]> err.no Git - linux-2.6/log
linux-2.6
17 years agokbuild: fix up printing of Linux C Library version in scripts/ver_linux
Jesper Juhl [Fri, 20 Jul 2007 00:49:23 +0000 (02:49 +0200)]
kbuild: fix up printing of Linux C Library version in scripts/ver_linux

I noticed, when running scripts/ver_linux on both a Gentoo system
and a Slackware system, that the line printing the C library
version looked a little odd. So I fixed it up to be in line with
all the rest.

  Old output:
    Linux C Library        > libc.2.5
  New output:
    Linux C Library        2.5

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
17 years agokbuild: do not do section mismatch checks on vmlinux in 2nd pass
Sam Ravnborg [Fri, 20 Jul 2007 20:36:56 +0000 (22:36 +0200)]
kbuild: do not do section mismatch checks on vmlinux in 2nd pass

We already check and warn about section mismatches from vmlinux
(build as vmlinux.o) during first pass so skip the checks
during the 2nd pass where we process modules.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
17 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Wed, 25 Jul 2007 18:28:19 +0000 (11:28 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/selinux-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/selinux-2.6:
  SELinux: null-terminate context string in selinux_xfrm_sec_ctx_alloc

17 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Wed, 25 Jul 2007 18:28:00 +0000 (11:28 -0700)]
Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI: Kconfig: remove CONFIG_ACPI_SLEEP from source
  ACPI: quiet ACPI Exceptions due to no _PTC or _TSS
  ACPI: Remove references to ACPI_STATE_S2 from acpi_pm_enter
  ACPI: Kconfig: always enable CONFIG_ACPI_SLEEP on X86
  ACPI: Kconfig: fold /proc/acpi/sleep under CONFIG_ACPI_PROCFS
  ACPI: Kconfig: CONFIG_ACPI_PROCFS now defaults to N
  ACPI: autoload modules - Create __mod_acpi_device_table symbol for all ACPI drivers
  ACPI: autoload modules - Create ACPI alias interface
  ACPI: autoload modules - ACPICA modifications
  ACPI: asus-laptop: Fix failure exits
  ACPI: fix oops due to typo in new throttling code
  ACPI: ignore _PSx method for hotplugable PCI devices
  ACPI: Use ACPI methods to select PCI device suspend state
  ACPI, PNP: hook ACPI D-state to PNP suspend/resume
  ACPI: Add acpi_pm_device_sleep_state helper routine
  ACPI: Implement the set_target() callback from pm_ops

17 years agom68knommu: make BOOTPARAM setup common
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: make BOOTPARAM setup common

Currently most of the m68knommu cpu/board setup files are handling
the setup of fixed boot parameters (via CONFIG_BOOTPARAM) themselves.
Move all this into the common setup code.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: fix reset register address casting
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: fix reset register address casting

Fix types used for reset register address setup.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: add configure support for Intec boards
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: add configure support for Intec boards

Add configure support for the Intec Wildfire and WildFireMod boards.

Signed-Off-By: Steve Bennett <steveb@workware.net.au>
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: simplify ColdFire resume code
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: simplify ColdFire resume code

It is useless to preserve THREAD_SR in `resume'.  The real user's sr
is actually in the stack.  We also don't need to disable interrupts :
we'll never be in an invalid state, the sp switch is atomic.

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: create hw_irq.h
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: create hw_irq.h

Need an include/asm-m68knommu/hw_irq.h for kernel/hrtimer.c

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: fix definition of MCFDMA_DIR_INV
Philippe De Muyter [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: fix definition of MCFDMA_DIR_INV

Fix a small typo in the definition of MCFDMA_DIR_INV (MCF5272 specific).

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: add build support for more Arcturus boards
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: add build support for more Arcturus boards

Add build support for the new Arcturus boards.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: need to include linux/device.h in dma.c
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: need to include linux/device.h in dma.c

Need to explicitly include linux/device.h.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: fix tick timer definition for coldfire
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: fix tick timer definition for coldfire

CLOCK_TICK_RATE should give the underlying frequency of the tick timer,
to make ntp happy.  For Coldfires, that's the main clock.

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: add configure support for more Arcturus boards
David Wu [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: add configure support for more Arcturus boards

Add configure support for the Arcturus UC5272 and UC5282 boards.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: fix workqueues in 68328 serial driver
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: fix workqueues in 68328 serial driver

Fix workqueues in 68328 serial driver.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agom68knommu: remove legacy power managament from 68328 serial driver
Greg Ungerer [Wed, 25 Jul 2007 12:07:20 +0000 (22:07 +1000)]
m68knommu: remove legacy power managament from 68328 serial driver

Remove the legacy power management code from the 68328 serial driver.
It is not used, and there is no current kernel support for power
management on the 68328.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm
Linus Torvalds [Wed, 25 Jul 2007 18:01:10 +0000 (11:01 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm:
  KVM: disable writeback for 0x0f 0x01 instructions.
  KVM: Fix removal of nx capability from guest cpuid
  Revert "KVM: Avoid useless memory write when possible"
  KVM: Fix unlikely kvm_create vs decache_vcpus_on_cpu race
  KVM: Correctly handle writes crossing a page boundary

17 years agoCache xtime every call to update_wall_time
john stultz [Wed, 25 Jul 2007 01:38:34 +0000 (18:38 -0700)]
Cache xtime every call to update_wall_time

This avoids xtime lag seen with dynticks, because while 'xtime' itself
is still not updated often, we keep a 'xtime_cache' variable around that
contains the approximate real-time that _is_ updated each time we do a
'update_wall_time()', and is thus never off by more than one tick.

