[PATCH] detect soft lockups

This patch adds a new kernel debug feature: CONFIG_DETECT_SOFTLOCKUP.

When enabled then per-CPU watchdog threads are started, which try to run
once per second.  If they get delayed for more than 10 seconds then a
callback from the timer interrupt detects this condition and prints out a
warning message and a stack dump (once per lockup incident).  The feature
is otherwise non-intrusive, it doesnt try to unlock the box in any way, it
only gets the debug info out, automatically, and on all CPUs affected by
the lockup.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Signed-Off-By: Matthias Urlichs <smurf@smurf.noris.de>
Signed-off-by: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] FUTEX_WAKE_OP: pthread_cond_signal() speedup

ATM pthread_cond_signal is unnecessarily slow, because it wakes one waiter
(which at least on UP usually means an immediate context switch to one of
the waiter threads).  This waiter wakes up and after a few instructions it
attempts to acquire the cv internal lock, but that lock is still held by
the thread calling pthread_cond_signal.  So it goes to sleep and eventually
the signalling thread is scheduled in, unlocks the internal lock and wakes
the waiter again.

Now, before 2003-09-21 NPTL was using FUTEX_REQUEUE in pthread_cond_signal
to avoid this performance issue, but it was removed when locks were
redesigned to the 3 state scheme (unlocked, locked uncontended, locked
contended).

Following scenario shows why simply using FUTEX_REQUEUE in
pthread_cond_signal together with using lll_mutex_unlock_force in place of
lll_mutex_unlock is not enough and probably why it has been disabled at
that time:

The number is value in cv->__data.__lock.
        thr1            thr2            thr3
0       pthread_cond_wait
1       lll_mutex_lock (cv->__data.__lock)
0       lll_mutex_unlock (cv->__data.__lock)
0       lll_futex_wait (&cv->__data.__futex, futexval)
0                       pthread_cond_signal
1                       lll_mutex_lock (cv->__data.__lock)
1                                       pthread_cond_signal
2                                       lll_mutex_lock (cv->__data.__lock)
2                                         lll_futex_wait (&cv->__data.__lock, 2)
2                       lll_futex_requeue (&cv->__data.__futex, 0, 1, &cv->__data.__lock)
                          # FUTEX_REQUEUE, not FUTEX_CMP_REQUEUE
2                       lll_mutex_unlock_force (cv->__data.__lock)
0                         cv->__data.__lock = 0
0                         lll_futex_wake (&cv->__data.__lock, 1)
1       lll_mutex_lock (cv->__data.__lock)
0       lll_mutex_unlock (cv->__data.__lock)
          # Here, lll_mutex_unlock doesn't know there are threads waiting
          # on the internal cv's lock

Now, I believe it is possible to use FUTEX_REQUEUE in pthread_cond_signal,
but it will cost us not one, but 2 extra syscalls and, what's worse, one of
these extra syscalls will be done for every single waiting loop in
pthread_cond_*wait.

We would need to use lll_mutex_unlock_force in pthread_cond_signal after
requeue and lll_mutex_cond_lock in pthread_cond_*wait after lll_futex_wait.

Another alternative is to do the unlocking pthread_cond_signal needs to do
(the lock can't be unlocked before lll_futex_wake, as that is racy) in the
kernel.

I have implemented both variants, futex-requeue-glibc.patch is the first
one and futex-wake_op{,-glibc}.patch is the unlocking inside of the kernel.
 The kernel interface allows userland to specify how exactly an unlocking
operation should look like (some atomic arithmetic operation with optional
constant argument and comparison of the previous futex value with another
constant).

It has been implemented just for ppc*, x86_64 and i?86, for other
architectures I'm including just a stub header which can be used as a
starting point by maintainers to write support for their arches and ATM
will just return -ENOSYS for FUTEX_WAKE_OP.  The requeue patch has been
(lightly) tested just on x86_64, the wake_op patch on ppc64 kernel running
32-bit and 64-bit NPTL and x86_64 kernel running 32-bit and 64-bit NPTL.

