I strongly recommend that this patch be applied.

Without this patch, running tcpdump on an SMP machine with the
pegasus adapter locks up 9 times out of 10.  This is what happens:

CPU0				CPU1
pegasus_start_xmit
	netif_stop_queue
				pegasus_set_multicast
					netif_stop_queue
					netif_wake_queue

Which crashes if another packet is sent to pegasus_start_xmit before
the first one is finished.

If there are other drivers doing this, please fix them too.

PS I submitted this to Petkan years ago but he probably lost it.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

For some reason, gcc-2.95.4 dies horridly on those asmlinkage
declarations.

I was unable to work out _why_ those functions have asmlinkage, as there
seem to be no instances of them - nobody calls
pwc_register_decompressor().
 What's up with that?
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

took me a while to get my external usb drive running under linux-2.6.6,
appears to need the appended fix. according to the header in unusual_devs.h
one should send them to you, so here you go.

the funny thing is that it works fine unter linux-2.6.0-test11 on different
hardware, but then i'm out of my depth concerning what might have been
changed in the usb driver in the meantime. i don't really follow the
changes. i didn't find any real documentation about what these flags do,
only a couple of "my usb drive worked with 2.4 but doesn't work any more
with 2.6" messages on the web, and some of them pointed to unusual_devs.h. i
do not claim to know what i've done with this fix, but i'd like to see it
officially included. :-)


T:  Bus=04 Lev=01 Prnt=01 Port=02 Cnt=01 Dev#=  2 Spd=480 MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs=  1
P:  Vendor=0402 ProdID=5621 Rev= 1.03
S:  Product=USB 2.0 Storage Device
S:  SerialNumber=00042222200000113608
C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr=  0mA
I:  If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Thanks to Gordon Elam <gbe@shoremicro.com> for the information needed
to do this.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Attached you will find patches that will bring the PWC driver in the kernel
up to version 9.0.1 from 8.12 (9.0 wasn't accepted at first). Patches are
against 2.4.26 and 2.6.7. The main difference with 9.0 is that the
video_relase() routine is now hopefully in line with kernel requirements.
I've also added one more ioctl() call, upon request.

I'm also slightly in the dark on the status of the PWC in the 2.6 kernel;
I've seen two patches: the first was a bad one, since it would crash your
kernel when you unplug the cam. I've seen a second patch to reverse the
first one, but I don't know if that went in. Either way, you might get a
conflict in pwc-if.c in and about the pwc_video_release() routine; this
patch was generated against the clean 2.6.7 kernel source. Should you need
or want to fix it manually, this patch should remove the
pwc_video_release() in the 2.6 kernel.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch causes the usb-storage driver to return an error indication
(DID_ERROR together with SUGGEST_RETRY) when the amount transferred by a
SCSI command is smaller than the "underflow" amount.  Some devices, like
the iRiver H100 series, occasionally transfer 0 bytes with GOOD status.
This change will make the sd driver aware that something unexpected has
happened so it can retry the command.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Matthew Dharm <mdharm-usb@one-eyed-alien.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

The kaweth usb ethernet driver is pretty noisy.  It generates over 40 lines
during a modprobe.  The following patch reduces the prints generated by
this driver.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch adds AIO support to gadgetfs, letting user mode programs use
"libaio" to manage USB I/O concurrency with the same API as they may
already be using for disk files.  In particular, it's practical to
stream isochronous data to/from userspace using this API, by keeping
an endpoint's I/O queue from emptying.

Each AIO "iocb" in userspace corresponds directly to one "usb_request"
(and one kiocb) in the kernel.  Their lifecycles, including cancelation,
overlap completely.  That's much of why the patch is so small (surprised
the heck out of me!); that, and using copy_{to,from}_user() rather than
trying fancy dma mapping tricks to attain zerocopy nirvana.

The kernel AIO module forgot to export kick_iocb(), so this adds the
missing declaration ... needed when using gadgetfs as a module.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

kaweth fails to cleanly shut down operations upon shutdown of the
controller.
Signed-Off-By: Oliver Neukum <oliver@neukum.name>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

you are using GFP_KERNEL in irq. That's illegal. GFP_ATOMIC must
be used.
Signed-Off-By: Oliver Neukum <oliver@neukum.name>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Some USB device controllers make it easy to handle all the various ways
hosts interpret the USB spec about when control-IN transfers need to send
a ZLP ... they can just send one if the host asks, or start the status
stage whenever the host thinks it's time.  Other controllers make it hard
to be forgiving in those cases.