IOW, this restores the original semantics for 'xtime' users, as long as
you use the proper abstraction functions (ie 'current_kernel_time()' or
'get_seconds()' depending on whether you want a timespec or just the
seconds field).

[ Updated Patch.  As penance for my sins I've also yanked another #ifdef
  that was added to avoid the xtime lag w/ hrtimers.  ]

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoCleanup non-arch xtime uses, use get_seconds() or current_kernel_time().
john stultz [Wed, 25 Jul 2007 00:47:43 +0000 (17:47 -0700)]
Cleanup non-arch xtime uses, use get_seconds() or current_kernel_time().

This avoids use of the kernel-internal "xtime" variable directly outside
of the actual time-related functions.  Instead, use the helper functions
that we already have available to us.

This doesn't actually change any behaviour, but this will allow us to
fix the fact that "xtime" isn't updated very often with CONFIG_NO_HZ
(because much of the realtime information is maintained as separate
offsets to 'xtime'), which has caused interfaces that use xtime directly
to get a time that is out of sync with the real-time clock by up to a
third of a second or so.

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSELinux: null-terminate context string in selinux_xfrm_sec_ctx_alloc
Venkat Yekkirala [Tue, 24 Jul 2007 14:53:23 +0000 (09:53 -0500)]
SELinux: null-terminate context string in selinux_xfrm_sec_ctx_alloc

xfrm_audit_log() expects the context string to be null-terminated
which currently doesn't happen with user-supplied contexts.

Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com>
Acked-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: James Morris <jmorris@namei.org>
17 years agoKVM: disable writeback for 0x0f 0x01 instructions.
Aurelien Jarno [Wed, 25 Jul 2007 08:19:54 +0000 (10:19 +0200)]
KVM: disable writeback for 0x0f 0x01 instructions.

0x0f 0x01 instructions (ie lgdt, lidt, smsw, lmsw and invlpg) does
not use writeback. This patch set no_wb=1 when emulating those
instructions.

This fixes a regression booting the FreeBSD kernel on AMD.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Avi Kivity <avi@qumranet.com>
17 years agoKVM: Fix removal of nx capability from guest cpuid
Avi Kivity [Wed, 25 Jul 2007 06:22:12 +0000 (09:22 +0300)]
KVM: Fix removal of nx capability from guest cpuid

Testing the wrong bit caused kvm not to disable nx on the guest when it is
disabled on the host (an mmu optimization relies on the nx bits being the
same in the guest and host).

This allows Windows to boot when nx is disabled on te host (e.g. when
host pae is disabled).

Signed-off-by: Avi Kivity <avi@qumranet.com>
17 years agoRevert "KVM: Avoid useless memory write when possible"
Avi Kivity [Mon, 23 Jul 2007 15:33:14 +0000 (18:33 +0300)]
Revert "KVM: Avoid useless memory write when possible"

This reverts commit a3c870bdce4d34332ebdba7eb9969592c4c6b243.  While it
does save useless updates, it (probably) defeats the fork detector, causing
a massive performance loss.

Signed-off-by: Avi Kivity <avi@qumranet.com>
17 years agoKVM: Fix unlikely kvm_create vs decache_vcpus_on_cpu race
Rusty Russell [Mon, 23 Jul 2007 07:08:21 +0000 (17:08 +1000)]
KVM: Fix unlikely kvm_create vs decache_vcpus_on_cpu race

We add the kvm to the vm_list before initializing the vcpu mutexes,
which can be mutex_trylock()'ed by decache_vcpus_on_cpu().

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Avi Kivity <avi@qumranet.com>
17 years agoKVM: Correctly handle writes crossing a page boundary
Avi Kivity [Sun, 22 Jul 2007 15:48:54 +0000 (18:48 +0300)]
KVM: Correctly handle writes crossing a page boundary

Writes that are contiguous in virtual memory may not be contiguous in
physical memory; so split writes that straddle a page boundary.

Thanks to Aurelien for reporting the bug, patient testing, and a fix
to this very patch.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Avi Kivity <avi@qumranet.com>
17 years agoPull auto-load-modules into release branch
Len Brown [Wed, 25 Jul 2007 05:36:53 +0000 (01:36 -0400)]
Pull auto-load-modules into release branch

17 years agoPull d-states into release branch
Len Brown [Wed, 25 Jul 2007 05:36:31 +0000 (01:36 -0400)]
Pull d-states into release branch

Conflicts:

drivers/acpi/sleep/main.c

Signed-off-by: Len Brown <len.brown@intel.com>
17 years agoPull kconfig into release branch
Len Brown [Wed, 25 Jul 2007 05:35:25 +0000 (01:35 -0400)]
Pull kconfig into release branch

17 years agoACPI: Kconfig: remove CONFIG_ACPI_SLEEP from source
Len Brown [Wed, 25 Jul 2007 02:26:33 +0000 (22:26 -0400)]
ACPI: Kconfig: remove CONFIG_ACPI_SLEEP from source

As it was a synonym for (CONFIG_ACPI && CONFIG_X86),
the ifdefs for it were more clutter than they were worth.

For ia64, just add a few stubs in anticipation of future
S3 or S4 support.

