Currently only the address of the pre-allocated ELF header is passed with
the elfcorehdr= kernel parameter. In order to reserve memory for the header
in the 2nd kernel also the size is required. Current kdump architecture
backends use different methods to do that, e.g. x86 uses the memmap= kernel
parameter. On s390 there is no easy way to transfer this information.
Therefore the elfcorehdr kernel parameter is extended to also pass the size.
This now can also be used as standard mechanism by all future kdump
architecture backends.

The syntax of the kernel parameter is extended as follows:

elfcorehdr=[size[KMG]@]offset[KMG]

This change is backward compatible because elfcorehdr=size is still allowed.
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

On s390 there is a different KEXEC_CONTROL_MEMORY_LIMIT for the normal and
the kdump kexec case. Therefore this patch introduces a new macro
KEXEC_CRASH_CONTROL_MEMORY_LIMIT. This is set to
KEXEC_CONTROL_MEMORY_LIMIT for all architectures that do not define
KEXEC_CRASH_CONTROL_MEMORY_LIMIT.
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

_TIF_SINGLE_STEP is incorrectly defined as 1<<TIF_FREEZE.  Fix it.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Print a message in case we do not receive an IRQ in time (for internal
I/O). Also print the ID of the last used channel path, since it is
possible that not the device itself but this specific path might have
a defect.
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Use the ENTRY macro for the system call wrapper sys_setns_wrapper
similarly to the other wrappers.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The rcu page table free code uses a couple of bits in the page table
pointer passed to tlb_remove_table to discern the different page table
types. __tlb_remove_table extracts the type with an incorrect mask which
leads to memory leaks. The correct mask is ((FRAG_MASK << 4) | FRAG_MASK).

Cc: stable@kernel.org
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

git commit 5e9a2692 "[S390] ptrace cleanup" introduced a regression
for the case when both a user PER set (e.g. a storage alteration trace) and
PTRACE_SINGLESTEP are active. The new code will overrule the user PER set
with a instruction-fetch PER set over the whole address space for ptrace
single stepping. The inferior process will be stopped after each instruction
with an instruction fetch event. Any other events that may have occurred
concurrently are not reported (e.g. storage alteration event) because the
control bits for them are not set. The solution is to merge the PER control
bits of the user PER set with the PER_EVENT_IFETCH control bit for
PTRACE_SINGLESTEP.

Cc: stable@kernel.org
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Fix this warning:
WARNING: vmlinux.o(.text+0x199b6): Section mismatch in reference from
the function alloc_masks() to the function .init.text:__alloc_bootmem()
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The .start function and indirectly the .next function of the show_cpuinfo
sequential operation uses NR_CPUS as limit instead of nr_cpu_ids.
This can cause warnings like this:

WARNING: at /usr/src/linux/include/linux/cpumask.h:107
Process lscpu (pid: 575, task: 000000007deb4338, ksp: 000000007794f588)
Krnl PSW : 0704000180000000 0000000000106db4 (show_cpuinfo+0x108/0x234)
           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:0 CC:0 PM:0 EA:3
Krnl GPRS: 0000000000000003 0000000000791988 000000000071b478 0000000000000004
           0000000000000001 0000000000000000 000000007d139500 0000000000000400
           0000000000000000 000000000070e24c 000000007d48d600 0000000000000005
           000000007d48d600 00000000004dfa10 0000000000106cf8 000000007794fcc0
Krnl Code: 0000000000106da8: 95001000           cli     0(%r1),0
           0000000000106dac: a774ffac           brc     7,106d04
           0000000000106db0: a7f40001           brc     15,106db2
          >0000000000106db4: 92011000           mvi     0(%r1),1
           0000000000106db8: a7f4ffa6           brc     15,106d04
           0000000000106dbc: c0e5000065b4       brasl   %r14,113924
           0000000000106dc2: c09000303a45       larl    %r9,70e24c
           0000000000106dc8: c020001eefd4       larl    %r2,4e4d70

Replacing NR_CPUS with nr_cpu_ids fixes it.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Current IRQ statistics support does not show detail counts for I/O
interrupts which are processed internally only. The result is a
summation count which is way off such as this one:

