pandora-kernel.git
15 years agoMerge branch 'master' of /home/davem/src/GIT/linux-2.6/
David S. Miller [Thu, 26 Mar 2009 22:23:24 +0000 (15:23 -0700)]
Merge branch 'master' of /home/davem/src/GIT/linux-2.6/

Conflicts:
drivers/net/wimax/i2400m/usb-notif.c

15 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Thu, 26 Mar 2009 18:20:23 +0000 (11:20 -0700)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev: (35 commits)
  [libata] Improve timeout handling
  [libata] Drain data on errors
  pata_sc1200: Activate secondary channel
  pata_artop: Serializing support
  [libata] ahci: correct enclosure LED state save
  [libata] More robust parsing for IDENTIFY DEVICE multi_count field
  sata_mv: fix LED blinking for SoC+NCQ
  sata_mv: optimize IRQ coalescing for 8-port chips
  sata_mv: implement IRQ coalescing (v2)
  sata_mv: cosmetic preparations for IRQ coalescing
  pata-rb532-cf: platform_get_irq() fix ignored failure
  pata_efar: fix *dma_mask
  pata_radisys: fix mwdma_mask to exclude mwdma0
  [libata] convert drivers to use ata.h mode mask defines
  include/linux/ata.h: add some more transfer masks
  ahci: Blacklist HP Compaq 6720s that spins off disks during ACPI power off
  [libata] sata_mv: Implement direct FIS transmission via mv_qc_issue_fis().
  [libata] Export ata_pio_queue_task() so that it can be used from sata_mv.
  [libata] sata_mv: Add a new mv_sff_check_status() function to sata_mv.
  [libata] sata_mv: Tighten up interrupt masking in mv_qc_issue()
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:17:39 +0000 (11:17 -0700)]
Merge git://git./linux/kernel/git/gregkh/usb-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (97 commits)
  USB: qcserial: add device id for HP devices
  USB: isp1760: Add a delay before reading the SKIPMAP registers in isp1760-hcd.c
  USB: allow malformed LANGID descriptors
  USB: pxa27x_udc: typo fixes and code cleanups
  USB: gadget: gadget zero uses new suspend/resume hooks
  USB: gadget: composite device-level suspend/resume hooks
  USB: r8a66597-hcd: suspend/resume support
  USB: more u32 conversion after transfer_buffer_length and actual_length
  USB: Fix cp2101 USB serial device driver termios functions for console use
  USB: CP2101 New Device ID
  USB: ipaq: handle 4 endpoint devices
  USB: S3C: Move usb-control.h to platform include
  USB: ohci-hcd: Add ARCH_S3C24XX to the ohci-s3c2410.c glue
  USB: pedantic: spelling correction in comment for ch9.h
  USB: host: fix sparse warning: Using plain integer as NULL pointer
  USB: ohci-s3c2410: fix name of bus clock
  USB: ohci-s3c2410: remove <mach/hardware.h> include
  USB: serial: rename cp2101 driver to cp210x
  USB: CP2101 Reduce Error Logging
  USB: CP2101 Support AN205 baud rates
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:17:04 +0000 (11:17 -0700)]
Merge git://git./linux/kernel/git/gregkh/driver-core-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6: (61 commits)
  Dynamic debug: fix pr_fmt() build error
  Dynamic debug: allow simple quoting of words
  dynamic debug: update docs
  dynamic debug: combine dprintk and dynamic printk
  sysfs: fix some bin_vm_ops errors
  kobject: don't block for each kobject_uevent
  sysfs: only allow one scheduled removal callback per kobj
  Driver core: Fix device_move() vs. dpm list ordering, v2
  Driver core: some cleanup on drivers/base/sys.c
  Driver core: implement uevent suppress in kobject
  vcs: hook sysfs devices into object lifetime instead of "binding"
  driver core: fix passing platform_data
  driver core: move platform_data into platform_device
  sysfs: don't block indefinitely for unmapped files.
  driver core: move knode_bus into private structure
  driver core: move knode_driver into private structure
  driver core: move klist_children into private structure
  driver core: create a private portion of struct device
  driver core: remove polling for driver_probe_done(v5)
  sysfs: reference sysfs_dirent from sysfs inodes
  ...

Fixed conflicts in drivers/sh/maple/maple.c manually

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:13:06 +0000 (11:13 -0700)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6: (53 commits)
  ide: use try_to_identify() in ide_driveid_update()
  ide: clear drive IRQ after re-enabling local IRQs in ide_driveid_update()
  ide: sanitize SELECT_MASK() usage in ide_driveid_update()
  ide: classify device type in do_probe()
  ide: remove broken EXABYTENEST support
  ide: shorten timeout value in ide_driveid_update()
  ide: propagate AltStatus workarounds to ide_driveid_update()
  ide: fix kmalloc() failure handling in ide_driveid_update()
  mn10300: remove <asm/ide.h>
  frv: remove <asm/ide.h>
  ide: remove pciirq argument from ide_pci_setup_ports()
  ide: fix ->init_chipset method to return 'int' value
  ide: remove try_to_identify() wrapper
  ide: remove no longer needed IRQ auto-probing from try_to_identify() (v2)
  ide: remove no longer needed IRQ fallback code from hwif_init()
  amd74xx: remove no longer needed ->init_hwif method
  ide: remove no longer needed IDE_HFLAG[_FORCE]_LEGACY_IRQS
  ide: use ide_pci_is_in_compatibility_mode() in ide_pci_init_{one,two}()
  ide: use pci_get_legacy_ide_irq() in ide_pci_init_{one,two}()
  ide: handle IDE_HFLAG[_FORCE]_LEGACY_IRQS in ide_pci_init_{one,two}()
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:11:23 +0000 (11:11 -0700)]
Merge git://git./linux/kernel/git/lethal/sh-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (96 commits)
  sh: add support for SMSC Polaris platform
  sh: fix the HD64461 level-triggered interrupts handling
  sh: sh-rtc wakeup support
  sh: sh-rtc invalid time rework
  sh: sh-rtc carry interrupt rework
  sh: disallow kexec virtual entry
  sh: kexec jump: fix for ftrace.
  sh: kexec: Drop SR.BL bit toggling.
  sh: add kexec jump support
  sh: rework kexec segment code
  sh: simplify kexec vbr code
  sh: Flush only the needed range when unmapping a VMA.
  sh: Update debugfs ASID dumping for 16-bit ASID support.
  sh: tlb-pteaex: Kill off legacy PTEA updates.
  sh: Support for extended ASIDs on PTEAEX-capable SH-X3 cores.
  sh: sh7763rdp: Change IRQ number for sh_eth of sh7763rdp
  sh: espt-giga board support
  sh: dma: Make G2 DMA configurable.
  sh: dma: Make PVR2 DMA configurable.
  sh: Move IRQ multi definition of DMAC to defconfig
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-nmw
Linus Torvalds [Thu, 26 Mar 2009 18:08:47 +0000 (11:08 -0700)]
Merge git://git./linux/kernel/git/steve/gfs2-2.6-nmw

