Currently with 2.5, when I suspend and resume my powerbook, I find
that the USB subsystem no longer sees root hub events, i.e. it doesn't
notice when I plug in a new USB device (it doesn't notice when I
unplug a device either but of course the driver for the device sees
that it is no longer responding).

It turns out that what happens is that the root hub timer goes off
after the OHCI driver has done its suspend stuff.  The timer routine
sees that the HCD is not running at the moment and doesn't schedule
another timeout.  Hence the series of timeouts stops.

This patch exposes DMA addresses in URBs.  It exposes new APIs that
let drivers be a bit smarter in terms of DMA, reducing USB overhead
on some platforms (but not commodity pcs).  As discussed with DaveM,
and on the usb-devel list.

Supporting patches are still needed.  Of course, there's teaching HCDs
to use _these_ addresses when they're provided (easy).  There's also
teaching drivers (like hid) to use the new usb_buffer_alloc() support,;
can happen incrementally.  And adding scatterlist support, which will
be desirable for usb-storage and hpusbscanner.  But this is the start
needed to get all of that going.

This patch makes EHCI

  - Share the same TD queueing code for control, bulk,
    and interrupt traffic;
  - Queue interrupt transfers, modifying the code for
    urb submit/unlink/complete;
  - Thinner, by removing lots of nasty fatty special case
    logic for interrupt transfers (size, no queueing, etc);
  - Grow some "automagic resubmit" logic, ready to be
    ripped out soonish;
  - Package its interrupt scheduling so it can be called
    from more places.

This patch teaches the ehci driver how to dump its schedule
through two driverfs files:

  "sched-async" shows the control/bulk queue.  usually has
     no tds, often the idle queue head is showing.
  "sched-periodic" shows the iso/interrupt schedule.

Since this is for debugging, the files aren't normally
configured into the driver.

Things are indeed as conjectured, and I can reproduce the situation
where usb_string() returns -EPIPE. Now that this is an internal
error code for the USB subsystem, and not meant to get out to the
user, I made these driverfs files empty in case of error.
(While if there is no error but the string has length 0,
the file will consist of a single '\n'.)

One fewer random memory corruption. Unfortunately, there are more.

Andries

The recent "unlink cleanups" patch had a problem
with bitmask byteswapping.

This patch is the first part of fixing the EHCI driver to queue
interrupt transactions, handle larger requests, and basically treat
interrupt as just "bulk that lives on the periodic schedule".  One
more patch should wrap that up.

     qh processing cleanup
	- split "append tds to qh" logic out of "put on async schedule",
	  so it can be used with "put on periodic schedule" too
     interrupt transfer cleanup
	- save rest of scheduling params in the qh
	- calculate scheduling params only once
     other cleanup
	- use new container_of()
	- minor code shrinkage (avoid pipe bitops, conditionals, etc)
	- rename variable (will track endpoints, not urbs)
	- free_config() logic

As requested by Adam Richter.

various compilers

This patch removes an artifact of code left over from the 2.0 days when
the kernel didn't use flat segments.

The preceding memcpy() already takes care of the NULL ptr mapping.

Unfortunately, someone changed the inb and friends definitions and
broke ARM.  This patch allows ARM to build NE2000 based drivers
again.

The following patch adds support for CONFIG_GENERIC_ISA_DMA, which went
into the 2.4-ac kernel series prior to 2.5 happening.

The following patch allows architectures to decide whether they want
the generic ISA DMA functionality provided by kernel/dma.c and other
supporting files.

In addition, we move the procfs "/proc/dma" support code out of fs/proc
into kernel/dma.c, and adapt it to use the new seq_file code.

The following patch is required so that munmap(0x8000, *) does not cause
ARM kernels to crash.  The problem is that the machine vectors are at
virtual address 0.  Unfortunately, when free_pgtables() is called, it
clears first level page tables starting at 0 in this case, and takes
out the machine vectors.  This then results in an unrecoverable hang.

We already have FIRST_USER_PGD_NR to define the first entry in the pgd
which may be cleared, so we use this to clamp "start_index"
appropriately.  The following patch does this.  Tested on ARM.

Bitops are used with on the timer_alive variable.  Therefore, timer_alive
needs to be "unsigned long" not "int".

Trivial patch; removes cruft that got left over from the TLB shootdown
changes.  "dir" is never used in zap_page_range

Various modules (eg, rd.c) use simple_strtol.  This patch exports the
symbol so these modules can be loaded.

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

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

This patch (against 2.5.30) contains all of the changes to generic code
needed by UML.

CREDITS -
	Changes my address and adds a UML credit for Lars Brinkoff.

