Merge do_boot_cpu() into the new style __cpu_up().

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

On multiprocessor systems the BogoMIPS for each CPU was reported was
the value for the last CPU having calibrated it's delay loop.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Define MAP_BASE for IP27

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Resurrect Cobalt support for 2.6.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Enable/disable irq's only if needed.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

-nostdlib boilerplate.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fix compilation, and bring 32/64 bit variants more in line.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fix wchan implementation, based on earlier by from Atsushi Nemoto.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Improved modules loader, more robust and works on 64bit kernels.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Until I figure out why NFS filesystems are having problems with
the 'load_irix_binary' and having kernel faults, Irix support is
disabled. I suspect locking of some sort, but I will now have to
investigate further.

Static IRIX binaries are now being detected properly and are using the
ELF interpreter found in this file.

Signed-off-by: Steven J. Hill <sjhill@realitydiluted.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Initialize iomem_resource.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Handle addresses beyond VMALLOC_END correctly.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Use intermediate variable.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

O2 doesn't have _that_ much RAM.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Changed all Au1x boards to noncoherent again.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

32-bit compatibility for various timer-related system calls.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Update to match the native siginfo structure and code.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

32-bit compatibility for ptrace GETEVENTMSG operation.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Implement 32-bit compatibility for waitid(2).

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Waitid(2) now has 5 arguments.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Use generic compat_sys_wait4 to implement 32-bit wait4(2).

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Allocate break code 513 to KDB.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Replace the complicated and broken attempt to clean interrupt by
something simple - doesn't need to be fast, after all.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

If you want RM7000 better fix it to build first ...

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

It's unwise to disable all interrupts of the boot node ;-)

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Rewrite to avoid the use of $at.  Unfortunately binutils 2.15 and CVS
binutils are broken and don't warn about this use of $at even though
gas is in .set noat mode so this for now is an accident waiting to
happen.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Enable RM7000 secondary cache for Atlas and Malta boards.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Moves a test which determines if we actually need to perform a
cacheflush to the right place.  That's a bug which is harmless on UP
but a severe bug on SMP.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Update MIPS to use the 4-level pagetable code thereby getting rid of
the compacrapability headers.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Document why calling smp_call_function will deadlock when called with
interrupts disabled.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fix ptrace aliasing issue in copy_from_user_page / copy_to_user_page.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

25Kf is also physically indexed.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Add a few more PrId vendor IDs.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

20Kc and SB1 don't suffer from aliases.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Delete duplicate copy of fixrange_init.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

rm9000_init() really is __init code.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Use preempt_schedule_irq.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

More oprofile bits for MIPS32-style performance counters.  The code to
bolt this into the actual hardware interrupt is yet missing from this
commit.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
 FEXPORT(ret_from_fork)

It works better when including arch/mips/sgi-ip27/Kconfig ...

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Move missplaced code line to the right place.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Provide functions to access cop0 config4-7 registers

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Use hardware mechanism to deal with cache aliases in the 24K.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Remove old wrong bits of cache code.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Clean up SEAD interrupt initialization.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Mask and ack CPU interrupts upon initialization.  Keep the state
of software interrupts when unmasking.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Move Sibyte Kconfig stuff into it's own Kconfig.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Replace deprecated interruptible_sleep_on() function call with direct
wait-queue usage.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Define __raw_read_can_lock / __raw_write_can_lock.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Update descriptions for MIPS Technologies evaluation boards.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Reenable EARLY_PRINTK for the DECstation.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Formatting fixes.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fix compilation; by Manish Lachwani.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fixes for system controllers for Atlas/Malta core cards.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Actually route cPCI interrupts.  Update inaccurate comments.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Fix TASK_SIZE for 32-bit processes on 64-bit kernels.

Fix register layout in o32 core dumps on 64-bit systems.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Get rid of early_init.  There's more need to make this form of
initialization actually useful and as is certainly unmergable with
upstream.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Dummy ISA DMA functions for systems that don't have ISA but share
drivers with ISA such as legacy free PCI.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fix build with SMP disabled and preemption enabled.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Move the invalid pmd and pte tables from .data to .bss. Fix alignment.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Remove unused arguments from preempt_{start,stop}/local_irq_{en,dis}able.
Don't clobber the preloaded TI_FLAGS in a2 needlessly.
Unexport local functions.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

Fix typos and formatting.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

De-optimize and decomplicate the spurious interrupt handler.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

[PATCH] type fix in arm/boot/compressed/misc.c

spot the typo...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] bluetooth hidp is broken on s390

Bluetooth HIDP selects INPUT and it really needs it to be there - module
depends on input core.  And input core is never built on s390...

