pci-dma: powerpc: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: parisc: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Kyle McMartin <kyle@mcmartin.ca>
Cc: Helge Deller <deller@gmx.de>
Cc: James Bottomley <James.Bottomley@suse.de>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: mips: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: ia64: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: frv: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: David Howells <dhowells@redhat.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: arm: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: alpha: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Acked-by: Matt Turner <mattst88@gmail.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: x86: use include/linux/pci-dma.h

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pci-dma: add include/linux/pci-dma.h

This patch adds include/linux/pci-dma.h that defines the pci_unmap state
API:

DECLARE_PCI_UNMAP_ADDR(ADDR_NAME)
DECLARE_PCI_UNMAP_LEN(LEN_NAME)
pci_unmap_addr(PTR, ADDR_NAME)
pci_unmap_addr_set(PTR, ADDR_NAME, VAL)
pci_unmap_len(PTR, LEN_NAME)
pci_unmap_len_set(PTR, LEN_NAME, VAL)

This enables us to remove lots of the duplication in architecture
implementations since there are only two ways to define the API.

If architectures define CONFIG_NEED_DMA_MAP_STATE, they get the real
definition of pci_unmap state API.  If not, they get the noop definition.

In the long term, it's better to replace the API with the generic device
model API such as DECLARE_DMA_UNMAP_ADDR.  We can map the API to the
generic one (like dma-mapping-compat.h does).  This patch also makes the
migration process easier.  We can remove this file after the migration.

It might be simpler to add the API to include/linux/pci.h but looks it's
already too large.  We'll remove pci-dma.h after finishing moving to the
generic device model.  So I put the API to a separate file.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: David Howells <dhowells@redhat.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Helge Deller <deller@gmx.de>
Cc: James Bottomley <James.Bottomley@suse.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Chris Zankel <chris@zankel.net>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

um: remove dma_sync_single_range

dma_sync_single_for_cpu/for_device supports a partial sync so there is no
point to have dma_sync_single_range (also dma_sync_single was obsoleted
long ago, replaced with dma_sync_single_for_cpu/for_device).

There is no user of dma_sync_single_range() in mainline and only Alpha
architecture supports dma_sync_single_range().  So it's unlikely that
someone out of the tree uses it.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Jeff Dike <jdike@addtoit.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

alpha: remove dma_sync_single_range

dma_sync_single_for_cpu/for_device supports a partial sync so there is no
point to have dma_sync_single_range (also dma_sync_single was obsoleted
long ago, replaced with dma_sync_single_for_cpu/for_device).

There is no user of dma_sync_single_range() in mainline and only Alpha
architecture supports dma_sync_single_range().  So it's unlikely that
someone out of the tree uses it.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Acked-by: Matt Turner <mattst88@gmail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

DMA-API.txt: remove dma_sync_single_range description

dma_sync_single_for_cpu/for_device supports a partial sync so there is no
point to have dma_sync_single_range (also dma_sync_single was obsoleted
long ago, replaced with dma_sync_single_for_cpu/for_device).

There is no user of dma_sync_single_range() in mainline and only Alpha
architecture supports dma_sync_single_range().  So it's unlikely that
someone out of the tree uses it.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: David Miller <davem@davemloft.net>
Acked-by: James Bottomley <James.Bottomley@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

DMA-API.txt: add dma_sync_single/sg API description

This adds the description of the following eight function:

dma_sync_single_for_cpu
pci_dma_sync_single_for_cpu
dma_sync_single_for_device
pci_dma_sync_single_for_device
dma_sync_sg_for_cpu
pci_dma_sync_sg_for_cpu
dma_sync_sg_for_device
pci_dma_sync_sg_for_device

It was unclear that the API permits a partial sync (some network drivers
already do though).  I made it clear that the sync_single API can do a
partial sync but the sync_sg API can't.

We could do a partial sync with the sync_sg API too, however, it's
difficult for driver writers to correctly use the sync_sg API for a
partial sync since the scatterlists passed in to the mapping API can't be
modified.  It's unlikely that driver writers want to do a partial sync
with the sync_sg API (because the sync_sg API are usually used for block
drivers).  So I think that it's better to forbid a partial sync with the
sync_sg API.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: David Miller <davem@davemloft.net>
Acked-by: James Bottomley <James.Bottomley@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/DMA-API.txt: remove deprecated function descriptions

dma_sync_single(), pci_dma_sync_single(), dma_sync_sg(), and
pci_dma_sync_sg() are deprecated.  We should not advertise them.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: David S. Miller <davem@davemloft.net>
Acked-by: Joerg Roedel <joro@8bytes.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac: e752x: add dram scrubbing support

Add support to scrub DRAM using the e752x integrated memory scrubbing
engine.  The e7320/7520/e7525 chipsets support scrubbing at one rate while
the i3100 chipset supports a normal and fast rate.

A similar patch was originally sent back in 2008:
http://sourceforge.net/mailarchive/forum.php?thread_name=1204835866.25206.70.camel@localhost.localdomain&forum_name=bluesmoke-devel

This version has the following updates:
- Use 16-bit PCI config cycles to access MCHSCRB register
    e7320/7520/e7525 docs say register is 16bits wide, i3100 says 8.  I
    tested 16bits on the i3100 to be safe.
- Recalcuate and round actual scrub rates

The changes have been tested on an i3100-based board.

Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac: e752x fsb ecc

FSB parity is only supported on the Xeon processor.  Previously it was
incorrectly enabled for the Celeron as well.

Signed-off-by: Konstantin Olifer <kolifer@gmail.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac: mpc85xx use resource_size instead of raw math

Use resource_size() instead of arithmetic.

Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Acked-by: Dave Jiang <djiang@mvista.com>
Cc: Peter Tyser <ptyser@xes-inc.com>
Cc: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac: mpc85xx improve SDRAM error reporting

Add the ability to detect the specific data line or ECC line which failed
when printing out SDRAM single-bit errors.  An example of a single-bit
SDRAM ECC error is below:

  EDAC MPC85xx MC1: Err Detect Register: 0x80000004
  EDAC MPC85xx MC1: Faulty data bit: 59
  EDAC MPC85xx MC1: Expected Data / ECC:  0x7f80d000_409effa0 / 0x6d
  EDAC MPC85xx MC1: Captured Data / ECC:  0x7780d000_409effa0 / 0x6d
  EDAC MPC85xx MC1: Err addr: 0x00031ca0
  EDAC MPC85xx MC1: PFN: 0x00000031

Knowning which specific data or ECC line caused an error can be useful in
tracking down hardware issues such as improperly terminated signals, loose
pins, etc.

Note that this feature is only currently enabled for 64-bit wide data
buses, 32-bit wide bus support should be added.

I don't have any 32-bit wide systems to test on.  If someone has one and
is willing to give this patch a shot with the check for a 64-bit data bus
removed it would be much appreciated and I can re-submit with both 32 and
64 bit buses supported.

Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Kumar Gala <galak@gate.crashing.org>
Cc: Dave Jiang <djiang@mvista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

edac: mpc85xx mask ecc syndrome correctly

With a 64-bit wide data bus only the lowest 8-bits of the ECC syndrome are
relevant.  With a 32-bit wide data bus only the lowest 16-bits are
relevant on most architectures.

Without this change, the ECC syndrome displayed can be mildly confusing,
eg:

  EDAC MPC85xx MC1: syndrome: 0x25252525

When in reality the ECC syndrome is 0x25.

A variety of Freescale manuals say a variety of different things about how
to decode the CAPTURE_ECC (syndrome) register.  I don't have a system with
a 32-bit bus to test on, but I believe the change is correct.  It'd be
good to get an ACK from someone at Freescale about this change though.

Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Kumar Gala <galak@gate.crashing.org>
Cc: Dave Jiang <djiang@mvista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

nsproxy: remove INIT_NSPROXY()

Remove INIT_NSPROXY(), use C99 initializer.
Remove INIT_IPC_NS(), INIT_NET_NS() while I'm at it.

Note: headers trim will be done later, now it's quite pointless because
results will be invalidated by merge window.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pid_ns: zap_pid_ns_processes: use SEND_SIG_NOINFO instead of force_sig()

zap_pid_ns_processes() uses force_sig(SIGKILL) to ensure SIGKILL will be
delivered to sub-namespace inits as well.  This is correct, but we are
going to change force_sig_info() semantics.  See
http://bugzilla.kernel.org/show_bug.cgi?id=15395#c31

We can use send_sig_info(SEND_SIG_NOINFO) instead, since
614c517d7c00af1b26ded20646b329397d6f51a1 ("signals: SEND_SIG_NOINFO should
be considered as SI_FROMUSER()") SEND_SIG_NOINFO means "from user" and
therefore send_signal() will get the correct from_ancestor_ns = T flag.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipmi: remove ipmi_smi.h self-include

There is no need for linux/ipmi_smi.h to include itself.

Signed-off-by: Corey Minyard <minyard@acm.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipmi: fix slave_addrs setting to actually work

Actually use the slave_addrs module parameter if it is specified, and make
things consistent about passing zero in for the slave address for the
default.

Signed-off-by: Bela Lubkin <blubkin@vmware.com>
Signed-off-by: Corey Minyard <minyard@acm.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipmi: add parameter to limit CPU usage in kipmid

In some cases kipmid can use a lot of CPU.  This adds a way to tune the
CPU used by kipmid to help in those cases.  By setting kipmid_max_busy_us
to a value between 100 and 500, it is possible to bring down kipmid CPU
load to practically 0 without loosing too much ipmi throughput
performance.  Not setting the value, or setting the value to zero,
operation is unaffected.

Signed-off-by: Martin Wilck <martin.wilck@ts.fujitsu.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipc: use rlimit helpers

Make sure compiler won't do weird things with limits.  E.g.  fetching them
twice may return 2 different values after writable limits are implemented.

I.e.  either use rlimit helpers added in
3e10e716abf3c71bdb5d86b8f507f9e72236c9cd ("resource: add helpers for
fetching rlimits") or ACCESS_ONCE if not applicable.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

copy_signal() cleanup: clean tty_audit_fork()

Remove unneeded initialization in tty_audit_fork().  It is called only via
copy_signal() and is useless after the kmem_cache_zalloc() was used.

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

copy_signal() cleanup: clean thread_group_cputime_init()

Remove unneeded initializations in thread_group_cputime_init() and in
posix_cpu_timers_init_group().  They are useless after kmem_cache_zalloc()
was used in copy_signal().

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

copy_signal() cleanup: kill taskstats_tgid_init() and acct_init_pacct()

Kill unused functions taskstats_tgid_init() and acct_init_pacct() because
we don't use them anywhere after using kmem_cache_zalloc() in
copy_signal().

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

copy_signal() cleanup: use zalloc and remove initializations

Use kmem_cache_zalloc() on signal creation and remove unneeded
initialization lines in copy_signal().

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m32r: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't, which is consistent with all architectures using the
modern ptrace code.