Signed-off-by: Len Brown <len.brown@intel.com>
17 years agoACPI: quiet ACPI Exceptions due to no _PTC or _TSS
Len Brown [Wed, 25 Jul 2007 04:57:46 +0000 (00:57 -0400)]
ACPI: quiet ACPI Exceptions due to no _PTC or _TSS

ACPI Exception (processor_throttling-0084): AE_NOT_FOUND, Evaluating _PTC [20070126]
ACPI Exception (processor_throttling-0147): AE_NOT_FOUND, Evaluating _TSS [20070126]

These methods are optional, so Linux should not
alarm users when they are not found.

http://bugzilla.kernel.org/show_bug.cgi?id=8802

Signed-off-by: Len Brown <len.brown@intel.com>
Acked-by: Luming Yu <luming.yu@intel.com>
17 years agoMerge branch 'for_paulus' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerpc
Linus Torvalds [Wed, 25 Jul 2007 03:26:25 +0000 (20:26 -0700)]
Merge branch 'for_paulus' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerpc

* 'for_paulus' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerpc: (25 commits)
  [POWERPC] 85xx: Added needed MPC85xx PCI device IDs
  [POWERPC] Add Freescale PCI VENDOR ID and 8641 device IDs
  [POWERPC] 85xxCDS: MPC8548 DTS cleanup.
  [POWERPC] 85xxCDS: Misc 8548 PCI Corrections.
  [POWERPC] 85xxCDS: Delay 8259 cascade hookup.
  [POWERPC] 85xxCDS: Make sure restart resets the PCI bus.
  [POWERPC] 85xxCDS: Allow 8259 cascade to share an MPIC interrupt line.
  [POWERPC] FSL: Add support for PCI-X controllers
  [POWERPC] Make sure virtual P2P bridge registers are setup on PCIe PHB
  [POWERPC] Provide ability to setup P2P bridge registers from struct resource
  [POWERPC] Add basic PCI/PCI Express support for 8544DS board
  [POWERPC] Make endianess of cfg_addr for indirect pci ops runtime
  [POWERPC] Removed setup_indirect_pci_nomap
  [POWERPC] 85xx: Add quirk to ignore bogus FPGA on CDS
  [POWERPC] 85xx: Added 8568 PCIe support
  [POWERPC] Fixup resources on pci_bus for PCIe PHB when no device is connected
  [POWERPC] Add basic PCI node for mpc8568mds board
  [POWERPC] Use Freescale pci/pcie common code for 85xx boards
  [POWERPC] Update PCI nodes in the 83xx/85xx boards device tree
  [POWERPC] Add 8548 CDS PCI express controller node and PCI-X device node
  ...

17 years agoocfs2: bad kunmap_atomic()
Jens Axboe [Tue, 24 Jul 2007 08:17:50 +0000 (10:17 +0200)]
ocfs2: bad kunmap_atomic()

kunmap_atomic() takes the virtual address, not the mapped page as
argument.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Cc: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 24 Jul 2007 23:00:26 +0000 (16:00 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] Fix marge error due to conflict in arch/mips/kernel/head.S
  [MIPS] ARC: Remove unused arch/mips/arc/console.c
  [MIPS] SNI: sniprom
  [MIPS] Jazz: remove unneeded reset functions
  [MIPS] Whitespace cleanup.
  [MIPS] Make resources for ds1742 "static __initdata"
  [MIPS] Replace __attribute_used__ with __used
  [MIPS] Jazz: Remove unused arch/mips/jazz/io.c
  [MIPS] Mark prom_free_prom_memory as __init_refok
  [MIPS] MIPSsim: Fix cflags

17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 24 Jul 2007 22:58:51 +0000 (15:58 -0700)]
Merge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  pata_hpt37x: Fix 2.6.22 clock PLL regression
  pata_ali: Correct HP detect
  ata_piix: fix suspend/resume for some TOSHIBA laptops
  PCI: export __pci_reenable_device()

17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Tue, 24 Jul 2007 22:58:15 +0000 (15:58 -0700)]
Merge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: (26 commits)
  netdev: i82596 Ethernet needs <asm/cacheflush.h>
  forcedeth: mcp73 device addition
  forcedeth: new device ids in pci_ids.h
  atl1: make atl1_init_ring_ptrs static
  eHEA: net_poll support
  drivers/net/acenic.c: fix check-after-use
  defxx: Use __maybe_unused rather than a local hack
  Fix error checking in Vitesse IRQ config
  ps3: reduce allocation size of rx skb buffers
  atl1: use kernel provided ethernet length constants
  atl1: fix typo in dma_req_block
  atl1: change cmb write threshold
  atl1: fix typo in DMA engine setup
  atl1: change tpd_avail function name
  ps3: fix rare issue that reenabling rx DMA fails
  ps3: removed calling netif_poll_enable() in open()
  ps3: use ethX as the name of irq
  ps3: use net_device_stats of net_device structure
  ps3: removed conditional ethtool support
  ps3: removed defines no longer used
  ...

17 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Tue, 24 Jul 2007 22:57:43 +0000 (15:57 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Mark most of initial bootup asm as .text.init.ref_ok
  [SPARC32]: Fix bug in sparc optimized memset.
  [SPARC64]: Update defconfig.
  [SPARC]: Add missing NOTES section.
  [SPARC32]: missing exports
  [SPARC32]: Deal with rtc/sun_mostek_rtc conflict.

