[PATCH] sparse: trivial parts of fs/* annotation

fs/*, fs/proc/*, fs/nfsd/*, fs/autofs4/*

The only interdependent group is the fs/autofs4/* stuff.

[PATCH] sparse: aio annotation

[PATCH] sparse: sg.c annotation

rest of pointers in sg ioctl arguments made __user.

[PATCH] sparse: scsi_ioctl __user annotation

scsi_ioctl() and scsi_ioctl_send_command() switched to __user *;
scsi/scsi_ioctl.c annotated.

[PATCH] sparse: scsi_cmd_ioctl __user annotation

scsi_cmd_ioctl() switched to __user *, block/scsi_ioctl.c annotated.

[PATCH] sparse: make sg_io_hdr->cmdp a __user pointer

copying moved into sg_io()

[PATCH] sparse: kernel/sysctl.c annotation and cleanup

[PATCH] sparse: trivial part of cdrom.c annotation

[PATCH] sparse: trivial part of drivers/scsi/* annotation

[PATCH] sparse: cdrom_generic_command split into kernel/user

The kernel instance (packet_command) has kernel pointers, userland one -
__user ones. Surprise.

[PATCH] more sparse checking of do_utimes()

Speaking of conflicts...  Here's another half of a do_utimes() patch,
same story as on ppc64:

Add __user pointer annotations to ppc64 code

Merge bk://cifs.bkbits.net/linux-2.5cifs
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge bk://bk.linux1394.org/ieee1394-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

adb.c needs <linux/devfs_fs_kernel.h>

Merge bk://kernel.bkbits.net/davem/net-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge bk://kernel.bkbits.net/davem/sparc-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

[PATCH] Missed RTC_IRQ patchlet

oh, hell - I missed this one when carving the thing out of big patch ;-/

Merge davem@nuts.davemloft.net:/disk1/BK/net-2.6
into kernel.bkbits.net:/home/davem/net-2.6

[PATCH] typo in OSS sparse __user annotations

Fix a missing initialization in the sparse __user annotation work for OSS.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Add __user pointer annotations to fs/compat.c

[NET]: Do net_todo_list empty check under semaphore.

Merge davem@nuts.davemloft.net:/disk1/BK/net-2.6
into kernel.bkbits.net:/home/davem/net-2.6

Merge davem@nuts.davemloft.net:/disk1/BK/sparc-2.6
into kernel.bkbits.net:/home/davem/sparc-2.6

Add __user pointer annotations to arch/ppc64/kernel/sys_ppc32.c

This also fixes a terminally broken sys32_utimes() implementation.

Fix sparse complaint about badly typed constant

[IPSEC]: Fix ref counting in __xfrmN_bundle_create().

[NET]: Simplify netdev_sysfs_xxx if SYSFS is not configured.

Don't need all the network sysfs code if CONFIG_SYSFS is not enabled.
Also:
* netdev_sysfs_unregister is declaration mismatch
* if netdev_sysfs_register fails print a warning.
  Need to still mark it as registered so the unregister_netdevice works,
  but we will probably end up leaking memory in that case.

[IRDA]: Fix NULL-ptr dereference in irlmp_get_saddr().

[IPV4]: Fix skb leak in igmpv3_newpack.

[IPV4,6]: Fix off-by-one in max protocol-type check

[IPV6]: Fix memory leak in ah6.c

[NET]: Prevent future missed updates of FOO_MAX macros.

[AF_KEY]: Set family for state selector.

As the comment in the code suggests, nobody really uses this so nobody
will notice the difference :) Still, let's make it correct while we're
at it.

[SPARC64]: compat select and futex need %o4 zero-extended.

Add __user annotations to arch/ppc64/kernel/process.c

Add __user annotations to arch/ppc64/kernel/syscalls.c

[TCP]: More sysctl tweakings for rcvbuf stuff.

1) Add sysctl to control rcvbuf moderation, off for now.
2) Set default winscale to zero.