With the following benchmark on UP x86-64 I get:

for i in nptl-orig nptl-requeue nptl-wake_op; do echo time elf/ld.so --library-path .:$i /tmp/bench; \
for j in 1 2; do echo ( time elf/ld.so --library-path .:$i /tmp/bench ) 2>&1; done; done
time elf/ld.so --library-path .:nptl-orig /tmp/bench
real 0m0.655s user 0m0.253s sys 0m0.403s
real 0m0.657s user 0m0.269s sys 0m0.388s
time elf/ld.so --library-path .:nptl-requeue /tmp/bench
real 0m0.496s user 0m0.225s sys 0m0.271s
real 0m0.531s user 0m0.242s sys 0m0.288s
time elf/ld.so --library-path .:nptl-wake_op /tmp/bench
real 0m0.380s user 0m0.176s sys 0m0.204s
real 0m0.382s user 0m0.175s sys 0m0.207s

The benchmark is at:
http://sourceware.org/ml/libc-alpha/2005-03/txt00001.txt
Older futex-requeue-glibc.patch version is at:
http://sourceware.org/ml/libc-alpha/2005-03/txt00002.txt
Older futex-wake_op-glibc.patch version is at:
http://sourceware.org/ml/libc-alpha/2005-03/txt00003.txt
Will post a new version (just x86-64 fixes so that the patch
applies against pthread_cond_signal.S) to libc-hacker ml soon.

Attached is the kernel FUTEX_WAKE_OP patch as well as a simple-minded
testcase that will not test the atomicity of the operation, but at least
check if the threads that should have been woken up are woken up and
whether the arithmetic operation in the kernel gave the expected results.

Acked-by: Ingo Molnar <mingo@redhat.com>
Cc: Ulrich Drepper <drepper@redhat.com>
Cc: Jamie Lokier <jamie@shareable.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Yoichi Yuasa <yuasa@hh.iij4u.or.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] 3c59x PM fixes

This patch adds some missing pci-related calls to the suspend and resume
routines of the 3c59x driver.  It also makes the driver free/request IRQ on
suspend/resume, in accordance with the proposal at:
http://lists.osdl.org/pipermail/linux-pm/2005-May/000955.html

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

[PATCH] swsusp: update documentation

This updates documentation a bit (mostly removing obsolete stuff), and
marks swsusp as no longer experimental in config.

Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: Fix off by one in e820_mapped

This allows a valid iommu placed immediately after memory to work, to be
recognized as after the last byte of memory and not overlapping it.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: create sysfs entries for cpu only for present cpus

Need to create sysfs only for cpus that are present.  Without which we see
NR_CPUS entries created when we have CONFIG_HOTPLUG and CONFIG_HOTPLUG_CPU
enabled.

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Acked-by: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: Fix cluster mode send_IPI_allbutself to use get_cpu()/put_cpu()

Need to ensure we dont get prempted when we clear ourself from mask when using
clustered mode genapic code.

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Acked-by: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: prefetchw() can fall back to prefetch() if !3DNOW

This is a multi-part message in MIME format.  If the cpu lacks 3DNOW
feature, we can use a normal prefetcht0 instruction instead of NOP5.
"prefetchw (%rxx)" and "prefetcht0 (%rxx)" have the same length, ranging
from 3 to 5 bytes depending on the register.  So this patch even helps
AMD64, shortening the length of the code.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86_64: print processor number in show_regs

Up to date I've been using the GS value to determine the processor number
in dumps from show_regs, however this can be cumbersome to do if you don't
have the vmlinux to verify with the address of cpu_pda, how about the
following?  I considered using hard_smp_processor_id for robustness but we
already dereference current so we're already relying on MSR_GS_BASE being
sane.

Signed-off-by: Zwane Mwaikambo <zwane@arm.linux.org.uk>
Acked-by: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] x86/x86_64: deferred handling of writes to /proc/irqxx/smp_affinity

When handling writes to /proc/irq, current code is re-programming rte
entries directly. This is not recommended and could potentially cause
chipset's to lockup, or cause missing interrupts.

CONFIG_IRQ_BALANCE does this correctly, where it re-programs only when the
interrupt is pending. The same needs to be done for /proc/irq handling as well.
Otherwise user space irq balancers are really not doing the right thing.