17 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 24 Jul 2007 22:57:12 +0000 (15:57 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [8021Q]: vlan_ioctl_handler: fix return value
  [GENETLINK]: Correctly report errors while registering a multicast group
  [GENETLINK]: Fix adjustment of number of multicast groups
  [GENETLINK]: Fix race in genl_unregister_mc_groups()
  [NETFILTER]: Clean up duplicate includes in net/netfilter/
  [NETFILTER]: Clean up duplicate includes in net/bridge/
  [NETFILTER]: Fix logging regression
  [TCPv6] MD5SIG: Ensure to reset allocation count to avoid panic.

17 years ago[8021Q]: vlan_ioctl_handler: fix return value
Andrew Morton [Tue, 24 Jul 2007 22:37:11 +0000 (15:37 -0700)]
[8021Q]: vlan_ioctl_handler: fix return value

net/8021q/vlan.c: In function 'vlan_ioctl_handler':
net/8021q/vlan.c:700: warning: 'err' may be used uninitialized in this function

The warning is incorrect, but from my reading this ioctl will return -EINVAL
on success.

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[GENETLINK]: Correctly report errors while registering a multicast group
Thomas Graf [Tue, 24 Jul 2007 22:34:53 +0000 (15:34 -0700)]
[GENETLINK]: Correctly report errors while registering a multicast group

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[GENETLINK]: Fix adjustment of number of multicast groups
Thomas Graf [Tue, 24 Jul 2007 22:33:51 +0000 (15:33 -0700)]
[GENETLINK]: Fix adjustment of number of multicast groups

The current calculation of the maximum number of genetlink
multicast groups seems odd, fix it.

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[GENETLINK]: Fix race in genl_unregister_mc_groups()
Thomas Graf [Tue, 24 Jul 2007 22:32:46 +0000 (15:32 -0700)]
[GENETLINK]: Fix race in genl_unregister_mc_groups()

family->mcast_groups is protected by genl_lock so it must
be held while accessing the list in genl_unregister_mc_groups().
Requires adding a non-locking variant of genl_unregister_mc_group().

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: Clean up duplicate includes in net/netfilter/
Jesper Juhl [Tue, 24 Jul 2007 22:31:05 +0000 (15:31 -0700)]
[NETFILTER]: Clean up duplicate includes in net/netfilter/

This patch cleans up duplicate includes in
net/netfilter/

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: Clean up duplicate includes in net/bridge/
Jesper Juhl [Tue, 24 Jul 2007 22:30:30 +0000 (15:30 -0700)]
[NETFILTER]: Clean up duplicate includes in net/bridge/

This patch cleans up duplicate includes in
net/bridge/

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: Fix logging regression
Patrick McHardy [Tue, 24 Jul 2007 22:29:55 +0000 (15:29 -0700)]
[NETFILTER]: Fix logging regression

Loading one of the LOG target fails if a different target has already
registered itself as backend for the same family. This can affect the
ipt_LOG and ipt_ULOG modules when both are loaded.

Reported and tested by: <t.artem@mailcity.com>

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCPv6] MD5SIG: Ensure to reset allocation count to avoid panic.
YOSHIFUJI Hideaki [Tue, 24 Jul 2007 22:27:30 +0000 (15:27 -0700)]
[TCPv6] MD5SIG: Ensure to reset allocation count to avoid panic.

After clearing all passwords for IPv6 peers, we need to
set allocation count to zero as well as we free the storage.
Otherwise, we panic when a user trys to (re)add a password.

Discovered and fixed by MIYAJIMA Mitsuharu <miyajima.mitsuharu@anchor.jp>.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Mark most of initial bootup asm as .text.init.ref_ok
David S. Miller [Tue, 24 Jul 2007 22:17:33 +0000 (15:17 -0700)]
[SPARC64]: Mark most of initial bootup asm as .text.init.ref_ok

We can't mark the whole thing init because there are dependencies
in bootloaders that assume that _start, or whatever the image
entry value, is 2 instructions before the "HdrS" signature.

In fact, TILO assumes this entry is always at 0x4000, yikes!

Also, right after the bootloader info area there are OBP strings and
values that get used later in the boot process, and those are not all
provably .init yet.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agopata_hpt37x: Fix 2.6.22 clock PLL regression
Alan Cox [Tue, 24 Jul 2007 14:17:48 +0000 (15:17 +0100)]
pata_hpt37x: Fix 2.6.22 clock PLL regression

Just one version of Linux ago
The PLL code broke - oh no!
But set the right mode
And fix up the code
Makes the PLL timing sync go

[whatever happened to the sailor from Nantucket, hero of many limericks?  -jg]

Closes-bug: #8791
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agopata_ali: Correct HP detect
Alan Cox [Fri, 20 Jul 2007 14:34:49 +0000 (15:34 +0100)]
pata_ali: Correct HP detect

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoata_piix: fix suspend/resume for some TOSHIBA laptops
Tejun Heo [Tue, 10 Jul 2007 06:55:43 +0000 (15:55 +0900)]
ata_piix: fix suspend/resume for some TOSHIBA laptops

ACPI implementations in several TOSHIBA laptops are weird and burn cpu
cycles for tens of seconds while trying to suspend if the PCI device
for the ATA controller is disabled when the ACPI suspend is called.

This patch uses DMI to match those machines and bypass device disable
on those machines during suspend.  As the device needs to be put into
enabled state on resume without affecting PCI enable count, matching
resume callback uses __pci_reenable_device().