           CPU0       CPU1       CPU2
I/O:       1331        710        442
[...]
QAI:         15         16         16   [I/O] QDIO Adapter Interrupt
QDI:          1          0          0   [I/O] QDIO Interrupt
DAS:        706        645        381   [I/O] DASD
C15:         26         10          0   [I/O] 3215
C70:          0          0          0   [I/O] 3270
TAP:          0          0          0   [I/O] Tape
VMR:          0          0          0   [I/O] Unit Record Devices
LCS:          0          0          0   [I/O] LCS
CLW:          0          0          0   [I/O] CLAW
CTC:          0          0          0   [I/O] CTC
APB:          0          0          0   [I/O] AP Bus

Fix this by moving I/O interrupt accounting into the common I/O layer.
Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

* git://github.com/rustyrussell/linux:
  lguest: move process freezing before pending signals check
  lguest: don't allow KVM-detection cpuid.
  lguest: Allow running under paravirt-enabled KVM.

* 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6:
  ARM: mark empty gpio.h files empty
  gpio: Fix ARM versatile-express build failure
  of: include errno.h

* 'spi/next' of git://git.secretlab.ca/git/linux-2.6:
  drivercore: Add helper macro for platform_driver boilerplate
  spi: irq: Remove IRQF_DISABLED
  OMAP: SPI: Fix the trying to free nonexistent resource error
  spi/spi-ep93xx: add module.h include
  spi/tegra: fix compilation error in spi-tegra.c
  spi: spi-dw: fix all sparse warnings
  spi/spi-pl022: Call pl022_dma_remove(pl022) only if enable_dma is true
  spi/spi-pl022: calculate_effective_freq() must set rate <= requested rate
  spi/spi-pl022: Don't allocate more sg than required.
  spi/spi-pl022: Use GFP_ATOMIC for allocation from tasklet
  spi/spi-pl022: Resolve formatting issues

* 'gpio/next' of git://git.secretlab.ca/git/linux-2.6:
  h8300: Move gpio.h to gpio-internal.h
  gpio: pl061: add DT binding support
  gpio: fix build error in include/asm-generic/gpio.h
  gpiolib: Ensure struct gpio is always defined
  irq: Add EXPORT_SYMBOL_GPL to function of irq generic-chip
  gpio-ml-ioh: Use NUMA_NO_NODE not GFP_KERNEL
  gpio-pch: Use NUMA_NO_NODE not GFP_KERNEL
  gpio: langwell: ensure alternate function is cleared
  gpio-pch: Support interrupt function
  gpio-pch: Save register value in suspend()
  gpio-pch: modify gpio_nums and mask
  gpio-pch: support ML7223 IOH n-Bus
  gpio-pch: add spinlock in suspend/resume processing
  gpio-pch: Delete invalid "restore" code in suspend()
  gpio-ml-ioh: Fix suspend/resume issue
  gpio-ml-ioh: Support interrupt function
  gpio-ml-ioh: Delete unnecessary code
  gpio/mxc: add chained_irq_enter/exit() to mx3_gpio_irq_handler()
  gpio/nomadik: use genirq core to track enablement
  gpio/nomadik: disable clocks when unused

It is generally a better idea to make intentionally empty files
contain the human-readable /* empty */ comment, also it makes
the files play nice with "make distclean".
Reported-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

A missing mach/gpio.h prevents building gpiolib on versatile express.

  CC      drivers/gpio/gpiolib.o
In file included from /.../linux/include/linux/gpio.h:18:0,
                 from /.../linux/drivers/gpio/gpiolib.c:10:
/.../linux/arch/arm/include/asm/gpio.h:5:23: fatal error: mach/gpio.h: No such file or directory
compilation terminated.
make[3]: *** [drivers/gpio/gpiolib.o] Error 1
make[2]: *** [drivers/gpio] Error 2
make[1]: *** [drivers] Error 2
make: *** [sub-make] Error 2
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

When compiling ath6kl for beagleboard (omap2plus_defconfig plus
CONFIG_ATH6KL, CONFIG_OF disable) with current linux-next compilation
fails:

include/linux/of.h:269: error: 'ENOSYS' undeclared (first use in this function)
include/linux/of.h:276: error: 'ENOSYS' undeclared (first use in this function)
include/linux/of.h:289: error: 'ENOSYS' undeclared (first use in this function)

Fix this by including errno.h from of.h.
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (204 commits)
  [SCSI] qla4xxx: export address/port of connection (fix udev disk names)
  [SCSI] ipr: Fix BUG on adapter dump timeout
  [SCSI] megaraid_sas: Fix instance access in megasas_reset_timer
  [SCSI] hpsa: change confusing message to be more clear
  [SCSI] iscsi class: fix vlan configuration
  [SCSI] qla4xxx: fix data alignment and use nl helpers
  [SCSI] iscsi class: fix link local mispelling
  [SCSI] iscsi class: Replace iscsi_get_next_target_id with IDA
  [SCSI] aacraid: use lower snprintf() limit
  [SCSI] lpfc 8.3.27: Change driver version to 8.3.27
  [SCSI] lpfc 8.3.27: T10 additions for SLI4
  [SCSI] lpfc 8.3.27: Fix queue allocation failure recovery
  [SCSI] lpfc 8.3.27: Change algorithm for getting physical port name
  [SCSI] lpfc 8.3.27: Changed worst case mailbox timeout
  [SCSI] lpfc 8.3.27: Miscellanous logic and interface fixes
  [SCSI] megaraid_sas: Changelog and version update
  [SCSI] megaraid_sas: Add driver workaround for PERC5/1068 kdump kernel panic
  [SCSI] megaraid_sas: Add multiple MSI-X vector/multiple reply queue support
  [SCSI] megaraid_sas: Add support for MegaRAID 9360/9380 12GB/s controllers
  [SCSI] megaraid_sas: Clear FUSION_IN_RESET before enabling interrupts
  ...

* 'for-linus' of git://ceph.newdream.net/git/ceph-client:
  libceph: fix double-free of page vector
  ceph: fix 32-bit ino numbers
  libceph: force resend of osd requests if we skip an osdmap
  ceph: use kernel DNS resolver
  ceph: fix ceph_monc_init memory leak
  ceph: let the set_layout ioctl set single traits
  Revert "ceph: don't truncate dirty pages in invalidate work thread"
  ceph: replace leading spaces with tabs
  libceph: warn on msg allocation failures
  libceph: don't complain on msgpool alloc failures
  libceph: always preallocate mon connection
  libceph: create messenger with client
  ceph: document ioctls
  ceph: implement (optional) max read size
  ceph: rename rsize -> rasize
  ceph: make readpages fully async

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (549 commits)
  ALSA: hda - Fix ADC input-amp handling for Cx20549 codec
  ALSA: hda - Keep EAPD turned on for old Conexant chips
  ALSA: hda/realtek - Fix missing volume controls with ALC260
  ASoC: wm8940: Properly set codec->dapm.bias_level
  ALSA: hda - Fix pin-config for ASUS W90V
  ALSA: hda - Fix surround/CLFE headphone and speaker pins order
  ALSA: hda - Fix typo
  ALSA: Update the sound git tree URL
  ALSA: HDA: Add new revision for ALC662
  ASoC: max98095: Convert codec->hw_write to snd_soc_write
  ASoC: keep pointer to resource so it can be freed
  ASoC: sgtl5000: Fix wrong mask in some snd_soc_update_bits calls
  ASoC: wm8996: Fix wrong mask for setting WM8996_AIF_CLOCKING_2
  ASoC: da7210: Add support for line out and DAC
  ASoC: da7210: Add support for DAPM
  ALSA: hda/realtek - Fix DAC assignments of multiple speakers
  ASoC: Use SGTL5000_LINREG_VDDD_MASK instead of hardcoded mask value
  ASoC: Set sgtl5000->ldo in ldo_regulator_register
  ASoC: wm8996: Use SND_SOC_DAPM_AIF_OUT for AIF2 Capture
  ASoC: wm8994: Use SND_SOC_DAPM_AIF_OUT for AIF3 Capture
  ...

* 'next-rebase' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci:
  PCI: Clean-up MPS debug output
  pci: Clamp pcie_set_readrq() when using "performance" settings
  PCI: enable MPS "performance" setting to properly handle bridge MPS
  PCI: Workaround for Intel MPS errata
  PCI: Add support for PASID capability
  PCI: Add implementation for PRI capability
  PCI: Export ATS functions to modules
  PCI: Move ATS implementation into own file
  PCI / PM: Remove unnecessary error variable from acpi_dev_run_wake()
  PCI hotplug: acpiphp: Prevent deadlock on PCI-to-PCI bridge remove
  PCI / PM: Extend PME polling to all PCI devices
  PCI quirk: mmc: Always check for lower base frequency quirk for Ricoh 1180:e823
  PCI: Make pci_setup_bridge() non-static for use by arch code
  x86: constify PCI raw ops structures
  PCI: Add quirk for known incorrect MPSS
  PCI: Add Solarflare vendor ID and SFC4000 device IDs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc: (83 commits)
  mmc: fix compile error when CONFIG_BLOCK is not enabled
  mmc: core: Cleanup eMMC4.5 conditionals
  mmc: omap_hsmmc: if multiblock reads are broken, disable them
  mmc: core: add workaround for controllers with broken multiblock reads
  mmc: core: Prevent too long response times for suspend
  mmc: recognise SDIO cards with SDIO_CCCR_REV 3.00
  mmc: sd: Handle SD3.0 cards not supporting UHS-I bus speed mode
  mmc: core: support HPI send command
  mmc: core: Add cache control for eMMC4.5 device
  mmc: core: Modify the timeout value for writing power class
  mmc: core: new discard feature support at eMMC v4.5
  mmc: core: mmc sanitize feature support for v4.5
  mmc: dw_mmc: modify DATA register offset
  mmc: sdhci-pci: add flag for devices that can support runtime PM
  mmc: omap_hsmmc: ensure pbias configuration is always done
  mmc: core: Add Power Off Notify Feature eMMC 4.5
  mmc: sdhci-s3c: fix potential NULL dereference
  mmc: replace printk with appropriate display macro
  mmc: core: Add default timeout value for CMD6
  mmc: sdhci-pci: add runtime pm support
  ...

Merge branch 'devel-stable' of http://ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur/linux-2.6-arm

* 'devel-stable' of http://ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur/linux-2.6-arm: (178 commits)
  ARM: 7139/1: fix compilation with CONFIG_ARM_ATAG_DTB_COMPAT and large TEXT_OFFSET
  ARM: gic, local timers: use the request_percpu_irq() interface
  ARM: gic: consolidate PPI handling
  ARM: switch from NO_MACH_MEMORY_H to NEED_MACH_MEMORY_H
  ARM: mach-s5p64x0: remove mach/memory.h
  ARM: mach-s3c64xx: remove mach/memory.h
  ARM: plat-mxc: remove mach/memory.h
  ARM: mach-prima2: remove mach/memory.h
  ARM: mach-zynq: remove mach/memory.h
  ARM: mach-bcmring: remove mach/memory.h
  ARM: mach-davinci: remove mach/memory.h
  ARM: mach-pxa: remove mach/memory.h
  ARM: mach-ixp4xx: remove mach/memory.h
  ARM: mach-h720x: remove mach/memory.h
  ARM: mach-vt8500: remove mach/memory.h
  ARM: mach-s5pc100: remove mach/memory.h
  ARM: mach-tegra: remove mach/memory.h
  ARM: plat-tcc: remove mach/memory.h
  ARM: mach-mmp: remove mach/memory.h
  ARM: mach-cns3xxx: remove mach/memory.h
  ...

Fix up mostly pretty trivial conflicts in:
 - arch/arm/Kconfig
 - arch/arm/include/asm/localtimer.h
 - arch/arm/kernel/Makefile
 - arch/arm/mach-shmobile/board-ap4evb.c
 - arch/arm/mach-u300/core.c
 - arch/arm/mm/dma-mapping.c
 - arch/arm/mm/proc-v7.S
 - arch/arm/plat-omap/Kconfig
largely due to some CONFIG option renaming (ie CONFIG_PM_SLEEP ->
CONFIG_ARM_CPU_SUSPEND for the arm-specific suspend code etc) and
addition of NEED_MACH_MEMORY_H next to HAVE_IDE.

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue: (21 commits)
  leases: fix write-open/read-lease race
  nfs: drop unnecessary locking in llseek
  ext4: replace cut'n'pasted llseek code with generic_file_llseek_size
  vfs: add generic_file_llseek_size
  vfs: do (nearly) lockless generic_file_llseek
  direct-io: merge direct_io_walker into __blockdev_direct_IO
  direct-io: inline the complete submission path
  direct-io: separate map_bh from dio
  direct-io: use a slab cache for struct dio
  direct-io: rearrange fields in dio/dio_submit to avoid holes
  direct-io: fix a wrong comment
  direct-io: separate fields only used in the submission path from struct dio
  vfs: fix spinning prevention in prune_icache_sb
  vfs: add a comment to inode_permission()
  vfs: pass all mask flags check_acl and posix_acl_permission
  vfs: add hex format for MAY_* flag values
  vfs: indicate that the permission functions take all the MAY_* flags
  compat: sync compat_stats with statfs.
  vfs: add "device" tag to /proc/self/mountstats
  cleanup: vfs: small comment fix for block_invalidatepage
  ...

