GitHub/mt8127/android_kernel_alcatel_ttab.git
16 years agoMerge branch 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Wed, 20 Feb 2008 01:48:51 +0000 (17:48 -0800)]
Merge branch 'drm-patches' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/sis: add pciid for SiS 662/671 chipset
  drm: add new rv380 pciid
  drm: add support for passing state into the suspend hooks.
  drm/i915: Fix hibernate save/restore of VGA attribute regs
  drm/i915 more registers for S3 (DSPCLK_GATE_D, CACHE_MODE_0, MI_ARB_STATE)
  drm/i915: restore pipeconf regs unconditionally
  drm/i915: save/restore interrupt state
  drm: convert drm from nopage to fault.
  i915: wrap chipset types requiring hw status set ioctl
  drm/radeon: add initial rs690 support to drm.

16 years agodrm/sis: add pciid for SiS 662/671 chipset
Chaoyu Chen [Wed, 20 Feb 2008 00:12:39 +0000 (10:12 +1000)]
drm/sis: add pciid for SiS 662/671 chipset

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agodrm: add new rv380 pciid
Mirko [Wed, 20 Feb 2008 00:07:57 +0000 (10:07 +1000)]
drm: add new rv380 pciid

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agodrm: add support for passing state into the suspend hooks.
Dave Airlie [Wed, 20 Feb 2008 00:02:20 +0000 (10:02 +1000)]
drm: add support for passing state into the suspend hooks.

fix i915 driver to use state for hibernate save avoidance.

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86
Linus Torvalds [Tue, 19 Feb 2008 23:56:05 +0000 (15:56 -0800)]
Merge git://git./linux/kernel/git/x86/linux-2.6-x86

* git://git.kernel.org/pub/scm/linux/kernel/git/x86/linux-2.6-x86: (32 commits)
  x86: fix page_is_ram() thinko
  x86: fix WARN_ON() message: teach page_is_ram() about the special 4Kb bios data page
  x86: i8259A: remove redundant irq_descinitialization
  x86: fix vdso_install breaks user "make install"
  x86: change IO delay back to 0x80
  x86: lds - Use THREAD_SIZE instead of numeric constant
  x86: lds - Use PAGE_SIZE instead of numeric constant
  x86 cleanup: suspend_asm_64.S - use X86_CR4_PGE instead of numeric value
  x86: docs fixes to Documentation/i386/IO-APIC.txt
  x86: fix printout ugliness in cpu info printk
  x86: clean up csum-wrappers_64.c some more
  x86: coding style fixes in arch/x86/lib/csum-wrappers_64.c
  x86: coding style fixes in arch/x86/lib/io_64.c
  x86: exclude vsyscall files from stackprotect
  x86: add pgd_large() on 64-bit, for consistency
  x86: minor cleanup of comments in processor.h
  x86: annotate pci/common.s:pci_scan_bus_with_sysdata with __devinit
  x86: fix section mismatch in head_64.S:initial_code
  x86: fix section mismatch in srat_64.c:reserve_hotadd
  x86: fix section mismatch warning in topology.c:arch_register_cpu
  ...

16 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 19 Feb 2008 23:54:22 +0000 (15:54 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] BCM47XX: Use new SSB SPROM data structure
  [MIPS] WGT634U: Register MTD as platform device.
  [MIPS] BCM47xx: Add defconfig file.
  [MIPS] RM: fix EISA=n compilation
  [MIPS] PCI: Coding style fixes for pcibios_enable_resources.
  [MIPS] PCI: Port i386 PCI fixes.
  [MIPS] Qemu: finish platform removal
  [MIPS] Wire up the timerfd_*() o32 system calls
  [MIPS] IP28: Add defconfig file
  [MIPS] SB1: Fix CONFIG_SIBYTE_DMA_PAGEOPS build failure.
  [MIPS] BCM1480: Remove stray function call resulting in infinite recursion
  [MIPS] Fix buggy invocations of kmap_coherent()
  [MIPS] Fix broken rm7000/rm9000 interrupt handling
  [MIPS] Handle I-cache coherency in flush_cache_range()
  [MIPS] IP27: Add missing ~ in DMA code.
  [MIPS] Use find_task_by_vpid in system calls

16 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/czankel/xtens...
Linus Torvalds [Tue, 19 Feb 2008 23:53:02 +0000 (15:53 -0800)]
Merge branch 'release' of git://git./linux/kernel/git/czankel/xtensa-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/czankel/xtensa-2.6: (29 commits)
  [XTENSA] Allow debugger to modify the WINDOWBASE register.
  [XTENSA] Fix makefile to work with binutils-2.18.
  [XTENSA] Fix register corruption for certain processor configurations
  [XTENSA] Fix cache flush macro for D$/I$ aliasing/non-aliasing
  [XTENSA] Exclude thread-global registers from the xtregs structures.
  [XTENSA] Add support for the sa_restorer function
  [XTENSA] Add support for configurable registers and coprocessors
  [XTENSA] Clean up stat structs.
  [XTENSA] Use preprocessor to generate the linker script for the ELF boot image
  [XTENSA] Add missing RELOCATE_ENTRY for debug vector
  [XTENSA] Add volatile keyword to asm statements accessing counter registers
  [XTENSA] Remove unused code
  [XTENSA] Fix modules for non-exec processor configurations
  [XTENSA] Add missing cast in elf.h ELF_CORE_COPY_REGS()
  [XTENSA] Fix comments regarding the number of frames to save
  [XTENSA] Add missing a2 register restore in register spill routine
  [XTENSA] adjust boot linker script start addresses
  [XTENSA] Remove oldmask from sigcontext and fix register flush
  [XTENSA] Clean up elf-gregset.
  [XTENSA] Fix icache flush for cache aliasing
  ...

16 years agodm-raid1.c: fix NULL dereferences
Adrian Bunk [Tue, 19 Feb 2008 19:44:19 +0000 (19:44 +0000)]
dm-raid1.c: fix NULL dereferences

This patch fixes two NULL dereferences introduced by commit
06386bbfd2441416875d0403d405c56822f6ebac and spotted by the Coverity
checker.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMN10300: Update asb2303_defconfig
David Howells [Tue, 19 Feb 2008 18:59:04 +0000 (18:59 +0000)]
MN10300: Update asb2303_defconfig

Update the ASB2303 default configuration.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMN10300: Make the kernel jump into gdbstub on a BUG
David Howells [Tue, 19 Feb 2008 18:58:59 +0000 (18:58 +0000)]
MN10300: Make the kernel jump into gdbstub on a BUG

Make the kernel jump into gdbstub (if configured) on a BUG with the register
set from the BUG rather than interpolating another illegal instruction and
leaving gdbstub's idea of the process counter in unsupported_syscall() where
the original BUG was detected.

With this patch, gdbstub reports a SIGABRT to the compiler and reports the
program counter at the original BUG, allowing the execution state at the time
of the BUG to be examined with GDB.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMN10300: Introduce barriers to replace removed volatiles in gdbstub
David Howells [Tue, 19 Feb 2008 18:58:54 +0000 (18:58 +0000)]
MN10300: Introduce barriers to replace removed volatiles in gdbstub

Introduce into the MN10300 gdbstub a couple of barrier() calls to replace the
removed volatility of the input/output index variables for the Rx ring buffer.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMN10300: Call update_process_times() outside of the xtime_lock
David Howells [Tue, 19 Feb 2008 18:58:49 +0000 (18:58 +0000)]
MN10300: Call update_process_times() outside of the xtime_lock

Call update_process_times() outside of the xtime_lock.  Somewhere somewhere
inside one of the functions called by that, xtime_lock is readlocked, which
ends up in a deadlock situation.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Tue, 19 Feb 2008 23:50:11 +0000 (15:50 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/mthca: Free correct MPT on error exit from mthca_fmr_alloc()
  IPoIB/cm: Fix ipoib_cm_dev_stop() cleanup when drain times out

16 years agodrm/i915: Fix hibernate save/restore of VGA attribute regs
Jesse Barnes [Tue, 29 Jan 2008 05:05:22 +0000 (21:05 -0800)]
drm/i915: Fix hibernate save/restore of VGA attribute regs

In hibernate, we may end up calling the VGA save regs function twice, so we need to make sure it's idempotent.  That means leaving ARX in index mode after the first save operation.  Fixes hibernate on 965.

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agodrm/i915 more registers for S3 (DSPCLK_GATE_D, CACHE_MODE_0, MI_ARB_STATE)
Keith Packard [Sun, 17 Feb 2008 03:19:29 +0000 (19:19 -0800)]
drm/i915 more registers for S3 (DSPCLK_GATE_D, CACHE_MODE_0, MI_ARB_STATE)

Failing to preserve the MI_ARB_STATE register was causing FIFO underruns on
the VGA output on my HP 2510p after resume.

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agodrm/i915: restore pipeconf regs unconditionally
Jesse Barnes [Fri, 8 Feb 2008 01:33:28 +0000 (17:33 -0800)]
drm/i915: restore pipeconf regs unconditionally

On many chipsets, the checks for DPLL enable or VGA mode will prevent the pipeconf regs from being restored, which could result in a blank display or X failing to come back after resume.  So restore them unconditionally along with actually restoring pipe B's palette correctly.

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agodrm/i915: save/restore interrupt state
Jesse Barnes [Tue, 19 Feb 2008 23:39:58 +0000 (09:39 +1000)]
drm/i915: save/restore interrupt state

On resume, if the interrupt state isn't restored correctly, we may end
up with a flood of unexpected or ill-timed interrupts, which could cause
the kernel to disable the interrupt or vblank events to happen at the
wrong time.  So save/restore them properly.

Signed-off-by: Dave Airlie <airlied@linux.ie>
16 years agodrm: convert drm from nopage to fault.
Nick Piggin [Thu, 7 Feb 2008 06:20:50 +0000 (16:20 +1000)]
drm: convert drm from nopage to fault.

Remove redundant vma range checks.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dave Airlie <airlied@linux.ie>
16 years agoi915: wrap chipset types requiring hw status set ioctl
Zhenyu Wang [Tue, 19 Feb 2008 10:59:09 +0000 (20:59 +1000)]
i915: wrap chipset types requiring hw status set ioctl

Also applys to recent added new chipset.

Signed-off-by: Zhenyu Wang <zhenyu.z.wang@intel.com>
Signed-off-by: Dave Airlie <airlied@linux.ie>
16 years agodrm/radeon: add initial rs690 support to drm.
Maciej Cencora [Tue, 19 Feb 2008 11:32:45 +0000 (21:32 +1000)]
drm/radeon: add initial rs690 support to drm.

This adds support for configuring the RS690 GART.

Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agoAdd missing init section definitions
Sam Ravnborg [Tue, 19 Feb 2008 20:00:18 +0000 (21:00 +0100)]
Add missing init section definitions

When adding __devinitconst etc. the __initconst variant
were missed.
Add this one and proper definitions for .head.text for use
in .S files.
The naming .head.text is preferred over .text.head as the
latter will conflict for a function named head when introducing
-ffunctions-sections.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agokbuild: fix reversed symbol name order in modpost
Geert Uytterhoeven [Sun, 17 Feb 2008 13:12:10 +0000 (14:12 +0100)]
kbuild: fix reversed symbol name order in modpost

XXXINIT_TO_INIT and XXXEXIT_TO_EXIT warnings use the reversed symbol name order
in the suggestion, e.g.:

    WARNING: vmlinux.o(.meminit.text+0x36c): Section mismatch in reference from the function free_area_init_core() to the function .init.text:setup_usemap()
    The function __meminit free_area_init_core() references
    a function __init setup_usemap().
    If free_area_init_core is only used by setup_usemap then
    annotate free_area_init_core with a matching annotation.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agoIB/mthca: Free correct MPT on error exit from mthca_fmr_alloc()
Roland Dreier [Tue, 19 Feb 2008 18:42:50 +0000 (10:42 -0800)]
IB/mthca: Free correct MPT on error exit from mthca_fmr_alloc()

When mthca_fmr_alloc() returns an error, it should free the MPT at the
index key, not mr->ibmr.lkey, since the lkey has been mangled by
hw_index_to_key() and no longer is the real index.  This bug causes
corruption of the MPT table free bitmap when mthca_fmr_alloc() fails.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
16 years agoIPoIB/cm: Fix ipoib_cm_dev_stop() cleanup when drain times out
Pradeep Satyanarayana [Tue, 12 Feb 2008 23:00:59 +0000 (15:00 -0800)]
IPoIB/cm: Fix ipoib_cm_dev_stop() cleanup when drain times out

Commit efcd9971 ("IPoIB/cm: Factor out ipoib_cm_free_rx_reap_list()")
introduced a bug in ipoib_cm_dev_stop() when the receive drain times
out.  In that case, the function moves all the pending rx stuff into a
private list but then calls ipoib_cm_free_rx_reap_list(), which
handles a different list.

Fix this by moving everything to the rx_reap_list that will actually
get freed up.

This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=906>.

Signed-off-by: Pradeep Satyanarayana <pradeeps@linux.vnet.ibm.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
16 years agoRevert "SLUB: Alternate fast paths using cmpxchg_local"
Linus Torvalds [Tue, 19 Feb 2008 17:08:49 +0000 (09:08 -0800)]
Revert "SLUB: Alternate fast paths using cmpxchg_local"

This reverts commit 1f84260c8ce3b1ce26d4c1d6dedc2f33a3a29c0c, which is
suspected to be the reason for some very occasional and hard-to-trigger
crashes that usually look related to memory allocation (mostly reported
in networking, but since that's generally the most common source of
shortlived allocations - and allocations in interrupt contexts - that in
itself is not a big clue).

See for example
http://bugzilla.kernel.org/show_bug.cgi?id=9973
http://lkml.org/lkml/2008/2/19/278
etc.

One promising suspicion for what the root cause of bug is (which also
explains why it's so hard to trigger in practice) came from Eric
Dumazet:

   "I wonder how SLUB_FASTPATH is supposed to work, since it is affected
    by a classical ABA problem of lockless algo.

    cmpxchg_local(&c->freelist, object, object[c->offset]) can succeed,
    while an interrupt came (on this cpu), and several allocations were
    done, and one free was performed at the end of this interruption, so
    'object' was recycled.

    c->freelist can then contain the previous value (object), but
    object[c->offset] was changed by IRQ.

    We then put back in freelist an already allocated object."

but another reason for the revert is simply that everybody agrees that
this code was the main suspect just by virtue of the pattern of oopses.

Cc: Torsten Kaiser <just.for.lkml@googlemail.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years ago[MIPS] BCM47XX: Use new SSB SPROM data structure
Aurelien Jarno [Mon, 18 Feb 2008 10:04:31 +0000 (11:04 +0100)]
[MIPS] BCM47XX: Use new SSB SPROM data structure

Switch the BCM47XX code to the new SPROM data structure now that the old
one has been removed.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] WGT634U: Register MTD as platform device.
Aurelien Jarno [Thu, 7 Feb 2008 02:17:16 +0000 (03:17 +0100)]
[MIPS] WGT634U: Register MTD as platform device.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] BCM47xx: Add defconfig file.
Aurelien Jarno [Mon, 18 Feb 2008 11:23:27 +0000 (12:23 +0100)]
[MIPS] BCM47xx: Add defconfig file.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] RM: fix EISA=n compilation
Adrian Bunk [Sun, 17 Feb 2008 21:59:48 +0000 (23:59 +0200)]
[MIPS] RM: fix EISA=n compilation

This patch fixes the following build error with CONFIG_EISA=n caused by
commit 231a35d37293ab88d325a9cb94e5474c156282c0:

<--  snip -->

...
  LD      .tmp_vmlinux1
arch/mips/sni/built-in.o: In function `snirm_a20r_setup_devinit':
a20r.c:(.init.text+0x42c): undefined reference to `sni_eisa_root_init'
a20r.c:(.init.text+0x42c): relocation truncated to fit: R_MIPS_26 against `sni_eisa_root_init'
arch/mips/sni/built-in.o: In function `snirm_setup_devinit':
rm200.c:(.init.text+0x52c): undefined reference to `sni_eisa_root_init'
rm200.c:(.init.text+0x52c): relocation truncated to fit: R_MIPS_26 against `sni_eisa_root_init'
make[1]: *** [.tmp_vmlinux1] Error 1

<--  snip  -->

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] PCI: Coding style fixes for pcibios_enable_resources.
Ralf Baechle [Tue, 19 Feb 2008 16:01:20 +0000 (16:01 +0000)]
[MIPS] PCI: Coding style fixes for pcibios_enable_resources.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] PCI: Port i386 PCI fixes.
Ralf Baechle [Tue, 19 Feb 2008 15:59:33 +0000 (15:59 +0000)]
[MIPS] PCI: Port i386 PCI fixes.

The MIPS version of pcibios_enalbe_resources did not have the fixes
from ed6d14f9760857c745206c978b80352fc09cfd19 yet which under circumstances
similar to x86 might result in failures.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] Qemu: finish platform removal
Adrian Bunk [Sun, 17 Feb 2008 20:09:51 +0000 (22:09 +0200)]
[MIPS] Qemu: finish platform removal

Two files were omitted from the recent removal of the qemu platform.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] Wire up the timerfd_*() o32 system calls
Dmitri Vorobiev [Mon, 18 Feb 2008 21:02:37 +0000 (00:02 +0300)]
[MIPS] Wire up the timerfd_*() o32 system calls

This patch enables the system calls timerfd_create(), timerfd_settime()
and timerfd_gettime() for MIPS architecture.

Please see the following Bugzilla entry for more details:

http://bugzilla.kernel.org/show_bug.cgi?id=10038

This was tested using a Malta 4Kc board in both little-endian and
big-endian modes. The unit test program is available from the URL
above.

Signed-off-by: Dmitri Vorobiev <dmitri.vorobiev@gmail.com>
[Ralf: Added N64, N32 and O32 bits on 64-bit kernels.]
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] IP28: Add defconfig file
Thomas Bogendoerfer [Mon, 18 Feb 2008 21:38:32 +0000 (22:38 +0100)]
[MIPS] IP28: Add defconfig file

Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] SB1: Fix CONFIG_SIBYTE_DMA_PAGEOPS build failure.
Thiemo Seufer [Thu, 14 Feb 2008 16:58:25 +0000 (16:58 +0000)]
[MIPS] SB1: Fix CONFIG_SIBYTE_DMA_PAGEOPS build failure.

Fix type mismatch warnings for 64-bit kernel builds which trigger -Werror.
The problem affects only SB-1 kernels with CONFIG_SIBYTE_DMA_PAGEOPS
enabled.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] BCM1480: Remove stray function call resulting in infinite recursion
Ralf Baechle [Mon, 18 Feb 2008 18:49:48 +0000 (18:49 +0000)]
[MIPS] BCM1480: Remove stray function call resulting in infinite recursion

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] Fix buggy invocations of kmap_coherent()
Ralf Baechle [Sat, 16 Feb 2008 22:34:25 +0000 (22:34 +0000)]
[MIPS] Fix buggy invocations of kmap_coherent()

kmap_coherent will only work correctly if the page it is called on is
not marked dirty.  If it's dirty the kernel address of the page should
be used instead of a temporary mapping.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] Fix broken rm7000/rm9000 interrupt handling
Thomas Koeller [Mon, 11 Feb 2008 22:42:12 +0000 (23:42 +0100)]
[MIPS] Fix broken rm7000/rm9000 interrupt handling

Properly acknowledge RM7K and RM9K interrupts. Before this, interrupts were
permanently masked after their first occurrence, making them non-functional.

Signed-off-by: Thomas Koeller <thomas.koeller@baslerweb.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] Handle I-cache coherency in flush_cache_range()
Ralf Baechle [Mon, 11 Feb 2008 14:51:40 +0000 (14:51 +0000)]
[MIPS] Handle I-cache coherency in flush_cache_range()

So far flush_cache_range() did't consider the I-cache largely because it
did rarely ever matter to real world code.  This was working primarily
because normally code and data are don't share the same pages - with the
exception of MIPS16 code which uses address constants embedded between
the code.   The following sequence of events may break the code:

 o MIPS16 executable being loaded
 o dynamic linker relocates the address constants embedded into the code:
 o   Uses mprotect(2) to make code pages PROT_READ|PROT_WRITE
 o   Performs the actual relocations by writing to the pages which likely
     are COW.  Because no PROT_EXEC is set I-cache coherence will not be
     considered.
 o   Uses mprotect(2) to switch code pages back to PROT_READ|PROT_EXEC.
     This results in a call to flush_cache_range() which also does not
     consider I-caches.
 o => executing the page just having been relocated may now result in the
   I-cache getting refilled with stale data from memory.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] IP27: Add missing ~ in DMA code.
Ralf Baechle [Fri, 8 Feb 2008 13:13:27 +0000 (13:13 +0000)]
[MIPS] IP27: Add missing ~ in DMA code.

Harmless since this function is not being called on I/O coherent systems
such as IP27.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years ago[MIPS] Use find_task_by_vpid in system calls
Pavel Emelyanov [Tue, 5 Feb 2008 07:44:24 +0000 (23:44 -0800)]
[MIPS] Use find_task_by_vpid in system calls

There are some places left in mips, that lookup task in initial namespace,
while the code doing so gets the pid from the user space and thus must
treat it as virtual.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
16 years agoMerge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Tue, 19 Feb 2008 16:04:23 +0000 (08:04 -0800)]
Merge branch 'for-linus' of git://git390.osdl.marist.edu/linux-2.6

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] qdio: FCP/SCSI write I/O stagnates on LPAR
  [S390] Fix futex_atomic_cmpxchg_std inline assembly.
  [S390] dcss: Fix Unlikely(x) != y
  [S390] sclp: clean up send/receive naming scheme
  [S390] etr: fix compile error on !SMP
  [S390] qdio: fix qdio_activate timeout handling.
  [S390] Initialize per cpu lowcores on cpu hotplug.
  [S390] find bit corner case.
  [S390] dasd: fix locking in __dasd_device_process_final_queue
  [S390] Make sure enabled wait psw is loaded in default_idle.
  [S390] Let NR_CPUS default to 32/64 on s390/s390x.
  [S390] cio: Do timed recovery on workqueue.
  [S390] cio: Remember to initialize recovery_lock.

16 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Tue, 19 Feb 2008 16:03:02 +0000 (08:03 -0800)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  libata: implement drain buffers
  libata: eliminate the home grown dma padding in favour of
  block: clear drain buffer if draining for write command
  block: implement request_queue->dma_drain_needed
  block: add request->raw_data_len
  block: update bio according to DMA alignment padding
  libata: update ATAPI overflow draining
  elevator: make elevator_get() attempt to load the appropriate module
  cfq-iosched: add hlist for browsing parallel to the radix tree
  block: make blk_rq_map_user() clear ->bio if it unmaps it
  fs/block_dev.c: remove #if 0'ed code
  make struct def_blk_aops static
  make blk_settings_init() static
  make blk_ioc_init() static
  make blk-core.c:request_cachep static again

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Tue, 19 Feb 2008 16:02:04 +0000 (08:02 -0800)]
Merge git://git./linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  [WATCHDOG] HP ProLiant WatchDog driver
  [WATCHDOG] blackfin Watchdog driver: relocate all strings used in __init functions to __initdata
  [WATCHDOG] Convert mtx1 wdt to be a platform device and use generic GPIO API
  [WATCHDOG] Add support for SB1 hardware watchdog

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-genirq
Linus Torvalds [Tue, 19 Feb 2008 15:54:02 +0000 (07:54 -0800)]
Merge git://git./linux/kernel/git/tglx/linux-2.6-genirq

* git://git.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-genirq:
  genirq: do not leave interupts enabled on free_irq
  genirq: spurious.c: use time_* macros

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Tue, 19 Feb 2008 15:53:28 +0000 (07:53 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Add regs_return_value().
  [SPARC64]: Kill pcic_present().
  [SPARC]: Kill 'prom_palette'.
  [ATYFB]: Kill 'prom_palette' sparc code.
  [SPARC64]: Kill 'prom_keyboard'.
  [SPARC]: Kill extern decl of 'panic_setup'.
  [SPARC64]: Delete 'boot_flags'.
  [SPARC64]: Kill unused function 'kernel_enter_debugger'.
  [SPARC64] arch/sparc64/kernel/unaligned.c: Use time_* macros
  [SPARC64]: Always register a PROM based early console.
  [SPARC64]: Update defconfig.
  [SPARC64]: Add -mtune=ultrasparc3 if possible.
  [SPARC64]: Remove Makefile code for ancient gcc and binutils.
  [SPARC64]: Remove DEBUG_BOOTMEM.
  [SPARC64]: Use shorter "get_zeroed_page" call.
  [SPARC]: Use shorter form of "get_zeroed_page".
  [SPARC]: video/cg14.c and video/sbuslib.c build fixes

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 19 Feb 2008 15:52:45 +0000 (07:52 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (60 commits)
  [NIU]: Bump driver version and release date.
  [NIU]: Fix BMAC alternate MAC address indexing.
  net: fix kernel-doc warnings in header files
  [IPV6]: Use BUG_ON instead of if + BUG in fib6_del_route.
  [IPV6]: dst_entry leak in ip4ip6_err. (resend)
  bluetooth: do not move child device other than rfcomm
  bluetooth: put hci dev after del conn
  [NET]: Elminate spurious print_mac() calls.
  [BLUETOOTH] hci_sysfs.c: Kill build warning.
  [NET]: Remove MAC_FMT
  net/8021q/vlan_dev.c: Use print_mac.
  [XFRM]: Fix ordering issue in xfrm_dst_hash_transfer().
  [BLUETOOTH] net/bluetooth/hci_core.c: Use time_* macros
  [IPV6]: Fix hardcoded removing of old module code
  [NETLABEL]: Move some initialization code into __init section.
  [NETLABEL]: Shrink the genl-ops registration code.
  [AX25] ax25_out: check skb for NULL in ax25_kick()
  [TCP]: Fix tcp_v4_send_synack() comment
  [IPV4]: fix alignment of IP-Config output
  Documentation: fix tcp.txt
  ...

16 years agoSmack: unlabeled outgoing ambient packets
Casey Schaufler [Fri, 15 Feb 2008 23:24:25 +0000 (15:24 -0800)]
Smack: unlabeled outgoing ambient packets

Smack uses CIPSO labeling, but allows for unlabeled packets by
specifying an "ambient" label that is applied to incoming unlabeled
packets.

Because the other end of the connection may dislike IP options, and ssh
is one know application that behaves thus, it is prudent to respond in
kind.

This patch changes the network labeling behavior such that an outgoing
packet that would be given a CIPSO label that matches the ambient label
is left unlabeled.  An "unlbl" domain is added and the netlabel
defaulting mechanism invoked rather than assuming that everything is
CIPSO.  Locking has been added around changes to the ambient label as
the mechanisms used to do so are more involved.

Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
Acked-by: Paul Moore <paul.moore@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agox86: fix page_is_ram() thinko
Arjan van de Ven [Mon, 18 Feb 2008 17:58:45 +0000 (09:58 -0800)]
x86: fix page_is_ram() thinko

page_is_ram() has a special case for the 640k-1M bios area, however
due to a thinko the special case checks the e820 table entry and
not the memory the user has asked for. This patch fixes the bug.

[ mingo@elte.hu: this too is better solved in the e820 space, but those
  fixes are too intrusive for v2.6.25. ]

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix WARN_ON() message: teach page_is_ram() about the special 4Kb bios data page
Arjan van de Ven [Mon, 18 Feb 2008 17:54:33 +0000 (09:54 -0800)]
x86: fix WARN_ON() message: teach page_is_ram() about the special 4Kb bios data page

This patch teaches page_is_ram() about the fact that the first
4Kb of memory are special on x86, even though the E820 table
normally doesn't exclude it.

This fixes the WARN_ON() reported by Laurent Riffard who was also
very helpful in diagnosing the issue.

[ mingo@elte.hu: we are working on doing this properly in the e820
  space, but for 2.6.25 this is the better fix. ]

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: i8259A: remove redundant irq_descinitialization
Ahmed S. Darwish [Sun, 17 Feb 2008 22:59:54 +0000 (00:59 +0200)]
x86: i8259A: remove redundant irq_descinitialization

Remove redundant irq_desc[NR_IRQS] element initialization in
init_ISA_irqs(). irq_desc[NR_IRQS] is already statically
initialized with the same values in kernel/irq/handle.c .

besides the clean-up value this also saves some space:

      text    data     bss     dec     hex filename
      1389     356      14    1759     6df i8259_32.o.before
      1325     356      14    1695     69f i8259_32.o.after

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix vdso_install breaks user "make install"
Sam Ravnborg [Sat, 16 Feb 2008 11:36:10 +0000 (12:36 +0100)]
x86: fix vdso_install breaks user "make install"

I suggest to make the vdso_install step independent as
in following patch.

This solves the issue at ahnd and still gives us the posibility
to install the files should they be needed.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: change IO delay back to 0x80
Ingo Molnar [Sun, 17 Feb 2008 19:20:24 +0000 (20:20 +0100)]
x86: change IO delay back to 0x80

change back the IO delay to 0x80.

Alan says that 0xed is known to break some older boxes, and given that
the get-rid-of-outb-APIs efforts are well underway we should just let
them be finished.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: lds - Use THREAD_SIZE instead of numeric constant
Cyrill Gorcunov [Sun, 17 Feb 2008 15:17:18 +0000 (18:17 +0300)]
x86: lds - Use THREAD_SIZE instead of numeric constant

Though we use PDA for regular task stack but that
is not acceptable for init_task wich is special
one. We still have to allocate init_task's stack
in that manner.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: lds - Use PAGE_SIZE instead of numeric constant
Cyrill Gorcunov [Sun, 17 Feb 2008 15:17:17 +0000 (18:17 +0300)]
x86: lds - Use PAGE_SIZE instead of numeric constant

It's much better to use PAGE_SIZE then magic 4096
(though it's almost synonym in most cases on x86 but
not for *all* cases ;)

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86 cleanup: suspend_asm_64.S - use X86_CR4_PGE instead of numeric value
Cyrill Gorcunov [Fri, 8 Feb 2008 21:49:13 +0000 (00:49 +0300)]
x86 cleanup: suspend_asm_64.S - use X86_CR4_PGE instead of numeric value

By including <asm/processor-flags.h> we're allowed to use
X86_CR4_PGE instead of numeric constant.

md5 sums of compiled files are differ due to this inclusion
but .text section remains the same.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: docs fixes to Documentation/i386/IO-APIC.txt
Nick Andrew [Sun, 17 Feb 2008 07:01:42 +0000 (18:01 +1100)]
x86: docs fixes to Documentation/i386/IO-APIC.txt

Clean up spelling and grammar of IO-APIC.txt

Signed-off-by: Nick Andrew <nick@nick-andrew.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix printout ugliness in cpu info printk
Marcin Slusarz [Fri, 1 Feb 2008 20:31:51 +0000 (21:31 +0100)]
x86: fix printout ugliness in cpu info printk

fix print_cpu_info, because it produced on boot:

  CPU: <6>AMD Athlon(tm) 64 Processor 3200+ stepping 00

instead of:

  CPU: AMD Athlon(tm) 64 Processor 3200+ stepping 00

(broken since 04e1ba852132c9ad006affcd5b8c8606295170b0 -
 x86: cleanup kernel/setup_64.c)

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: clean up csum-wrappers_64.c some more
Ingo Molnar [Sun, 17 Feb 2008 15:48:25 +0000 (16:48 +0100)]
x86: clean up csum-wrappers_64.c some more

no code changed:

arch/x86/lib/csum-wrappers_64.o:
   text    data     bss     dec     hex filename
    839       0       0     839     347 csum-wrappers_64.o.before
    839       0       0     839     347 csum-wrappers_64.o.after
md5:
b31994226c33e0b52bef5a0e110b84b0  csum-wrappers_64.o.before.asm
b31994226c33e0b52bef5a0e110b84b0  csum-wrappers_64.o.after.asm

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: coding style fixes in arch/x86/lib/csum-wrappers_64.c
Paolo Ciarrocchi [Sun, 17 Feb 2008 13:56:50 +0000 (14:56 +0100)]
x86: coding style fixes in arch/x86/lib/csum-wrappers_64.c

no code changed:

arch/x86/lib/csum-wrappers_64.o:
   text    data     bss     dec     hex filename
    839       0       0     839     347 csum-wrappers_64.o.before
    839       0       0     839     347 csum-wrappers_64.o.after
md5:
b31994226c33e0b52bef5a0e110b84b0  csum-wrappers_64.o.before.asm
b31994226c33e0b52bef5a0e110b84b0  csum-wrappers_64.o.after.asm

Signed-off-by: Paolo Ciarrocchi <paolo.ciarrocchi@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: coding style fixes in arch/x86/lib/io_64.c
Paolo Ciarrocchi [Sun, 17 Feb 2008 13:41:16 +0000 (14:41 +0100)]
x86: coding style fixes in arch/x86/lib/io_64.c

This simple patch makes the file error free (according to
checkpatch.pl)

no code changed:

arch/x86/lib/io_64.o:
   text    data     bss     dec     hex filename
    308       0       0     308     134 io_64.o.before
    308       0       0     308     134 io_64.o.after
md5:
3c64f9ed83d091678e849b36ca27bee3  io_64.o.before.asm
3c64f9ed83d091678e849b36ca27bee3  io_64.o.after.asm

Signed-off-by: Paolo Ciarrocchi <paolo.ciarrocchi@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: exclude vsyscall files from stackprotect
Ingo Molnar [Thu, 14 Feb 2008 07:38:49 +0000 (08:38 +0100)]
x86: exclude vsyscall files from stackprotect

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: add pgd_large() on 64-bit, for consistency
H. Peter Anvin [Tue, 19 Feb 2008 15:18:32 +0000 (16:18 +0100)]
x86: add pgd_large() on 64-bit, for consistency

In order to have it at all levels, add pgd_large() which only
returns 0.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: minor cleanup of comments in processor.h
Mike Travis [Fri, 8 Feb 2008 23:37:42 +0000 (15:37 -0800)]
x86: minor cleanup of comments in processor.h

Signed-off-by: Mike Travis <travis@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Jack Steiner <steiner@sgi.com>
Cc: linux-mm@kvack.org
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: annotate pci/common.s:pci_scan_bus_with_sysdata with __devinit
Sam Ravnborg [Sun, 17 Feb 2008 12:23:00 +0000 (13:23 +0100)]
x86: annotate pci/common.s:pci_scan_bus_with_sysdata with __devinit

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix section mismatch in head_64.S:initial_code
Sam Ravnborg [Sun, 17 Feb 2008 12:22:59 +0000 (13:22 +0100)]
x86: fix section mismatch in head_64.S:initial_code

initial_code are initially used to hold a function pointer
from __init and later from __cpuinit. This confuses modpost
and changing initial_code to REFDATA silence the warning.
(But now we do not discard the variable anymore).

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix section mismatch in srat_64.c:reserve_hotadd
Sam Ravnborg [Sun, 17 Feb 2008 12:22:58 +0000 (13:22 +0100)]
x86: fix section mismatch in srat_64.c:reserve_hotadd

reserve_hotadd() are only used by __init acpi_numa_memory_affinity_init().
Annotate reserve_hotadd() with __init is the trivial fix.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix section mismatch warning in topology.c:arch_register_cpu
Sam Ravnborg [Sun, 17 Feb 2008 12:22:49 +0000 (13:22 +0100)]
x86: fix section mismatch warning in topology.c:arch_register_cpu

arch_register_cpu() is only defined for HOTPLUG_CPU code
so simple fix is to ignore references by annotating the
function __ref.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix section mismatch in setup_64.c:srat_detect_node
Sam Ravnborg [Sun, 17 Feb 2008 12:22:47 +0000 (13:22 +0100)]
x86: fix section mismatch in setup_64.c:srat_detect_node

srat_detect_node() is only used by __cpuinit init_intel().
So the trivial fix is to annotate srat_detect_node() with __cpuinit.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix section mismatch warning in setup_64.c:nearby_node
Sam Ravnborg [Sun, 17 Feb 2008 12:22:46 +0000 (13:22 +0100)]
x86: fix section mismatch warning in setup_64.c:nearby_node

nearby_node() were only used by __cpuinit amd_detect_cmp()
So annotating nearby_node() __cpuinit was the trivial fix.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: fix compile warning building without CONFIG_SYSCTL
Li Zefan [Sun, 3 Feb 2008 07:40:30 +0000 (15:40 +0800)]
x86: fix compile warning building without CONFIG_SYSCTL

arch/x86/kernel/nmi_64.c:50: warning: 'unknown_nmi_panic_callback' declared 'static' but never defined

