]> pilppa.com Git - linux-2.6-omap-h63xx.git/log
linux-2.6-omap-h63xx.git
17 years agoicside: use ec->dma directly
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:32 +0000 (00:32 +0200)]
icside: use ec->dma directly

* hwif->hwif_data contains pointer to struct expansion_card so use ec->dma
  directly instead of caching it in hwif->hw.dma.

* Remove no longer needed hw_regs_t.dma and NO_DMA define.

Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: fix ide_register_hw() to check hwif->io_ports[]
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:32 +0000 (00:32 +0200)]
ide: fix ide_register_hw() to check hwif->io_ports[]

hwif->hw.io_ports[] and hwif->io_ports[] should be the same but "4drives"
support and scc_pata host driver set only hwif->io_ports[].

To compensate for this check hwif->io_ports[] instead of hwif->hw.io_ports[]
in ide_register_hw() (instead of fixing "4drives" and scc_pata because hwif->hw
is to be removed).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add CONFIG_IDE_ARCH_OBSOLETE_INIT
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:32 +0000 (00:32 +0200)]
ide: add CONFIG_IDE_ARCH_OBSOLETE_INIT

Add CONFIG_IDE_ARCH_OBSOLETE_INIT to drivers/ide/Kconfig and use it instead
of defining IDE_ARCH_OBSOLETE_INIT in <arch/ide.h>.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: remove redundant comments from ide.h
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:31 +0000 (00:32 +0200)]
ide: remove redundant comments from ide.h

There is better documentation for these functions in drivers/ide/.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add ide_find_port() helper
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:31 +0000 (00:32 +0200)]
ide: add ide_find_port() helper

* Add ide_find_port() helper.

* Convert icside, rapide and ide_platform host drivers to use it.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add ide_device_add()
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:31 +0000 (00:32 +0200)]
ide: add ide_device_add()

* Add ide_device_add() helper and convert host drivers to use it
  instead of open-coded variants.

* Make ide_pci_setup_ports() and do_ide_setup_pci_device()
  take 'u8 *idx' argument instead of 'ata_index_t *index'.

* Remove no longer needed ata_index_t.

* Unexport probe_hwif_init() and make it static.

* Unexport ide_proc_register_port().

There should be no functionality changes caused by this patch
(sgiioc4.c: ide_proc_register_port() requires hwif->present
 to be set and it won't be set if probe_hwif_init() fails).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add ->fixup method to ide_hwif_t
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:31 +0000 (00:32 +0200)]
ide: add ->fixup method to ide_hwif_t

* Add ->fixup method to ide_hwif_t.

* Set hwif->fixup in ide_pci_setup_ports() to d->fixup.

* Use hwif->fixup in probe_hwif().

* Use probe_hwif_init() instead of probe_hwif_init_with_fixup() in
  ide_setup_pci_device().

* Add 'fixup' argument to ide_register_hw() and use it to set hwif->fixup,
  update all ide_register_hw() users accordingly.

* Convert ide-cs/delkin_cb host drivers to use ide_register_hw().

* Restore hwif->fixup in ide_hwif_restore().

* Remove ide_register_hw_with_fixup(), probe_hwif_init_with_fixup()
  and 'fixup' argument from probe_hwif().

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agosiimage: separate PATA and SATA methods
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:31 +0000 (00:32 +0200)]
siimage: separate PATA and SATA methods

* Split off sil_sata_udma_filter() from sil_udma_filter()
  and rename sil_udma_filter() to sil_pata_udma_filter().

* Rename siimage_busproc() to sil_sata_busproc().

* Rename siimage_reset_poll() to sil_sata_reset_poll()
  and in init_hwif_siimage() set ->reset_poll method only
  for SATA controllers.

* Rename siimage_pre_reset() to sil_sata_pre_reset(),
  in init_hwif_siimage() set ->pre_reset method only for
  SATA controllers and remove redundant is_sata() call.

* Add CONFIG_BLK_DEV_IDE_SATA #ifdef/#endif to pdev_is_sata()
  so compiler will know to throw out unused SATA code for
  CONFIG_BLK_DEV_IDE_SATA=n case (830 bytes saved on x86-32).

* Bump driver version.

Some minor cleanups while at it:

* Convert sil_{pata,sata}_udma_filter() to use ATA_UDMA* defines.

* In siimage_mmio_ide_dma_test_irq() move 'base' variable
  under 'if (SATA_ERROR_REG)' block.

* Simplify sil_sata_reset_poll() a bit.

* Cache is_sata() result in init_hwif_siimage()

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoaec62xx: remove aec62xx_dma_lost_irq()
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:31 +0000 (00:32 +0200)]
aec62xx: remove aec62xx_dma_lost_irq()

* Remove aec62xx_dma_lost_irq() (generic ide_dma_lost_irq() will be used now).

* Bump driver version.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoalim15x3: use ->host_flags and ->udma_mask fields from ide_pci_device_t
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:30 +0000 (00:32 +0200)]
alim15x3: use ->host_flags and ->udma_mask fields from ide_pci_device_t

* Make a local copy of ali15x3_chipset in alim15x3_init_one() and set
  ->host_flags / ->udma_mask according to the controller capabilities.

* Cleanup init_hwif_common_ali15x3().

* Bump driver version.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoalim15x3: fix CD_ROM DMA and PIO FIFO settings setup
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:30 +0000 (00:32 +0200)]
alim15x3: fix CD_ROM DMA and PIO FIFO settings setup

* Setup CD_ROM DMA and PIO FIFO settings in init_chipset_ali15x3() instead
  of ata66_ali15x3().  The latter is called from init_hwif_common_ali15x3()
  only if DMA base exists (which insists m5529_revision > 0x20).

  This changes makes CD_ROM DMA / PIO FIFO bits being set only once
  and also when "idex=ata66" kernel parameter is used.

* While at it move also chip_is_1543c_e setup from ata66_ali15x3() to
  init_chipset_ali15x3() and check if isa_dev exists before accessing it.

* Bump driver version.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add IDE_HFLAG_{IO_32BIT,UNMASK_IRQS} host flags
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:30 +0000 (00:32 +0200)]
ide: add IDE_HFLAG_{IO_32BIT,UNMASK_IRQS} host flags

Add IDE_HFLAG_{IO_32BIT,UNMASK_IRQS} host flag to tell ide_pci_setup_ports()
to set drive->{io_32bit,unmask} for both drives on the interface.  Convert
amd74xx, sl82c105 and via82cxxx host drivers to use these new host flags.