Fix up trivial conflict in fs/gfs2/file.c (llseek changes)

* http://sucs.org/~rohan/git/gfs2-3.0-nmw: (24 commits)
  GFS2: Move readahead of metadata during deallocation into its own function
  GFS2: Remove two unused variables
  GFS2: Misc fixes
  GFS2: rewrite fallocate code to write blocks directly
  GFS2: speed up delete/unlink performance for large files
  GFS2: Fix off-by-one in gfs2_blk2rgrpd
  GFS2: Clean up ->page_mkwrite
  GFS2: Correctly set goal block after allocation
  GFS2: Fix AIL flush issue during fsync
  GFS2: Use cached rgrp in gfs2_rlist_add()
  GFS2: Call do_strip() directly from recursive_scan()
  GFS2: Remove obsolete assert
  GFS2: Cache the most recently used resource group in the inode
  GFS2: Make resource groups "append only" during life of fs
  GFS2: Use rbtree for resource groups and clean up bitmap buffer ref count scheme
  GFS2: Fix lseek after SEEK_DATA, SEEK_HOLE have been added
  GFS2: Clean up gfs2_create
  GFS2: Use ->dirty_inode()
  GFS2: Fix bug trap and journaled data fsync
  GFS2: Fix inode allocation error path
  ...

* '3.2-without-smb2' of git://git.samba.org/sfrench/cifs-2.6: (52 commits)
  Fix build break when freezer not configured
  Add definition for share encryption
  CIFS: Make cifs_push_locks send as many locks at once as possible
  CIFS: Send as many mandatory unlock ranges at once as possible
  CIFS: Implement caching mechanism for posix brlocks
  CIFS: Implement caching mechanism for mandatory brlocks
  CIFS: Fix DFS handling in cifs_get_file_info
  CIFS: Fix error handling in cifs_readv_complete
  [CIFS] Fixup trivial checkpatch warning
  [CIFS] Show nostrictsync and noperm mount options in /proc/mounts
  cifs, freezer: add wait_event_freezekillable and have cifs use it
  cifs: allow cifs_max_pending to be readable under /sys/module/cifs/parameters
  cifs: tune bdi.ra_pages in accordance with the rsize
  cifs: allow for larger rsize= options and change defaults
  cifs: convert cifs_readpages to use async reads
  cifs: add cifs_async_readv
  cifs: fix protocol definition for READ_RSP
  cifs: add a callback function to receive the rest of the frame
  cifs: break out 3rd receive phase into separate function
  cifs: find mid earlier in receive codepath
  ...

* 'for-linus' of git://oss.sgi.com/xfs/xfs: (69 commits)
  xfs: add AIL pushing tracepoints
  xfs: put in missed fix for merge problem
  xfs: do not flush data workqueues in xfs_flush_buftarg
  xfs: remove XFS_bflush
  xfs: remove xfs_buf_target_name
  xfs: use xfs_ioerror_alert in xfs_buf_iodone_callbacks
  xfs: clean up xfs_ioerror_alert
  xfs: clean up buffer allocation
  xfs: remove buffers from the delwri list in xfs_buf_stale
  xfs: remove XFS_BUF_STALE and XFS_BUF_SUPER_STALE
  xfs: remove XFS_BUF_SET_VTYPE and XFS_BUF_SET_VTYPE_REF
  xfs: remove XFS_BUF_FINISH_IOWAIT
  xfs: remove xfs_get_buftarg_list
  xfs: fix buffer flushing during unmount
  xfs: optimize fsync on directories
  xfs: reduce the number of log forces from tail pushing
  xfs: Don't allocate new buffers on every call to _xfs_buf_find
  xfs: simplify xfs_trans_ijoin* again
  xfs: unlock the inode before log force in xfs_change_file_space
  xfs: unlock the inode before log force in xfs_fs_nfs_commit_metadata
  ...

In setlease, we use i_writecount to decide whether we can give out a
read lease.

In open, we break leases before incrementing i_writecount.