- Changed pending_irq_balance_cpumask to pending_irq_migrate_cpumask for
  lack of a generic name.
- added move_irq out of IRQ_BALANCE, and added this same to X86_64
- Added new proc handler for write, so we can do deferred write at irq
  handling time.
- Display of /proc/irq/XX/smp_affinity used to display CPU_MASKALL, instead
  it now shows only active cpu masks, or exactly what was set.
- Provided a common move_irq implementation, instead of duplicating
  when using generic irq framework.

Tested on i386/x86_64 and ia64 with CONFIG_PCI_MSI turned on and off.
Tested UP builds as well.

MSI testing: tbd: I have cards, need to look for a x-over cable, although I
did test an earlier version of this patch.  Will test in a couple days.

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Acked-by: Zwane Mwaikambo <zwane@holomorphy.com>
Grudgingly-acked-by: Andi Kleen <ak@muc.de>
Signed-off-by: Coywolf Qi Hunt <coywolf@lovecn.org>
Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc32: add missing sysfs node for ocp_func_emac_data.phy_feat_exc

Add sysfs node for ocp_func_emac_data.phy_feat_exc field.

Signed-off-by: Eugene Surovegin <ebs@ebshome.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ppc32: fix ocp_device_suspend to use pm_message_t instead of u32

Recent "u32 -> pm_message_t" change triggered hidden bug in
ocp_device_suspend.  Fix it to correctly use pm_message_t instead of u32.

Signed-off-by: Eugene Surovegin <ebs@ebshome.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge /pub/scm/linux/kernel/git/paulus/ppc64-2.6

Merge branch 'upstream' of /linux/kernel/git/jgarzik/netdev-2.6

Merge master.kernel.org:/home/rmk/linux-2.6-arm

Merge /pub/scm/linux/kernel/git/sam/kbuild

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

Merge /pub/scm/linux/kernel/git/gregkh/i2c-2.6

Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

[PATCH] remove linux/version.h include from arch/ppc64

Changing CONFIG_LOCALVERSION rebuilds too much, for no apparent reason.

Use system_utsname for progress and debug header.

Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] Invert sense of SLB class bit

Currently, we set the class bit in kernel SLB entries, and clear it on
user SLB entries.  On POWER5, ERAT entries created in real mode have
the class bit clear.  So to avoid flushing kernel ERAT entries on each
context switch, this patch inverts our usage of the class bit, setting
it on user SLB entries and clearing it on kernel SLB entries.

Booted on POWER5 and G5.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[SPARC64]: Kconfig fix (GEN_RTC dependencies)

Yet another architecture not coverd by GEN_RTC - sparc64 never picked
it until now and it doesn't have asm/rtc.h to go with it, so it
wouldn't compile anyway (or have these ioctls in the user-visible
headers, for that matter).

FWIW, I'm very tempted to introduce ARCH_HAS_GEN_RTC and have it set
in arch/*/Kconfig for architectures that know what to do with this
stuff - for something supposedly generic the list of architectures
where it doesn't work is getting too long...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SUNSU]: Compile fixes.

sunsu had been broken by ->stop_tx/->start_tx API changes.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SPARC64]: Don't include drivers/firmware/Kconfig

It's really not relevant for this platform in any
way, after all.

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

[RTC]: Use SA_SHIRQ in sparc specific code.

Based upon a report from Jason Wever.

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

[MOXA]: Fix this driver properly.

Actually, proper fix of that breakage is embarrassingly simple - it's yet
another gratitious leftover include of asm/segment.h, so incremental to the
previos would be removal of that BROKEN and removal of bogus include from
mxser.c itself.

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

[IEEE80211]: Use correct size_t printf format string in ieee80211_rx.c

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

[PATCH] ppc64: Fix build with oprofile disabled

Fix build with oprofile disabled.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Move oprofile_model into cpu feature struct

Move oprofile_model into cpu feature struct.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Move oprofile_impl.h into include/asm-ppc64

Move oprofile_impl.h into include/asm-ppc64 in preparation for moving
oprofile_model into cpu feature struct.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Add oprofile cpu_type to cpu feature struct

Add oprofile cpu_type to cpu feature struct.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Use num_pmcs in oprofile code

Change oprofile to use num_pmcs from the cpu feature struct.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: remove CPU_FTR_PMC8

Remove the CPU_FTR_PMC8 feature now we encode the number of PMCs
directly.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: add number of PMCs to cputable

Add a field in the cputable struct to store the number of PMCs.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[IPW2200]: ipw2200.h needs linux/dma-mapping.h

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

[PATCH] ppc64: Allow world readable /proc/ppc64/lparcfg

I would like to be able to read the lparcfg data from any user so we
can make "intelligent" decisions based on underlying attributes when
running in lpars.  Yes there's software that likes to do this :) and
runs as non-root.