While at it:

* Add IDE_HFLAGS_AMD define (amd74xx host driver).

* Add IDE_HFLAGS_VIA define (via82cxxx host driver).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add IDE_HFLAG_RQSIZE_256 host flag
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:30 +0000 (00:32 +0200)]
ide: add IDE_HFLAG_RQSIZE_256 host flag

Add IDE_HFLAG_RQSIZE_256 host flag to tell ide_pci_setup_ports() to set
hwif->rqsize to 256 sectors.  Convert pdc202xx_old host driver to use it.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add IDE_HFLAG_FORCE_LEGACY_IRQS host flag
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:30 +0000 (00:32 +0200)]
ide: add IDE_HFLAG_FORCE_LEGACY_IRQS host flag

Add IDE_HFLAG_FORCE_LEGACY_IRQS host flag to tell ide_pci_setup_ports()
to always set hwif->irq to legacy IRQ 14/15 and convert generic IDE PCI
and via82cxxx host drivers to use it.

While at it:

* Add IDE_HFLAGS_UMC define (generic IDE PCI host driver).

* Remove no longer needed init_hwif_generic() (generic IDE PCI host driver).

* Set d->udma_mask instead of hwif->ultra_mask (via82cxxx host driver).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add ->chipset field to ide_pci_device_t
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:30 +0000 (00:32 +0200)]
ide: add ->chipset field to ide_pci_device_t

Add ->chipset field to ide_pci_device_t and use it in ide_hwif_configure()
to set hwif->chipset.  Convert cmd64x, cy82c693, rz1000 and trm290 host
drivers to use this new ability.

While at it define hwif_chipset_t as u8 to save some space in hw_regs_t,
ide_hwif_t and ide_pci_device_t instances.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: remove unused ->next field from ide_pci_device_t
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
ide: remove unused ->next field from ide_pci_device_t

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: add hwif_register_devices() helper
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
ide: add hwif_register_devices() helper

Add hwif_register_devices() helper to fix code duplication between
probe_hwif_init_with_fixup() and ideprobe_init().  Also remove stale
comment while at it.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoserverworks: remove dead code from svwks_set_dma_mode()
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
serverworks: remove dead code from svwks_set_dma_mode()

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agorz1000: set serialized flag only if mate interface exists
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
rz1000: set serialized flag only if mate interface exists

Setting hwif->serialized makes sense only if the mate interface exists.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide: fix disabled ports reporting for PCI controllers
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
ide: fix disabled ports reporting for PCI controllers

Report all disabled ports in ide_pci_setup_ports() (prevents the bogus
warning when ide_hwif_configure()->ide_match_hwif() fails to find free
ide_hwifs[] slots).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agocmd64x: always set hwif->chipset for CMD646
Bartlomiej Zolnierkiewicz [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
cmd64x: always set hwif->chipset for CMD646

hwif->chipset should be set to ide_cmd646 also when DMA base is invalid.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoide/pci/sis5513.c: add missing "else"
Adrian Bunk [Fri, 19 Oct 2007 22:32:29 +0000 (00:32 +0200)]
ide/pci/sis5513.c: add missing "else"

This patch adds a missing "else" that was missing in
commit c77a89cd98d99819f23a4a08e5e17ee1f13f6e4d.

Spotted by the Coverity checker.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>,
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
17 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Fri, 19 Oct 2007 20:12:46 +0000 (13:12 -0700)]
Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (41 commits)
  ACPICA: hw: Don't carry spinlock over suspend
  ACPICA: hw: remove use_lock flag from acpi_hw_register_{read, write}
  ACPI: cpuidle: port idle timer suspend/resume workaround to cpuidle
  ACPI: clean up acpi_enter_sleep_state_prep
  Hibernation: Make sure that ACPI is enabled in acpi_hibernation_finish
  ACPI: suppress uninitialized var warning
  cpuidle: consolidate 2.6.22 cpuidle branch into one patch
  ACPI: thinkpad-acpi: skip blanks before the data when parsing sysfs
  ACPI: AC: Add sysfs interface
  ACPI: SBS: Add sysfs alarm
  ACPI: SBS: Add ACPI_PROCFS around procfs handling code.
  ACPI: SBS: Add support for power_supply class (and sysfs)
  ACPI: SBS: Make SBS reads table-driven.
  ACPI: SBS: Simplify data structures in SBS
  ACPI: SBS: Split host controller (ACPI0001) from SBS driver (ACPI0002)
  ACPI: EC: Add new query handler to list head.
  ACPI: Add acpi_bus_generate_event4() function
  ACPI: Battery: add sysfs alarm
  ACPI: Battery: Add sysfs support
  ACPI: Battery: Misc clean-ups, no functional changes
  ...

Fix up conflicts in drivers/misc/thinkpad_acpi.[ch] manually

17 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/mingo/linux-2.6-sched
Linus Torvalds [Fri, 19 Oct 2007 19:07:03 +0000 (12:07 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/mingo/linux-2.6-sched

* git://git.kernel.org/pub/scm/linux/kernel/git/mingo/linux-2.6-sched:
  sched: fix guest time accounting going faster than user time accounting

17 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Fri, 19 Oct 2007 19:01:22 +0000 (12:01 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] Delete totally outdated Documentation/mips/time.README
  [MIPS] Kill duplicated setup_irq() for cp0 timer
  [MIPS] Sibyte: Finish conversion to modern time APIs.
  [MIPS] time: Helpers to compute clocksource/event shift and mult values.
  [MIPS] SMTC: Build fix.
  [MIPS] time: Delete dead code.
  [MIPS] MIPSsim: Strip defconfig file to the bones.

17 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Fri, 19 Oct 2007 19:00:58 +0000 (12:00 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6: (51 commits)
  [CIFS] log better errors on failed mounts
  [CIFS] Return better error when server requires signing but client forbids
  [CIFS] fix typo
  [CIFS] acl support part 4
  [CIFS] Fix minor problems noticed by scan
  [CIFS] fix bad handling of EAGAIN error on kernel_recvmsg in cifs_demultiplex_thread
  [CIFS] build break
  [CIFS] endian fixes
  [CIFS] endian fixes in new acl code
  [CIFS] Fix some endianness problems in new acl code
  [CIFS] missing #endif from a previous patch
  [CIFS] formatting fixes
  [CIFS] Break up unicode_sessetup string functions
  [CIFS] parse server_GUID in SPNEGO negProt response
  [CIFS]
  [CIFS] Fix endian conversion problem in posix mkdir
  [CIFS] fix build break when lanman not enabled
  [CIFS] remove two sparse warnings
  [CIFS] remove compile warnings when debug disabled
  [CIFS] CIFS ACL support part 3
  ...