The old code only disables the breakpoints on PTRACE_KILL, while after
this patch this also happens for PTRACE_CONT and PTRACE_SYSCALL which
matches the behaviour of the other architetures.  I think this is a
bugfixes, but please double verify this is correct.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cris arch-v32: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

The way breakpoints are disabled is entirely inconsistent currently, I
tried to make some sense of it, but I suspect all of the content of
ptrace_disable should be moved into user_disable_single_step, this
defintively needs some revisting as the current patch changes behaviour in
not quite designed ways.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Mikael Starvik <starvik@axis.com>
Cc: Jesper Nilsson <jesper.nilsson@axis.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cris arch-v10: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT and
PTRACE_KILL.  This also makes PTRACE_SINGLESTEP return -EIO while it
previously succeeded despite not actually causing any kind of single
stepping.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Mikael Starvik <starvik@axis.com>
Cc: Jesper Nilsson <jesper.nilsson@axis.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

xtensa: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Chris Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

um: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

XXX: I'm not sure arch_has_single_step() is placed in the exactly correct
location, please verify in which of the ptrace headers it should really
be.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mips: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT and
PTRACE_KILL.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

microblaze: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT and
PTRACE_KILL.  This also makes PTRACE_SINGLESTEP return -EIO while it
previously succeeded despite not actually causing any kind of single
stepping.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Acked-by: Michal Simek <monstr@monstr.eu>
Cc: John Williams <john.williams@petalogix.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m68knommu: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  m68knommu already defines the
nessecary user_enable_single_step and user_disable_single_step functions
for this.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

h8300: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

avr32: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't which is consistent with all architectures using the
modern ptrace code.

Currently avr32 doesn't implement any code to disable single stepping when
one of the non-syscall requests is called which seems wrong, but I've left
it as-is for now.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Acked-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arm: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't and the single stepping disable only happens if the
tracee process isn't a zombie yet, which is consistent with all
architectures using the modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

alpha: use generic ptrace_resume code

Use the generic ptrace_resume code for PTRACE_SYSCALL, PTRACE_CONT,
PTRACE_KILL and PTRACE_SINGLESTEP.  This implies defining
arch_has_single_step in <asm/ptrace.h> and implementing the
user_enable_single_step and user_disable_single_step functions, which also
causes the breakpoint information to be cleared on fork, which could be
considered a bug fix.

Also the TIF_SYSCALL_TRACE thread flag is now cleared on PTRACE_KILL which
it previously wasn't, which is consistent with all architectures using the
modern ptrace code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Richard Henderson <rth@twiddle.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ptrace: move user_enable_single_step & co prototypes to linux/ptrace.h

While in theory user_enable_single_step/user_disable_single_step/
user_enable_blockstep could also be provided as an inline or macro there's
no good reason to do so, and having the prototype in one places keeps code
size and confusion down.

Roland said:

  The original thought there was that user_enable_single_step() et al
  might well be only an instruction or three on a sane machine (as if we
  have any of those!), and since there is only one call site inlining
  would be beneficial.  But I agree that there is no strong reason to care
  about inlining it.

  As to the arch changes, there is only one thought I'd add to the
  record.  It was always my thinking that for an arch where
  PTRACE_SINGLESTEP does text-modifying breakpoint insertion,
  user_enable_single_step() should not be provided.  That is,
  arch_has_single_step()=>true means that there is an arch facility with
  "pure" semantics that does not have any unexpected side effects.
  Inserting a breakpoint might do very unexpected strange things in
  multi-threaded situations.  Aside from that, it is a peculiar side
  effect that user_{enable,disable}_single_step() should cause COW
  de-sharing of text pages and so forth.  For PTRACE_SINGLESTEP, all these
  peculiarities are the status quo ante for that arch, so having
  arch_ptrace() itself do those is one thing.  But for building other
  things in the future, it is nicer to have a uniform "pure" semantics
  that arch-independent code can expect.

  OTOH, all such arch issues are really up to the arch maintainer.  As
  of today, there is nothing but ptrace using user_enable_single_step() et
  al so it's a distinction without a practical difference.  If/when there
  are other facilities that use user_enable_single_step() and might care,
  the affected arch's can revisit the question when someone cares about
  the quality of the arch support for said new facility.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ptrace: use ptrace_request() in the remaining architectures

Use ptrace_request() in the three remaining architectures that didn't use it
(m68knommu, h8300, microblaze).  This means:

 - ptrace_request now handles PTRACE_{PEEK,POKE}{TEXT,DATA} and PTRACE_DETATCH
   calls that were previously called directly, or in case of h8300 even open
   coded.
 - adds new support for PTRACE_SETOPTIONS/PTRACE_GETEVENTMSG/
   PTRACE_GETSIGINFO/PTRACE_SETSIGINFO

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Michal Simek <monstr@monstr.eu>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Acked-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

nodemask: fix the declaration of NODEMASK_ALLOC()

we can't declarate two variable at the same scope by NODEMASK_ALLOC().

This patch fixes it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: update maintainer list

Nishimura-san have been working for memcg very good.  His review and tests
give us much improvements and account migraiton which he is now
challenging is really important.

He is a stakeholder.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: fix oom kill behavior

In current page-fault code,

handle_mm_fault()
-> ...
-> mem_cgroup_charge()
-> map page or handle error.
-> check return code.

If page fault's return code is VM_FAULT_OOM, page_fault_out_of_memory() is
called.  But if it's caused by memcg, OOM should have been already
invoked.

Then, I added a patch: a636b327f731143ccc544b966cfd8de6cb6d72c6.  That
patch records last_oom_jiffies for memcg's sub-hierarchy and prevents
page_fault_out_of_memory from being invoked in near future.

But Nishimura-san reported that check by jiffies is not enough when the
system is terribly heavy.

This patch changes memcg's oom logic as.
 * If memcg causes OOM-kill, continue to retry.
 * remove jiffies check which is used now.
 * add memcg-oom-lock which works like perzone oom lock.
 * If current is killed(as a process), bypass charge.

Something more sophisticated can be added but this pactch does
fundamental things.
TODO:
 - add oom notifier
 - add permemcg disable-oom-kill flag and freezer at oom.
 - more chances for wake up oom waiter (when changing memory limit etc..)