This patch updates all the gadget drivers to explicitly set the req->zero
flag to reflect whether the USB spec says a ZLP "must" be sent by the
device.  "Forgiving" drivers won't notice the change, but the others need
to see this information passed down from the gadget driver.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

In a recent thread it was mentioned that some architectures, such as PPC,
do not guarantee relative ordering between different varieties of
processor/device data transfers.  This patch adds an mb() instruction to
the UHCI driver, to insure that the data structures in memory (cached by
the CPU and accessed by DMA) are fully flushed before the controller is
started (by regular bus IO).
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

A tricky problem the dummy_hcd driver has to solve is keeping track of the
usb_device structure that corresponds to a registered gadget.  Right now
that's not done very robustly.  This patch stores the address of the
structure when a new URB is submitted and also acquires a reference to
make sure that completing the final URB won't deallocate the structure
before dummy_hcd is through with it.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Thanks to Ludovic Aubry for this patch.  It changes some local variables
used by the UHCI driver to store IO addresses from unsigned int to
unsigned long.  This is vitally necessary on 64-bit platforms.




From: Ludovic Aubry <laubrycomm@free.fr>
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This fixes a long-standing bug in the hid-tmff driver that
causes it to not work at all.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Build fixes for ethernet gadget on PXA.

From:          Frank Neuber <fn@kernelport.de>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Various tweaks to EHCI IRQ handling, these may affact some systems.

 - Delays enabling IRQs until the root hub is more fully set up, so
   any "resume detect" IRQs can be handled properly.  (Craig Nadler)

 - Power down ports on driver shutdown.  (Craig Nadler)

 - Remove some duplicate irq-sharing logic that somehow crept in; check
   only once, and return IRQ_NONE to detect IRQ storms better (db)

 - Minor comment fix re integrated TTs.  (db)

From:          Craig Nadler <craig.nadler@arc.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

I overlooked that setting control lines in open can fail.

  - check for error doing control transfers
Signed-Off-By: Oliver Neukum <oliver@neukum.name>
Signed-off-by: Vojtech Pavlik <vojtech@suse.cz>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

  - races with urb->current, union header evaluation, DMA handling
Signed-off-By: Oliver Neukum <oliver@neukum.name>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch fixes another bug in the dummy_hcd driver.  When a gadget
driver unregisters (simulating a device disconnect), dummy_hcd stops the
timer that it uses for processing outstanding URBs.  Unfortunately this
means that those URBs will never be given back.  The patch lets the timer
continue running; when it expires all the remaining URBs will fail so the
timer won't be reinitialized.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch implements the missing functionality necessary to get device
resets working fully.  It adds a bit-array of ports with logical
connect-changes pending to the hub structure, so that the hub driver can
recognize that these ports should be treated as though they had
disconnected and re-connected.  This is how we will handle devices that
"morph" (i.e., change their descriptors) following a reset, as might
happen with a firmware upload.

There is also a lot of additional kerneldoc and a few small changes to
some log messages.  An important restriction is that usb_reset_device()
will refuse to work if the device is suspended.  Trying to reset a
suspended device leads to several problems, not least of which is that the
device's parent hub might be suspended as well.

With this patch the device reset code is pretty much complete.  However it
won't always work correctly until the device locking is straightened out.
That's coming up next.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

Also clean it all up - use get_cpu()/put_cpu() instead of
playing games by hand with preemption.

Here is a simple patch which drops some out-of-date code in the w83781d
and w83627hf i2c chip drivers. These bits are left over from the times
when chip drivers were setting default limits at init.


Signed-off-by: Jean Delvare <khali at linux-fr dot org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

into kroah.com:/home/greg/linux/BK/driver-2.6

Al missed this one in his sparse fixes
Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

From: Mika Kukkonen <mika@osdl.org>
Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

From: Mika Kukkonen <mika@osdl.org>
Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

From: Mika Kukkonen <mika@osdl.org>
Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Split ata_eng_timeout into the main part (ata_qc_timeout)
and the part called by the SCSI layer (ata_eng_timeout).

Zero behavior changes.