Previously, sysfs read() and write() calls looked for sysfs_ops in the struct 
sysfs_dir, in the kobject. Since objects belong to a subsystem, and is a member
of a group of like devices, the sysfs_ops have been moved to struct subsystem,
and are referenced from there.

The only remaining member of struct sysfs_dir is the dentry of the object's 
directory. That is moved out of the dir struct and directly into struct kobject.
That saves us 4 bytes/object.

All of the sysfs functions that referenced the struct have been changed to just
reference the dentry.

A struct subsystem is basically a collection of objects of a certain type,
and some callbacks to operate on objects of that type. 

subsystems contain embedded kobjects themselves, and have a similar set of 
library routines that kobjects do, which are mostly just wrappers for the
correlating kobject routines. 

kobjects are inserted in depth-first order into their subsystem's list of 
objects. Orphan kobjects are also given foster parents that point to their
subsystem. This provides a bit more rigidity in the hierarchy, and disallows
any orphan kobjects.

When an object is unregistered, it is removed from its subsystem's list. When
the objects' refcount hits 0, the subsystem's ->release() callback is called. 

Documentation describing the objects and the interfaces has also been added.

It's now

int sysfs_create_link(struct kobject * kobj, struct kobject * target, char * name)

So, the caller doesn't have to determine the path of the target nor the depth of
the object we're creating the symlink for; it's all taken care of.

into osdl.org:/home/mochel/src/kernel/devel/linux-2.5-kobject

This removes a small thinko (2 of: n = *pos) in kernel/module.c's
s_start() function.

Ok, with this simple fix loop builds and loop works no worse than the
old shm/loop approach (better - it does boot).

However, both 2.4.44+ and this animal have the same and rather odd
breakage in loop-over-device.  Loop-over-file works fine, but loop over
device gives random crap on reads.  And no, it's not a problem with
underlying device itself - something happens in loop.c.

I'm going down right now, so I'll be back in several hours and will try
to debug it.  Hopefully by the morning I'll have a fix.

This ensures we deal properly with multiple perfctr overflow interrupts
under high load.

Make the test for supported GCC options into a macro, and add new checks
for -march={winchip-c6,winchip2,c3}.

The new resource interface foolishly replicated the (obsolete,
racy) spirit of the check_region call as check_resource.  You
should use request_resource/release_resource instead.

Fix an off-by-one error in the simplified SD_MAJOR macro which not only
botches up the sd majors but also steals away the major used by my
favorite slow-than-dirt SCSI RAID controller: cpqarray. Please, can I
have it back?

 - add lo->lo_blocksize
 - kill lo_get_bs() - great name, but...
 - set ->lo_device only if we do have a block device
 - pull determination of ->lo_blocksize into both branches - bdev
   variant gets it from lo_device and file one uses ->i_blocksize.
 - switched the ioctl getting information about underlying object
   to lo->lo_device ? stat.rdev : stat.dev
 - i.e. st_rdev of underlying object if it's a device and st_dev - if it's
   a file.
 - reverted the bogosity in shmem.c

This works on obviating the need for a separate data type to describe a sysfs
directory (which was renamed from struct driver_dir_entry to struct sysfs_dir).

All sysfs creation and removal functions now take a struct kobject, instead
of a struct sysfs_dir. This kobject is embedded in ->d_fsdata of the directory.

sysfs_create_dir() takes only 1 parameter now: the object that we're creating
the directory for. The parent dentry is derived by looking at the object's 
parent. 

sysfs_create_file() takes the object as the first parameter, and the attribute
as the second, which makes more sense from an API perspective. 

sysfs_remove_file() now takes an attribute as a second parameter, to be 
consistent with the creation function.

sysfs_remove_link() is created, which is basically the old sysfs_remove_file().
(symlinks don't have an attribute associated with them; only a name, which was
prohibiting the previous change). 

open() and close() look for a kobject now, and do refcounting directly on the
object. Because of that, we don't need the ->open() and ->close() callbacks
in struct sysfs_ops, so they've been removed. 

read() and write() also now look for a kobject now. 

The comments have been updated, too. 

This is not meant to be fancy; just something simple for which we can 
control the refcount and other common functionality using common code. 
   
The basic operations for registration and reference count manipulation are
included. 

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

into redhat.com:/home/jgarzik/repo/misc-2.5

Patch from Hugh Dickins

Added shmem_readpage and shmem_prepare_write so tmpfs files can be used
by the loop driver (together with simple_commit_write).  shmem_getpage
extended to accept file page passed in, which may have to be copied
over from swap page.