Reviewed-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Tested-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: fix typos in memcg_test.txt

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Reviewed-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: update memcg_test.txt to describe memory thresholds

Decription of sanity check for memory thresholds.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: add simple listener of cgroup events to documentation

An example of cgroup notification API usage.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: remove events before destroying subsystem state objects

Events should be removed after rmdir of cgroup directory, but before
destroying subsystem state objects.  Let's take reference to cgroup
directory dentry to do that.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hioryu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: fix race between userspace and kernelspace

Notify userspace about cgroup removing only after rmdir of cgroup
directory to avoid race between userspace and kernelspace.

eventfd are used to notify about two types of event:
 - control file-specific, like crossing memory threshold;
 - cgroup removing.

To understand what really happen, userspace can check if the cgroup still
exists.  To avoid race beetween userspace and kernelspace we have to
notify userspace about cgroup removing only after rmdir of cgroup
directory.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: handle panic_on_oom=always case

Presently, if panic_on_oom=2, the whole system panics even if the oom
happend in some special situation (as cpuset, mempolicy....).  Then,
panic_on_oom=2 means painc_on_oom_always.

Now, memcg doesn't check panic_on_oom flag. This patch adds a check.

BTW, how it's useful ?

kdump+panic_on_oom=2 is the last tool to investigate what happens in
oom-ed system.  When a task is killed, the sysytem recovers and there will
be few hint to know what happnes.  In mission critical system, oom should
never happen.  Then, panic_on_oom=2+kdump is useful to avoid next OOM by
knowing precise information via snapshot.

TODO:
 - For memcg, it's for isolate system's memory usage, oom-notiifer and
   freeze_at_oom (or rest_at_oom) should be implemented. Then, management
   daemon can do similar jobs (as kdump) or taking snapshot per cgroup.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Reviewed-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: update memcg_test.txt

Update memcg_test.txt to describe how to test the move-charge feature.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg : share event counter rather than duplicate

Memcg has 2 eventcountes which counts "the same" event.  Just usages are
different from each other.  This patch tries to reduce event counter.

Now logic uses "only increment, no reset" counter and masks for each
checks.  Softlimit chesk was done per 1000 evetns.  So, the similar check
can be done by !(new_counter & 0x3ff).  Threshold check was done per 100
events.  So, the similar check can be done by (!new_counter & 0x7f)

ALL event checks are done right after EVENT percpu counter is updated.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: update threshold and softlimit at commit

Presently, move_task does "batched" precharge.  Because res_counter or
css's refcnt are not-scalable jobs for memcg, try_charge_()..  tend to be
done in batched manner if allowed.

Now, softlimit and threshold check their event counter in try_charge, but
the charge is not a per-page event.  And event counter is not updated at
charge().  Moreover, precharge doesn't pass "page" to try_charge() and
softlimit tree will be never updated until uncharge() causes an event."

So the best place to check the event counter is commit_charge().  This is
per-page event by its nature.  This patch move checks to there.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: use generic percpu instead of private implementation

When per-cpu counter for memcg was implemneted, dynamic percpu allocator
was not very good.  But now, we have good one and useful macros.  This
patch replaces memcg's private percpu counter implementation with generic
dynamic percpu allocator.

The benefits are
- We can remove private implementation.
- The counters will be NUMA-aware. (Current one is not...)
- This patch makes sizeof struct mem_cgroup smaller. Then,
  struct mem_cgroup may be fit in page size on small config.
        - About basic performance aspects, see below.

 [Before]
 # size mm/memcontrol.o
   text    data     bss     dec     hex filename
  24373    2528    4132   31033    7939 mm/memcontrol.o

 [page-fault-throuput test on 8cpu/SMP in root cgroup]
 # /root/bin/perf stat -a -e page-faults,cache-misses --repeat 5 ./multi-fault-fork 8

 Performance counter stats for './multi-fault-fork 8' (5 runs):

       45878618  page-faults                ( +-   0.110% )
      602635826  cache-misses               ( +-   0.105% )

   61.005373262  seconds time elapsed   ( +-   0.004% )

 Then cache-miss/page fault = 13.14

 [After]
 #size mm/memcontrol.o
   text    data     bss     dec     hex filename
  23913    2528    4132   30573    776d mm/memcontrol.o
 # /root/bin/perf stat -a -e page-faults,cache-misses --repeat 5 ./multi-fault-fork 8

 Performance counter stats for './multi-fault-fork 8' (5 runs):

       48179400  page-faults                ( +-   0.271% )
      588628407  cache-misses               ( +-   0.136% )

   61.004615021  seconds time elapsed   ( +-   0.004% )

  Then cache-miss/page fault = 12.22

 Text size is reduced.
 This performance improvement is not big and will be invisible in real world
 applications. But this result shows this patch has some good effect even
 on (small) SMP.

Here is a test program I used.

 1. fork() processes on each cpus.
 2. do page fault repeatedly on each process.
 3. after 60secs, kill all childredn and exit.

(3 is necessary for getting stable data, this is improvement from previous one.)