This bug is reported in bugzilla bug 7780.

  http://bugzilla.kernel.org/show_bug.cgi?id=7780

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoPCI: export __pci_reenable_device()
Tejun Heo [Tue, 10 Jul 2007 06:36:05 +0000 (15:36 +0900)]
PCI: export __pci_reenable_device()

Some odd ACPI implementations choke if certain controller is disabled
when ACPI suspend is invoked but we still need to make sure the PCI
device is enabled during resume.  Simply using pci_enable_device()
unbalances device enable count.  Export __pci_reenable_device().

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years ago[SPARC32]: Fix bug in sparc optimized memset.
Alexander Shmelev [Tue, 24 Jul 2007 20:41:44 +0000 (13:41 -0700)]
[SPARC32]: Fix bug in sparc optimized memset.

Sparc optimized memset (arch/sparc/lib/memset.S) does not fill last
byte of the memory area, if area size is less than 8 bytes and start
address is not word (4-bytes) aligned.

Here is code chunk where bug located:
/* %o0 - memory address, %o1 - size, %g3 - value */
8:
     add    %o0, 1, %o0
    subcc    %o1, 1, %o1
    bne,a    8b
     stb %g3, [%o0 - 1]

This code should write byte every loop iteration, but last time delay
instruction stb is not executed because branch instruction sets
"annul" bit.

Patch replaces bne,a by bne instruction.

Error can be reproduced by simple kernel module:

--------------------
#include <linux/module.h>
#include <linux/config.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <string.h>

static void do_memset(void **p, int size)
{
        memset(p, 0x00, size);
}

static int __init memset_test_init(void)
{
    char fooc[8];
    int *fooi;
    memset(fooc, 0xba, sizeof(fooc));

    do_memset((void**)(fooc + 3), 1);

    fooi = (int*) fooc;
    printk("%08X %08X\n", fooi[0], fooi[1]);

    return -1;
}

static void __exit memset_test_cleanup(void)
{
    return;
}

module_init(memset_test_init);
module_exit(memset_test_cleanup);

MODULE_LICENSE("GPL");
EXPORT_NO_SYMBOLS;
--------------------

Signed-off-by: Alexander Shmelev <ashmelev@task.sun.mcst.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agonetdev: i82596 Ethernet needs <asm/cacheflush.h>
Geert Uytterhoeven [Fri, 20 Jul 2007 16:40:45 +0000 (18:40 +0200)]
netdev: i82596 Ethernet needs <asm/cacheflush.h>

netdev: i82596 Ethernet needs <asm/cacheflush.h> on m68k

drivers/net/82596.c: In function 'init_rx_bufs':
drivers/net/82596.c:552: error: implicit declaration of function 'cache_clear'
drivers/net/82596.c: In function 'i596_start_xmit':
drivers/net/82596.c:1104: error: implicit declaration of function 'cache_push'

The driver still compiles on ia32 (CONFIG_APRICOT=y)

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoforcedeth: mcp73 device addition
Ayaz Abdulla [Mon, 23 Jul 2007 00:43:26 +0000 (20:43 -0400)]
forcedeth: mcp73 device addition

This patch contains new device settings for MCP73 chipset.

Signed-Off-By: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoforcedeth: new device ids in pci_ids.h
Ayaz Abdulla [Mon, 23 Jul 2007 00:43:15 +0000 (20:43 -0400)]
forcedeth: new device ids in pci_ids.h

This patch contains new device ids for MCP73 chipset.

Signed-Off-By: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: make atl1_init_ring_ptrs static
Chris Snook [Mon, 23 Jul 2007 20:38:39 +0000 (16:38 -0400)]
atl1: make atl1_init_ring_ptrs static

Sparse (correctly) thinks this function should be static.  Make it so.

Signed-off-by: Chris Snook <csnook@redhat.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoeHEA: net_poll support
Jan-Bernd Themann [Mon, 23 Jul 2007 14:05:03 +0000 (16:05 +0200)]
eHEA: net_poll support

net_poll support for eHEA added

Signed-off-by: Jan-Bernd Themann <themann@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agodrivers/net/acenic.c: fix check-after-use
Adrian Bunk [Mon, 23 Jul 2007 08:02:32 +0000 (10:02 +0200)]
drivers/net/acenic.c: fix check-after-use

The Coverity checker noted that we've already dereferenced "dev" when we
check whether it's NULL.

Since it's impossible that "dev" is NULL at this place this patch
removes the NULL check.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agodefxx: Use __maybe_unused rather than a local hack
Maciej W. Rozycki [Fri, 20 Jul 2007 12:14:07 +0000 (13:14 +0100)]
defxx: Use __maybe_unused rather than a local hack

 This is a change to remove a local hack in favour to __maybe_unused that
has been recently added.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoFix error checking in Vitesse IRQ config
Andy Fleming [Wed, 18 Jul 2007 06:06:28 +0000 (01:06 -0500)]
Fix error checking in Vitesse IRQ config

phy_read() returns a negative number if there's an error, but the
error-checking code in the Vitesse driver's config_intr function
triggers if phy_read() returns non-zero.  Correct that.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: reduce allocation size of rx skb buffers
Masakazu Mokuno [Fri, 20 Jul 2007 08:39:25 +0000 (17:39 +0900)]
ps3: reduce allocation size of rx skb buffers

Reduced allocation size for rx skb buffers, from 2308 bytes to
1356 per buffer.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: use kernel provided ethernet length constants
Jay Cliburn [Thu, 19 Jul 2007 23:45:14 +0000 (18:45 -0500)]
atl1: use kernel provided ethernet length constants

Use constants already provided by the kernel for ethernet related lengths.

Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: fix typo in dma_req_block
Jay Cliburn [Thu, 19 Jul 2007 23:45:13 +0000 (18:45 -0500)]
atl1: fix typo in dma_req_block

s/dam/dma

Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: change cmb write threshold
Jay Cliburn [Thu, 19 Jul 2007 23:45:12 +0000 (18:45 -0500)]
atl1: change cmb write threshold

Change the threshold number of descriptors used to trigger CMB writes.
The vendor reports that under certain conditions this will reduce the
number of unnecessary tx interrupts and improve rx performance.

This change is lifted directly from vendor version 1.2.40.2 of the L1
driver.

Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: fix typo in DMA engine setup
Jay Cliburn [Thu, 19 Jul 2007 23:45:11 +0000 (18:45 -0500)]
atl1: fix typo in DMA engine setup

The DMA engine setup contains a typo that can result in an incorrect
dmaw_block setting.

Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: change tpd_avail function name
Jay Cliburn [Thu, 19 Jul 2007 23:45:10 +0000 (18:45 -0500)]
atl1: change tpd_avail function name

Change tpd_avail() to atl1_tpd_avail().

Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: fix rare issue that reenabling rx DMA fails
Masakazu Mokuno [Fri, 20 Jul 2007 08:35:54 +0000 (17:35 +0900)]
ps3: fix rare issue that reenabling rx DMA fails

Fixed rare issue that 'lv1_net_start_rx_dma failed, status=-9" was shown
in dmesg.  This meant restarting rx DMA had been rejected by the hypervisor.
This issue would caused if the guest os requested starting DMA when
the hypervisor thought the DMA was in progress.
The state machine for DMA status of the hypervisor would be updated
by processing interrupt in the hypervisor.
Thus we should wait for the interrupt delivery before restarting
DMA.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: removed calling netif_poll_enable() in open()
Masakazu Mokuno [Fri, 20 Jul 2007 08:34:07 +0000 (17:34 +0900)]
ps3: removed calling netif_poll_enable() in open()

Removed use of netif_poll_enable() in open function.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: use ethX as the name of irq
Masakazu Mokuno [Fri, 20 Jul 2007 08:33:02 +0000 (17:33 +0900)]
ps3: use ethX as the name of irq

Use net_device name for registration of irq as many network drivers do.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: use net_device_stats of net_device structure
Masakazu Mokuno [Fri, 20 Jul 2007 08:31:07 +0000 (17:31 +0900)]
ps3: use net_device_stats of net_device structure

Removed the statistics information from private structre.
Instead, use net_device_stats in net_device structure.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: removed conditional ethtool support
Masakazu Mokuno [Fri, 20 Jul 2007 08:28:59 +0000 (17:28 +0900)]
ps3: removed conditional ethtool support

Removed conditional ethtool support.  Always enabled.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: removed defines no longer used
Masakazu Mokuno [Fri, 20 Jul 2007 08:27:12 +0000 (17:27 +0900)]
ps3: removed defines no longer used

Removed defines no longer used.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: tx descriptor handling cleanup
Masakazu Mokuno [Fri, 20 Jul 2007 08:24:56 +0000 (17:24 +0900)]
ps3: tx descriptor handling cleanup

gelic: TX descriptor handling cleanup

        - Emitted return value of NETDEV_TX_LOCKED when DMA map or kick
          failure.
          Now it would free the skb, update drop packet statistics
          and return OK. Requested from Jeff Garzik.
        - Enable tx queue if number of free descriptors are more than 2
        - Fixed descriptor leak if dma map for second descriptor failed
        - Stopped calling xmit handler from interrupt handler in order
          to recheck tx queue.  Instead, call appropriate helper functions.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: some minor cleanups
Masakazu Mokuno [Fri, 20 Jul 2007 08:22:51 +0000 (17:22 +0900)]
ps3: some minor cleanups

- Removed the embarrassing definition which was used in only one
  place.
- Fixed wrong initialization of dmac_cmd_status.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agops3: fix wrong calculation of rx descriptor address
Masakazu Mokuno [Fri, 20 Jul 2007 08:20:54 +0000 (17:20 +0900)]
ps3: fix wrong calculation of rx descriptor address

Fixed the bug that calculation of the address of rx descriptor was
wrong.

Signed-off-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agonetxen: Load firmware during probe, dma watchdog fix.
dhananjay@netxen.com [Thu, 19 Jul 2007 09:11:09 +0000 (14:41 +0530)]
netxen: Load firmware during probe, dma watchdog fix.

The firmware should be loaded after resetting hardware during PCI probe,
besides module unload. This fixes issue with 2nd port of multiport adapter
on powerpc blades. This patch also fixes a bug that PCI resources are not
freed if dma watchdog shutdown failed. The dma watchdog poll messages
during module unload are also suppressed.

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: Milan Bag <mbag@netxen.com>
Signed-off-by: Wen Xiong <wenxiong@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoucc_geth: add support to netif message level
Li Yang [Thu, 19 Jul 2007 03:48:29 +0000 (11:48 +0800)]
ucc_geth: add support to netif message level

Adds netif message level control to the ucc_geth driver.  The level can
be set by module parameter and ethtool.

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoucc_geth: add ethtool support
Li Yang [Thu, 19 Jul 2007 03:47:47 +0000 (11:47 +0800)]
ucc_geth: add ethtool support

The patch enables statistics in ucc_geth and adds ethtool support to
ucc_geth driver.