MAINTAINERS -
	Adds a UML entry.

Makefile -
	When ARCH=um is on the command line to do a UML build, the old value
	of ARCH is saved in $(SUBARCH) for the later use of the UML build.

drivers/char/Makefile -
	Sets KEYMAP, KEYBD, CONSOLE empty to prevent hardware drivers from
	being compiled in.

drivers/net/setup.c -
	Backs out a UML hook which leaked into your tree which UML no
	longer uses.

This a preparatory patch which allows UML to avoid changing linux/linkage.h.

It restructures linkage.h so that all of the arch-specific stuff is in
asm-*/linkage.h.  linux/linkage.h is now arch independent.  It should be
functionally unchanged.

Five arches, i386, ia64, sh, m68k, and arm, have non-empty linkage.h files.
The other arch linkage.h files are all empty.

Also, __ALIGN_STR is no longer defined independently of __ALIGN.  It is now
derived by stringizing __ALIGN.

Following a suggestion by Keith Owens, stringify.h was generalized to allow
commas in its argument, linkage.h now includes it, and no longer defines its
own stringify macros.

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

From Rusty.
The old form of designated initializers are obsolete: we need to
replace them with the ISO C forms before 2.6.  Gcc has always supported
both forms anyway.

(rmk - cleaned up spacing to be more reasonable)

From: (via Rusty) Art Haas <ahaas@neosoft.com>
Here are patches for files in fs/adfs. Patches are against 2.5.26.

(rmk - updated for 2.5.30, cleaned up spacing to be more reasonable)

From: (via Rusty) James Mayer <james@cobaltmountain.com>

3c503, hp-plus, smc-ultra, smc-ultra32, wd

Feedback from review by Arjan @ Redhat:
o Cleanup: Removed unneccesary #ifdef/#endif wrappers for features 
  already part of kernel.
o Bug fix: Use kernel funcs pci_[clear|set]_mwi rather than using direct
  calls to pci_write_config_word.
o Bug fix: Added read-behind-write calls to post writes before delays.
o Bug fix: removed calls to mdelay in interrupt context.
o Clean up: removed calls to memset followed by mb() and replaced with
  direct clearing of descriptor bits.

o Bug fix: fixed jumbo frames sized from 1514 to 2048.
o ASF support: disable ARP when driver is loaded or resumed; enable when 
  driver is removed or suspended.
o Bug fix: changed default setting for RxIntDelay to 0 for 82542/3/4
  controllers to workaround h/w errata where controller will hang when
  RxIntDelay <> 0 under certian network conditions.
o Bug fix: perform controller reset using I/O rather than mmio because
  some chipsets try to perform a 64-bit write, but the controller ignores
  the upper 32-bit write once the reset is intiated by the lower 32-bit
  write, causing a master abort.
o Clean up: removed unused and undocumented user-settable settings for
  PHY.
o Bug fix: ethtool GEEPROM was using byte address rather than word
  addressing.
o Feature: added support for ethtool SEEPROM.
o Feature: added support for entropy pool.

Feedback from review by Arjan @ Redhat:
o Cleanup: Removed unneccesary #ifdef/#endif wrappers for features 
  already part of kernel.
o Bug fix: Added read-behind-write calls to post writes before delays.
o Cleanup: added "e100" prefix to all driver messages.
o Cleanup: removed unnecessary barrier() call because spin_unlock
  already implies a barrier.
o Cleanup: removed __NO_VERSION__
o Cleanup: fixed some spelling mistakes.
o Cleanup: removed some proprietary /proc entries.

o Added device ID support for Dell LOM.
o Added device ID support for 82511QM mobile nics.
o Bug fix: ethtool get/set EEPROM routines modified to use byte
  addressing rather than word addressing.
o Feature: added MDIX mode support for 82550 and up.
o Bug fix: added reboot notifier to setup WOL settings when
  shutting system down.
o Cleanup: removed yield() redefinition (Andrew Morton,
  akpm@zip.com.au).
o Bug fix: flow control now working when link partner is
  autoneg capable but not flow control capable.
o Bug fix: added check for corrupted EEPROM
o Bug fix: don't report checksum offloading for the older
  controllers that don't support the feature.
o Bug fix: calculate cable diagnostics when link goes down
  rather than when queuering /proc file.
o Cleanup: move mdi_access_lock to local get/set mdi routines.

unaligned accesses.

to after C+ command register Rx/Tx enable.

Verified by Nathaniel <wfilardo@fuse.net>

This patch fixes memory leak of driverfs_mknod(), and changes odd
error code.