Use bdget and sb_set_blocksize so loop can see our preferred blocksize
PAGE_CACHE_SIZE.  Use copy_highpage, removed from highmem.h in 2.4.17:
restore it but with kmap_atomics.  Restore (a simple) copy_page to
asm-sparc64/page.h, which alone of all arches discarded it.

Patch from Hugh Dickins

From Matthew Wilcox <willy@debian.org>: shmem_getpage must
flush_dcache_page after allocating and clearing a new page.

Patch from Hugh Dickins

Checked shmem_file_write against recent filemap source, and against
2.4 and 2.4-ac: folded in missing fixes, mostly related to far file
positions.  Plus the new kmap_atomic copying technique.  But for now,
as before, no mark_page_accessed or SetPageReferenced in shmem.c: add
those, or whatever, later on when akpm has reviewed usage elsewhere.

Patch from Hugh Dickins

Added shmem_file_sendfile to allow sendfile from tmpfs.  Checked
do_shmem_file_read and shmem_file_read against filemap equivalents
to add in any recent fixes (-EINVAL when count < 0 was missing).

Patch from Hugh Dickins

Remove obsolete shmem_fs_type: we were in some doubt whether safe yet
to do so, then found 2.5.4 typo changed it from 2.4's "shm" to "shmem"
ever since: nobody complained, delete it - we're "tmpfs" since 2.4.4.
Use libfs' simple_empty and simple_sync_file instead of homegrown.
Remove exit_shmem_fs, it fools people that this might be a module.
Allow for faint possibility that shm_mnt could not be initialized.

Patch from Hugh Dickins

Here I intended a patch to remove the unsatisfactory shmem_recalc_inode
(which tries to work out when vmscan has freed undirtied hole pages,
to relax its blocks-in-use limit; but can only do so per-inode when it
needs to be per-super).  I had hoped to use the releasepage method, but
it looks like it can't quite be bent to this task, the page might still
be rebusied after release.  2.4-ac uses a removepage method dedicated to
this, but I'm reluctant to ask for such a minor address_space_operation.

So, leave shmem_recalc_inode as is, but avoid the issue as much as
possible, by letting shmem_getpage use the empty_zero_page instead of
allocating when a hole is read (but this cannot be done when it's being
mapped, nowadays the nopage doesn't know if page will be copied or not).
Whereupon shmem_getpage(,,,SGP_READ) can do partial trunc's holdpage.

Patch from Hugh Dickins

The last set of tmpfs patches left shmem_getpage with an inadequate
next_index test to guard against races with truncation.  Now remove
that check and settle the issue with checks against i_size within
shmem_swp_alloc, which needs to know whether reading or writing.

Patch from Hugh Dickins

shmem_getpage does need to lock its page (to secure it against
shmem_writepage), but it's easier for its callers if it unlocks
before returning.  The only caller who appeared to be using the
page lock was shmem_file_write, but it wasn't actually protecting
against anything - i_sem prevents concurrent writes and truncates,
and do_shmem_file_read was dropping the lock before copying anyway.

 - name_to_kdev_t() turned into name_to_dev_t(), callers updated.
 - table of names is gone, we use driverfs instead.

 - root name is converted to dev_t only at prepare_namespace() time - we
   use to do it in setup and we need it after driver initialization.  So
   setup only stores the root name and leaves the work to
   prepare_namespace().

 - disk names for rd and cm206 changed to match the old behaviour of
   root= parser: ramdisks have ram<n> in ->disk_name now (instead of
   rd<n>) and cm206 - cm206cd (instead of cm206).

This uses better constraints that do not go through the register
unneccessarily.

Do not oops when no swapfile is available and make it compile on
DISCONTIGMEM machines.

This patch does a few cleanups/fixes with __verify_write:
 - Only compile it when needed.
 - Move test for KERNEL_DS out of line.
 - The mmap semaphore is needed to access the vma list.
 - Use fixmap for the WP test.
 - Removes an obsolete comment in fixmap.h

It was copied from i386 and is unused.

This fixes PTRACE_GETFPREGS to initilize the fpu struct correctly on
cpus with fxsr, as well as removing redundant code.

 - fixes an idiocy with floppy_find() et.al. - they forgot to set
   *part to 0.  As the result, open() on anything other than fd0 had
   lead to interesting effects...

 - fixes off-by-1 in set_disk_ro().

implementation. The optimized versions only work
for large areas, make sure we don't use them for
anything else.

configuration system, and instead just cleanly show
the dependency.

The ifdef <actual cpu selection> is very bad style, we usually introduce
feature CONFIG_ options in config.in instead.

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux