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

The spin_unlock() in rotate_reclaimable_page() is not a sufficient memory
barrier.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

"clear_buffer_locked()" is just a nicer name for a clear-bit, and
on x86 (and others) we don't need any expensive memory barriers
after that.

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

kfifo_alloc tries to round up the buffer size to the next power of two.

But it accidently uses the original size when calling kfifo_init,
which will BUG.
Acked-by: Stelian Pop <stelian@popies.net>
Signed-off-by: Martin Waitz <tali@admingilde.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This is a relatively small changeset, to address small nagging problems.
Andrew pointed me at the double registration specifically, so I had to do
something about it. At least now Fabio's box won't collapse if he configures
UB by mistake. Also, a few people complained that the help text was
misleading.

I have not done anything about the oops on disconnect which happens to
Martin Schlemmer. It's next. After that I can get to Peter Jones' CD
burning patch and doing resets.

- Fix double kobject registration and eventual oops on unplug if a device
  is not cooperating.
- Add a reference to usb-storage into the configuration help.
- Just upping timeouts fixes my ZIP drive.
- Max out the diag (trace) file size.
- Set capacity to zero in case the media is absent. It works in 2.4.10-rc1.
  Either Al fixed the block layer, or the whole thing was a bug in ub.c
  (and sd.c) and a big misunderstanding.
Signed-off-by: Pete Zaitcev <zaitcev@redhat.com>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This clarifies more of the x86 caller/callee stack ownership
issues by making the exception and interrupt handler assembler
interfaces use register calling conventions.

System calls still use the stack.

Tested with "crashme" on UP/SMP.

We need to open the gap between congestion on/off a little bit, or we
risk burning many cycles continually putting processes on a wait queue
only to wake them up again immediately. This was observed with CFQ at
least, which showed way excessive sys time.

Patch is from Arjan.
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Patch is from Daniel Drake <dsd@gentoo.org>
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Daniel Drake <dsd@gentoo.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

Signed-off-by: David S. Miller <davem@davemloft.net>

Using idr_get_new_above in init_new_context lets us get rid of an
awkward init function which wasn't running early enough in boot
anyway.
Signed-off-by: Nathan Lynch <nathanl@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

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

The routine ext2_readdir() will, when reading a directory page
returns an error, try the next page, without reporting the
error to user space. That is bad, and the patch below changes that.

In my case the filesystem was damaged, and ext2_readdir wanted
to read 60000+ pages and wrote as many error messages to syslog
("attempt to access beyond end"), not what one wants.

[no doubt a similar patch is appropriate for ext3]

This preserves 2.4.x behavior.
Signed-off-by: David S. Miller <davem@davemloft.net>

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

This patch adds the cypress_m8 usb-serial driver for the Delorme Earthmate usb gps and the Cypress hid->com rs232 adapter to the kernel tree.
Signed-off-by: Lonnie Mendez <lmendez19@austin.rr.com>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

The patch below adds support for the KC2190 usb-to-usb networking
device; the version I have reports itself as:

Bus 001 Device 003: ID 050f:0190 KC Technology, Inc.

I was under the impression that support for this had been added a long
time ago, but searching through old kernel versions all I could find was
a comment about the chip, with no support. Patch is against 2.6.9 but is
pretty minimal.

I don't have a Windows box around to test interoperability with that
driver, but the patch appears to make it work perfectly between 2 Linux
boxes.
Signed-off-by: Jonathan McDowell <noodles@earth.li>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

It is updated outside of locks, so...
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

On OMAP-1510 the 48MHz DPLL used by USB is now set up earlier in
the boot process.  On Innovator, both host ports work by default.
(Although USB2 conflicts with ttyS2, which you might prefer.)
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch updates the serial gadget driver:

 - The bulk is Al Borcher's version 2.0 updates:

     * Use the new utilities to build config descriptors from tables.

     * Add CDC ACM support, which brings lots of goodness including
       interop with Windows.

     * Work better given PXA 25x hardware automagic.

 - I have a few minor tweaks.

     * Match the recent tty API change:  from_user is gone!!

     * Define new product ID for the CDC ACM version, so Windows
       registry won't get confused by the non-ACM version.

     * Add OTG descriptor

     * Allocates packet buffers more efficiently on platforms like
       OMAP: use kmalloc not the dma-coherent allocator.

Al has some docs on how to talk to Windows this, including an INF file.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This has minor patches to the ethernet, file_storage, and zero
gadget drivers.

 - Use longer "manufacturer" strings; CONFIG_LOCALVERSION would overflow.

 - Remove compile warning on file_storage with non-net2280

 - Oops fix in ethernet gadget, on boards that don't report VBUS
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This is the second of the patches from Dale Farnsworth to
support OHCI implementations that use big-endian byte sex
for registers ... all register accesses changed, that's why
it's so large.  (It also includes part of an update from
his third patch, so that it compiles/runs by itself.)