Marked as broken on s390, for now; if somebody has better ideas, feel
free to fix it and remove dependency...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] idmouse cleanup and overflow fix

switched to simple_read_from_buffer(), killed broken use of min().
Incidentally, that use of min() had been fixed once, only to be
reintroduced in commit 4244f72436ab77c3c29a6447af81734ab3925d85:

    [PATCH] USB: upgrade of the idmouse driver

[snip]
-       if (count > IMGSIZE - *ppos)
-               count = IMGSIZE - *ppos;
+       count = min ((loff_t)count, IMGSIZE - (*ppos));

Note the lovely use of cast to shut the warning about misuse of min()
up...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] amikbd fix

it's input_allocate_device(), not input_dev_allocate()...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] missing exports of do_settimeofday() variants

frv, sh64, ia64 and sparc64 do not have do_settimeofday() exported (the
last two are using variant in kernel/time.c).  Exports added to match
the rest of architectures.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] missing bits in sparkspkr conversion

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] arguments out of order in class_device_create() call (s390)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] missing include in infiniband

use of IS_ERR/PTR_ERR in infiniband/core/agent.c, without a portable
chain of includes pulling err.h (breaks on a bunch of platforms).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] sata_sil24 iomem annotations and fixes

trivial iomem annotations + missing memcpy_fromio() caught by those

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s2io iomem annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fix ->signal->live leak in copy_process()

exit_signal() (called from copy_process's error path) should decrement
->signal->live, otherwise forking process will miss 'group_dead' in
do_exit().

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] blk: fix merge bug in as-iosched

as-iosched deals with aliased requests differently from other ioscheds.

It links together aliased requests using rq->queuelist instead of
spilling alises to dispatch queue like other ioscheds do.  Requests
linked in this way cannot be merged.

Unfortunately, generic q->last_merge handling patch didn't take this
into account and q->last_merge could be set to an aliased request
resulting in Badness, corrupt list and eventually panic.

This explicitly marks aliased requests to be unmergeable.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge branch 'master' of /linux/kernel/git/torvalds/linux-2.6

[IPV4]: Fix issue reported by Coverity in ipv4/fib_frontend.c

fib_del_ifaddr() dereferences ifa->ifa_dev, so the code already assumes that
ifa->ifa_dev is non-NULL, the check is unnecessary.

Signed-off-by: Jayachandran C. <c.jayachandran at gmail.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>

[ETH]: ether address compare

Expose faster ether compare for use by protocols and other
driver. And change name to be more consistent with other ether
address manipulation routines in same file

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>

[TCP]: fix tcp_tso_win_divisor documentation

The default value for tcp_tso_win_divisor is 3.

Signed-off-by: Akinobu Mita <mita@miraclelinux.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>

Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

Merge /pub/scm/linux/kernel/git/gregkh/usb-2.6

USB: fix up some odd parts due to partial merges

Thanks to Andrew for doing the hard work on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Merge /pub/scm/linux/kernel/git/sridhar/lksctp-2.6

[PATCH] USB: add nokia_dku2 driver

This driver comes from the gnokii project.
Was further cleaned up by me to match recent usb-serial core changes.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] USB: add new device id to ftdi_sio module

This is a patch to get the ELV FHZ1000 Home Automation control device to
work with Linux. The patch adds a new device ID to the ftdi_sio driver.
It is for kernel version 2.6.13.4.

Signed-off-by: Martin Hagelin <martin.hagelin@home.se>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] USB: fix correct wording in drivers/usb/net/KConfig

Signed-off-by: Koen Kooi <koen@handhelds.org>

[PATCH] USB: fix pm patches with CONFIG_PM off part 2

With CONFIG_PM=n:

drivers/built-in.o(.text+0x1098c): In function `hub_thread':
drivers/usb/core/hub.c:2673: undefined reference to `.dpm_runtime_resume'
drivers/built-in.o(.text+0x10998):drivers/usb/core/hub.c:2674: undefined reference to `.dpm_runtime_resume'

Please, never ever ever put extern decls into .c files.  Use the darn header
files :(

Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] USB: fix pm patches with CONFIG_PM off part 1

With CONFIG_PM=n:

drivers/built-in.o(.text+0x2a69c): In function `ohci_hub_control':
drivers/usb/host/ohci-hub.c:539: undefined reference to `.usb_hcd_resume_root_hub'
drivers/built-in.o(.text+0x2b920): In function `ohci_irq':
drivers/usb/host/ohci-hcd.c:726: undefined reference to `.usb_hcd_resume_root_hub'

Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] usbcore: Fix handling of sysfs strings and other attributes

This patch (as592) makes a few small improvements to the way device
strings are handled, and it fixes some bugs in a couple of other sysfs
attribute routines.  (Look at show_configuration_string() to see what I
mean.)