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-nmw:
  GFS2: Fix freeze issue
  Fix a minor bug in the previous patch
  GFS2: Clean up of glops.c
  GFS2: Fix locking bug in failed shared to exclusive conversion
  GFS2: Pagecache usage optimization on GFS2
  GFS2: fix sparse warning: Should it be static?
  GFS2: fix sparse warnings: constant is so big it is ...
  GFS2: Support quota/noquota mount arguments
  GFS2: Fix alignment issue and tidy gfs2_bitfit
  GFS2: Add a "demote a glock" interface to sysfs
  GFS2: Expose UUID via sysfs/uevent
  GFS2: Support generation of discard requests
  GFS2: Fix deadlock on journal flush
  GFS2: Fix error path ref counting for root inode
  GFS2: Remove unused field from glock
  GFS2: Merge lock_dlm module into GFS2
  GFS2: Remove "double" locking in quota
  GFS2: change gfs2_quota_scan into a shrinker
  GFS2: Bring back lvb-related stuff to lock_nolock to support quotas
  GFS2: Fix remount argument parsing

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/hskinnemoen/avr32-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:05:58 +0000 (11:05 -0700)]
Merge git://git./linux/kernel/git/hskinnemoen/avr32-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/hskinnemoen/avr32-2.6:
  avr32: remove duplicated #include
  avr32: fix out-of-range rjmp instruction on large kernels
  avr32: Fix out-of-range rcalls in large kernels

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:05:17 +0000 (11:05 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6: (430 commits)
  ALSA: hda - Add quirk for Acer Ferrari 5000
  ALSA: hda - Use cached calls to get widget caps and pin caps
  ALSA: hda - Don't create empty/single-item input source
  ALSA: hda - Fix the wrong pin-cap check in patch_realtek.c
  ALSA: hda - Cache pin-cap values
  ALSA: hda - Avoid output amp manipulation to digital mic pins
  ALSA: hda - Add function id to proc output
  ALSA: pcm - Safer boundary checks
  ALSA: hda - Detect digital-mic inputs on ALC663 / ALC272
  ALSA: sound/ali5451: typo: s/resouces/resources/
  ALSA: hda - Don't show the current connection for power widgets
  ALSA: Fix wrong pointer to dev_err() in arm/pxa2xx-ac97-lib.c
  ASoC: Declare Headset as Mic and Headphone widgets for SDP3430
  ASoC: OMAP: N810: Add more jack functions
  ASoC: OMAP: N810: Mark not connected input pins
  ASoC: Add FLL support for WM8400
  ALSA: hda - Don't reset stream at each prepare callback
  ALSA: hda - Don't reset BDL unnecessarily
  ALSA: pcm - Fix delta calculation at boundary overlap
  ALSA: pcm - Reset invalid position even without debug option
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Thu, 26 Mar 2009 18:04:34 +0000 (11:04 -0700)]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (29 commits)
  crypto: sha512-s390 - Add missing block size
  hwrng: timeriomem - Breaks an allyesconfig build on s390:
  nlattr: Fix build error with NET off
  crypto: testmgr - add zlib test
  crypto: zlib - New zlib crypto module, using pcomp
  crypto: testmgr - Add support for the pcomp interface
  crypto: compress - Add pcomp interface
  netlink: Move netlink attribute parsing support to lib
  crypto: Fix dead links
  hwrng: timeriomem - New driver
  crypto: chainiv - Use kcrypto_wq instead of keventd_wq
  crypto: cryptd - Per-CPU thread implementation based on kcrypto_wq
  crypto: api - Use dedicated workqueue for crypto subsystem
  crypto: testmgr - Test skciphers with no IVs
  crypto: aead - Avoid infinite loop when nivaead fails selftest
  crypto: skcipher - Avoid infinite loop when cipher fails selftest
  crypto: api - Fix crypto_alloc_tfm/create_create_tfm return convention
  crypto: api - crypto_alg_mod_lookup either tested or untested
  crypto: amcc - Add crypt4xx driver
  crypto: ansi_cprng - Add maintainer
  ...

15 years agoMerge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Thu, 26 Mar 2009 18:04:08 +0000 (11:04 -0700)]
Merge branch 'next' of git://git./linux/kernel/git/davej/cpufreq

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq: (35 commits)
  [CPUFREQ] Prevent p4-clockmod from auto-binding to the ondemand governor.
  [CPUFREQ] Make cpufreq-nforce2 less obnoxious
  [CPUFREQ] p4-clockmod reports wrong frequency.
  [CPUFREQ] powernow-k8: Use a common exit path.
  [CPUFREQ] Change link order of x86 cpufreq modules
  [CPUFREQ] conservative: remove 10x from def_sampling_rate
  [CPUFREQ] conservative: fixup governor to function more like ondemand logic
  [CPUFREQ] conservative: fix dbs_cpufreq_notifier so freq is not locked
  [CPUFREQ] conservative: amend author's email address
  [CPUFREQ] Use swap() in longhaul.c
  [CPUFREQ] checkpatch cleanups for acpi-cpufreq
  [CPUFREQ] powernow-k8: Only print error message once, not per core.
  [CPUFREQ] ondemand/conservative: sanitize sampling_rate restrictions
  [CPUFREQ] ondemand/conservative: deprecate sampling_rate{min,max}
  [CPUFREQ] powernow-k8: Always compile powernow-k8 driver with ACPI support
  [CPUFREQ] Introduce /sys/devices/system/cpu/cpu*/cpufreq/cpuinfo_transition_latency
  [CPUFREQ] checkpatch cleanups for powernow-k8
  [CPUFREQ] checkpatch cleanups for ondemand governor.
  [CPUFREQ] checkpatch cleanups for powernow-k7
  [CPUFREQ] checkpatch cleanups for speedstep related drivers.
  ...

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Thu, 26 Mar 2009 18:03:39 +0000 (11:03 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/security-testing-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/security-testing-2.6: (71 commits)
  SELinux: inode_doinit_with_dentry drop no dentry printk
  SELinux: new permission between tty audit and audit socket
  SELinux: open perm for sock files
  smack: fixes for unlabeled host support
  keys: make procfiles per-user-namespace
  keys: skip keys from another user namespace
  keys: consider user namespace in key_permission
  keys: distinguish per-uid keys in different namespaces
  integrity: ima iint radix_tree_lookup locking fix
  TOMOYO: Do not call tomoyo_realpath_init unless registered.
  integrity: ima scatterlist bug fix
  smack: fix lots of kernel-doc notation
  TOMOYO: Don't create securityfs entries unless registered.
  TOMOYO: Fix exception policy read failure.
  SELinux: convert the avc cache hash list to an hlist
  SELinux: code readability with avc_cache
  SELinux: remove unused av.decided field
  SELinux: more careful use of avd in avc_has_perm_noaudit
  SELinux: remove the unused ae.used
  SELinux: check seqno when updating an avc_node
  ...

15 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Thu, 26 Mar 2009 18:02:50 +0000 (11:02 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  MIPS: R2: Fix problem with code that incorrectly modifies ebase.
  MIPS: Change {set,clear,change}_c0_<foo> to return old value.
  MIPS: compat: Remove duplicated #include
  MIPS: VR5500: Enable prefetch
  MIPS: Fix oops in dma_unmap_page on not coherent mips platforms

15 years agowriteback: double the dirty thresholds
Wu Fengguang [Mon, 23 Mar 2009 00:57:38 +0000 (08:57 +0800)]
writeback: double the dirty thresholds

Enlarge default dirty ratios from 5/10 to 10/20.  This fixes [Bug
#12809] iozone regression with 2.6.29-rc6.

The iozone benchmarks are performed on a 1200M file, with 8GB ram.

  iozone -i 0 -i 1 -i 2 -i 3 -i 4 -r 4k -s 64k -s 512m -s 1200m -b tmp.xls
  iozone -B -r 4k -s 64k -s 512m -s 1200m -b tmp.xls

The performance regression is triggered by commit 1cf6e7d83bf3(mm: task
dirty accounting fix), which makes more correct/thorough dirty
accounting.

