log message:

      Reverting last change and putting in new file.c (not sure why most of
      these other files also got hit, but there are no changes in them).
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-03 13:26:06 -0700 (Mon, 03 Jun 2002)
Revision: 1588
Log message:

      Added inode-specific name to chunk pathname resolution...
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-03 13:54:16 -0700 (Mon, 03 Jun 2002)
Revision: 1589
Log message:

      Adding general.h include
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-03 13:55:27 -0700 (Mon, 03 Jun 2002)
Revision: 1590
Log message:

      again..
      

Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-03 17:02:02 -0700 (Mon, 03 Jun 2002)
Revision: 1591
Log message:

      Fixed bug in dir.c in mojave_get_sbi. Now ls and other directory operations
      work just fine. Still having problems with the file operations though.
      

Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-04 01:02:00 -0700 (Tue, 04 Jun 2002)
Revision: 1592
Log message:

      
      I think I identified one of the problems in mojave_file_open. I only added a
      comment for now and I will try to fix it tomorrow morning. It refers to the
      fact that when a new file is created the chunk table associated with that
      file needs to be written as well.
      
      Another obvious problem is that ls will return the size of the chunk table
      rather than the size of the file (or something like that) and we need to fix
      that too.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-04 13:17:07 -0700 (Tue, 04 Jun 2002)
Revision: 1593
Log message:

      Added MOD_DEC_USE_COUNT to our put_super function.
      

Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-04 13:23:59 -0700 (Tue, 04 Jun 2002)
Revision: 1594
Log message:

      Updated the Makefile with run and rm options
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-04 13:48:53 -0700 (Tue, 04 Jun 2002)
Revision: 1595
Log message:

      Add functionality to dir_create so that it actually creates a chunk
      table header for the new file.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-04 13:52:51 -0700 (Tue, 04 Jun 2002)
Revision: 1596
Log message:

      Actually build these files now
      

Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-05 00:28:24 -0700 (Wed, 05 Jun 2002)
Revision: 1597
Log message:

      
      I finally fixed the locking problem. Here are the files we worked on today.
      I only modified _create, so that it passes the write of the header. However,
      something else is probably going wrong since the size of the file is still 0.
      
      I will try to look over it tomorrow morning.
      

Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-05 01:02:16 -0700 (Wed, 05 Jun 2002)
Revision: 1598
Log message:

      
      I added get_fs and set_fs around read and write calls but I still get:
      
      Wrote 646  out of 24 bytes
      Enter mojave_file_open!
      
      when I issue a touch.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-05 13:25:36 -0700 (Wed, 05 Jun 2002)
Revision: 1599
Log message:

      Small changes..
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-05 13:31:58 -0700 (Wed, 05 Jun 2002)
Revision: 1600
Log message:

      Small changes
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-05 19:21:57 -0700 (Wed, 05 Jun 2002)
Revision: 1601
Log message:

      Syncing my tree up (not sure why bin/Makefile is touched)
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 04:21:25 -0700 (Thu, 06 Jun 2002)
Revision: 1602
Log message:

      Chunk entries no longer rely on being looked up via pathname (e.g. in
      $MNT/.chunks). Unfortunately I'm still having trouble creating them, but
      we'll see how this goes. We might end up having to create the same
      structure (with the .chunks dir), but actually do it in a sane way, e.g.
      by looking everything up via dentrys (and having valid dentrys at that)
      rather than by strings. Currently the dentries are faked, with the only
      legitimate field being dentry->d_inode.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 10:28:32 -0700 (Thu, 06 Jun 2002)
Revision: 1603
Log message:

      Still running into NULL pointer problems when trying to write
      chunk...Here's the latest sync of the tree (not sure how different it
      is).
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 11:20:55 -0700 (Thu, 06 Jun 2002)
Revision: 1604
Log message:

      I changed the chunk granularity to be 0x100 so that we can really see
      whether or not our write/read is working. Our reading/writing is getting
      closer to working. Currently the problem is that the header does not
      appear to be written (it says it is, but when I try to read it later I
      still get num_entries = 0).
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 13:24:06 -0700 (Thu, 06 Jun 2002)
Revision: 1605
Log message:

      Properly build file struct in dir.c. Who knows what I changed in file.c
      (I suppose it wouldn't be too hard to check...).
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 13:37:08 -0700 (Thu, 06 Jun 2002)
Revision: 1606
Log message:

      Do a bare amount of clean up if we fail in a read or write.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 13:52:08 -0700 (Thu, 06 Jun 2002)
Revision: 1607
Log message:

      Small changes to tree.
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-06 16:11:17 -0700 (Thu, 06 Jun 2002)
Revision: 1608
Log message:

      We now appear to be writing the chunk tables correctly (for new files at
      least, I haven't tried modifying existing files as it will be a bit
      trickier). For some reason, however, mojave_file_read is never called
      when I try doing cat or cp on one of these files. My guess is that a
      page containing the file is being cached in memory, and that VFS tries
      to use that before actually trying to read off disk (I could be totally
      wrong).
      
      Even if that is the case, we might be able to get around it by replacing
      some ops (perhaps inode->i_mapping->a_ops??). For the moment, however,
      we do have a bit of progress.
      
      With regards to f_pos, *off, etc, I have done the following: I never
      change file->f_pos, but do modify the passed *off (e.g. I do
      *off += bytes_written to reflect a successful read). This appears to be
      the expected behaviour. There might be some weird stuff about this when
      we are modifying existing files (e.g. see comments about f_pos at
      http://www.cse.unsw.edu.au/~neilb/oss/linux-commentary/vfs-5.html)
      
      -jason
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-07 07:24:55 -0700 (Fri, 07 Jun 2002)
Revision: 1609
Log message:

      Latest sync of tree. Arrangement of reading/writing now appears to be
      correct. Apparently, however, I was mistaken about how much iget() will
      do for us. I think we may need to do a bit of work before reading will
      actually give us data back (i.e. store more data in each ct_entry).
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-07 07:34:57 -0700 (Fri, 07 Jun 2002)
Revision: 1610
Log message:

      Remove warning on compile of dir.c
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-07 10:09:35 -0700 (Fri, 07 Jun 2002)
Revision: 1611
Log message:

      Latest sync. Not sure what actually changed (nothing interesting)...
      

Changes by: Cristian Tapus (crt at cs.caltech.edu)
Date: 2002-06-07 12:22:44 -0700 (Fri, 07 Jun 2002)
Revision: 1612
Log message:

      Version 0.01 of slides for final presentation
      

Changes by: buckweat (buckweat at unknown.email)
Date: 2002-06-07 13:20:28 -0700 (Fri, 07 Jun 2002)
Revision: 1613
Log message:

      Today's slides