Add __user annotations to ppc64 signal.c

[PATCH] sparse: trivial part of kernel/* __user annotation

[PATCH] sparse: partial mm/* __user annotation

[PATCH] sparse: rtc.c __user annotation

Also replacement of #if RTC_IRQ with #ifdef, moved inlined function in
front of all uses.

[PATCH] sparse: trivial part of drivers/ide __user annotation

[PATCH] sparse: ide-proc.c fixes

Fixed dereferencing userland pointers, general idiocy in parsing.

[PATCH] sparse: OSS __user annotation

caddr_t is gone (most of the instances were void __user * in disguise,
the rest - void *)

Fixed b0rken ioctls in mpu401 and pss - they used to dereference
userland pointers with no checks.

[PATCH] sparse: partial serial_core.c __user annotation

[PATCH] sparse: drivers/pci __user annotation

[PATCH] sparse: partial drivers/input __user annotation

[PATCH] sparse: trivial part of drivers/block __user annotation

[PATCH] sparse: trivial part of drivers/char __user annotation

[PATCH] sparse: acpi __user annotation

[PATCH] sparse: ipc __user annotation

[PATCH] sparse: ext2 __user annotation (ioctl)

[PATCH] sparse: nfs __user annotation (client only, and not touching RPC)

[PATCH] sparse: parport __user annotation (sysctls)

[PATCH] sparse: oprofile __user annotation

[PATCH] nfsd: missing dget()

From: Neil Brown <neilb@cse.unsw.edu.au>

The recentish change to fh_compose not consuming a reference to the passed
dentries missed this needed dget.

Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] vt_ioctl() comment fix

From: <Andries.Brouwer@cwi.nl>

The "Probably init should be changed" happened almost ten years ago.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] CREDITS file update

From: Robert Love <rml@ximian.com>

Update Robert's email address

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] CREDITS file update

From: <Valdis.Kletnieks@vt.edu>

Thomas Dunbar used to be working in one of the Dean's offices here.  He's
now down the hall from me (and has been for several years, actually).  So
let's fix the crufty pointers.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fs/dcache.c: avoid race when updating nr_unused count of unused dentries.

From: NeilBrown <neilb@cse.unsw.edu.au>

d_count == 1 is no guarantee that dentry is on the dentry_unused list, even
if it has just been incremented inside dcache_lock, as dput can decrement
at any time.

This test from Greg Banks is much safer, and is more transparently correct.

Cc: Maneesh Soni <maneesh@in.ibm.com>, Dipankar Sarma <dipankar@in.ibm.com>
From: Greg Banks <gnb@melbourne.sgi.com>
Signed-off-by: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] epoll events send fix

From: Davide Libenzi <davidel@xmailserver.org>

We're now sending epi->event.events while it should be sending
epi->revents.

Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ADT746X MAINTAINERS update

From: Colin Leroy <colin@colino.net>

I'm maintaining this driver i wrote :)

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fix unchecked return value in register_disk()

From: "Luiz Fernando N. Capitulino" <lcapitulino@prefeitura.sp.gov.br>

Signed-off-by: Luiz Capitulino <lcapitulino@prefeitura.sp.gov.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] asm-i386/timer.h docu cleanup

From: Adam Lackorzynski <adam@os.inf.tu-dresden.de>

This patch fixes a typo (intruupt), a possible confusion (ms should be
microseconds here, not milli), and adds descriptions for the missing
functions.

Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] use SLAB_PANIC for general caches

From: Brian Gerst <bgerst@didntduck.org>

Initialize the general caches using SLAB_PANIC instead of BUG().