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years ago[SPARC64]: Update defconfig.
David S. Miller [Mon, 23 Jul 2007 02:25:03 +0000 (19:25 -0700)]
[SPARC64]: Update defconfig.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC]: Add missing NOTES section.
David S. Miller [Mon, 23 Jul 2007 02:19:05 +0000 (19:19 -0700)]
[SPARC]: Add missing NOTES section.

This fixes boot failures when the build-id LD option is
actually used, because without it we end up with multiple
PT_LOAD sections which the SILO boot loader cannot handle.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC32]: missing exports
Al Viro [Sun, 22 Jul 2007 07:38:24 +0000 (00:38 -0700)]
[SPARC32]: missing exports

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC32]: Deal with rtc/sun_mostek_rtc conflict.
Al Viro [Sun, 22 Jul 2007 07:37:48 +0000 (00:37 -0700)]
[SPARC32]: Deal with rtc/sun_mostek_rtc conflict.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoaacraid: fix security hole
Alan Cox [Mon, 23 Jul 2007 13:51:05 +0000 (14:51 +0100)]
aacraid: fix security hole

On the SCSI layer ioctl path there is no implicit permissions check for
ioctls (and indeed other drivers implement unprivileged ioctls). aacraid
however allows all sorts of very admin only things to be done so should
check.

Signed-off-by: Alan Cox <alan@redhat.com>
Acked-by: Mark Salyzyn <mark_salyzyn@adaptec.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Tue, 24 Jul 2007 19:32:12 +0000 (12:32 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/selinux-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/selinux-2.6:
  SELinux: fix memory leak in security_netlbl_cache_add()

17 years agoMerge branch 'request-queue-t' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Tue, 24 Jul 2007 19:26:44 +0000 (12:26 -0700)]
Merge branch 'request-queue-t' of git://git.kernel.dk/linux-2.6-block

* 'request-queue-t' of git://git.kernel.dk/linux-2.6-block:
  [BLOCK] Add request_queue_t and mark it deprecated
  [BLOCK] Get rid of request_queue_t typedef

17 years agoslab: correctly handle __GFP_ZERO
Andrew Morton [Tue, 24 Jul 2007 19:02:40 +0000 (12:02 -0700)]
slab: correctly handle __GFP_ZERO

Use the correct local variable when calling into the page allocator.  Local
`flags' can have __GFP_ZERO set, which causes us to pass __GFP_ZERO into the
page allocator, possibly from illegal contexts.  The page allocator will later
do prep_zero_page()->kmap_atomic(..., KM_USER0) from irq contexts and will
then go BUG.

Cc: Mike Galbraith <efault@gmx.de>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoloop.h build fix
Arnd Bergmann [Tue, 24 Jul 2007 01:44:00 +0000 (18:44 -0700)]
loop.h build fix

include/linux/loop.h:66: error: expected specifier-qualifier-list before 'request_queue_t'

Cc: Sebastian Siewior <sebastian@breakpoint.cc>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofix hugetlb page allocation leak
Ken Chen [Tue, 24 Jul 2007 01:44:00 +0000 (18:44 -0700)]
fix hugetlb page allocation leak

dequeue_huge_page() has a serious memory leak upon hugetlb page
allocation.  The for loop continues on allocating hugetlb pages out of
all allowable zone, where this function is supposedly only dequeue one
and only one pages.

Fixed it by breaking out of the for loop once a hugetlb page is found.

Signed-off-by: Ken Chen <kenchen@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agominor gpio doc update
David Brownell [Tue, 24 Jul 2007 01:43:57 +0000 (18:43 -0700)]
minor gpio doc update

Fix doc bug noted by Uwe Kleine-König:  gpio_set_direction() is long
gone, replaced by gpio_direction_input() and gpio_direction_output().

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: <ukleinek@informatik.uni-freiburg.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agolink lguest example launcher non-static
Rusty Russell [Tue, 24 Jul 2007 01:43:56 +0000 (18:43 -0700)]
link lguest example launcher non-static

S.Caglar Onur points out that many distributions don't ship a static
zlib.  Unfortunately the launcher currently maps virtual device memory
where shared libraries want to go.

The solution is to pre-scan the args to figure out how much memory we
have, then allocate devices above that, rather than down from the top
possible address.  This also turns out to be simpler.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofault_inject: silence a warning
Stephen Rothwell [Tue, 24 Jul 2007 01:43:55 +0000 (18:43 -0700)]
fault_inject: silence a warning

lib/fault-inject.c:168: warning: 'debugfs_create_ul_MAX_STACK_TRACE_DEPTH' defined but not used

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFix failure exits in asus-laptop
Al Viro [Tue, 24 Jul 2007 01:43:54 +0000 (18:43 -0700)]
Fix failure exits in asus-laptop

Fallout from f8a7c6fe14f556ca8eeddce258cb21392d0c3a2f.  However, looking
at it shows that checks done in ASUS_LED_UNREGISTER() can't trigger
at all (we never get to asus_led_exit() if registration fails) and
if that registration fails, we actually leak stuff.  IOW, it's worse
than just replacing class_dev with dev in there - the tests themselves
had been papering over the lousy cleanup logics.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Michal Piotrowski <michal.k.k.piotrowski@gmail.com>
Cc: Corentin Chary <corentincj@iksaif.net>
Cc: Karol Kozimor <sziwan@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoinclude/asm-xtensa/io.h must #include <asm/page.h>
Adrian Bunk [Tue, 24 Jul 2007 01:43:54 +0000 (18:43 -0700)]
include/asm-xtensa/io.h must #include <asm/page.h>