17 years agoMerge ssh://master.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-x86
Linus Torvalds [Fri, 19 Oct 2007 18:57:05 +0000 (11:57 -0700)]
Merge ssh://master.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-x86

* ssh://master.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-x86:
  x86: fix global_flush_tlb() bug

17 years agoMerge branch 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6
Linus Torvalds [Fri, 19 Oct 2007 18:55:50 +0000 (11:55 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6

* 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6:
  [XFS] cleanup fid types mess
  [XFS] fixups after behavior removal merge into mainline git

17 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Fri, 19 Oct 2007 18:54:39 +0000 (11:54 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [NET]: Fix possible dev_deactivate race condition
  [INET]: Justification for local port range robustness.
  [PACKET]: Kill unused pg_vec_endpage() function
  [NET]: QoS/Sched as menuconfig
  [NET]: Fix bug in sk_filter race cures.
  [PATCH] mac80211: make ieee802_11_parse_elems return void

17 years agoNULL terminate the pci_device_ids in pasemi_edac
Stephen Rothwell [Fri, 19 Oct 2007 06:41:11 +0000 (23:41 -0700)]
NULL terminate the pci_device_ids in pasemi_edac

Fixes:
drivers/edac/pasemi_edac: struct pci_device_id is 32 bytes.  The last of 1 is:
0x00 0x00 0x19 0x59 0x00 0x00 0xa0 0x0a 0xff 0xff 0xff 0xff 0xff 0xff 0xff 0xff
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
FATAL: drivers/edac/pasemi_edac: struct pci_device_id is not terminated with a NULL entry!

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Douglas Thompson <dougthompson@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUninline fork.c/exit.c
Alexey Dobriyan [Fri, 19 Oct 2007 06:41:10 +0000 (23:41 -0700)]
Uninline fork.c/exit.c

Save ~650 bytes here.

add/remove: 4/0 grow/shrink: 0/7 up/down: 430/-1088 (-658)
function                                     old     new   delta
__copy_fs_struct                               -     202    +202
__put_fs_struct                                -     112    +112
__exit_fs                                      -      58     +58
__exit_files                                   -      58     +58
exit_files                                    58       2     -56
put_fs_struct                                112       5    -107
exit_fs                                      161       2    -159
sys_unshare                                  774     590    -184
copy_process                                4031    3840    -191
do_exit                                     1791    1597    -194
copy_fs_struct                               202       5    -197

No difference in lmbench lat_proc tests on 2-way Opteron 246.
Smaaaal degradation on UP P4 (within errors).

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru>
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agokernel/fork.c: remove unneeded variable initialization in copy_process()
Mariusz Kozlowski [Fri, 19 Oct 2007 06:41:09 +0000 (23:41 -0700)]
kernel/fork.c: remove unneeded variable initialization in copy_process()

This initialization of is not needed so just remove it.

Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoLinux Kernel Markers - Documentation
Mathieu Desnoyers [Fri, 19 Oct 2007 06:41:08 +0000 (23:41 -0700)]
Linux Kernel Markers - Documentation

Here is some documentation explaining what is/how to use the Linux
Kernel Markers.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoLinux Kernel Markers - Samples
Mathieu Desnoyers [Fri, 19 Oct 2007 06:41:08 +0000 (23:41 -0700)]
Linux Kernel Markers - Samples

Module example showing how to use the Linux Kernel Markers.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agomarkers: Add samples subdir
Mathieu Desnoyers [Fri, 19 Oct 2007 06:41:07 +0000 (23:41 -0700)]
markers: Add samples subdir

Begin infrastructure for kernel code samples in the samples/ directory.
Add its Kconfig and Kbuild files.
Source its Kconfig file in all arch/ Kconfigs.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoLinux Kernel Markers
Mathieu Desnoyers [Fri, 19 Oct 2007 06:41:06 +0000 (23:41 -0700)]
Linux Kernel Markers

The marker activation functions sits in kernel/marker.c.  A hash table is used
to keep track of the registered probes and armed markers, so the markers
within a newly loaded module that should be active can be activated at module
load time.

marker_query has been removed. marker_get_first, marker_get_next and
marker_release should be used as iterators on the markers.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Mike Mason <mmlnx@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoCombine instrumentation menus in kernel/Kconfig.instrumentation
Mathieu Desnoyers [Fri, 19 Oct 2007 06:41:05 +0000 (23:41 -0700)]
Combine instrumentation menus in kernel/Kconfig.instrumentation

Quoting Randy:

"It seems sad that this patch sources Kconfig.marker, a 7-line file,
20-something times.  Yes, you (we) don't want to put those 7 lines into
20-something different files, so sourcing is the right thing.

However, what you did for avr32 seems more on the right track to me: make
_one_ Instrumentation support menu that includes PROFILING, OPROFILE, KPROBES,
and MARKERS and then use (source) that in all of the arches."

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-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>
17 years agoChange struct marker users
Mathieu Desnoyers [Fri, 19 Oct 2007 06:41:04 +0000 (23:41 -0700)]
Change struct marker users

Prior to use struct marker in the linux kernel markers, we need to clean
two drivers which use this structure name.

Change bonding driver types :
- struct marker to struct bond_marker.
- marker_t to bond_marker_t.
- marker_header to bond_marker_header.
- marker_header_t to bond_marker_header_t.

Change qla4xxx struct marker_entry usage :
- Change struct marker_entry for struct qla4_marker_entry.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Chad Tindel <ctindel@users.sourceforge.net>
Cc: Jay Vosburgh <fubar@us.ibm.com>
Cc: David Somayajulu <david.somayajulu@qlogic.com>
Cc: James Bottomley <James.Bottomley@SteelEye.com>
Cc: Ravi Anand <ravi.anand@qlogic.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoHook up group scheduler with control groups
Srivatsa Vaddagiri [Fri, 19 Oct 2007 06:41:03 +0000 (23:41 -0700)]
Hook up group scheduler with control groups

Enable "cgroup" (formerly containers) based fair group scheduling.  This
will let administrator create arbitrary groups of tasks (using "cgroup"
pseudo filesystem) and control their cpu bandwidth usage.

[akpm@linux-foundation.org: fix cpp condition]
Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Signed-off-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Balbir Singh <balbir@in.ibm.com>
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>
17 years agoAdd documentation for extended crashkernel syntax
Bernhard Walle [Fri, 19 Oct 2007 06:41:02 +0000 (23:41 -0700)]
Add documentation for extended crashkernel syntax

This adds the documentation for the extended crashkernel syntax into
Documentation/kdump/kdump.txt.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse extended crashkernel command line on sh
Bernhard Walle [Fri, 19 Oct 2007 06:41:02 +0000 (23:41 -0700)]
Use extended crashkernel command line on sh

This patch removes the crashkernel parsing from arch/sh/kernel/machine_kexec.c
and calls the generic function, introduced in the generic patch, in
setup_bootmem_allocator().

This is necessary because the amount of System RAM must be known in this
function now because of the new syntax.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse extended crashkernel command line on ppc64
Bernhard Walle [Fri, 19 Oct 2007 06:41:01 +0000 (23:41 -0700)]
Use extended crashkernel command line on ppc64

This patch adapts the ppc64 code to use the generic parse_crashkernel()
function introduced in the generic patch of that series.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse extended crashkernel command line on ia64
Bernhard Walle [Fri, 19 Oct 2007 06:41:00 +0000 (23:41 -0700)]
Use extended crashkernel command line on ia64

This patch adapts IA64 to use the generic parse_crashkernel() function instead
of its own parsing for the crashkernel command line.

Because the total amount of System RAM must be known when calling this
function, efi_memmap_init() is modified to return its accumulated total_memory
variable.

Also, the crashkernel handling is moved in an own function in
arch/ia64/kernel/setup.c to make the code more readable.

[kamalesh@linux.vnet.ibm.com: build fix]
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse extended crashkernel command line on x86_64
Bernhard Walle [Fri, 19 Oct 2007 06:40:59 +0000 (23:40 -0700)]
Use extended crashkernel command line on x86_64

This patch removes the crashkernel parsing from
arch/x86_64/kernel/machine_kexec.c and calls the generic function, introduced
in the last patch, in setup_bootmem_allocator().

This is necessary because the amount of System RAM must be known in this
function now because of the new syntax.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse extended crashkernel command line on i386
Bernhard Walle [Fri, 19 Oct 2007 06:40:59 +0000 (23:40 -0700)]
Use extended crashkernel command line on i386

This patch removes the crashkernel parsing from
arch/i386/kernel/machine_kexec.c and calls the generic function, introduced in
the last patch, in setup_bootmem_allocator().

This is necessary because the amount of System RAM must be known in this
function now because of the new syntax.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoExtended crashkernel command line
Bernhard Walle [Fri, 19 Oct 2007 06:40:58 +0000 (23:40 -0700)]
Extended crashkernel command line

This patch adds a extended crashkernel syntax that makes the value of reserved
system RAM dependent on the system RAM itself:

    crashkernel=<range1>:<size1>[,<range2>:<size2>,...][@offset]
    range=start-[end]

For example:

    crashkernel=512M-2G:64M,2G-:128M

The motivation comes from distributors that configure their crashkernel
command line automatically with some configuration tool (YaST, you know ;)).
Of course that tool knows the value of System RAM, but if the user removes
RAM, then the system becomes unbootable or at least unusable and error
handling is very difficult.

This series implements this change for i386, x86_64, ia64, ppc64 and sh.  That
should be all platforms that support kdump in current mainline.  I tested all
platforms except sh due to the lack of a sh processor.

This patch:

This is the generic part of the patch.  It adds a parse_crashkernel() function
in kernel/kexec.c that is called by the architecture specific code that
actually reserves the memory.  That function takes the whole command line and
looks itself for "crashkernel=" in it.

If there are multiple occurrences, then the last one is taken.  The advantage
is that if you have a bootloader like lilo or elilo which allows you to append
a command line parameter but not to remove one (like in GRUB), then you can
add another crashkernel value for testing at the boot command line and this
one overwrites the command line in the configuration then.

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Vivek Goyal <vgoyal@in.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoIPC: fix error case when idr-cache is empty in ipcget()
Pierre Peiffer [Fri, 19 Oct 2007 06:40:57 +0000 (23:40 -0700)]
IPC: fix error case when idr-cache is empty in ipcget()

With the use of idr to store the ipc, the case where the idr cache is
empty, when idr_get_new is called (this may happen even if we call
idr_pre_get() before), is not well handled: it lets
semget()/shmget()/msgget() return ENOSPC when this cache is empty, what 1.
does not reflect the facts and 2.  does not conform to the man(s).

This patch fixes this by retrying the whole process of allocation in this case.

Signed-off-by: Pierre Peiffer <pierre.peiffer@bull.net>
Cc: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agovirtualization of sysv msg queues is incomplete
Kirill Korotaev [Fri, 19 Oct 2007 06:40:56 +0000 (23:40 -0700)]
virtualization of sysv msg queues is incomplete

Virtualization of sysv msg queues is incomplete: msg_hdrs and msg_bytes
variables visible from userspace are global.  Let's make them
per-namespace.

Signed-off-by: Alexey Kuznetsov <alexey@openvz.org>
Signed-off-by: Kirill Korotaev <dev@openvz.org>
Cc: Pierre Peiffer <pierre.peiffer@bull.net>
Cc: Nadia Derbey <Nadia.Derbey@bull.net>
Cc: Serge Hallyn <serue@us.ibm.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoIPC: cleanup some code and wrong comments about semundo list managment
Pierre Peiffer [Fri, 19 Oct 2007 06:40:55 +0000 (23:40 -0700)]
IPC: cleanup some code and wrong comments about semundo list managment

Some comments about sem_undo_list seem wrong.
About the comment above unlock_semundo:
"... If task2 now exits before task1 releases the lock (by calling
unlock_semundo()), then task1 will never call spin_unlock(). ..."

This is just wrong, I see no reason for which task1 will not call
spin_unlock... The rest of this comment is also wrong... Unless I
miss something (of course).

Finally, (un)lock_semundo functions are useless, so remove them
for simplification. (this avoids an useless if statement)

Signed-off-by: Pierre Peiffer <pierre.peiffer@bull.net>
Cc: Nadia Derbey <Nadia.Derbey@bull.net>
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>
17 years agoipc: remove unneeded parameters
Nadia Derbey [Fri, 19 Oct 2007 06:40:55 +0000 (23:40 -0700)]
ipc: remove unneeded parameters

Remvoe the unneeded parameters from ipc_checkid() and ipc_buildid()
interfaces.

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofix idr_find() locking
Nadia Derbey [Fri, 19 Oct 2007 06:40:54 +0000 (23:40 -0700)]
fix idr_find() locking

This is a patch that fixes the way idr_find() used to be called in ipc_lock():
in all the paths that don't imply an update of the ipcs idr, it was called
without the idr tree being locked.

The changes are:
  . in ipc_ids, the mutex has been changed into a reader/writer semaphore.
  . ipc_lock() now takes the mutex as a reader during the idr_find().
  . a new routine ipc_lock_down() has been defined: it doesn't take the
    mutex, assuming that it is being held by the caller. This is the routine
    that is now called in all the update paths.

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Acked-by: Jarek Poplawski <jarkao2@o2.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: fix wrong comments
Nadia Derbey [Fri, 19 Oct 2007 06:40:53 +0000 (23:40 -0700)]
ipc: fix wrong comments

This patch fixes the wrong / obsolete comments in the ipc code.  Also adds
a missing lock around ipc_get_maxid() in shm_get_stat().

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: inline ipc_buildid()
Nadia Derbey [Fri, 19 Oct 2007 06:40:52 +0000 (23:40 -0700)]
ipc: inline ipc_buildid()

This is a trivial patch that changes the ipc_buildid() routine into a static
inline.

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: introduce the ipcid_to_idx macro
Nadia Derbey [Fri, 19 Oct 2007 06:40:52 +0000 (23:40 -0700)]
ipc: introduce the ipcid_to_idx macro

This is a trivial patch that changes all the (id % SEQ_MULTIPLIER) into a call
to the ipcid_to_idx(id) macro.

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoStoring ipcs into IDRs
Nadia Derbey [Fri, 19 Oct 2007 06:40:51 +0000 (23:40 -0700)]
Storing ipcs into IDRs

This patch converts casts of struct kern_ipc_perm to
   . struct msg_queue
   . struct sem_array
   . struct shmid_kernel
into the equivalent container_of() macro. It improves code maintenance
because the code need not change if kern_ipc_perm is no longer at the
beginning of the containing struct.

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: integrate ipc_checkid() into ipc_lock()
Nadia Derbey [Fri, 19 Oct 2007 06:40:51 +0000 (23:40 -0700)]
ipc: integrate ipc_checkid() into ipc_lock()

This patch introduces a new ipc_lock_check() routine interface:
   . each time ipc_checkid() is called, this is done after calling ipc_lock().
     ipc_checkid() is now called from inside ipc_lock_check().

[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: fix RCU locking]
Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: remove the ipc_get() routine
Nadia Derbey [Fri, 19 Oct 2007 06:40:50 +0000 (23:40 -0700)]
ipc: remove the ipc_get() routine

This is a trivial patch that removes the ipc_get() routine: it is replaced
by a call to idr_find().

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: unify the syscalls code
Nadia Derbey [Fri, 19 Oct 2007 06:40:49 +0000 (23:40 -0700)]
ipc: unify the syscalls code

This patch introduces a change into the sys_msgget(), sys_semget() and
sys_shmget() routines: they now share a common code, which is better for
maintainability.

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoipc: store ipcs into IDRs
Nadia Derbey [Fri, 19 Oct 2007 06:40:48 +0000 (23:40 -0700)]
ipc: store ipcs into IDRs

This patch introduces ipcs storage into IDRs. The main changes are:
  . This ipc_ids structure is changed: the entries array is changed into a
    root idr structure.
  . The grow_ary() routine is removed: it is not needed anymore when adding
    an ipc structure, since we are now using the IDR facility.
  . The ipc_rmid() routine interface is changed:
       . there is no need for this routine to return the pointer passed in as
         argument: it is now declared as a void
       . since the id is now part of the kern_ipc_perm structure, no need to
         have it as an argument to the routine

Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoAdd irq protection in the percpu-counters cpu-hotplug-callback path
Gautham R Shenoy [Fri, 19 Oct 2007 06:40:47 +0000 (23:40 -0700)]
Add irq protection in the percpu-counters cpu-hotplug-callback path

Some of the per-cpu counters and thus their locks are accessed from IRQ
contexts.  This can cause a deadlock if it interrupts a cpu-offline thread
which is transferring a dead-cpu's counts to the global counter.

Add appropriate IRQ protection in the cpu-hotplug callback path.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.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>
17 years agoCPU HOTPLUG: avoid hotadd when proper possible_map isn't specified
KAMEZAWA Hiroyuki [Fri, 19 Oct 2007 06:40:47 +0000 (23:40 -0700)]
CPU HOTPLUG: avoid hotadd when proper possible_map isn't specified

cpu-hot-add should be fail if cpu is not set in cpu_possible_map.  If go
ahead, the system will panic soon.

Especially, arch which requires additional_cpus= parameter should handle
this.  Tested on ia64.

[akpm@linux-foundation.org: coding-style fixes]
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>
17 years agohotplug cpu: migrate a task within its cpuset
Cliff Wickman [Fri, 19 Oct 2007 06:40:46 +0000 (23:40 -0700)]
hotplug cpu: migrate a task within its cpuset

When a cpu is disabled, move_task_off_dead_cpu() is called for tasks that have
been running on that cpu.

Currently, such a task is migrated:
 1) to any cpu on the same node as the disabled cpu, which is both online
    and among that task's cpus_allowed
 2) to any cpu which is both online and among that task's cpus_allowed

It is typical of a multithreaded application running on a large NUMA system to
have its tasks confined to a cpuset so as to cluster them near the memory that
they share.  Furthermore, it is typical to explicitly place such a task on a
specific cpu in that cpuset.  And in that case the task's cpus_allowed
includes only a single cpu.

This patch would insert a preference to migrate such a task to some cpu within
its cpuset (and set its cpus_allowed to its entire cpuset).

With this patch, migrate the task to:
 1) to any cpu on the same node as the disabled cpu, which is both online
    and among that task's cpus_allowed
 2) to any online cpu within the task's cpuset
 3) to any cpu which is both online and among that task's cpus_allowed

In order to do this, move_task_off_dead_cpu() must make a call to
cpuset_cpus_allowed_locked(), a new subset of cpuset_cpus_allowed(), that will
not block.  (name change - per Oleg's suggestion)

Calls are made to cpuset_lock() and cpuset_unlock() in migration_call() to set
the cpuset mutex during the whole migrate_live_tasks() and
migrate_dead_tasks() procedure.

[akpm@linux-foundation.org: build fix]
[pj@sgi.com: Fix indentation and spacing]
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Paul Jackson <pj@sgi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoControl groups: Replace "cont" with "cgrp" and other misc renaming
Paul Menage [Fri, 19 Oct 2007 06:40:44 +0000 (23:40 -0700)]
Control groups: Replace "cont" with "cgrp" and other misc renaming

Replace "cont" with "cgrp" and other misc renaming

This patch finishes some of the names that got missed in the great
"task containers" -> "control groups" rename. Primarily it renames
the local variable "cont" to "cgrp" in a number of places, and renames
the CONT_* enum members to CGRP_*.

This patch is not intended to have any effect on the generated code;
the output of "objdump -d kernel/cgroup.o" is unchanged.

Signed-off-by: Paul Menage <menage@google.com>
Acked-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse task_pid_nr() instead of pid_nr(task_pid())
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:44 +0000 (23:40 -0700)]
Use task_pid_nr() instead of pid_nr(task_pid())

There are two places that do so - the cgroups subsystem and the autofs
code.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Ian Kent <raven@themaw.net>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse task_pid_nr() in ip_vs_sync.c
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:43 +0000 (23:40 -0700)]
Use task_pid_nr() in ip_vs_sync.c

The sync_master_pid and sync_backup_pid are set in set_sync_pid() and are
used later for set/not-set checks and in printk.  So it is safe to use the
global pid value in this case.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoRemove unused variables from fs/proc/base.c
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:43 +0000 (23:40 -0700)]
Remove unused variables from fs/proc/base.c

When removing the explicit task_struct->pid usage I found that
proc_readfd_common() and proc_pident_readdir() get this field, but do not
use it at all.  So this cleanup is a cheap help with the task_struct->pid
isolation.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse helpers to obtain task pid in printks (arch code)
Alexey Dobriyan [Fri, 19 Oct 2007 06:40:41 +0000 (23:40 -0700)]
Use helpers to obtain task pid in printks (arch code)

One of the easiest things to isolate is the pid printed in kernel log.
There was a patch, that made this for arch-independent code, this one makes
so for arch/xxx files.

It took some time to cross-compile it, but hopefully these are all the
printks in arch code.

Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse helpers to obtain task pid in printks
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:40 +0000 (23:40 -0700)]
Use helpers to obtain task pid in printks

The task_struct->pid member is going to be deprecated, so start
using the helpers (task_pid_nr/task_pid_vnr/task_pid_nr_ns) in
the kernel.

The first thing to start with is the pid, printed to dmesg - in
this case we may safely use task_pid_nr(). Besides, printks produce
more (much more) than a half of all the explicit pid usage.

[akpm@linux-foundation.org: git-drm went and changed lots of stuff]
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Dave Airlie <airlied@linux.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoIsolate the explicit usage of signal->pgrp
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:39 +0000 (23:40 -0700)]
Isolate the explicit usage of signal->pgrp

The pgrp field is not used widely around the kernel so it is now marked as
deprecated with appropriate comment.

The initialization of INIT_SIGNALS is trimmed because
a) they are set to 0 automatically;
b) gcc cannot properly initialize two anonymous (the second one
   is the one with the session) unions. In this particular case
   to make it compile we'd have to add some field initialized
   right before the .pgrp.

This is the same patch as the 1ec320afdc9552c92191d5f89fcd1ebe588334ca one
(from Cedric), but for the pgrp field.

Some progress report:

We have to deprecate the pid, tgid, session and pgrp fields on struct
task_struct and struct signal_struct.  The session and pgrp are already
deprecated.  The tgid value is close to being such - the worst known usage
in in fs/locks.c and audit code.  The pid field deprecation is mainly
blocked by numerous printk-s around the kernel that print the tsk->pid to
log.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Herbert Poetzl <herbert@13thfloor.at>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFix tsk->exit_state usage
Eugene Teo [Fri, 19 Oct 2007 06:40:38 +0000 (23:40 -0700)]
Fix tsk->exit_state usage

tsk->exit_state can only be 0, EXIT_ZOMBIE, or EXIT_DEAD.  A non-zero test
is the same as tsk->exit_state & (EXIT_ZOMBIE | EXIT_DEAD), so just testing
tsk->exit_state is sufficient.

Signed-off-by: Eugene Teo <eugeneteo@kernel.sg>
Cc: Roland McGrath <roland@redhat.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoproc: export a processes resource limits via /proc/pid
Neil Horman [Fri, 19 Oct 2007 06:40:37 +0000 (23:40 -0700)]
proc: export a processes resource limits via /proc/pid

Currently, there exists no method for a process to query the resource
limits of another process.  They can be inferred via some mechanisms but
they cannot be explicitly determined.  Given that this information can be
usefull to know during the debugging of an application, I've written this
patch which exports all of a processes limits via /proc/<pid>/limits.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoremove BITS_TO_TYPE macro
Jiri Slaby [Fri, 19 Oct 2007 06:40:37 +0000 (23:40 -0700)]
remove BITS_TO_TYPE macro

remove BITS_TO_TYPE macro

I realized, that it is actually the same as DIV_ROUND_UP, use it instead.

[akpm@linux-foundation.org: build fix]
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFlashPoint, use BIT instead of BITW
Jiri Slaby [Fri, 19 Oct 2007 06:40:36 +0000 (23:40 -0700)]
FlashPoint, use BIT instead of BITW

FlashPoint, use BIT instead of BITW

BITW was an ushort variant of BIT, use BIT instead

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodefine global BIT macro
Jiri Slaby [Fri, 19 Oct 2007 06:40:35 +0000 (23:40 -0700)]
define global BIT macro

define global BIT macro

move all local BIT defines to the new globally define macro.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Kumar Gala <galak@gate.crashing.org>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoget rid of input BIT* duplicate defines
Jiri Slaby [Fri, 19 Oct 2007 06:40:32 +0000 (23:40 -0700)]
get rid of input BIT* duplicate defines

get rid of input BIT* duplicate defines

use newly global defined macros for input layer. Also remove includes of
input.h from non-input sources only for BIT macro definiton. Define the
macro temporarily in local manner, all those local definitons will be
removed further in this patchset (to not break bisecting).
BIT macro will be globally defined (1<<x)

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: <dtor@mail.ru>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Cc: <lenb@kernel.org>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Cc: <perex@suse.cz>
Acked-by: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: <vernux@us.ibm.com>
Cc: <malattia@linux.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodefine first set of BIT* macros
Jiri Slaby [Fri, 19 Oct 2007 06:40:31 +0000 (23:40 -0700)]
define first set of BIT* macros

define first set of BIT* macros

- move BITOP_MASK and BITOP_WORD from asm-generic/bitops/atomic.h to
  include/linux/bitops.h and rename it to BIT_MASK and BIT_WORD
- move BITS_TO_LONGS and BITS_PER_BYTE to bitops.h too and allow easily
  define another BITS_TO_something (e.g. in event.c) by BITS_TO_TYPE macro
Remaining (and common) BIT macro will be defined after all occurences and
conflicts will be sorted out in the patches.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoamba-pl011, rename BIT macro
Jiri Slaby [Fri, 19 Oct 2007 06:40:31 +0000 (23:40 -0700)]
amba-pl011, rename BIT macro

amba-pl011, rename BIT macro

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agos2io, rename BIT macro
Jiri Slaby [Fri, 19 Oct 2007 06:40:29 +0000 (23:40 -0700)]
s2io, rename BIT macro

s2io, rename BIT macro

BIT macro will be global definiton of (1<<x)

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Ramkrishna Vepa <ram.vepa@neterion.com>
Cc: Rastapur Santosh <santosh.rastapur@neterion.com>
Cc: Sivakumar Subramani <sivakumar.subramani@neterion.com>
Cc: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoi2c-pxa, rename BIT macro to PXA_BIT
Jiri Slaby [Fri, 19 Oct 2007 06:40:28 +0000 (23:40 -0700)]
i2c-pxa, rename BIT macro to PXA_BIT

i2c-pxa, rename BIT macro to PXA_BIT

BIT macro will be global definiton of (1 << x)

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Nicolas Pitre <nico@cam.org>
Cc: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agocyber2000fb: checkpatch fixes
Krzysztof Helt [Fri, 19 Oct 2007 06:40:28 +0000 (23:40 -0700)]
cyber2000fb: checkpatch fixes

This patch fixes errors and warnings pointed out by the checkpatch.pl
script.

Antonino Daplas replaced BIT with ENCODE_BIT.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agocyber2000fb, rename BIT macro
Jiri Slaby [Fri, 19 Oct 2007 06:40:27 +0000 (23:40 -0700)]
cyber2000fb, rename BIT macro

cyber2000fb, rename BIT macro

BIT will be global macro for (1 << x)

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoforbid asm/bitops.h direct inclusion
Jiri Slaby [Fri, 19 Oct 2007 06:40:26 +0000 (23:40 -0700)]
forbid asm/bitops.h direct inclusion

forbid asm/bitops.h direct inclusion

Because of compile errors that may occur after bit changes if asm/bitops.h is
included directly without e.g.  linux/kernel.h which includes linux/bitops.h,
forbid direct inclusion of asm/bitops.h.  Thanks to Adrian Bunk.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoremove asm/bitops.h includes
Jiri Slaby [Fri, 19 Oct 2007 06:40:25 +0000 (23:40 -0700)]
remove asm/bitops.h includes

remove asm/bitops.h includes

including asm/bitops directly may cause compile errors. don't include it
and include linux/bitops instead. next patch will deny including asm header
directly.

Cc: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofs/select, remove unused macros
Jiri Slaby [Fri, 19 Oct 2007 06:40:25 +0000 (23:40 -0700)]
fs/select, remove unused macros

fs/select, remove unused macros

this is due to preparation for global BIT macro

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMisc: phantom, improved data passing
Jiri Slaby [Fri, 19 Oct 2007 06:40:24 +0000 (23:40 -0700)]
Misc: phantom, improved data passing

This new version guarantees amb_bit switch in small enough intervals, so that
the device won't stop working in the middle of a movement anymore.  However it
preserves old (openhaptics) functionality.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMisc: phantom, add comment about openhaptics
Jiri Slaby [Fri, 19 Oct 2007 06:40:24 +0000 (23:40 -0700)]
Misc: phantom, add comment about openhaptics

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMisc: phantom, synchronize_irq() on suspend
Jiri Slaby [Fri, 19 Oct 2007 06:40:23 +0000 (23:40 -0700)]
Misc: phantom, synchronize_irq() on suspend

Wait after disabling device's interrupt until the handler finishes its work if
still in progress.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFix cpusets update_cpumask
Paul Menage [Fri, 19 Oct 2007 06:40:22 +0000 (23:40 -0700)]
Fix cpusets update_cpumask

Cause writes to cpuset "cpus" file to update cpus_allowed for member tasks:

- collect batches of tasks under tasklist_lock and then call
  set_cpus_allowed() on them outside the lock (since this can sleep).

- add a simple generic priority heap type to allow efficient collection
  of batches of tasks to be processed without duplicating or missing any
  tasks in subsequent batches.

- make "cpus" file update a no-op if the mask hasn't changed

- fix race between update_cpumask() and sched_setaffinity() by making
  sched_setaffinity() post-check that it's not running on any cpus outside
  cpuset_cpus_allowed().

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Paul Menage <menage@google.com>
Cc: Paul Jackson <pj@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agocpusets: decrustify cpuset mask update code
Paul Jackson [Fri, 19 Oct 2007 06:40:21 +0000 (23:40 -0700)]
cpusets: decrustify cpuset mask update code

Decrustify the kernel/cpuset.c 'cpus' and 'mems' updating code.

Other than subtle improvements in the consistency of identifying
white space at the beginning and end of passed in masks, this
doesn't make any visible difference in behaviour.  But it's
one or two hundred kernel text bytes smaller, and easier to
understand.

[akpm@linux-foundation.org: coding-style fix]
Signed-off-by: Paul Jackson <pj@sgi.com>
Reviewed-by: 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>
17 years agocpuset sched_load_balance flag
Paul Jackson [Fri, 19 Oct 2007 06:40:20 +0000 (23:40 -0700)]
cpuset sched_load_balance flag

Add a new per-cpuset flag called 'sched_load_balance'.

When enabled in a cpuset (the default value) it tells the kernel scheduler
that the scheduler should provide the normal load balancing on the CPUs in
that cpuset, sometimes moving tasks from one CPU to a second CPU if the
second CPU is less loaded and if that task is allowed to run there.

When disabled (write "0" to the file) then it tells the kernel scheduler
that load balancing is not required for the CPUs in that cpuset.

Now even if this flag is disabled for some cpuset, the kernel may still
have to load balance some or all the CPUs in that cpuset, if some
overlapping cpuset has its sched_load_balance flag enabled.

If there are some CPUs that are not in any cpuset whose sched_load_balance
flag is enabled, the kernel scheduler will not load balance tasks to those
CPUs.