There is therefore a window between the break lease and the i_writecount
increment when setlease could add a new read lease.

This would leave us with a simultaneous write open and read lease, which
shouldn't happen.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

This makes NFS follow the standard generic_file_llseek locking scheme.

Cc: Trond.Myklebust@netapp.com
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

This gives ext4 the benefits of unlocked llseek.

Cc: tytso@mit.edu
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Add a generic_file_llseek variant to the VFS that allows passing in
the maximum file size of the file system, instead of always
using maxbytes from the superblock.

This can be used to eliminate some cut'n'paste seek code in ext4.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

The i_mutex lock use of generic _file_llseek hurts.  Independent processes
accessing the same file synchronize over a single lock, even though
they have no need for synchronization at all.

Under high utilization this can cause llseek to scale very poorly on larger
systems.

This patch does some rethinking of the llseek locking model:

First the 64bit f_pos is not necessarily atomic without locks
on 32bit systems. This can already cause races with read() today.
This was discussed on linux-kernel in the past and deemed acceptable.
The patch does not change that.

Let's look at the different seek variants:

SEEK_SET: Doesn't really need any locking.
If there's a race one writer wins, the other loses.

For 32bit the non atomic update races against read()
stay the same. Without a lock they can also happen
against write() now.  The read() race was deemed
acceptable in past discussions, and I think if it's
ok for read it's ok for write too.

=> Don't need a lock.

SEEK_END: This behaves like SEEK_SET plus it reads
the maximum size too. Reading the maximum size would have the
32bit atomic problem. But luckily we already have a way to read
the maximum size without locking (i_size_read), so we
can just use that instead.

Without i_mutex there is no synchronization with write() anymore,
however since the write() update is atomic on 64bit it just behaves
like another racy SEEK_SET.  On non atomic 32bit it's the same
as SEEK_SET.

=> Don't need a lock, but need to use i_size_read()

SEEK_CUR: This has a read-modify-write race window
on the same file. One could argue that any application
doing unsynchronized seeks on the same file is already broken.
But for the sake of not adding a regression here I'm
using the file->f_lock to synchronize this. Using this
lock is much better than the inode mutex because it doesn't
synchronize between processes.

=> So still need a lock, but can use a f_lock.

This patch implements this new scheme in generic_file_llseek.
I dropped generic_file_llseek_unlocked and changed all callers.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

This doesn't change anything for the compiler, but hch thought it would
make the code clearer.

I moved the reference counting into its own little inline.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Add inlines to all the submission path functions. While this increases
code size it also gives gcc a lot of optimization opportunities
in this critical hotpath.

In particular -- together with some other changes -- this
allows gcc to get rid of the unnecessary clearing of
sdio at the beginning and optimize the messy parameter passing.
Any non inlining of a function which takes a sdio parameter
would break this optimization because they cannot be done if the
address of a structure is taken.

Note that benefits are only seen with CONFIG_OPTIMIZE_INLINING
and CONFIG_CC_OPTIMIZE_FOR_SIZE both set to off.

This gives about 2.2% improvement on a large database benchmark
with a high IOPS rate.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Only a single b_private field in the map_bh buffer head is needed after
the submission path. Move map_bh separately to avoid storing
this information in the long term slab.

This avoids the weird 104 byte hole in struct dio_submit which also needed
to be memseted early.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

A direct slab call is slightly faster than kmalloc and can be better cached
per CPU. It also avoids rounding to the next kmalloc slab.

In addition this enforces cache line alignment for struct dio to avoid
any false sharing.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Fix most problems reported by pahole.

There is still a weird 104 byte hole after map_bh. I'm not sure what
causes this.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

There's nothing on the stack, even before my changes.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

This large, but largely mechanic, patch moves all fields in struct dio
that are only used in the submission path into a separate on stack
data structure. This has the advantage that the memory is very likely
cache hot, which is not guaranteed for memory fresh out of kmalloc.

This also gives gcc more optimization potential because it can easier
determine that there are no external aliases for these variables.

The sdio initialization is a initialization now instead of memset.
This allows gcc to break sdio into individual fields and optimize
away unnecessary zeroing (after all the functions are inlined)
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

We need to move the inode to the end of the list to actually make the
spinning prevention explained in the comment above it work.  With a
plain list_move it will simply stay in place as we're always reclaiming
from the head of the list.
Signed-off-by: Christoph Hellwig <hch@lst.de>