  CC      kernel/time/clocksource.o
In file included from
/home/bunk/linux/kernel-2.6/linux-2.6.22-rc6-mm1/include/linux/clocksource.h:18,
                 from
/home/bunk/linux/kernel-2.6/linux-2.6.22-rc6-mm1/kernel/time/clocksource.c:27:
include2/asm/io.h: In function 'virt_to_phys':
include2/asm/io.h:46: error: implicit declaration of function '__pa'
include2/asm/io.h: In function 'phys_to_virt':
include2/asm/io.h:51: error: implicit declaration of function '__va'
include2/asm/io.h:51: warning: return makes pointer from integer without a cast
make[3]: *** [kernel/time/clocksource.o] Error 1

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Christian Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agospusched: fix mismerge in spufs.h
Christoph Hellwig [Tue, 24 Jul 2007 01:43:53 +0000 (18:43 -0700)]
spusched: fix mismerge in spufs.h

spufs.h now has two enums for the sched_flags leading to identical
values for SPU_SCHED_WAS_ACTIVE and SPU_SCHED_NOTIFY_ACTIVE.  Merge
them into a single enum as they were in the IBM development tree.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agonfsd: fix possible oops on re-insertion of rpcsec_gss modules
J. Bruce Fields [Tue, 24 Jul 2007 01:43:52 +0000 (18:43 -0700)]
nfsd: fix possible oops on re-insertion of rpcsec_gss modules

The handling of the re-registration case is wrong here; the "test" that was
returned from auth_domain_lookup will not be used again, so that reference
should be put.  And auth_domain_lookup never did anything with "new" in
this case, so we should just clean it up ourself.

Thanks to Akinobu Mita for bug report, analysis, and testing.

Cc: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu>
Cc: Neil Brown <neilb@suse.de>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoFix Au1100 fb dependencies, add helptext
Ralf Baechle [Tue, 24 Jul 2007 01:43:50 +0000 (18:43 -0700)]
Fix Au1100 fb dependencies, add helptext

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agouml: more __init annotations
Jeff Dike [Tue, 24 Jul 2007 01:43:48 +0000 (18:43 -0700)]
uml: more __init annotations

2.6.23-rc1 turned up another batch of references from non-__init code to
__init code.  In most cases, these were missing __init annotations.  In one
case (os_drop_memory), the annotation was present but wrong.

init_maps is __init, but for some reason was being very careful about the
mechanism by which it allocated memory, checking whether it was OK to use
kmalloc (at this point in the boot, it definitely isn't) and using either
alloc_bootmem_low_pages or kmalloc/vmalloc.  So, the kmalloc/vmalloc code is
removed.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agouml: fix aio compilation bug
Jeff Dike [Tue, 24 Jul 2007 01:43:47 +0000 (18:43 -0700)]
uml: fix aio compilation bug

Restructure do_aio thanks to commments from Ulrich and Al.

Uli started this by seeing that UML's initialization of a struct iocb
initialized fields that it shouldn't.

Al followed up by adding the following cleanups:
eliminating a variable by just using an anonymous structure in
its place.
hoisting a duplicated line out of the switch.
simplifying the error checking at the end.

I added a severity to the printk.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Ulrich Drepper <drepper@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agouml: fix string exporting on UML/i386
Jeff Dike [Tue, 24 Jul 2007 01:43:46 +0000 (18:43 -0700)]
uml: fix string exporting on UML/i386

In 2.6.23-rc1, i386 fiddled its string support such that UML started getting
undefined references from modules.  The UML asm/string.h was including the
i386 string.h, which defined __HAVE_ARCH_STR*, but the corresponding
implementations weren't being pulled in.

This is fixed by adding arch/i386/lib/string.h to the list of host
architecture files to be pulled in to UML.

A complication is that the libc exports file assumed that the generic strlen
and strstr weren't in use (i.e.  __HAVE_ARCH_STR is defined), then they aren't
exported.  This is untrue for strlen, which is exported in either case, so
this logic is not needed.

Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agofallocate syscall interface deficiency
Ulrich Drepper [Tue, 24 Jul 2007 01:43:46 +0000 (18:43 -0700)]
fallocate syscall interface deficiency

The fallocate syscall returns ENOSYS in case the filesystem does not support
the operation and expects the userlevel code to fill in.  This is good in
concept.

The problem is that the libc code for old kernels should be able to
distinguish the case where the syscall is not at all available vs not
functioning for a specific mount point.  As is this is not possible and we
always have to invoke the syscall even if the kernel doesn't support it.

I suggest the following patch.  Using EOPNOTSUPP is IMO the right thing to do.

Cc: Amit Arora <aarora@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoUse resource_size_t for serial port IO addresses
Josh Boyer [Tue, 24 Jul 2007 01:43:44 +0000 (18:43 -0700)]
Use resource_size_t for serial port IO addresses

At present, various parts of the serial code use unsigned long to define
resource addresses.  This is a problem, because some 32-bit platforms have
physical addresses larger than 32-bits, and have mmio serial uarts located
above the 4GB point.

This patch changes the type of mapbase in both struct uart_port and struct
plat_serial8250_port to resource_size_t, which can be configured to be 64
bits on such platforms.  The mapbase in serial_struct can't safely be
changed, because that structure is user visible.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>