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] USB: Fix maxpacket length for ep0 on root hubs

This patch (as591) fixes a rather innocuous bug that has been around for
quite a long time: Virtual root hubs should have a maxpacket length of
64 for endpoint 0.  I didn't realize it was wrong until I started
looking through the endpoint attribute files.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] usbcore: Use kzalloc instead of kmalloc/memset

This patch (as590) fixes up all the remaining places where usbcore can
use kzalloc rather than kmalloc/memset.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] usbcore: Wrap lines before column 80

I can't stand text lines that wrap-around in my 80-column windows.  This
patch (as589) makes cosmetic changes to a couple of source files.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] usbcore: endpoint attributes track altsetting changes

This patch (as588) fixes the way endpoint attribute files are registered
and unregistered.  Now they will correctly track along with altsetting
changes.  This fixes bugzilla entry #5467.

In a separate but related change, when a usb_reset_configuration call
fails, the device state is not changed to USB_STATE_ADDRESS.  In the
first place, failure means that we don't know what the state is, not
that we know the device is unconfigured.  In the second place, doing
this can potentially lead to a memory leak, since usbcore might not
realize there still is a current configuration that needs to be
destroyed.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] usbcore: Improve endpoint sysfs file handling

This revised patch (as587b) improves the implementation of USB endpoint
sysfs files.  Instead of storing a whole bunch of attributes for every
single endpoint, each endpoint now gets its own kobject and they can
share a static list of attributes.  The number of extra fields added to
struct usb_host_endpoint has been reduced from 4 to 1.

The bEndpointAddress field is retained even though it is redundant (it
repeats the same information as the attributes' directory name).  The
code avoids calling kobject_register, to prevent generating unwanted
hotplug events.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] USB Storage: Expand range of Freecom unusual_devs entry

Freecom seems to be one of those vendors that can't get the GET CAPACITY
thing right. This expands the US_FL_FIX_CAPACITY flag for the entire
range of their fccd product line.

This is based on a patch sent by Stuart Black
<stuart_black@yahoo.co.uk>.

Signed-off-by: Phil Dibowitz <phil@ipom.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] mdc800: remove embrions of C++ exceptions

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] PATCH: usb-storage: use kthread API

This patch is originally from Alan Stern (as569).  It has been rediffed
against a current tree.

This patch converts usb-storage to use the kthread API for creating its
control and scanning threads.  The new code doesn't use kthread_stop
because the threads need (or will need in the future) to exit
asynchronously.

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 <gregkh@suse.de>

[PATCH] PATCH: usb-storage: implement minimal PM

This patch from Alan Stern started as as568.  It has been rediffed against
a current tree.

This patch adds minimal suspend/resume support to usb-storage.  Just enough
for it to qualify as PM-aware.

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 <gregkh@suse.de>

[PATCH] PATCH: usb-storage: allocate separate sense buffer

This patch is from Alan Stern (as560).  It has been rediffed against a
current tree.

This patch allocates a separate buffer for usb-storage to use when
auto-sensing.  Up to now we have been using the sense buffer embedded in a
scsi_cmnd struct, which is dangerous on hosts that (a) don't do
cache-coherent DMA or (b) have DMA alignment restrictions.

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 <gregkh@suse.de>

[PATCH] PATCH: usb-storage: move GetMaxLUN later in time

This patch is originally from Alan Stern (as557).  It has been re-diffed
against a current tree, and I also corrected a minor merging error.

Some time ago we introduced a delay before device scanning, because many
devices do not like to receive SCSI commands right after enumeration.
Now it turns out there's a device that doesn't like to receive
Get-Max-LUN right after enumeration either.  Accordingly this patch
delays the Get-Max-LUN request until the beginning of the scanning
procedure.  This fixes Bugzilla entry #5010.

Three things are worth noting.  First, I removed the locking code from
usb_stor_acquire_resources.  It's not needed, because the locking is to
protect against disconnect events and acquire_resources is only called
during probe (so the disconnect routine can't be called).  Second, I
initialized to 0 the buffer used for the Get-Max-LUN response.  It's not
really necessary, but it will prevent random values from showing up in
the debugging log when the request fails.  Third, I added a test against
the SINGLE_LUN flag.  This will allow us to use the flag to indicate
Bulk-only devices that can't handle Get-Max-LUN.

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 <gregkh@suse.de>