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dm: change maintainer

From: <thornber@redhat.com>

Change of device-mapper maintainer

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dm-table.c: proper usage of dm_vcalloc

From: Kevin Corry <kevcorry@us.ibm.com>

dm-table.c: Proper usage of dm_vcalloc. [Dave Olien]

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dm: add static and __init qualifiers

From: Kevin Corry <kevcorry@us.ibm.com>

DM: Add static and __init qualifiers. [Dave Olien]

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dm-ioctl: replace dm_[add|remove]_wait_queue() with dm_wait_event()

From: Kevin Corry <kevcorry@us.ibm.com>

Some testing of DM multipath has turned up a problem with the DEVICE_WAIT
command.  In the tests, while performing a DEVICE_WAIT on a multipath device,
the command sometimes returns immediately, even though the event-number is
correct and no path-failure has occurred to trigger an event.  The problem was
tracked down to the call to schedule() in dev_wait(), which would return even
though it was not woken up by a DM table event.

This patch moves the responsibility for waiting from the ioctl interface into
the core driver, and uses wait_event_interruptible() instead of relying on
wait-queues and schedule().

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dm.c: free cloned bio on error path

From: Kevin Corry <kevcorry@us.ibm.com>

In __map_bio(), if the target returns an error while mapping the I/O, the
cloned bio needs to be freed.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dm-ioctl.c: fix off-by-one error

From: Kevin Corry <kevcorry@us.ibm.com>

Fix an OB1 error when calculating an output buffer size, that could cause a
missing null termininator in the 'list devices' ioctl results.  [Steffan
Paletta]

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] i386: put irq stacks in .bss.page_aligned section

From: Matt Mackall <mpm@selenic.com>

I noticed that my vmlinux BSS had grown from 17k to 45k between 2.6.5 and
2.6.6.  8k was moving a pair of objects in head.S from the text section to
bss, 8k was the introduction of IRQ stacks, while the remainder (12k) was
page alignment slop, some of it spurious.  The following patch brings BSS
down to the expected 33k.

Throw the IRQ stacks into the page aligned section to avoided wasted BSS
space.  While we'd expect this to save up to 4k, this saves over 10k of BSS
here due to gcc3.3 spuriously aligning other objects in the vicinity.

Signed-off-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Add `make checkstack' target

Jorn Engel <joern@wohnheim.fh-wedel.de>

Adds a `make checkstack' target.  This disassembles and parses vmlinux and
*.ko, generating a sorted list of stack hogs, eg:

0xc0100663 huft_build:                                  1456
0xc01022d3 huft_build:                                  1456
0xc0103127 inflate_dynamic:                             1328
0xc0101487 inflate_dynamic:                             1324
0xc010131f inflate_fixed:                               1168
0xc0102fb7 inflate_fixed:                               1168
0x00000003 sha512_transform:                            984
0xc024aee3 Vpd:                                         952
0x00000003 twofish_setkey:                              804
0x000060c3 isd200_action:                               792
0x0000e6cb nfsd4_proc_compound:                         760
0xc0249087 SkPnmiGetStruct:                             712
....

Supported architectures are alleged to be arm, x86, ia64, mips, mips64, ppc,
ppc64 and s390x.

Signed-off-by: Jörn Engel <joern@wohnheim.fh-wedel.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc64: xmon fixes

From: Anton Blanchard <anton@samba.org>

A few xmon fixes:

- Check we have the set-indicator property before calling rtas.
- We were printing a cpu number in one place with %s, fix.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc32: reorg DMA API, add coherent alloc in irq

From: Matt Porter <mporter@kernel.crashing.org>

Reorg PPC DMA API. Use PCI DMA API wrapper and add coherent alloc in irq.

Signed-off-by: Matt Porter <mporter@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc32: fix PPC4xx for the handle_page_fault() change

From: Matt Porter <mporter@kernel.crashing.org>

Fixes PPC4xx from the do_page_fault()->handle_page_fault() change.

Signed-off-by: Matt Porter <mporter@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc32: ix compiling arch/ppc/8260_io/enet.c

From: Tom Rini <trini@kernel.crashing.org>

When the 826x enet driver was made to call register_netdev/free_netdev,
no one declared 'err' in the function.

Signed-off-by: Tom Rini <trini@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] use new msleep() in ADT746x driver

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>

This patch replaces schedule_timeout() with the new msleep().