It's very similar to say VM where you can get CP to provide feedback
of the real hardware inside a VM guest.

Signed-off-by: Wim Coekaerts <wim.coekaerts@oracle.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: remove use of asm/segment.h

Removed PPC64 architecture specific users of asm/segment.h.

Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc/ppc64: Merge more include files

This patch merges several include files from
asm-ppc and asm-ppc64 into the new asm-powerpc.

Signed-off-by: Jon Loeliger <jdl@freescale.com>
Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] Move 3 more headers to asm-powerpc

Merged several nearly-identical header files from asm-ppc and asm-ppc64
into asm-powerpc.

Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: speedup cmpxchg

cmpxchg has the following code:

__typeof__(*(ptr)) _o_ = (o);
__typeof__(*(ptr)) _n_ = (n);

Unfortunately it makes gcc 4.0 store and load the variables to the stack.
Eg in atomic_dec_and_test we get:

  stw     r10,112(r1)
  stw     r9,116(r1)
  lwz     r9,112(r1)
  lwz     r0,116(r1)

x86 is just casting the values so do that instead. Also change __xchg*
and __cmpxchg* to take unsigned values, removing a few sign extensions.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: poison initmem

Poison initmem after we free it so we catch use after free issues.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: systemcfg is now a pointer

The following patch fixes 2 issues:
  1) use PLATFORM_LPAR bit to test if running in LPAR mode
  2) systemcfg pointer is assigned from static data in
     arch/ppc64/kernel/pacaData.c.  The file arch/ppc64/kernel/head.S
     now refers to is using the GOT binding to the pointer and hence
     must deref it.

Signed-off-by: Jimi Xenidis <jimix@watson.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Consolidate early console and PPCDBG code

Consolidate the early console and PPCDBG code in udbg.c

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Remove old includes

Trim some no longer needed includes from udbg.c and friends.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Take udbg out of ppc_md

Take udbg out of ppc_md. Allows us to not overwrite early udbg inits
when assigning ppc_md.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Split SCC and 15550 udbg code

Split scc and 15550 functions from udbg each into their own file.
This makes them more symetric with the lpar and btext code.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Make udbg_init_uart set the ppc_md udbg methods.

make udbg_init_uart set the ppc_md udbg methods.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Clean up CR handling

Make the 16550 and real mode 16550 use tail recursion like the scc code
instead of repeating the routine except for the character sent.

Gcc recoginizes the tail recursion and handles it efficently without
stack allocations.  The maple real putc shrinks from 188 to 104 bytes
of instructions.  udbg_putc drops from 188 to 140 bytes.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Remove dummy getc routines

Now that xmon is fixed we should not need the dummy getc routines.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: dont bypass ppc_md.udbg* functions

udbg_getc_poll is a ppc_md function.   don't call directly into udbg.c

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[PATCH] ppc64: Add missing include

inline pmac_call_feature references ppc_md so include asm/machdep.h
in asm/pmac_feature.h

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

[TCP]: Fix TCP_OFF() bug check introduced by previous change.

The TCP_OFF assignment at the bottom of that if block can indeed set
TCP_OFF without setting TCP_PAGE.  Since there is not much to be
gained from avoiding this situation, we might as well just zap the
offset.  The following patch should fix it.

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

[NET]: 2.6.13 breaks libpcap (and tcpdump)

Patrick McHardy says:

  Never mind, I got it, we never fall through to the second switch
  statement anymore. I think we could simply break when load_pointer
  returns NULL. The switch statement will fall through to the default
  case and return 0 for all cases but 0 > k >= SKF_AD_OFF.