#define _GNU_SOURCE
#include <stdio.h>
#include <sched.h>
#include <sys/mman.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <stdlib.h>

/*
 * For avoiding contention in page table lock, FAULT area is
 * sparse. If FAULT_LENGTH is too large for your cpus, decrease it.
 */
#define FAULT_LENGTH (2 * 1024 * 1024)
#define PAGE_SIZE 4096
#define MAXNUM (128)

void alarm_handler(int sig)
{
}

void *worker(int cpu, int ppid)
{
void *start, *end;
char *c;
cpu_set_t set;
int i;

CPU_ZERO(&set);
CPU_SET(cpu, &set);
sched_setaffinity(0, sizeof(set), &set);

start = mmap(NULL, FAULT_LENGTH, PROT_READ|PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
if (start == MAP_FAILED) {
perror("mmap");
exit(1);
}
end = start + FAULT_LENGTH;

pause();
//fprintf(stderr, "run%d", cpu);
while (1) {
for (c = (char*)start; (void *)c < end; c += PAGE_SIZE)
*c = 0;
madvise(start, FAULT_LENGTH, MADV_DONTNEED);
}
return NULL;
}

int main(int argc, char *argv[])
{
int num, i, ret, pid, status;
int pids[MAXNUM];

if (argc < 2)
return 0;

setpgid(0, 0);
signal(SIGALRM, alarm_handler);
num = atoi(argv[1]);
pid = getpid();

for (i = 0; i < num; ++i) {
ret = fork();
if (!ret) {
worker(i, pid);
exit(0);
}
pids[i] = ret;
}
sleep(1);
kill(-pid, SIGALRM);
sleep(60);
for (i = 0; i < num; i++)
kill(pids[i], SIGKILL);
for (i = 0; i < num; i++)
waitpid(pids[i], &status, 0);
return 0;
}

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: typo in comment to mem_cgroup_print_oom_info()

s/mem_cgroup_print_mem_info/mem_cgroup_print_oom_info/

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: implement memory thresholds

It allows to register multiple memory and memsw thresholds and gets
notifications when it crosses.

To register a threshold application need:
- create an eventfd;
- open memory.usage_in_bytes or memory.memsw.usage_in_bytes;
- write string like "<event_fd> <memory.usage_in_bytes> <threshold>" to
  cgroup.event_control.

Application will be notified through eventfd when memory usage crosses
threshold in any direction.

It's applicable for root and non-root cgroup.

It uses stats to track memory usage, simmilar to soft limits. It checks
if we need to send event to userspace on every 100 page in/out. I guess
it's good compromise between performance and accuracy of thresholds.

[akpm@linux-foundation.org: coding-style fixes]
[nishimura@mxp.nes.nec.co.jp: fix documentation merge issue]
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Vladislav Buzov <vbuzov@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Alexander Shishkin <virtuoso@slind.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: rework usage of stats by soft limit

Instead of incrementing counter on each page in/out and comparing it with
constant, we set counter to constant, decrement counter on each page
in/out and compare it with zero.  We want to make comparing as fast as
possible.  On many RISC systems (probably not only RISC) comparing with
zero is more effective than comparing with a constant, since not every
constant can be immediate operand for compare instruction.

Also, I've renamed MEM_CGROUP_STAT_EVENTS to MEM_CGROUP_STAT_SOFTLIMIT,
since really it's not a generic counter.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Vladislav Buzov <vbuzov@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Alexander Shishkin <virtuoso@slind.org>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: extract mem_group_usage() from mem_cgroup_read()

Helper to get memory or mem+swap usage of the cgroup.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Vladislav Buzov <vbuzov@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Alexander Shishkin <virtuoso@slind.org>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroup: implement eventfd-based generic API for notifications

This patchset introduces eventfd-based API for notifications in cgroups
and implements memory notifications on top of it.

It uses statistics in memory controler to track memory usage.

Output of time(1) on building kernel on tmpfs:

Root cgroup before changes:
make -j2  506.37 user 60.93s system 193% cpu 4:52.77 total
Non-root cgroup before changes:
make -j2  507.14 user 62.66s system 193% cpu 4:54.74 total
Root cgroup after changes (0 thresholds):
make -j2  507.13 user 62.20s system 193% cpu 4:53.55 total
Non-root cgroup after changes (0 thresholds):
make -j2  507.70 user 64.20s system 193% cpu 4:55.70 total
Root cgroup after changes (1 thresholds, never crossed):
make -j2  506.97 user 62.20s system 193% cpu 4:53.90 total
Non-root cgroup after changes (1 thresholds, never crossed):
make -j2  507.55 user 64.08s system 193% cpu 4:55.63 total

This patch:

Introduce the write-only file "cgroup.event_control" in every cgroup.

To register new notification handler you need:
- create an eventfd;
- open a control file to be monitored. Callbacks register_event() and
  unregister_event() must be defined for the control file;
- write "<event_fd> <control_fd> <args>" to cgroup.event_control.
  Interpretation of args is defined by control file implementation;

eventfd will be woken up by control file implementation or when the
cgroup is removed.

To unregister notification handler just close eventfd.

If you need notification functionality for a control file you have to
implement callbacks register_event() and unregister_event() in the
struct cftype.

[kamezawa.hiroyu@jp.fujitsu.com: Kconfig fix]
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Paul Menage <menage@google.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: Dan Malek <dan@embeddedalley.com>
Cc: Vladislav Buzov <vbuzov@embeddedalley.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Alexander Shishkin <virtuoso@slind.org>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: improve performance in moving swap charge

Try to reduce overheads in moving swap charge by:

- Adds a new function(__mem_cgroup_put), which takes "count" as a arg and
  decrement mem->refcnt by "count".
- Removed res_counter_uncharge, css_put, and mem_cgroup_put from the path
  of moving swap account, and consolidate all of them into mem_cgroup_clear_mc.
  We cannot do that about mc.to->refcnt.

These changes reduces the overhead from 1.35sec to 0.9sec to move charges
of 1G anonymous memory(including 500MB swap) in my test environment.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: move charges of anonymous swap

This patch is another core part of this move-charge-at-task-migration
feature.  It enables moving charges of anonymous swaps.

To move the charge of swap, we need to exchange swap_cgroup's record.

In current implementation, swap_cgroup's record is protected by:

  - page lock: if the entry is on swap cache.
  - swap_lock: if the entry is not on swap cache.

This works well in usual swap-in/out activity.

But this behavior make the feature of moving swap charge check many
conditions to exchange swap_cgroup's record safely.