The default 5/10 dirty ratios were picked (a) with the old dirty logic
and (b) largely at random and (c) designed to be aggressive.  In
particular, that (a) means that having fixed some of the dirty
accounting, maybe the real bug is now that it was always too aggressive,
just hidden by an accounting issue.

The enlarged 10/20 dirty ratios are just about enough to fix the regression.

[ We will have to look at how this affects the old fsync() latency issue,
  but that probably will need independent work.  - Linus ]

Cc: Nick Piggin <npiggin@suse.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Reported-by: "Lin, Ming M" <ming.m.lin@intel.com>
Tested-by: "Lin, Ming M" <ming.m.lin@intel.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMake relatime default
Matthew Garrett [Thu, 26 Mar 2009 17:53:14 +0000 (17:53 +0000)]
Make relatime default

Change the default behaviour of the kernel to use relatime for all
filesystems. This can be overridden with the "strictatime" mount
option.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoAdd a strictatime mount option
Matthew Garrett [Thu, 26 Mar 2009 17:49:56 +0000 (17:49 +0000)]
Add a strictatime mount option

Add support for explicitly requesting full atime updates. This makes it
possible for kernels to default to relatime but still allow userspace to
override it.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoAllow relatime to update atime once a day
Matthew Garrett [Thu, 26 Mar 2009 17:32:14 +0000 (17:32 +0000)]
Allow relatime to update atime once a day

Allow atime to be updated once per day even with relatime. This lets
utilities like tmpreaper (which delete files based on last access time)
continue working, making relatime a plausible default for distributions.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Reviewed-by: Matthew Wilcox <willy@linux.intel.com>
Acked-by: Valerie Aurora Henson <vaurora@redhat.com>
Acked-by: Alan Cox <alan@redhat.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Thu, 26 Mar 2009 08:22:01 +0000 (01:22 -0700)]
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

15 years agoixgbe: Allow Priority Flow Control settings to survive a device reset
PJ Waskiewicz [Wed, 25 Mar 2009 22:10:42 +0000 (22:10 +0000)]
ixgbe: Allow Priority Flow Control settings to survive a device reset

When changing DCB parameters, ixgbe needs to have the MAC reset.  The way
the flow control code is setup today, PFC will be disabled on a reset.
This patch adds a new flow control type for PFC, and then has the netlink
layer take care of toggling which type of flow control to enable.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: core: remove unneeded include in net/core/utils.c.
Rami Rosen [Thu, 26 Mar 2009 08:11:48 +0000 (01:11 -0700)]
net: core: remove unneeded include in net/core/utils.c.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: update version number
Jesse Brandeburg [Wed, 25 Mar 2009 22:06:01 +0000 (22:06 +0000)]
e1000e: update version number

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: fix close interrupt race
Jesse Brandeburg [Wed, 25 Mar 2009 22:05:41 +0000 (22:05 +0000)]
e1000e: fix close interrupt race

As noticed by Alan Cox, it is possible for e1000e to exit its interrupt
handler or NAPI with interrupts enabled even when the driver is unloading or
being configured administratively down.

fix related to fix for: http://bugzilla.kernel.org/show_bug.cgi?id=12876

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: fix loss of multicast packets
Jesse Brandeburg [Wed, 25 Mar 2009 22:05:21 +0000 (22:05 +0000)]
e1000e: fix loss of multicast packets

e1000e (and e1000, igb, ixgbe, ixgb) all do a series of operations each
time a multicast address is added.  The flow goes something like

1) stack adds one multicast address
2) stack passes whole current list of unicast and multicast addresses to
   driver
3) driver clears entire list in hardware
4) driver programs each multicast address using iomem in a loop

This was causing multicast packets to be lost during the reprogramming
process.

reference with test program:
http://kerneltrap.org/mailarchive/linux-netdev/2009/3/14/5160514/thread

Thanks to Dave Boutcher for his report and test program.

This driver fix prepares an array all at once in memory and programs it in
one shot to the hardware, not requiring an "erase" cycle.  It would still
be possible for packets to be dropped while the receiver is off during
reprogramming.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: Dave Boutcher <daveboutcher@gmail.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: commonize tx cleanup routine to match e1000 & igb
Alexander Duyck [Wed, 25 Mar 2009 22:05:03 +0000 (22:05 +0000)]
e1000e: commonize tx cleanup routine to match e1000 & igb

This change updates the e1000e tx cleanup routine to more closely match
what already exists in igb and e1000.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonetfilter: fix nf_logger name in ebt_ulog.
Eric Leblond [Thu, 26 Mar 2009 08:04:28 +0000 (01:04 -0700)]
netfilter: fix nf_logger name in ebt_ulog.

This patch renames the ebt_ulog nf_logger from "ulog" to "ebt_ulog" to
be in sync with other modules naming. As this name was currently only
used for informational purpose, the renaming should be harmless.

Signed-off-by: Eric Leblond <eric@inl.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonetfilter: fix warning in ebt_ulog init function.
Eric Leblond [Thu, 26 Mar 2009 08:04:02 +0000 (01:04 -0700)]
netfilter: fix warning in ebt_ulog init function.

The ebt_ulog module does not follow the fixed convention about function
return. Loading the module is triggering the following message:

sys_init_module: 'ebt_ulog'->init suspiciously returned 1, it should follow 0/-E convention
sys_init_module: loading module anyway...
Pid: 2334, comm: modprobe Not tainted 2.6.29-rc5edenwall0-00883-g199e57b #146
Call Trace:
 [<c0441b81>] ? printk+0xf/0x16
 [<c02311af>] sys_init_module+0x107/0x186
 [<c0202cfa>] syscall_call+0x7/0xb

The following patch fixes the return treatment in ebt_ulog_init()
function.

Signed-off-by: Eric Leblond <eric@inl.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonetfilter: fix warning about invalid const usage
Eric Leblond [Thu, 26 Mar 2009 08:03:23 +0000 (01:03 -0700)]
netfilter: fix warning about invalid const usage

This patch fixes the declaration of the logger structure in ebt_log
and ebt_ulog: I forgot to remove the const option from their declaration
in the commit ca735b3aaa945626ba65a3e51145bfe4ecd9e222 ("netfilter:
use a linked list of loggers").

Pointed-out-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Eric Leblond <eric@inl.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000: fix close race with interrupt
Jesse Brandeburg [Wed, 25 Mar 2009 21:59:22 +0000 (21:59 +0000)]
e1000: fix close race with interrupt

this is in regards to
http://bugzilla.kernel.org/show_bug.cgi?id=12876

where it appears that e1000 can leave its interrupt enabled after
exiting the driver.  Fix the bug by making the interrupt enable
paths more aware of the driver exiting.

Thanks to Alan Cox for the poke and initial investigation.

CC: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000: cleanup clean_tx_irq routine so that it completely cleans ring
Alexander Duyck [Wed, 25 Mar 2009 21:59:04 +0000 (21:59 +0000)]
e1000: cleanup clean_tx_irq routine so that it completely cleans ring

The tx cleanup routine was stopping after 64 packets and this was causing
issues resulting in the ring not being completely cleaned.

This change updates the driver to clean the entire ring and if it doesn't
it then will retry on the next pass.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000: fix tx hang detect logic and address dma mapping issues
Alexander Duyck [Wed, 25 Mar 2009 21:58:45 +0000 (21:58 +0000)]
e1000: fix tx hang detect logic and address dma mapping issues

This patch changes the dma mapping to better support
skb_dma_map/skb_dma_unmap and addresses and redefines the tx hang logic to
be based off of time stamp instead of if the dma field is populated

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobridge: bad error handling when adding invalid ether address
Stephen Hemminger [Thu, 26 Mar 2009 04:01:47 +0000 (21:01 -0700)]
bridge: bad error handling when adding invalid ether address

This fixes an crash when empty bond device is added to a bridge.
If an interface with invalid ethernet address (all zero) is added
to a bridge, then bridge code detects it when setting up the forward
databas entry. But the error unwind is broken, the bridge port object
can get freed twice: once when ref count went to zeo, and once by kfree.
Since object is never really accessible, just free it.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobonding: select current active slave when enslaving device for mode tlb and alb
Jiri Pirko [Thu, 26 Mar 2009 00:23:38 +0000 (17:23 -0700)]
bonding: select current active slave when enslaving device for mode tlb and alb

I've hit an issue on my system when I've been using RealTek RTL8139D cards in
bonding interface in mode balancing-alb. When I enslave a card, the current
active slave (bond->curr_active_slave) is not set and the link is therefore
not functional.

----
# cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.5.0 (November 4, 2008)

Bonding Mode: adaptive load balancing
Primary Slave: None
Currently Active Slave: None
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth1
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:1f:1f:01:2f:22
----

The thing that gets it right is when I unplug the cable and then I put it back
into the NIC. Then the current active slave is set to eth1 and link is working
just fine. Here is dmesg log with bonding DEBUG messages turned on:
----
ADDRCONF(NETDEV_UP): bond0: link is not ready
event_dev: bond0, event: 1
IFF_MASTER
event_dev: bond0, event: 8
IFF_MASTER
bond_ioctl: master=bond0, cmd=35216
slave_dev=cac5d800:
slave_dev->name=eth1:
eth1: ! NETIF_F_VLAN_CHALLENGED
event_dev: eth1, event: 8
eth1: link up, 100Mbps, full-duplex, lpa 0xC5E1
event_dev: eth1, event: 1
event_dev: eth1, event: 8
IFF_SLAVE
Initial state of slave_dev is BOND_LINK_UP
bonding: bond0: enslaving eth1 as an active interface with an up link.
ADDRCONF(NETDEV_CHANGE): bond0: link becomes ready
event_dev: bond0, event: 4
IFF_MASTER
bond0: no IPv6 routers present

<<<<cable unplug>>>>

eth1: link down
event_dev: eth1, event: 4
IFF_SLAVE
bonding: bond0: link status definitely down for interface eth1, disabling it
event_dev: bond0, event: 4
IFF_MASTER

<<<<cable plug>>>>

eth1: link up, 100Mbps, full-duplex, lpa 0xC5E1
event_dev: eth1, event: 4
IFF_SLAVE
bonding: bond0: link status definitely up for interface eth1.
bonding: bond0: making interface eth1 the new active one.
event_dev: eth1, event: 8
IFF_SLAVE
event_dev: eth1, event: 8
IFF_SLAVE
bonding: bond0: first active interface up!
event_dev: bond0, event: 4
IFF_MASTER
----

The current active slave is set by calling bond_select_active_slave() function
from bond_miimon_commit() function when the slave (eth1) link goes to state up.

I also tested this on other machine with Broadcom NetXtreme II BCM5708
1000Base-T NIC and there all works fine. The thing is that this adapter is down
and goes up after few seconds after it is enslaved.

This patch calls bond_select_active_slave() in bond_enslave() function for modes
alb and tlb and makes sure that the current active slave is set up properly even
when the slave state is already up. Tested on both systems, works fine.

Notice: The same problem can maybe also occrur in mode 8023AD but I'm unable to
test that.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogianfar: reallocate skb when headroom is not enough for fcb
Li Yang [Tue, 24 Mar 2009 23:15:33 +0000 (23:15 +0000)]
gianfar: reallocate skb when headroom is not enough for fcb

Gianfar uses a hardware header FCB for offloading.  However when used
with bridging or IP forwarding, TX skb might not have enough headroom
for the FCB.  Reallocate skb for such cases.

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoBump release date to 25Mar2009 and version to 0.22
Florian Fainelli [Tue, 24 Mar 2009 23:34:38 +0000 (23:34 +0000)]
Bump release date to 25Mar2009 and version to 0.22

This patch bumps the driver release date to March 25th 2009
and release version to 0.22.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agor6040: Fix second PHY address
Florian Fainelli [Tue, 24 Mar 2009 23:34:35 +0000 (23:34 +0000)]
r6040: Fix second PHY address

This patch fixes the second PHY address which is strapped
to be at PHY address 3 instead of 2.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: fix wait_event_timeout handling
Heiko Carstens [Tue, 24 Mar 2009 20:57:19 +0000 (20:57 +0000)]
qeth: fix wait_event_timeout handling

wait_event_timeout just takes the numnber of jiffies to wait as
an argument. That value does not include jiffies itself.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: check for completion of a running recovery
Ursula Braun [Tue, 24 Mar 2009 20:57:18 +0000 (20:57 +0000)]
qeth: check for completion of a running recovery

When a recovery is started for a qeth device, additional invocations
to change a mac address, to configure a VLAN interface on top, or to
add multicast addresses should wait till recovery is finished,
otherwise recovery might fail.

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: unregister MAC addresses during recovery.
Klaus-Dieter Wacker [Tue, 24 Mar 2009 20:57:17 +0000 (20:57 +0000)]
qeth: unregister MAC addresses during recovery.

qeth: Unregister MAC addresses from device (layer 2) during
recovery cycle. When the device is set online the MAC
addresses are registered again on the device.

Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: remove EDDP
Frank Blaschka [Tue, 24 Mar 2009 20:57:16 +0000 (20:57 +0000)]
qeth: remove EDDP

Performance measurements showed EDDP does not lower CPU costs but increase
them. So we dump out EDDP code from qeth driver.

Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: add statistics for tx csum
Frank Blaschka [Tue, 24 Mar 2009 20:57:15 +0000 (20:57 +0000)]
qeth: add statistics for tx csum

Add statistics counter for software tx checksumming.

Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: struct device - replace bus_id with dev_name(), dev_set_name()
Kay Sievers [Tue, 24 Mar 2009 20:57:14 +0000 (20:57 +0000)]
qeth: struct device - replace bus_id with dev_name(), dev_set_name()

Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agorndis_wlan: Fix build with netdev_ops compat disabled.
David S. Miller [Wed, 25 Mar 2009 07:03:16 +0000 (00:03 -0700)]
rndis_wlan: Fix build with netdev_ops compat disabled.

Instead of storing a private ->set_multicast_list, just
have a private netdev ops.