Moreover the kernel will partition the 'sched domains' (non-overlapping
sets of CPUs over which load balancing is attempted) into the finest
granularity partition that it can find, while still keeping any two CPUs
that are in the same shed_load_balance enabled cpuset in the same element
of the partition.

This serves two purposes:
 1) It provides a mechanism for real time isolation of some CPUs, and
 2) it can be used to improve performance on systems with many CPUs
    by supporting configurations in which load balancing is not done
    across all CPUs at once, but rather only done in several smaller
    disjoint sets of CPUs.

This mechanism replaces the earlier overloading of the per-cpuset
flag 'cpu_exclusive', which overloading was removed in an earlier
patch: cpuset-remove-sched-domain-hooks-from-cpusets

See further the Documentation and comments in the code itself.

[akpm@linux-foundation.org: don't be weird]
Signed-off-by: Paul Jackson <pj@sgi.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUninline the task_xid_nr_ns() calls
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:19 +0000 (23:40 -0700)]
Uninline the task_xid_nr_ns() calls

Since these are expanded into call to pid_nr_ns() anyway, it's OK to move
the whole routine out-of-line.  This is a cheap way to save ~100 bytes from
vmlinux.  Together with the previous two patches, it saves half-a-kilo from
the vmlinux.

Un-inline other (currently inlined) functions must be done with additional
performance testing.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUninline find_pid etc set of functions
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:19 +0000 (23:40 -0700)]
Uninline find_pid etc set of functions

The find_pid/_vpid/_pid_ns functions are used to find the struct pid by its
id, depending on whic id - global or virtual - is used.

The find_vpid() is a macro that pushes the current->nsproxy->pid_ns on the
stack to call another function - find_pid_ns().  It turned out, that this
dereference together with the push itself cause the kernel text size to
grow too much.

Move all these out-of-line.  Together with the previous patch this saves a
bit less that 400 bytes from .text section.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoIsolate some explicit usage of task->tgid
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:18 +0000 (23:40 -0700)]
Isolate some explicit usage of task->tgid

With pid namespaces this field is now dangerous to use explicitly, so hide
it behind the helpers.

Also the pid and pgrp fields o task_struct and signal_struct are to be
deprecated.  Unfortunately this patch cannot be sent right now as this
leads to tons of warnings, so start isolating them, and deprecate later.

Actually the p->tgid == pid has to be changed to has_group_leader_pid(),
but Oleg pointed out that in case of posix cpu timers this is the same, and
thread_group_leader() is more preferable.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopid namespaces: remove the struct pid unneeded fields
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:17 +0000 (23:40 -0700)]
pid namespaces: remove the struct pid unneeded fields

Since we've switched from using pid->nr to pid->upids->nr some
fields on struct pid are no longer needed

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUninline find_task_by_xxx set of functions
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:16 +0000 (23:40 -0700)]
Uninline find_task_by_xxx set of functions

The find_task_by_something is a set of macros are used to find task by pid
depending on what kind of pid is proposed - global or virtual one.  All of
them are wrappers above the most generic one - find_task_by_pid_type_ns() -
and just substitute some args for it.

It turned out, that dereferencing the current->nsproxy->pid_ns construction
and pushing one more argument on the stack inline cause kernel text size to
grow.

This patch moves all this stuff out-of-line into kernel/pid.c.  Together
with the next patch it saves a bit less than 400 bytes from the .text
section.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopid namespaces: changes to show virtual ids to user
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:14 +0000 (23:40 -0700)]
pid namespaces: changes to show virtual ids to user

This is the largest patch in the set. Make all (I hope) the places where
the pid is shown to or get from user operate on the virtual pids.

The idea is:
 - all in-kernel data structures must store either struct pid itself
   or the pid's global nr, obtained with pid_nr() call;
 - when seeking the task from kernel code with the stored id one
   should use find_task_by_pid() call that works with global pids;
 - when showing pid's numerical value to the user the virtual one
   should be used, but however when one shows task's pid outside this
   task's namespace the global one is to be used;
 - when getting the pid from userspace one need to consider this as
   the virtual one and use appropriate task/pid-searching functions.

[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: nuther build fix]
[akpm@linux-foundation.org: yet nuther build fix]
[akpm@linux-foundation.org: remove unneeded casts]
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopid namespaces: destroy pid namespace on init's death
Sukadev Bhattiprolu [Fri, 19 Oct 2007 06:40:13 +0000 (23:40 -0700)]
pid namespaces: destroy pid namespace on init's death

Terminate all processes in a namespace when the reaper of the namespace is
exiting.  We do this by walking the pidmap of the namespace and sending
SIGKILL to all processes.

Signed-off-by: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopid namespaces: allow signalling cgroup-init
Sukadev Bhattiprolu [Fri, 19 Oct 2007 06:40:13 +0000 (23:40 -0700)]
pid namespaces: allow signalling cgroup-init

Only the global-init process must be special - any other cgroup-init
process must be killable to prevent run-away processes in the system.

TODO:  Ideally we should allow killing the cgroup-init only from parent
cgroup and prevent it being killed from within the cgroup.
But that is a more complex change and will be addressed by a follow-on
patch. For now allow the cgroup-init to be terminated by any process
with sufficient privileges.

Signed-off-by: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopid namespaces: create a slab-cache for 'struct pid_namespace'
Sukadev Bhattiprolu [Fri, 19 Oct 2007 06:40:12 +0000 (23:40 -0700)]
pid namespaces: create a slab-cache for 'struct pid_namespace'

This will help fixing memory leaks due to bad reference counting.

Signed-off-by: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopid namespaces: initialize the namespace's proc_mnt
Pavel Emelyanov [Fri, 19 Oct 2007 06:40:11 +0000 (23:40 -0700)]
pid namespaces: initialize the namespace's proc_mnt

The namespace's proc_mnt must be kern_mount-ed to make this pointer always
valid, independently of whether the user space mounted the proc or not.  This
solves raced in proc_flush_task, etc.  with the proc_mnt switching from NULL
to not-NULL.

The initialization is done after the init's pid is created and hashed to make
proc_get_sb() finr it and get for root inode.

Sice the namespace holds the vfsmnt, vfsmnt holds the superblock and the
superblock holds the namespace we must explicitly break this circle to destroy
all the stuff.  This is done after the init of the namespace dies.  Running a
few steps forward - when init exits it will kill all its children, so no
proc_mnt will be needed after its death.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Sukadev Bhattiprolu <sukadev@us.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>