So I changed modification of swap_cgroup's recored(swap_cgroup_record())
to use xchg, and define a new function to cmpxchg swap_cgroup's record.

This patch also enables moving charge of non pte_present but not uncharged
swap caches, which can be exist on swap-out path, by getting the target
pages via find_get_page() as do_mincore() does.

[kosaki.motohiro@jp.fujitsu.com: fix ia64 build]
[akpm@linux-foundation.org: fix typos]
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: avoid oom during moving charge

This move-charge-at-task-migration feature has extra charges on
"to"(pre-charges) and "from"(left-over charges) during moving charge.
This means unnecessary oom can happen.

This patch tries to avoid such oom.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: improve performance in moving charge

Try to reduce overheads in moving charge by:

- Instead of calling res_counter_uncharge() against the old cgroup in
  __mem_cgroup_move_account() everytime, call res_counter_uncharge() at the end
  of task migration once.
- removed css_get(&to->css) from __mem_cgroup_move_account() because callers
  should have already called css_get(). And removed css_put(&to->css) too,
  which was called by callers of move_account on success of move_account.
- Instead of calling __mem_cgroup_try_charge(), i.e. res_counter_charge(),
  repeatedly, call res_counter_charge(PAGE_SIZE * count) in can_attach() if
  possible.
- Instead of calling css_get()/css_put() repeatedly, make use of coalesce
  __css_get()/__css_put() if possible.

These changes reduces the overhead from 1.7sec to 0.6sec to move charges
of 1G anonymous memory in my test environment.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: move charges of anonymous page

This patch is the core part of this move-charge-at-task-migration feature.
 It implements functions to move charges of anonymous pages mapped only by
the target task.

Implementation:
- define struct move_charge_struct and a valuable of it(mc) to remember the
  count of pre-charges and other information.
- At can_attach(), get anon_rss of the target mm, call __mem_cgroup_try_charge()
  repeatedly and count up mc.precharge.
- At attach(), parse the page table, find a target page to be move, and call
  mem_cgroup_move_account() about the page.
- Cancel all precharges if mc.precharge > 0 on failure or at the end of
  task move.

[akpm@linux-foundation.org: a little simplification]
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: add interface to move charge at task migration

In current memcg, charges associated with a task aren't moved to the new
cgroup at task migration.  Some users feel this behavior to be strange.
These patches are for this feature, that is, for charging to the new
cgroup and, of course, uncharging from the old cgroup at task migration.

This patch adds "memory.move_charge_at_immigrate" file, which is a flag
file to determine whether charges should be moved to the new cgroup at
task migration or not and what type of charges should be moved.  This
patch also adds read and write handlers of the file.

This patch also adds no-op handlers for this feature.  These handlers will
be implemented in later patches.  And you cannot write any values other
than 0 to move_charge_at_immigrate yet.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: clean up cgroup_pidlist_find() a bit

Don't call get_pid_ns() before we locate/alloc the ns.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Acked-by: Paul Menage <menage@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: blkio subsystem as module

Modify the Block I/O cgroup subsystem to be able to be built as a module.
As the CFQ disk scheduler optionally depends on blk-cgroup, config options
in block/Kconfig, block/Kconfig.iosched, and block/blk-cgroup.h are
enhanced to support the new module dependency.

Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: fix CONTENTS in cgroups documentation

Add a forgotten item into CONTENTS.

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Acked-by: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: subsystem module unloading

Provides support for unloading modular subsystems.

This patch adds a new function cgroup_unload_subsys which is to be used
for removing a loaded subsystem during module deletion.  Reference
counting of the subsystems' modules is moved from once (at load time) to
once per attached hierarchy (in parse_cgroupfs_options and
rebind_subsystems) (i.e., 0 or 1).

Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: subsystem module loading interface

Add interface between cgroups subsystem management and module loading

This patch implements rudimentary module-loading support for cgroups -
namely, a cgroup_load_subsys (similar to cgroup_init_subsys) for use as a
module initcall, and a struct module pointer in struct cgroup_subsys.

Several functions that might be wanted by modules have had EXPORT_SYMBOL
added to them, but it's unclear exactly which functions want it and which
won't.

Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: revamp subsys array

This patch series provides the ability for cgroup subsystems to be
compiled as modules both within and outside the kernel tree.  This is
mainly useful for classifiers and subsystems that hook into components
that are already modules.  cls_cgroup and blkio-cgroup serve as the
example use cases for this feature.

It provides an interface cgroup_load_subsys() and cgroup_unload_subsys()
which modular subsystems can use to register and depart during runtime.
The net_cls classifier subsystem serves as the example for a subsystem
which can be converted into a module using these changes.

Patch #1 sets up the subsys[] array so its contents can be dynamic as
modules appear and (eventually) disappear.  Iterations over the array are
modified to handle when subsystems are absent, and the dynamic section of
the array is protected by cgroup_mutex.

Patch #2 implements an interface for modules to load subsystems, called
cgroup_load_subsys, similar to cgroup_init_subsys, and adds a module
pointer in struct cgroup_subsys.

Patch #3 adds a mechanism for unloading modular subsystems, which includes
a more advanced rework of the rudimentary reference counting introduced in
patch 2.

Patch #4 modifies the net_cls subsystem, which already had some module
declarations, to be configurable as a module, which also serves as a
simple proof-of-concept.

Part of implementing patches 2 and 4 involved updating css pointers in
each css_set when the module appears or leaves.  In doing this, it was
discovered that css_sets always remain linked to the dummy cgroup,
regardless of whether or not any subsystems are actually bound to it
(i.e., not mounted on an actual hierarchy).  The subsystem loading and
unloading code therefore should keep in mind the special cases where the
added subsystem is the only one in the dummy cgroup (and therefore all
css_sets need to be linked back into it) and where the removed subsys was
the only one in the dummy cgroup (and therefore all css_sets should be
unlinked from it) - however, as all css_sets always stay attached to the
dummy cgroup anyway, these cases are ignored.  Any fix that addresses this
issue should also make sure these cases are addressed in the subsystem
loading and unloading code.