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoax88796: Add method to take MAC from platform data
Daniel Mack [Wed, 25 Mar 2009 06:32:03 +0000 (23:32 -0700)]
ax88796: Add method to take MAC from platform data

Implement a way to provide the MAC address for ax88796 devices from
their platform data. Boards might decide to set the address
programmatically, taken from boot tags or other sources.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agodrivers/net/ax88796.c: take IRQ flags from platform_device
Daniel Mack [Wed, 25 Mar 2009 06:31:22 +0000 (23:31 -0700)]
drivers/net/ax88796.c: take IRQ flags from platform_device

This patch adds support to the ax88796 ethernet driver to take IRQ flags
given by the platform_device definition.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: add support for 82574 device ID 0x10F6
Bruce Allan [Wed, 25 Mar 2009 06:28:02 +0000 (23:28 -0700)]
e1000e: add support for 82574 device ID 0x10F6

Add device ID for a new variant of the 82574 adapter.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Acked-by: John Ronciak <john.ronciak@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years ago[libata] Improve timeout handling
Alan Cox [Tue, 24 Mar 2009 10:23:46 +0000 (10:23 +0000)]
[libata] Improve timeout handling

On a timeout call a device specific handler early in the recovery so that
we can complete and process successful commands which timed out due to IRQ
loss or the like rather more elegantly.

[Revised to exclude the timeout handling on a few devices that inherit from
 SFF but are not SFF enough to use the default timeout handler]

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agoipv6: Fix conflict resolutions during ipv6 binding
Vlad Yasevich [Tue, 24 Mar 2009 16:24:51 +0000 (16:24 +0000)]
ipv6: Fix conflict resolutions during ipv6 binding

The ipv6 version of bind_conflict code calls ipv6_rcv_saddr_equal()
which at times wrongly identified intersections between addresses.
It particularly broke down under a few instances and caused erroneous
bind conflicts.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6: Make v4-mapped bindings consistent with IPv4
Vlad Yasevich [Tue, 24 Mar 2009 16:24:50 +0000 (16:24 +0000)]
ipv6: Make v4-mapped bindings consistent with IPv4

Binding to a v4-mapped address on an AF_INET6 socket should
produce the same result as binding to an IPv4 address on
AF_INET socket.  The two are interchangable as v4-mapped
address is really a portability aid.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6: Allow ipv4 wildcard binds after ipv6 address binds
Vlad Yasevich [Tue, 24 Mar 2009 16:24:49 +0000 (16:24 +0000)]
ipv6: Allow ipv4 wildcard binds after ipv6 address binds

The IPv4 wildcard (0.0.0.0) address does not intersect
in any way with explicit IPv6 addresses.  These two should
be permitted, but the IPv4 conflict code checks the ipv6only
bit as part of the test.  Since binding to an explicit IPv6
address restricts the socket to only that IPv6 address, the
side-effect is that the socket behaves as v6-only.  By
explicitely setting ipv6only in this case, allows the 2 binds
to succeed.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6: Disallow binding to v4-mapped address on v6-only socket.
Vlad Yasevich [Tue, 24 Mar 2009 16:24:48 +0000 (16:24 +0000)]
ipv6: Disallow binding to v4-mapped address on v6-only socket.

A socket marked v6-only, can not receive or send traffic to v4-mapped
addresses.  Thus allowing binding to v4-mapped address on such a
socket makes no sense.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years ago[libata] Drain data on errors
Alan Cox [Tue, 24 Mar 2009 10:23:19 +0000 (10:23 +0000)]
[libata] Drain data on errors

If the device is signalling that there is data to drain after an error we
should read the bytes out and throw them away. Without this some devices
and controllers get wedged and don't recover.

Based on earlier work by Mark Lord

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata_sc1200: Activate secondary channel
Alan Cox [Tue, 24 Mar 2009 10:22:25 +0000 (10:22 +0000)]
pata_sc1200: Activate secondary channel

Implement serialize and turn on slave channel

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata_artop: Serializing support
Alan Cox [Tue, 24 Mar 2009 10:21:49 +0000 (10:21 +0000)]
pata_artop: Serializing support

Enable both ports on the 6210 and serialize them

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] ahci: correct enclosure LED state save
David Milburn [Fri, 20 Mar 2009 19:14:23 +0000 (14:14 -0500)]
[libata] ahci: correct enclosure LED state save

ahci_transmit_led_message saves off the led_state
with a value that includes the port number OR'd
in, this incorrect value maybe reported back
in ahci_led_store.

For instance, if you turn off all the leds for
port 1 and cat the value back it will report 1
instead of 0.

# echo 0 > /sys/class/scsi_host/host1/em_message
# cat /sys/class/scsi_host/host1/em_message
1

Signed-off-by: David Milburn <dmilburn@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] More robust parsing for IDENTIFY DEVICE multi_count field
Mark Lord [Thu, 19 Mar 2009 17:32:21 +0000 (13:32 -0400)]
[libata] More robust parsing for IDENTIFY DEVICE multi_count field

Make libata more robust when parsing the multi_count
field from a drive's identify data.  This prevents us from
attempting to use dubious multi_count values ad infinitum.

Reset dev->multi_count to zero and reprobe it each time
through this routine, as it can change on device reset.

Also ensure that the reported "maximum" value is valid
and is a power of two, and that the reported "count" value
is valid and also a power of two.  And that the "count"
value is not greater than the "maximum" value.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: fix LED blinking for SoC+NCQ
Mark Lord [Sun, 15 Mar 2009 15:33:19 +0000 (11:33 -0400)]
sata_mv: fix LED blinking for SoC+NCQ

For Marvell SoC chips, the HDD LED does not blink when there is
disk I/O if NCQ is enabled. Add a quirk that enables blink mode for
the LED while NCQ is enabled on any port of a SoC host controller.
Normal LED function is restored when NCQ is not enabled on any port.

The code to enable the blink mode is based on earlier code
and suggestions from Frans Pop, Saeed Bishara, and possibly others.

Signed-off-by: Mark Lord <mlord@pobox.com>
Tested-by: Frans Pop <elendil@planet.nl>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: optimize IRQ coalescing for 8-port chips
Mark Lord [Wed, 11 Mar 2009 04:56:00 +0000 (00:56 -0400)]
sata_mv: optimize IRQ coalescing for 8-port chips

Enable use of the "all ports" IRQ coalescing optimization
for GEN_II / GEN_IIE chips that have dual host-controllers (8-ports).
Currently only the 6081 chip qualifies, but other chips may come along someday.

Rather than each half of the chip having to satisfy a local set of coalescing thresholds,
use of this feature groups all ports together under a single set of thresholds.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: implement IRQ coalescing (v2)
Mark Lord [Wed, 11 Mar 2009 02:01:17 +0000 (22:01 -0400)]
sata_mv: implement IRQ coalescing (v2)

Add IRQ coalescing to sata_mv (off by default).

This feature can reduce total interrupt overhead for RAID setups
in some situations, by deferring the interrupt signal until one or both of:

 a) a specified io_count (completed SATA commands) is achieved, or
 b) a specified time interval elapses after an IO completion.

For now, module parameters are used to set the irq_coalescing_io_count
and irq_coalescing_usecs (timeout) globally.  These may eventually
be supplemented with sysfs attributes, so that thresholds can be set
on-the-fly and on a per-chip (or even per-host_controller) basis.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: cosmetic preparations for IRQ coalescing
Mark Lord [Tue, 10 Mar 2009 22:51:04 +0000 (18:51 -0400)]
sata_mv: cosmetic preparations for IRQ coalescing

Various cosmetic changes in preparation for the IRQ coalescing feature.

Note that the various MV_IRQ_COAL_* definitions are restored/renamed
in the folloup patch which adds IRQ coalescing to the driver.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata-rb532-cf: platform_get_irq() fix ignored failure
Florian Fainelli [Fri, 13 Mar 2009 14:41:43 +0000 (15:41 +0100)]
pata-rb532-cf: platform_get_irq() fix ignored failure

platform_get_irq() can return -ENXIO, but since 'irq' is an
unsigned int, it does not show when the IRQ resource wasn't found.
Make irq an int so that we can use a single variable to test the
platform_get_irq() return value.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Phil Sutter <n0-1@freewrt.org>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata_efar: fix *dma_mask
Erik Inge Bolsø [Sat, 14 Mar 2009 22:08:20 +0000 (23:08 +0100)]
pata_efar: fix *dma_mask

According to Alan:
>and yes the EFAR does UDMA66.

mwdma:
>Yep - wrong comment. The EFAR is a sort of clone of the PIIX and I
>copied the comment while EFAR don't appear to have copied the
>limitation

Signed-off-by: Erik Inge Bolsø <knan-lkml@anduin.net>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata_radisys: fix mwdma_mask to exclude mwdma0
Erik Inge Bolsø [Sat, 14 Mar 2009 22:07:33 +0000 (23:07 +0100)]
pata_radisys: fix mwdma_mask to exclude mwdma0

As noted by Alan:
>Your suspicions are correct here btw - the device can only do MWDMA1 and
>MWDMA2 (much like some PIIX devices)

Signed-off-by: Erik Inge Bolsø <knan-lkml@anduin.net>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] convert drivers to use ata.h mode mask defines
Erik Inge Bolsø [Sat, 14 Mar 2009 20:38:24 +0000 (21:38 +0100)]
[libata] convert drivers to use ata.h mode mask defines

No functional changes in this patch.

Signed-off-by: Erik Inge Bolsø <knan-lkml@anduin.net>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agoinclude/linux/ata.h: add some more transfer masks
Erik Inge Bolsø [Sat, 14 Mar 2009 20:37:48 +0000 (21:37 +0100)]
include/linux/ata.h: add some more transfer masks

Signed-off-by: Erik Inge Bolsø <knan-lkml@anduin.net>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agoahci: Blacklist HP Compaq 6720s that spins off disks during ACPI power off
Maciej Rutecki [Thu, 19 Mar 2009 23:06:46 +0000 (00:06 +0100)]
ahci: Blacklist HP Compaq 6720s that spins off disks during ACPI power off

Blacklist HP Compaq 6720s so that it doesn't play a "spin down,
spin up, spin down" ping-pong with the hard disk during system
power off.

Signed-off-by: Maciej Rutecki <maciej.rutecki@gmail.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] sata_mv: Implement direct FIS transmission via mv_qc_issue_fis().
Mark Lord [Wed, 25 Feb 2009 20:19:20 +0000 (15:19 -0500)]
[libata] sata_mv: Implement direct FIS transmission via mv_qc_issue_fis().

This is initially needed to work around NCQ errata,
whereby the READ_LOG_EXT command sometimes fails
when issued in the traditional (sff) fashion.

Portions of this code will likely be reused for
implementation of the target mode feature later on.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] Export ata_pio_queue_task() so that it can be used from sata_mv.
Mark Lord [Wed, 25 Feb 2009 20:18:32 +0000 (15:18 -0500)]
[libata] Export ata_pio_queue_task() so that it can be used from sata_mv.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] sata_mv: Add a new mv_sff_check_status() function to sata_mv.
Mark Lord [Wed, 25 Feb 2009 20:17:43 +0000 (15:17 -0500)]
[libata] sata_mv: Add a new mv_sff_check_status() function to sata_mv.
This is necessary for use with the upcoming "mv_qc_issue_fis()" patch,
but is being added separately here for easier code review.

When using command issue via the "mv_qc_issue_fis()" mechanism,
the initial ATA_BUSY bit does not show in the ATA status (shadow) register.
This can confuse libata!  So here we add a hook to fake ATA_BUSY
for that situation, until the first time a BUSY, DRQ, or ERR bit is seen.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] sata_mv: Tighten up interrupt masking in mv_qc_issue()
Mark Lord [Wed, 25 Feb 2009 20:15:39 +0000 (15:15 -0500)]
[libata] sata_mv: Tighten up interrupt masking in mv_qc_issue()
so that it doesn't miss any protocols.  Handle future cases where a
qc is specially marked for polled issue or where a particular chip
version prefers interrupts over polling for PIO.

This mimics the polling decision logic from ata_sff_qc_issue().

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] sata_mv: Enable use of (basic) DMA for ATAPI on GEN_IIE chips
Mark Lord [Wed, 25 Feb 2009 20:14:48 +0000 (15:14 -0500)]
[libata] sata_mv: Enable use of (basic) DMA for ATAPI on GEN_IIE chips

This also gets rid of any need for mv_mode_filter().

Using basic DMA on GEN_IIE requires setting an undocumented
bit in an undocumented register.  For safety, we clear that
bit again when switching back to EDMA mode.

To avoid a performance penalty when switching modes,
we cache the register in port_priv, as already done for other regs.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years ago[libata] sata_mv: cache frequently-accessed registers
Mark Lord [Wed, 25 Feb 2009 20:13:03 +0000 (15:13 -0500)]
[libata] sata_mv: cache frequently-accessed registers

Maintain a local (mv_port_priv) cache of frequently accessed registers,
to avoid having to re-read them (very slow) on every transistion
between EDMA and non-EDMA modes.  This speeds up things like
flushing the drive write cache, and anything using basic DMA transfers.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agoahci: drop intx manipulation on msi enable
Tejun Heo [Fri, 23 Jan 2009 02:31:39 +0000 (11:31 +0900)]
ahci: drop intx manipulation on msi enable

There's no need to turn off intx explicitly on msi enable.  This is
automatically handled by pci.  Drop it.

This might be needed on machines if the BIOS turns intx off during
boot.  However, there's no evidence of such behavior for ahci and
the only such case seems to be ICH5 PATA according to ata_piix.
Also, given the way ahci operates, it's highly unlikely BIOS ever
disables IRQ for the controller.  However, as this change has slight
possibility of introducing failure, please schedule it for #upstream.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata-rb532-cf: drop custom freeze and thaw
Phil Sutter [Tue, 27 Jan 2009 13:35:53 +0000 (14:35 +0100)]
pata-rb532-cf: drop custom freeze and thaw

I'm not quite sure what freezing and thawing is used for. Tests showed
that the port is being frozen at initialisation state and thawed right
afterwards, then the functions were not called anymore. Dropping the
complete custom code for handling the frozen state seems to work at
least for a standard use case including mounting a partition, copying
some files in it (in parallel) and finally removing them and unmounting
the partition.

Signed-off-by: Phil Sutter <n0-1@freewrt.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata-rb532-cf: use ata_sff_data_xfer32()
Phil Sutter [Tue, 27 Jan 2009 13:35:52 +0000 (14:35 +0100)]
pata-rb532-cf: use ata_sff_data_xfer32()

The biggest difference between rb532_pata_data_xfer() and
ata_sff_data_xfer32() is the call to ata_sff_pause() at the end of
rb532_pata_data_xfer() which I suppose to be unnecessary since it works
without. I've also tested using ata_sff_data_xfer() as replacement, but
since we know that the driver supports 32bit IO, using the optimised
version should be safe.

Signed-off-by: Phil Sutter <n0-1@freewrt.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata-rb532-cf: use ata_sff_exec_command()
Phil Sutter [Tue, 27 Jan 2009 13:35:51 +0000 (14:35 +0100)]
pata-rb532-cf: use ata_sff_exec_command()

The only difference between rb532_pata_exec_command() and
ata_sff_exec_command() is added debugging output, so it can be dropped
and the standard op used instead.

Signed-off-by: Phil Sutter <n0-1@freewrt.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata-rb532-cf: replace rb532_pata_finish_io()
Phil Sutter [Tue, 27 Jan 2009 13:35:50 +0000 (14:35 +0100)]
pata-rb532-cf: replace rb532_pata_finish_io()

Since the delay used internally is just the same as ata_sff_pause()
uses, rb532_pata_finish_io() does exactly the same as ata_sff_pause()
and thus can be replaced by the later one.

Signed-off-by: Phil Sutter <n0-1@freewrt.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: remove leftovers
Mark Lord [Sat, 31 Jan 2009 02:40:48 +0000 (21:40 -0500)]
sata_mv: remove leftovers

Remove redundant code left over from the earlier patch 04/07.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: optimize use of mv_edma_cfg
Mark Lord [Fri, 30 Jan 2009 23:52:58 +0000 (18:52 -0500)]
sata_mv: optimize use of mv_edma_cfg

Try and avoid unnecessary reconfiguration of the EDMA config register
on every single non-EDMA I/O operation, by moving the call to
mv_edma_cfg() into mv_stop_edma().  It must then also be invoked
from mv_hardreset() and from mv_port_start().

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: introduce support for ATAPI devices
Mark Lord [Fri, 30 Jan 2009 23:51:54 +0000 (18:51 -0500)]
sata_mv: introduce support for ATAPI devices

Add ATAPI support to sata_mv, using sff DMA for GEN_II chipsets,
and plain old PIO for GEN_IIE.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: mv_fill_sg fixes v2
Mark Lord [Sun, 1 Feb 2009 21:50:32 +0000 (16:50 -0500)]
sata_mv: mv_fill_sg fixes v2

Fix mv_fill_sg() to zero out the reserved word (required for ATAPI),
and to include a memory barrier.  This may also help with problems
reported by Jens on the PPC platform.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: update ata_qc_from_tag
Mark Lord [Fri, 30 Jan 2009 23:49:29 +0000 (18:49 -0500)]
sata_mv: update ata_qc_from_tag

Update the logic in ata_qc_from_tag() to match that used
in similar places elsewhere in libata.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: restructure mv_qc_issue
Mark Lord [Fri, 30 Jan 2009 23:48:41 +0000 (18:48 -0500)]
sata_mv: restructure mv_qc_issue

Rearrange logic in mv_qc_issue() to handle protocols
other than ATA_PROT_DMA, ATA_PROT_NCQ, and ATA_PROT_PIO.
This is in preparation for later enabling ATAPI support.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: rearrange mv_start_dma() and friends
Mark Lord [Fri, 30 Jan 2009 23:47:51 +0000 (18:47 -0500)]
sata_mv: rearrange mv_start_dma() and friends

Rearrange mv_start_dma() and friends, in preparation for adding
non-EDMA DMA modes, and non-EDMA interrupts, to the driver.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_mv: cleanup chipset GENeration FLAGS
Mark Lord [Fri, 30 Jan 2009 23:46:39 +0000 (18:46 -0500)]
sata_mv: cleanup chipset GENeration FLAGS

Clean up the chipset GENeration FLAGS, and rename them
for consistency with other uses of GEN_XX within sata_mv.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agobe2net: cleanup rx/tx rate calculations
Sathya Perla [Tue, 24 Mar 2009 23:40:13 +0000 (16:40 -0700)]
be2net: cleanup rx/tx rate calculations

Hi, Pls accept this patch to cleanup rx/tx rate calculations as follows:
- check for jiffies wraparound
- remove typecast of a denominator
- do rate calculation only in workqueue context periodically

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/kaber/nf-2.6
David S. Miller [Tue, 24 Mar 2009 23:38:53 +0000 (16:38 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/kaber/nf-2.6

15 years agoDynamic debug: fix pr_fmt() build error
Greg Banks [Wed, 11 Mar 2009 10:07:28 +0000 (21:07 +1100)]
Dynamic debug: fix pr_fmt() build error

When CONFIG_DYNAMIC_DEBUG is enabled, allow callers of pr_debug()
to provide their own definition of pr_fmt() even if that definition
uses tricks like

#define pr_fmt(fmt) "%s:" fmt, __func__

Signed-off-by: Greg Banks <gnb@sgi.com>
Cc: Jason Baron <jbaron@redhat.com>
Acked-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoDynamic debug: allow simple quoting of words
Greg Banks [Fri, 6 Feb 2009 01:54:26 +0000 (12:54 +1100)]
Dynamic debug: allow simple quoting of words

Allow simple quoting of words in the dynamic debug control language.

This allows more natural specification when using the control language
to match against printk formats, e.g

#echo -n 'format "Setting node for non-present cpu" +p' >
/mnt/debugfs/dynamic_debug/control

instead of

#echo -n 'format Setting\040node\040for\040non-present\040cpu +p' >
/mnt/debugfs/dynamic_debug/control

Adjust the dynamic debug documention to describe that and provide a
new example.  Adjust the existing examples in the documentation to
reflect the current whitespace escaping behaviour when reading the
control file.  Fix some minor documentation trailing whitespace.

Signed-off-by: Greg Banks <gnb@melbourne.sgi.com>
Acked-by: Jason Baron <jbaron@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agodynamic debug: update docs
Jason Baron [Thu, 5 Feb 2009 16:53:15 +0000 (11:53 -0500)]
dynamic debug: update docs

updates the documentation for 'dynamic debug' feature.

Signed-off-by: Greg Banks <gnb@sgi.com>
Signed-off-by: Jason Baron <jbaron@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agodynamic debug: combine dprintk and dynamic printk
Jason Baron [Thu, 5 Feb 2009 16:51:38 +0000 (11:51 -0500)]
dynamic debug: combine dprintk and dynamic printk

This patch combines Greg Bank's dprintk() work with the existing dynamic
printk patchset, we are now calling it 'dynamic debug'.

The new feature of this patchset is a richer /debugfs control file interface,
(an example output from my system is at the bottom), which allows fined grained
control over the the debug output. The output can be controlled by function,
file, module, format string, and line number.

for example, enabled all debug messages in module 'nf_conntrack':

echo -n 'module nf_conntrack +p' > /mnt/debugfs/dynamic_debug/control

to disable them:

echo -n 'module nf_conntrack -p' > /mnt/debugfs/dynamic_debug/control

A further explanation can be found in the documentation patch.

Signed-off-by: Greg Banks <gnb@sgi.com>
Signed-off-by: Jason Baron <jbaron@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agosysfs: fix some bin_vm_ops errors
Hugh Dickins [Mon, 23 Mar 2009 01:41:27 +0000 (01:41 +0000)]
sysfs: fix some bin_vm_ops errors

Commit 86c9508eb1c0ce5aa07b5cf1d36b60c54efc3d7a
"sysfs: don't block indefinitely for unmapped files" in linux-next
crashes the PowerMac G5 when X starts up.  It's caught out by the way
powerpc's pci_mmap of legacy_mem uses shmem_zero_setup(), substituting
a new vma->vm_file whose private_data no longer points to the bin_buffer
(substitution done because some versions of X crash if that mmap fails).

The fix to this is straightforward: the original vm_file is fput() in
that case, so this mmap won't block sysfs at all, so just don't switch
over to bin_vm_ops if vm_file has changed.

