uhci.c would call the completion callback when the call to submit_urb
failed. This is a rare situation.

This patch only calls the completion handler if the URB successfully
completed immediately (as in the case of talking to the virtual root
hub).

JE

My previous patch which cleaned up some of the spinlocks, moved one of
the spinlocks around a call to kmem_cache_alloc. It would sometimes
erroneously call it with GFP_KERNEL.

This patch fixes the problem by always calling it with GFP_ATOMIC.
Thanks to Greg for pointing this out to me.

JE

    - refer to "usbfs"
    - describe the /proc/bus/usb/BBB/DDD files
    - more info about the .../drivers and .../devices
    - ... generally, gives more information.
  
This is ever so slightly forward looking in how it describes
bandwidth requirements for high speed periodic transfers,
it's expecting a bugfix patch that's in my queue.  (That info            
is currently broken/meaningless.)

  
This updates the Philips tweak so that it also applies
to late-model ICH chips from Intel.  (Or so I'm told ... :)
That's at least three EHCI implementations known to
behave on Linux.  (And one hopes VIA soon too...)

It also cleans up a few comments relating to 64bit DMA;
recent API spec updates make it look like no games are
needed with the PCI DMA mask -- it doesn't change the
segment used by pci_pool allocations, so it just needs
to get turned on if the overall system does 64bit DMA.

I've a query in to see if those Philips/Intel tweaks are
chip quirks, or just something the EHCI 1.0 spec isn't
at all clear about.

inline cleanup to save a chunk of memory in usb-ohci

Comment and documentation cleanups

This patch completely punts on passing SLAB_POISON, I've
gotten burnt by it once too many.  Seems like the slab code
changed somewhere.  I've got a separate patch to make the
pci_pool code use CONFIG_DEBUG_SLAB, which I'll send
around.
    
Meanwhile, I needed this to get at least EHCI to intialize on
a 2.5.7-pre1 system.  Please merge.

This patch adds a missing "break", which prevented low and full
speed periodic transfers from getting through the "hcd" framework
to the driver (OHCI, for now).  Sigh... :)

added more printer quirks to the list

added HP DeskJet 959C to the quirks list

This patch remove unused variable/function/define, and small indent
cleanup.

The following patch moves MSDOS_SB() and MSDOS_I() into #define
__KERNEL__.

- Apply more patches from Vojtech Pavlik for the handling of host chip setup.
   Hopefully they are settled now.

- Kill unused CONFIG_BLK_DEV_MODES

- Push register addressing down in to task_vlb_sync.

- Make the taskfile parsing stuff actually readable. This is compressing the
   code by an incredible amount. We use just one function doing the whole
   scanning right now. This should make sure that the IRQ handler used by a
   particular command is always right.  I didn't introduce typos hopefully
   here.

- Don't call ide_handler_parser as argument for do_taskfile() any longer. We
   have killed this function by coalescing it's functionality with
   ide_cmd_type_parser() anyway.

- Kill unused SLC90E66 code, which Vojtech apparently missed in his patch.

- sync up with 2.5.7-pre2

Once again the actual patch is rather big mostly due to the removal of
some default configuration variables which are not used anylonger. So time for
the next patch stage.

- add joystick support for CS46xx driver
- Audigy code updates
- fix sound/core/Config.in (wrong dep_tristate usage)
- rawmidi interface fixes (memory leak)
- chang spinlock to rwlock in pcm_native.c (streams linking)
- further fixes of dependencies in Makefiles
- remove experimental time-sync support from sequencer
- fix/update for 32-bit -> 64-bit ioctl converter code
- wavefront driver cleanups
- CMIPCI driver updates
- update joystick support in CS4281
- add detection (not support) of M Audio Delta1010LT
- add AMD768 PCI ID to intel8x0 driver
- add joystick code to trident driver
- remove static variable initialization to zero

This adds initial driverfs support to ISAPNP driver.  It was approved by
the ISAPNP maintainer (Jaroslav Kysela).

Seperates affs_sb_info from struct super_block.

Seperates hfs_sb_info from struct super_block.

Seperates shmem_sb_info from struct super_block.

	file_system_typer-related code moved from fs/super.c to
fs/filesystems.c

into home.transmeta.com:/home/torvalds/v2.5/linux

Seperates udf_sb_info from struct super_block.

Seperates isofs_sb_info from struct super_block.

I got burnt one too many time by mismatches between
the pci_pool and "real" slabs... something changed in
mm/slab.c and broke a driver, so I'm going for the real
fix this time.  Having poisoning that _works_ is a huge
help in the innards of the USB host controller drivers.

This patch gets rid of some #ifdefs and makes the pci_pool
code poison memory if CONFIG_DEBUG_SLAB is set.
The functionality has always been there, but this makes
it simpler to get at.

into home.transmeta.com:/home/torvalds/v2.5/linux

exiting heathrow_modem_enable() with a lock held.

SI_DETHREAD, plus a couple of minor fixes.

into home.transmeta.com:/home/torvalds/v2.5/linux

Remove remnants of zlib_fs.

Remove FS_REQUIRES_DEV from JFFS2. We never really used the block device anyway.

From Arnaldo Carvalho de Melo.

added NEC printer to quirks list

  
Workaround for the ATEN switches

Several functions in the serial drivers can be called from bottom
half or interrupt context. They must use the GFP_ATOMIC flag for
calls to kmalloc() and usb_submit_urb().
  
Functions which must use GFP_ATOMIC:
1. All *_callback() functions.
2. Any code which is inside a spinlock.
3. write(), throttle(), unthrottle(), which may be called by
   the line discipline in bottom half context.
  
Functions which can use GFP_KERNEL:
1. open(), close(), startup(), shutdown(), set_termios().

  
Here is a patch to add support for F5U011 to catc.c driver. The
patch has been compile tested against 2.5.6 and 2.5.7pre1
(and tested against 2.5.5-dj1) and should apply cleanly.

  I've written a patch Vojtech and I discussed for enhancing the
hiddev code to optionally provide more detailed output on read().
The old functionality is still supported by default, and in
situations where HID usage codes are unique across reports, the
old method is still preferable due to its terseness.
    
  The new method provides the ability to determine exactly which
value has changed, in cases where the HID usage codes are not  
unique.  It also provides a means to optionally receive notification
when input reports are received from the device, whether or not
any of the values in the report have changed.

  The details of the changes are as follows:
  
  - All current code behaves identically

  - A new ioctl pair HIDIOCGFLAG/HIDIOCSFLAG gets and clears
    flags on the hiddev device.                             

  - If you set the flag HIDDEV_FLAG_UREF, the read() call switches
    from reading hiddev_event structures to hiddev_usage_ref
    structures.  The change takes effect immediately, even to
    already queued events that haven't been read() yet.  Here's
    an example of enabling FLAG_UREF:                          

    {
        int flag = HIDDEV_FLAG_UREF;
        if (ioctl(fd, HIDIOCSFLAG, &flag) != 0) {
                perror("ioctl");
                exit(1);
        }
    }
  
  - With the HIDDEV_FLAG_REPORT set (which is only allowed if
    HIDDEV_FLAG_UREF is also set), there is a special uref that
    will be read() in addition to the ones corresponding to
    changes in the device state: when uref.field_index is set to
    HID_FIELD_INDEX_NONE, this uref is a notification that the
    report referred to by report_type and report_id has been
    received from the device.  This can be useful in situations
    when the notification of the arrival of a report is useful
    even if there is no change in state.