This patch:

Make subsys[] able to be dynamically populated to support modular
subsystems

This patch reworks the way the subsys[] array is used so that subsystems
can register themselves after boot time, and enables the internals of
cgroups to be able to handle when subsystems are not present or may
appear/disappear.

Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroup: introduce coalesce css_get() and css_put()

Current css_get() and css_put() increment/decrement css->refcnt one by
one.

This patch add a new function __css_get(), which takes "count" as a arg
and increment the css->refcnt by "count".  And this patch also add a new
arg("count") to __css_put() and change the function to decrement the
css->refcnt by "count".

These coalesce version of __css_get()/__css_put() will be used to improve
performance of memcg's moving charge feature later, where instead of
calling css_get()/css_put() repeatedly, these new functions will be used.

No change is needed for current users of css_get()/css_put().

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: Paul Menage <menage@google.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroup: introduce cancel_attach()

Add cancel_attach() operation to struct cgroup_subsys.  cancel_attach()
can be used when can_attach() operation prepares something for the subsys,
but we should rollback what can_attach() operation has prepared if attach
task fails after we've succeeded in can_attach().

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Reviewed-by: Paul Menage <menage@google.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/email-clients.txt: update gmail information

Gmail web gui does not work for sending patches now even with firefox
"view source with" extension.  It will use windows style line breaks to
wrap lines automatically when sening email.

Rewrite the gmail web gui part of email client documentation.

Signed-off-by: Dave Young <hidave.darkstar@gmail.com>
Cc: Randy Dunlap <rdunlap@xenotime.net>
Cc: Martin Bligh <mbligh@mbligh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

scripts/kernel-doc: fix empty function description section

scripts/kernel-doc mishandles a function that has a multi-line function
short description and no function parameters.  The observed problem was
from drivers/scsi/scsi_netlink.c:

/**
 * scsi_netlink_init - Called by SCSI subsystem to intialize
 *  the SCSI transport netlink interface
 *
 **/

kernel-doc treated the " * " line as a Description: section with only a
newline character in the Description contents.  This caused
output_highlight() to complain: "output_highlight got called with no
args?", plus produce a perl call stack backtrace.

The fix is just to ignore Description sections if they only contain "\n".

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation: remove obsolete voyager.txt file

x86/Voyager support has been removed a year ago.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Inspired-by: Jonathan Corbet <corbet@lwn.net>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/SubmitChecklist: add rule #1

Add header file requirements.  Compliments of Stephen Rothwell.

Stephen calls this Rule #1, so I put it there, but I didn't want to demote
any of the others in the list, so I made one of them number 2b.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/vm/: split txt and source files

Documentation/vm/:
Expose example and tool source files in the Documentation/ directory in
their own files instead of being buried (almost hidden) in readme/txt files.
This should help to prevent bitrot.

This will make them more visible/usable to users who may need
to use them, to developers who may need to test with them, and
to anyone who would fix/update them if they were more visible.

Also, if any of these possibly should not be in the kernel tree at
all, it will be clearer that they are here and we can discuss if
they should be removed.

Also build the recently-added map_hugetlb.c.
Make several functions static to prevent linker warnings.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Acked-by: Eric B Munson <ebmunson@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/fs/: split txt and source files

Make dnotify_test.c source file and add it to Makefile so that
bitrot can be prevented.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/laptop/: split txt and source files

Documentation/laptops/laptop-mode.txt:
Expose example and tool source files in the Documentation/ directory in
their own files instead of being buried (almost hidden) in readme/txt files.
This should help to prevent bitrot.

This will make them more visible/usable to users who may need
to use them, to developers who may need to test with them, and
to anyone who would fix/update them if they were more visible.

Also, if any of these possibly should not be in the kernel tree at
all, it will be clearer that they are here and we can discuss if
they should be removed.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Documentation/timers/: split txt and source files

Documentation/timers/hpet.txt:
Expose example and tool source files in the Documentation/timers/ directory in
their own files instead of being buried (almost hidden) in readme/txt files.
This should help to prevent bitrot.

This will make them more visible/usable to users who may need
to use them, to developers who may need to test with them, and
to anyone who would fix/update them if they were more visible.

Also, if any of these possibly should not be in the kernel tree at
all, it will be clearer that they are here and we can discuss if
they should be removed.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ufs: make solaris fsck happy

Alex Viskovatoff let me know that after copying data to solaris's ufs from
linux, solaris's fsck sees some errors in cylinder summary information.
This is because of solaris expects find some data on another places, then
curernt implementation save it.  This patch fixes this issue.  It is
tested by me, and also Alex reported that it works for him.

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Reported-by: Alex Viskovatoff <viskovatoff@imap.cc>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/ufs: recognize Solaris-specific file system state

Recent releases of Solaris set the fs_clean state of an unmounted UFS file
system as FSLOG ("logging fs").  However, the Linux kernel currently does
not recognize the value which represents this state.  Thus, attempting to
mount such a file system rw produces the message

kernel: ufs_read_super: can't grok fs_clean 0xfffffffd

and the file system is mounted read-only.  This patch makes the kernel
recognize that value.

Signed-off-by: Alex Viskovatoff <viskovatoff@imap.cc>
Cc: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/char/mem.c: cleanups

- fix switch statement layout

- fix whitespace stuff

- fix comment layout

- remove unneeded inlining

- use __weak

- remove trailing whitespace

- move uncached_access() inside `#ifndef __HAVE_PHYS_MEM_ACCESS_PROT' - it
  is otherwise unused.

Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

/dev/mem: dont allow seek to last page

So as to return a uniform error -EOVERFLOW instead of a random one:

# kmem-seek 0xfffffffffffffff0
seek /dev/kmem: Device or resource busy
# kmem-seek 0xfffffffffffffff1
seek /dev/kmem: Block device required

Suggested by OGAWA Hirofumi.

Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: bfin-lq035q1-fb: remove duplicated #include

Remove duplicated #include('s) in
  drivers/video/bfin-lq035q1-fb.c

Signed-off-by: Huang Weiyi <weiyi.huang@gmail.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

video: fix first line of kernel-doc for a few functions

The function name must be followed by a space, hypen, space, and a short
description.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

obsolete config in kernel source: LWMON5

There was some conditionalizing for the LWMON5 boards in kernel source.
However infrastructure for enabling this isn't here so probably the
special case code can go as well.

Signed-off-by: Christoph Egger <siccegge@stud.informatik.uni-erlangen.de>
Acked-by: Alexander Shishkin <virtuoso@slind.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/video/via: fix continuation line formats

String constants that are continued on subsequent lines with \ will cause
spurious whitespace in the resulting output.

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

NUC900 LCD Controller Driver

An LCD controller driver for nuc900s.  The Linux LOGO is just fine and the
FB-Test application was ok, too.

Signed-off-by: Wang Qiang <rurality.linux@gmail.com>
Cc: Wang Zongshun <mcuos.com@gmail.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fbdev: remove obsolete CONFIG_FB_SOFT_CURSOR

The config Option FB_SOFT_BUFFER was removed in
c465e05a03209651078b95686158648fd7ed84c5 ("fbcon/fbdev: Move softcursor
out of fbdev to fbcon").

While moving to fbcon this single driver has it left as a select in
KConfig / #ifdef in source.  This last occurence is removed here so the
option is really gone

Signed-off-by: Christoph Egger <siccegge@stud.informatik.uni-erlangen.de>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

intelfb: new maintainer

It seems that Sylvain no longer maintain the intelfb driver.  Two weeks
ago I had a short mail conversation with Jean regarding who can replace
him.  Well I will do it, because I know the driver and use it very often.

Attached is a patch which update the maintainer file to make bug reporting
easier.

Signed-off-by: Maik Broemme <mbroemme@plusserver.de>
Sylvain Meyer <sylvain.meyer@worldonline.fr>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Eric Anholt <eric@anholt.net>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

broadsheetfb: support storing waveform

This patch adds waveform storing capability to broadsheetfb. It uses the
firmware class to retrieve the waveform, and the request to initiate the
waveform storing is done via a driver sysfs entry, loadstore_waveform.

Broadsheet is a framebuffer device.  It is slightly different from a
typical framebuffer controller that drives a normal TFT-LCD display.  Most
E-Ink display panels require a waveform in order to function.  That is, in
order to drive the state of a pixel to black, gray, or white, a specific
waveform is utilized.  Basically, that waveform represents the specific
E-field wiggling needed to get the pixel to its optimal state given
current temperature, and its previous state.  TN/IPS-LCDs use a similar
concept but the driving waveform is sufficiently simple that it is
internalized in the TFT source/gate driver.

These E-Ink waveforms are specific to a production batch.  That is, a
batch of display films are produced, then they get characterized and a
waveform is generated for that batch.  Broadsheet, typically, is attached
to its private SPI flash which is then flashed with this waveform.

Users won't be able to see the waveform and typically won't ever need to
know about it.  If however, the display panel attached to broadsheet is
changed out, then they will need to update their waveform.  That would
typically be done at a factory or repair facility rather than by a user.

[akpm@linux-foundation.org: fix printk warning]
Signed-off-by: Jaya Kumar <jayakumar.lkml@gmail.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

broadsheetfb: add MMIO hooks

Allow boards with GP-MMIO controllers to provide hooks to broadsheetfb in
order to offload cmd/data writes and data reads instead of relying only on
host based GPIO wiggling.

Signed-off-by: Jaya Kumar <jayakumar.lkml@gmail.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

viafb: support color depth 15 and 30

Add support for the color depth 15 on IGA1 and 30 on IGA1 and IGA2.  To
allow the usage of those the driver now refuses color depth that are
totally off and otherwise the selection in viafb_check_var is used.
Therefore the first call to this for the first framebuffer was delayed a
bit.  It only enables the new formats if they are requested exactly
(viafb_bpp=15|30).

As this is a new feature, no regressions are expected.  The color depth 15
was successfully tested.  Didn't get anything usable for 30 but that might
be the programs fault.  I would like to get some feedback whether it works
as expected or not if somebody knows a program/configuration where it
should.

Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

viafb: rework color setting

This is a rewritten version of viafb_setcolreg.  The hardware register
writes were split up and moved to hw.c where they belong as this is really
low level stuff.  It was made dual fb aware.

Furthermore viafb_setcmap was removed as the problem with 8bpp originated
from a bug in writing multiple color registers at once.  The removal of
viafb_setcmap might introduce a small performance regression but its
certainly better to receive the correct result a bit slower than a garbled
picture fast.  It should give us a working 8bpp mode and is more
extensible than the old hardcoded code.  No other regressions are expected
but as the hardware might be a bit picky it might cause some regressions
in 8bpp mode on some hardware although I doubt that.

Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

viafb: some virtual_xres handling fixes

Do not require the virtual_xres to be aligned as line length is for such
purposes.  Calculate always the smallest line length required.

Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

viafb: rework color checking

Make color checking a bit more tolerant in what values it allows and more
fine grained to later support 15 and 30 bits formats.  It splits the
filling of the color information in var to a seperate function and sets
some color related values in var that where previously untouched.

This could be a bug fix but at least I don't know any applications that
was fooled by not correctly setting the fields in var.  At least no
regressions are expected.

Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>