This patch adds support to the OHCI code for big-endian controllers,
maintaining the existing little-endian support.  This is done using
the following transforms:

        ohci_readl(p)   --> ohci_readl(ohci, p)
	writel(v, p)    --> ohci_writel(ohci, v, p)

	cpu_to_le16(v)  --> cpu_to_hc16(ohci, v)
	cpu_to_le16p(v) --> cpu_to_hc16p(ohci, v)
	cpu_to_le32(v)  --> cpu_to_hc32(ohci, v)
	cpu_to_le32p(v) --> cpu_to_hc32p(ohci, v)

	le16_to_cpu(v) --> hc16_to_cpu(ohci, v)
	le16_to_cpup(v) --> hc16_to_cpup(ohci, v)
	le32_to_cpu(v) --> hc32_to_cpu(ohci, v)
	le32_to_cpup(v) --> hc32_to_cpup(ohci, v)

	OHCI_FRAME_NO(ohci->hcca) --> ohci_frame_no(ohci)

There's no runtime penalty except for systems configured to handle
both byteorders at the same time.
Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This is the first of several patches from Dale Farnsworth,
laying groundwork for a patch that supports OHCI on more
chips (IBM STB04xxx and FreeScale MPC52xx).


Replace pre-byteswapped constants for in-memory data structures in ohci.h
with cpu-native constants, to help support "big-endian OHCI" hardware.
Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Collaborative work between David S. Miller and
Herbert Xu.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

On Fri, 29 Oct 2004, David Brownell wrote:

> Could you add a comment there that the HCD is
> now _required_ to return a pointer that usbcore
> can pass to kfree?  Since that's a layering
> violation (in anticipation of later changes to
> make it not be one!), it's surprising ... and
> in any case, the memory lifecycle is no longer
> described just by the calls in this header!

That's a reasonable request, and here's an extra patch to add the comment.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch removes the hcd release code from the final host controller
driver, dummy-hcd.  Please apply.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch changes the non-PCI-based OHCI-related host controller drivers,
removing the code that frees the driver-specific hcd structures.
Unfortunately I am not able to test it, because I don't have the necessary
hardware.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch removes the code for deallocating the usb_hcd structure from
the three PCI-based host controller drivers.  It also moves the embedded
struct usb_hcd member to the front of the larger driver-specific
structures, as required for the core to do its work.  Please apply.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch contains changes to usbcore making the core responsible for
deallocating memory for usb_hcd structures, rather than calling back into
the host controller drivers.  This solves a long-standing oops, since the
drivers may have been unloaded from memory by the time the release routine
is called.  (There probably are some Bugzilla entries caused by this, but
I don't know which ones.)

To really fix this up properly will require merging the HCD "glue" layer
into the higher core structures.  As things stand now, a usb_hcd contains
a usb_bus member, and the usb_bus is associated with a class_device: when
the class_device is released the usb_hcd gets deallocated.  It's awkward
in that neither the usb_hcd nor the usb_bus has its own reference count,
and the class_device is created when the bus is registered (not when it is
initialized).
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Neither requests to suspended devices nor control request stalls should
ever log fault messages, they happen routinely; this stops such logging.
There's also a better message for control/bulk timeout -- but after the
HCD finishes unlinking the URB, not from 'swapper' in the timer callback!
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Delete the second pci_set_master(), not the first one.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch adds equal support for interrupt out transfers to the usb serial
core to match the current interrupt in support.  It also improves a few
debug messages, nothing major.
Signed-off-by: Lonnie Mendez <lmendez19@austin.rr.com>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch reorganizes the startup and shutdown code in the dummy-hcd
driver to make it consistent with all the other host controller drivers.
For example, the platform device representing the HC hardware is
allocated separately and given as an argument to the probe() and remove()
routines, as it would be for a real device.

This paves the way to making widespread and uniform changes in all the
HCDs at once.  Please apply.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Although this is the first patch in a series of six, it's not closely
related to the others.  This includes a whole bunch of simple cleanups
for the dummy-hcd driver, all of which fall into the following categories:

	Convert explicit container_of() to type-safe inline functions,

	Minimize reliance on global variables.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

bits_wanted is expanded to bytes using the wrong shift value
(when iovp_shift != PAGE_SHIFT), resulting in an explosion of
used iommu resources.

This potentially results in mistakenly running out of DMA mapping
resources when the system is under *heavy* i/o load.
Signed-off-by: Nigel Croxon <nigel.croxon@hp.com>
Signed-off by: Alex Williamson <alex.williamson@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

Signed-off-by: David S. Miller <davem@davemloft.net>

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

into nuts.davemloft.net:/disk1/BK/net-2.6