Here's a patch to do just that.

I left BPF_MSH alone because it's really a hack to calculate the IP
header length, which makes no sense when applied to the special data.

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

[NET]: Do not protect sysctl_optmem_max with CONFIG_SYSCTL

The ipv4 and ipv6 protocols need to access it unconditionally.
SYSCTL=n build failure reported by Russell King.

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

[NETFILTER] remove bogus hand-coded htonll() from nenetlink_queue

htonll() is nothing else than cpu_to_be64(), so we'd rather call the
latter.

Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: IrDA prototype fixes

Every file should #include the header files containing the prototypes
of it's global functions.

In this case this showed that the prototype of irlan_print_filter()
was wrong which is also corrected in this patch.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SCTP]: net/sctp/sysctl.c should #include <net/sctp/sctp.h>

Every file should #include the header files containing the prototypes of
it's global functions.

sctp.h contains the prototypes of sctp_sysctl_{,un}register().

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER]: net/netfilter/nfnetlink*: make functions static

This patch makes needlessly global functions static.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV4]: net/ipv4/ipconfig.c should #include <linux/nfs_fs.h>

Every file should #include the header files containing the prototypes of
it's global functions.

nfs_fs.h contains the prototype of root_nfs_parse_addr().

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[ATM]: net/atm/ioctl.c should #include "common.h"

Every file should #include the header files containing the prototypes
of it's global functions.

common.h contains the prototype for vcc_ioctl().

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TG3]: Update driver version and release date.

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

[TG3]: Use status tag to check for new events

Use the status tag to determine if there are new events in
tg3_interrupt_tagged(). We discussed about this a while ago with Grant
Grundler and DaveM. This scheme makes it unnecessary to clear the
updated bit in the status block when using tagged mode, and only
a simple comparison is needed to determine if there are new events.

The tp->lock around netif_rx_complete() and tg3_restart_ints() is also
removed. It is unnecessary with DaveM's new locking scheme.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TG3]: Remove status block access in tg3_msi() and add prefetches

Remove unnecessary status block accesses in tg3_msi(). Since MSI is
not shared, it is unnecessary to read the status block to determine if
there are any new events in the MSI handler. It is also unnecessary to
clear the updated bit in the status block.

Since the poll list is per-cpu, tg3_poll() will be scheduled to run on
the same CPU that received the MSI. Prefetches for the status block
and the next rx descriptors are added in tg3_msi() to improve their
access times when tg3_poll() runs.

In the non-MSI irq handlers, we need to check the status block because
interrupts may be shared. Only prefetches for the next rx descriptors
are added.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TG3]: Add PHY loopback test

Improve ethtool loopback self test by adding PHY loopback to the
existing MAC loopback test.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TG3]: Add ethtool -p support

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TG3]: Minor 5780 and 5752 fixes

Minor SerDes bug fixes for 5780S and nvram bug fixes for 5780 and
5752.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[PATCH] klist: fix klist to have the same klist_add semantics as list_head

at the moment, the list_head semantics are

list_add(node, head)

whereas current klist semantics are

klist_add(head, node)

This is bound to cause confusion, and since klist is the newcomer, it
should follow the list_head semantics.

I also added missing include guards to klist.h

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Driver core: small cleanup; remove check for NULL before kfree() in driver core

Remove needless checking of variable for NULL before calling kfree() on it.
Applies to 2.6.13-rc6-git9

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Driver core: hande sysdev suspend failure

This patch adds the return value check for sysdev suspend and does
restore in failure case. Send the patch to pm-list, but seems lost, so I
resend it.

Signed-off-by: Shaohua Li<shaohua.li@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Driver core: Documentation: use S_IRUSR | ... in stead of 0644

Change filemode to use defines in stead of 0644,
based on suggestions by Walter Harms and Domen Puncer.

Signed-off-by: Jan Veldeman <Jan.Veldeman@advalvas.be>
Signed-off-by: Domen Puncer <domen@coderock.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Driver core: Documentation: fix whitespace between parameters

Fix whitespace after comma between parameters.

Signed-off-by: Jan Veldeman <Jan.Veldeman@advalvas.be>
Signed-off-by: Domen Puncer <domen@coderock.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Floppy: add cmos attribute to floppy driver tidy

Fiddle with coding style a bit.

Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Floppy: Add cmos attribute to floppy driver

Currently only a device 'fdX' shows up in sysfs; the other possible
device for this drive (like fd0h1440 etc) must be guessed from there.

This patch corrects the floppy driver to create a platform device for
each floppy found; each platform device also has an attribute 'cmos'
which represents the cmos type for this drive. From this attribute the
other possible device types can be computed.

From: Hannes Reinecke <hare@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Driver core: link device and all class devices derived from it.

Driver core: link device and all class devices derived from it.

To ease the task of locating class devices derived from a certain
device create symlinks from parent device to its class devices.
Change USB host class device name from usbX to usb_hostX to avoid
conflict when creating aforementioned links.

Tweaked by Greg to have the symlink be "class_name:class_device_name" in
order to prevent duplicate links.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

[PATCH] Fix manual binding infinite loop

Fix for manual binding of drivers to devices.  Problem is if you pass in
a valid device id, but the driver refuses to bind.  Infinite loop as
write() tries to resubmit the data it just sent.

Thanks to Michal Ostrowski <mostrows@watson.ibm.com> for pointing the
problem out.

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

[PATCH] orinoco: New driver - spectrum_cs.

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree dee4f325520d4ea29397dd67ca657b7235bb1790 (from c88faac230cc9775445e5c644991c352e35c72a1)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 17:46:39 2005 -0400

    New driver - spectrum_cs.

    Driver for 802.11b cards using RAM-loadable Symbol firmware, such as
    Symbol Wireless Networker LA4100, CompactFlash cards by Socket
    Communications and Intel PRO/Wireless 2011B.

    The driver implements Symbol firmware download.  The rest is handled
    in hermes.c and orinoco.c.

    Utilities for downloading the Symbol firmware are available at
    http://sourceforge.net/projects/orinoco/
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] orinoco: New driver - orinoco_nortel.

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree dce61aef99ceb57370b70222dc34d788666c0ac3 (from ceb6695092be8dcdfe2dec6ee5097d613011489d)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 15:50:55 2005 -0400

    New driver - orinoco_nortel.

    This is a driver for Nortel emobility PCI adaptors, which consist of an
    Orinoco compatible PCMCIA card and a simple PCI-to-PCMCIA bridge.  The
    driver initializes the device and uses Orinoco core driver for actual
    wireless networking.
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] orinoco: Remove EXPERIMENTAL mark from PLX_HERMES, TMD_HERMES and PCI_HERMES.

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree ceb6695092be8dcdfe2dec6ee5097d613011489d (from 6b39374a27eb4be7e9d82145ae270ba02ea90dc8)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 14:50:10 2005 -0400

    Remove EXPERIMENTAL mark from PLX_HERMES, TMD_HERMES and PCI_HERMES.

    Those drivers have been used for a long time, and there have been very
    few problem reports.
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] orinoco: Optimize orinoco_join_ap()

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree cb289b9f9b2a0f3ae7070a008f22e383b37526ee (from 56bfcdb38b3d04c1f8c1fd705e411f4be53b663c)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 19:05:16 2005 -0400

    Optimize orinoco_join_ap() - break from loop once the requested
BSSID
    is found.
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] orinoco: Fix memory leak on error in processing hostscan frames.

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree ca955293cdfd3139e150d3b4fed3922a7eb651fb (from cb289b9f9b2a0f3ae7070a008f22e383b37526ee)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 19:08:00 2005 -0400

    Fix memory leak on error in processing hostscan frames.
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] orinoco: Remove entry for Intel PRO/Wireless 2011B.

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree c88faac230cc9775445e5c644991c352e35c72a1 (from dce61aef99ceb57370b70222dc34d788666c0ac3)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 17:09:45 2005 -0400

    Remove entry for Intel PRO/Wireless 2011B.

    It is not supported by this driver because it has no firmware in
    flash.  spectrum_cs is needed for this device.
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] orinoco: Change orinoco_translate_scan() to return error code on error.