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Colin Leroy <colin@colino.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fix non-existent /dev/adb

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>

This patch fixes the lack of /dev/adb in kernel 2.6.7-rc1.  The call to
devfs_mk_cdev() has probably been removed too soon.  Hope this one is
better than the last one ;)

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Colin Leroy <colin@colino.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Fix typo in pmac_zilog

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>

This patch fixes a typo preventing channel B from working on the Rx path of
pmac zilog (never calling tty_flip_*).  I think I never tested channel B :)

Thanks to Hollis Blanchard for spotting the bug.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc64: remove silly debug path from get_vsid()

From: David Gibson <david@gibson.dropbear.id.au>

Currently the get_vsid() and get_kernel_vsid() functions have a test which
enables a different VSID algorithm for debugging.  Using a dumb VSID
algorithm for stressing the hash table is a reasonable debugging tool, but
switching it at runtime makes no sense at all.  Plus it adds another test
and memory access to the performance critical SLB miss path.

This patch removes the test, replacing it with a compile time switch.  It
seems to make a measurable, although small speedup to the SLB miss path
(maybe 0.5%).

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] d_bucket initialisation simplification

The recent nfsd fix was too fancy.  When initialising d_bucket to an "invalid"
value, all we need is a NULL in there.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge bk://linux.bkbits.net/linux-2.5
into hostme.bitkeeper.com:/repos/c/cifs/linux-2.5cifs

Missing spin lock init

[PATCH] radeonfb iBook & IGP fixes

This patch to radeonfb fixes support for the latest iBook models along
with an initialisation problem on some IGP chipsets. Please apply.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[sbp2]: Fix use-after-free bug.

Signed-off-by: Luiz Capitulino <lcapitulino@prefeitura.sp.gov.br>
Signed-off-by: Ben Collins <bcollins@debian.org>

Merge bk://bk.arm.linux.org.uk/linux-2.6-rmk
into ppc970.osdl.org:/home/torvalds/v2.6/linux

[ARM] Fix sparse complaint

Sparse found an instance where we were directly dereferencing a
userspace pointer in the signal handling code.  This change
fixes that.

Merge bk://kernel.bkbits.net/gregkh/linux/pci-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge bk://kernel.bkbits.net/gregkh/linux/i2c-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge bk://kernel.bkbits.net/gregkh/linux/usb-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Merge bk://kernel.bkbits.net/gregkh/linux/driver-2.6
into ppc970.osdl.org:/home/torvalds/v2.6/linux

Revert 3c905C initialization changes that cause problems

Cset exclude: akpm@osdl.org[torvalds]|ChangeSet|20040510205830|02933

[WATCHDOG] v2.6.6 linux/fs.h-patch

From: Christoph Hellwig <hch@lst.de>

All watchdog drivers need linux/fs.h

[WATCHDOG] v2.6.6 sc520_wdt.c-patch2

This patch also removes the cbar usage which is unnecessary. The MMCR is
always available at 0xfffef000; there is no need to use the cbar register
(if mmcr aliasing is enabled, then the MMCR is _also_ available at
another address set by CBAR).

[WATCHDOG] v2.6.6 sc520_wdt.c-patch1

Clean-up (general stuff: comments, keep module parameters together, ...)
Added clear definitions for the Watchdog Timer Control Register bits
Made start, stop and keepalive return 0 if successful
Fixed nowayout behaviour so that it is consistent with other watchdog drivers
Fixed release behaviour so that it is consistent with other watchdog drivers
Added wdt_set_heartbeat function to set the timeout/heartbeat of the watchdog
Made sure that memory remapping (wdtmrctl) is done before misc_register is started
MMCR_BASE_DEFAULT was wrong (Bug 2497 reported by Sean Young)

Tested by Sean Young

[WATCHDOG] v2.6.6 w83627hf_wdt.c

When drivers starts show the correct watchdog driver info.

Merge kroah.com:/home/greg/linux/BK/bleed-2.6
into kroah.com:/home/greg/linux/BK/i2c-2.6

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