But more fixes made before realizing that was the problem:-

It should not be an error if bin_page_mkwrite() finds no underlying
page_mkwrite().

Check that a file already mmap'ed has the same underlying vm_ops
_before_ pointing vma->vm_ops at bin_vm_ops.

If the file being mmap'ed is a shmem/tmpfs file, don't fail the mmap
on CONFIG_NUMA=y, just because that has a set_policy and get_policy:
provide bin_set_policy, bin_get_policy and bin_migrate.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Eric Biederman <ebiederm@aristanetworks.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agokobject: don't block for each kobject_uevent
Arjan van de Ven [Thu, 19 Mar 2009 16:09:05 +0000 (09:09 -0700)]
kobject: don't block for each kobject_uevent

Right now, the kobject_uevent code blocks for each uevent that's being
generated, due to using (for hystoric reasons) UHM_WAIT_EXEC as flag to
call_usermode_helper().  Specifically, the effect is that each uevent
that is being sent causes the code to wake up keventd, then block until
keventd has processed the work. Needless to say, this happens many times
during the system boot.

This patches changes that to UHN_NO_WAIT (brilliant name for a constant
btw) so that we only schedule the work to fire the uevent message, but
do not wait for keventd to process the work.

This removes one of the bottlenecks during boot; each one of them is
only a small effect, but the sum of them does add up.

[Note, distros that need this are broken, they should be setting
CONFIG_UEVENT_HELPER_PATH to "", that way this code path will never be
excuted at all -- gregkh]

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agosysfs: only allow one scheduled removal callback per kobj
Alex Chiang [Fri, 13 Mar 2009 18:07:36 +0000 (12:07 -0600)]
sysfs: only allow one scheduled removal callback per kobj

The only way for a sysfs attribute to remove itself (without
deadlock) is to use the sysfs_schedule_callback() interface.

Vegard Nossum discovered that a poorly written sysfs ->store
callback can repeatedly schedule remove callbacks on the same
device over and over, e.g.

$ while true ; do echo 1 > /sys/devices/.../remove ; done

If the 'remove' attribute uses the sysfs_schedule_callback API
and also does not protect itself from concurrent accesses, its
callback handler will be called multiple times, and will
eventually attempt to perform operations on a freed kobject,
leading to many problems.

Instead of requiring all callers of sysfs_schedule_callback to
implement their own synchronization, provide the protection in
the infrastructure.

Now, sysfs_schedule_callback will only allow one scheduled
callback per kobject. On subsequent calls with the same kobject,
return -EAGAIN.

This is a short term fix. The long term fix is to allow sysfs
attributes to remove themselves directly, without any of this
callback hokey pokey.

[cornelia.huck@de.ibm.com: s390 ccwgroup bits]

Reported-by: vegard.nossum@gmail.com
Signed-off-by: Alex Chiang <achiang@hp.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoDriver core: Fix device_move() vs. dpm list ordering, v2
Cornelia Huck [Wed, 4 Mar 2009 11:44:00 +0000 (12:44 +0100)]
Driver core: Fix device_move() vs. dpm list ordering, v2

dpm_list currently relies on the fact that child devices will
be registered after their parents to get a correct suspend
order. Using device_move() however destroys this assumption, as
an already registered device may be moved under a newly registered
one.

This patch adds a new argument to device_move(), allowing callers
to specify how dpm_list should be adapted.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoDriver core: some cleanup on drivers/base/sys.c
Zhenwen Xu [Tue, 3 Mar 2009 10:36:02 +0000 (18:36 +0800)]
Driver core: some cleanup on drivers/base/sys.c

do some cleanup on drivers/base/sys.c

Signed-off-by: Zhenwen Xu <helight.xu@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoDriver core: implement uevent suppress in kobject
Ming Lei [Sun, 1 Mar 2009 13:10:49 +0000 (21:10 +0800)]
Driver core: implement uevent suppress in kobject

This patch implements uevent suppress in kobject and removes it
from struct device, based on the following ideas:

1,Uevent sending should be one attribute of kobject, so suppressing it
in kobject layer is more natural than in device layer. By this way,
we can do it for other objects embedded with kobject.

2,It may save several bytes for each instance of struct device.(On my
omap3(32bit ARM) based box, can save 8bytes per device object)

This patch also introduces dev_set|get_uevent_suppress() helpers to
set and query uevent_suppress attribute in case to help kobject
as private part of struct device in future.

[This version is against the latest driver-core patch set of Greg,please
ignore the last version.]

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agovcs: hook sysfs devices into object lifetime instead of "binding"
Kay Sievers [Mon, 9 Mar 2009 13:18:52 +0000 (14:18 +0100)]
vcs: hook sysfs devices into object lifetime instead of "binding"

During bootup performance tracing I noticed many occurrences of
vca* device creation and removal, leading to the usual userspace
uevent processing, which are, in this case, rather pointless.

A simple test showing the kernel timing (not including all the
work userspace has to do), gives us these numbers:
  $ time for i in `seq 1000`; do echo a > /dev/tty2; done
  real    0m1.142s
  user    0m0.015s
  sys     0m0.540s

If we move the hook for the vcs* driver core devices from the
tty "binding" to the vc allocation/deallocation, which is what
the vcs* devices represent, we get the following numbers:
  $ time for i in `seq 1000`; do echo a > /dev/tty2; done
  real    0m0.152s
  user    0m0.030s
  sys     0m0.072s

Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agodriver core: fix passing platform_data
Ming Lei [Fri, 13 Mar 2009 15:06:59 +0000 (23:06 +0800)]
driver core: fix passing platform_data

We will remove platform_data field from struct device until
all platform devices pass its specific data from platfom_device
and all platform drivers use platform specific data passed by
platform_device->platform_data. This kind of conversion will
need a long time, for thousands of files is affected.

To make the conversion easily, we allow platform specific data
passed by struct device or struct platform_device and platform
driver may use it from struct device or struct platform_device.

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agodriver core: move platform_data into platform_device
Ming Lei [Sun, 8 Mar 2009 15:13:32 +0000 (23:13 +0800)]
driver core: move platform_data into platform_device

This patch moves platform_data from struct device into
struct platform_device, based on the two ideas:

1. Now all platform_driver is registered by platform_driver_register,
   which makes probe()/release()/... of platform_driver passed parameter
   of platform_device *, so platform driver can get platform_data from
   platform_device;

2. Other kind of devices do not need to use platform_data, we can
   decrease size of device if moving it to platform_device.

Taking into consideration of thousands of files to be fixed and they
can't be finished in one night(maybe it will take a long time), so we
keep platform_data in device to allow two kind of cases coexist until
all platform devices pass its platfrom data from
platform_device->platform_data.

All patches to do this kind of conversion are welcome.

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agosysfs: don't block indefinitely for unmapped files.
Eric W. Biederman [Wed, 4 Mar 2009 19:57:20 +0000 (11:57 -0800)]
sysfs: don't block indefinitely for unmapped files.

Modify sysfs bin files so that we can remove the bin file while they are
still mapped.  When the kobject is removed we unmap the bin file and
arrange for future accesses to the mapping to receive SIGBUS.

Implementing this prevents a nasty DOS when pci devices are hot plugged
and unplugged.  Where if any of their resources were mmaped the kernel
could not free up their pci resources or release their pci data
structures.

[akpm@linux-foundation.org: remove unused var]
Signed-off-by: Eric W. Biederman <ebiederm@aristanetworks.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>