Signed-off-by: Pavel Roskin <proski@gnu.org>
diff-tree 8fc038ec51acf5f777fade80c5e38112b766aeee (from ca955293cdfd3139e150d3b4fed3922a7eb651fb)
Author: Pavel Roskin <proski@gnu.org>
Date:   Thu Sep 1 19:10:12 2005 -0400

    Change orinoco_translate_scan() to return error code on error.
    Adjust the caller to check for errors and clean up if needed.
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] Kconfig fix (PHYLIB vs. s390)

drivers/net/phy/phy.c is broken on s390; it uses enable_irq() and friends
and these do not exist on s390.  Marked as broken for now.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] (15/22) Kconfig fix (82596)

driver is non-modular

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] mv643xx: add netpoll api support

Add support for the netpoll api for use by netconsole, kgdb, etc.

Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] mv643xx: Fix promiscuous mode handling

mv643xx_eth_get_config_reg() was reading the wrong register.
mv643xx_eth_set_config_reg() was or'ing instead of setting the
register.  These functions are trivial and both are called only from
mv643xx_eth_set_rx_mode() when changing to/from promiscuous mode.
Remove both functions and do the operations directly in
mv643xx_eth_set_rx_mode().

Also, maintain promiscuous mode setting across port resets.

Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] mv643xx: Disable per port bandwidth limits

The mv643xx chips support per port bandwith limits.  This patch
disables the bandwidth limits by clearing the MTU register.

Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] mv643xx: fix outstanding tx skb counter

This patch corrects the accounting of outstanding tx skbs.  It fixes
a bug that causes "Error on Queue Full" messages seen since scatter-gather
was enabled by using the hardware tcp/udp checksum generator.

Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] mv643xx: fix skb memory leak

This patch fixes an skb memory leak under heavy receive load
(whenever the more packets have been received than the NAPI budget
allows to be processed).

Signed-off-by: Dale Farnsworth <dale@farnsworth.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sis190: basic sis191 support

The sis191 is the gigabit brother of the sis190. SiS's driver suggests
that the register set is backward compatible: this should hopefully
give a basic driver.

The device should allow the usual features from a modern ethernet
adapter (802.1q, SG, Jumbo frames, TSO, checksum offload). So far
the relevant register layout is not documented. SiS's driver does
not provide these features either (at least not for Linux).

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sis190: RGMII Tx internal delay fiddling

Don't ask.
The patch is based on SiS's GPLed driver.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sis190: make 10Mbps the default when handling the StationControl register

This patch does three things:
- widen the access to the StationControl register (note the SIS_W16
  versus SIS_W32 change);
- default to 10Mbps half duplex when the LPA can not be evaluated
  (reg31->ctl is identical for both). It can be argued that it makes
  sense as the lowest common denominator when everything else failed.
  Btw it works better than the current code. :o)
- remove some enums: they do not document anymore.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sis190: recent chipsets from SiS include a RGMII

Extracted from SiS's GPLed driver. From the few pdf available at SiS's,
it seems that the 965 and the 966 south bridge include this interface
whereas the 965L (and anything below) does not. It is expected to be a
sis191 related feature and should not hurt the existing sis190 driver.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] sis190: unmask the link change events

link changes reporting does not work when the driver masks its irq event

Signed-off-by: Arnaud Patard <apatard@mandriva.com>
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] 'mdio_bus_exit' in discarded section .text.exit

When building with  CONFIG_PHYLIB=y on Itanium, I see:
 `mdio_bus_exit' referenced in section `.init.text' of
drivers/built-in.o: defined in discarded section `.exit.text' of
drivers/built-in.o

I believe that mdio_bus_exit should not be declared __exit, because it is
referencesd from __init sections in, say, phy_init().

Signed-off-by: Peter Chubb <peterc@gelato.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] drivers/net/ne3210.c: cleanups

- make two needlessly global functions static

- kill an ancient version variable

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] (7/7) __user annotations (ethtool)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] (6/7) missing include (ipw2200)

added missing include of dma-mapping.h

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] (5/7) iomem annotations, NULL noise removal (ipw2100)

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

[PATCH] (4/7) missing include (uli526x)

added missing include of dma-mapping.h, removed bogus ptrace.h (what the
hell was it doing there, in the first place?)

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>