This patch also fixes nmi_32.c

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: CPA: remove BUG_ON for LRU/Compound pages
Andi Kleen [Mon, 11 Feb 2008 09:50:21 +0000 (10:50 +0100)]
x86: CPA: remove BUG_ON for LRU/Compound pages

New implementation does not use lru for anything so there is no need
to reject pages that are in the LRU. Similar for compound pages (which
were checked because they also use page->lru)

[ tglx@linutronix.de: removed unused variable ]

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: don't make swapper_pg_fixmap global
Adrian Bunk [Wed, 13 Feb 2008 21:29:55 +0000 (23:29 +0200)]
x86: don't make swapper_pg_fixmap global

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Ian Campbell <ijc@hellion.org.uk>
Cc: hpa@zytor.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: don't make irq_return global
Adrian Bunk [Wed, 13 Feb 2008 21:29:53 +0000 (23:29 +0200)]
x86: don't make irq_return global

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: hpa@zytor.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: unexport io_delay_type
Adrian Bunk [Wed, 13 Feb 2008 21:29:42 +0000 (23:29 +0200)]
x86: unexport io_delay_type

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: hpa@zytor.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: make mxcsr_feature_mask static again
Adrian Bunk [Wed, 13 Feb 2008 21:29:33 +0000 (23:29 +0200)]
x86: make mxcsr_feature_mask static again

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Roland McGrath <roland@redhat.com>
Cc: hpa@zytor.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: sparse warning in efi.c
Harvey Harrison [Wed, 13 Feb 2008 21:26:13 +0000 (13:26 -0800)]
x86: sparse warning in efi.c

Yes, it should.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86: sparse error in efi_32.c
Harvey Harrison [Wed, 13 Feb 2008 21:26:39 +0000 (13:26 -0800)]
x86: sparse error in efi_32.c

arch/x86/kernel/efi_32.c:42:6: warning: symbol 'efi_call_phys_prelog' was not declared. Should it be static?
arch/x86/kernel/efi_32.c:84:6: warning: symbol 'efi_call_phys_epilog' was not declared. Should it be static?

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agox86, kprobes: remove sparse warnings from x86
Harvey Harrison [Thu, 14 Feb 2008 23:23:53 +0000 (15:23 -0800)]
x86, kprobes: remove sparse warnings from x86

arch/x86/kernel/kprobes.c:584:16: warning: symbol 'kretprobe_trampoline_holder' was not declared. Should it be static?
arch/x86/kernel/kprobes.c:676:6: warning: symbol 'trampoline_handler' was not declared. Should it be static?

Make them static and add the __used attribute, approach taken from the
arm kprobes implementation.

kretprobe_trampoline_holder uses inline assemly to define the global
symbol kretprobe_trampoline, but nothing ever calls the holder explicitly.

trampoline handler is only called from inline assembly in the same file,
mark it used and static.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years agoremove mca-pentium
Adrian Bunk [Thu, 14 Feb 2008 21:23:37 +0000 (23:23 +0200)]
remove mca-pentium

This patch removes the mca-pentium boot option that was a noop.

besides the source code cleanup factor, this saves some text as well:

   arch/x86/kernel/cpu/bugs.o:
      text    data     bss     dec     hex filename
       651      77       4     732     2dc bugs.o.before
       631      53       4     688     2b0 bugs.o.after

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
16 years ago[S390] qdio: FCP/SCSI write I/O stagnates on LPAR
Ursula Braun [Tue, 19 Feb 2008 14:29:35 +0000 (15:29 +0100)]
[S390] qdio: FCP/SCSI write I/O stagnates on LPAR

If running on LPAR, qdio might overlook an incoming buffer in certain
scenarios. The patch makes sure that incoming buffers are detected
immediately in all situations.

Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Fix futex_atomic_cmpxchg_std inline assembly.
Heiko Carstens [Tue, 19 Feb 2008 14:29:34 +0000 (15:29 +0100)]
[S390] Fix futex_atomic_cmpxchg_std inline assembly.

Add missing exception table entry so that the kernel can handle
proctection exceptions as well on the cs instruction. Currently only
specification exceptions are handled correctly.
The missing entry allows user space to crash the kernel.

Cc: stable <stable@kernel.org>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] dcss: Fix Unlikely(x) != y
Roel Kluin [Tue, 19 Feb 2008 14:29:33 +0000 (15:29 +0100)]
[S390] dcss: Fix Unlikely(x) != y

Fix Unlikely(x) != y

Cc: Gerald Schaefer <geraldsc@de.ibm.com>
Cc: Stefan Weinhuber <wein@de.ibm.com>
Cc: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] sclp: clean up send/receive naming scheme
Peter Oberparleiter [Tue, 19 Feb 2008 14:29:32 +0000 (15:29 +0100)]
[S390] sclp: clean up send/receive naming scheme

Make state change events adjust the correct mask by cleaning up
naming inconsistencies. Also remove chance for lockup by removing
unnecessary mask related check before reading events.

Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] etr: fix compile error on !SMP
Heiko Carstens [Tue, 19 Feb 2008 14:29:31 +0000 (15:29 +0100)]
[S390] etr: fix compile error on !SMP

Since a5fbb6d1064be885d2a6b82f625186753cf74848
"KVM: fix !SMP build error" smp_call_function isn't a define anymore
that folds into nothing but a define that calls up_smp_call_function
with all parameters. Hence we cannot #ifdef out the unused code
anymore...
This seems to be the preferred method, so do this for s390 as well.

arch/s390/kernel/time.c: In function 'etr_sync_clock':
arch/s390/kernel/time.c:825: error: 'clock_sync_cpu_start' undeclared
arch/s390/kernel/time.c:862: error: 'clock_sync_cpu_end' undeclared

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] qdio: fix qdio_activate timeout handling.
Heiko Carstens [Tue, 19 Feb 2008 14:29:30 +0000 (15:29 +0100)]
[S390] qdio: fix qdio_activate timeout handling.

Current code in qdio_activate waits for at least 5 seconds
until it returns. It may return earlier if an error occurs,
but not if everything is ok. This large timeout value
became visible with commit dfa77f611ff295598e218aa0eb6efa73a5cf26d0
"qdio: set QDIO_ACTIVATE_TIMEOUT to 5s", which intended to
fix the timeout value which was zero. In turn setting an
FCP adapter online took 5 seconds.

In practice waiting for 5ms before continuing is sufficient
as pointed out by Utz Bacher and Cornelia Huck.

Cc: Utz Bacher <utz.bacher@de.ibm.com>
Cc: Jan Glauber <jan.glauber@de.ibm.com>
Cc: Ursula Braun <braunu@de.ibm.com>
Cc: Martin Peschke <mp3@de.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Initialize per cpu lowcores on cpu hotplug.
Heiko Carstens [Tue, 19 Feb 2008 14:29:29 +0000 (15:29 +0100)]
[S390] Initialize per cpu lowcores on cpu hotplug.

Just copy the first 512 read-only bytes of the current cpu lowcore if
a new cpu gets onlined. The rest is zeroed out and must be explicitly
initialized. Current code just copies the entire lowcore and
initializes the needed fields.
This should reveal bugs in future enhancements quite early.
Also when the lowcore of the first cpu is replaced this is now done
atomically (no interrupts, no machine checks).

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] find bit corner case.
Martin Schwidefsky [Tue, 19 Feb 2008 14:29:28 +0000 (15:29 +0100)]
[S390] find bit corner case.

Fix [ext2_]find_first_[zero_]bit for the corner case of an all clear
or all set bit field by always handling that last word of the bit field
with __ffz_word/__ffs_word.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] dasd: fix locking in __dasd_device_process_final_queue
Stefan Weinhuber [Tue, 19 Feb 2008 14:29:27 +0000 (15:29 +0100)]
[S390] dasd: fix locking in __dasd_device_process_final_queue

After setting the status of the cqr and releasing the lock for the
block cqr queue, we call the cqr callback function, which will usually
just trigger the dasd_block_tasklet. But when the tasklet is already
running the cqr might be processed before we invoke the callback
function. In rare cases the callback pointer may already be invalid
by the time we want to call it, which will result in a panic.
Solution: Call the callback function first and then release the lock.

Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Make sure enabled wait psw is loaded in default_idle.
Heiko Carstens [Tue, 19 Feb 2008 14:29:26 +0000 (15:29 +0100)]
[S390] Make sure enabled wait psw is loaded in default_idle.

If both NO_IDLE_HZ and VIRT_TIMER are disabled default_idle won't load
an enabled wait psw and busy loop instead. This is because the
idle_chain is empty and the return value of atomic_notifier_call_chain
will be NOTIFY_DONE, which causes default_idle to return instead of
loading an enabled wait psw.
Fix this by calling __atomic_notifier_call_chain instead and add proper
return value handling.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Let NR_CPUS default to 32/64 on s390/s390x.
Heiko Carstens [Tue, 19 Feb 2008 14:29:25 +0000 (15:29 +0100)]
[S390] Let NR_CPUS default to 32/64 on s390/s390x.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] cio: Do timed recovery on workqueue.
Cornelia Huck [Tue, 19 Feb 2008 14:29:24 +0000 (15:29 +0100)]
[S390] cio: Do timed recovery on workqueue.

We can't do our recovery in softirq context, so we schedule it from
our timer function.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] cio: Remember to initialize recovery_lock.
Cornelia Huck [Tue, 19 Feb 2008 14:29:23 +0000 (15:29 +0100)]
[S390] cio: Remember to initialize recovery_lock.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years agolibata: implement drain buffers
James Bottomley [Tue, 19 Feb 2008 10:36:57 +0000 (11:36 +0100)]
libata: implement drain buffers

This just updates the libata slave configure routine to take advantage
of the block layer drain buffers.  It also adjusts the size lengths in
the atapi code to add the drain buffer to the DMA length so the driver
knows it can rely on it.

I suspect I should also be checking for AHCI as well as ATA_DEV_ATAPI,
but I couldn't see how to do that easily.

tj: * atapi_drain_needed() added such that draining is applied to only
      misc ATAPI commands.
    * q->bounce_gfp used when allocating drain buffer.
    * Now duplicate ATAPI PIO drain logic dropped.
    * ata_dev_printk() used instead of sdev_printk().

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
16 years agolibata: eliminate the home grown dma padding in favour of
James Bottomley [Tue, 19 Feb 2008 10:36:56 +0000 (11:36 +0100)]
libata: eliminate the home grown dma padding in favour of

that provided by the block layer

ATA requires that all DMA transfers begin and end on word boundaries.
Because of this, a large amount of machinery grew up in ide to adjust
scatterlists on this basis.  However, as of 2.5, the block layer has a
dma_alignment variable which ensures both the beginning and length of a
DMA transfer are aligned on the dma_alignment boundary.  Although the
block layer does adjust the beginning of the transfer to ensure this
happens, it doesn't actually adjust the length, it merely makes sure
that space is allocated for transfers beyond the declared length.  The
upshot of this is that scatterlists may be padded to any size between
the actual length and the length adjusted to the dma_alignment safely
knowing that memory is allocated in this region.

Right at the moment, SCSI takes the default dma_aligment which is on a
512 byte boundary.  Note that this aligment only applies to transfers
coming in from user space.  However, since all kernel allocations are
automatically aligned on a minimum of 32 byte boundaries, it is safe to
adjust them in this manner as well.

tj: * Adjusting sg after padding is done in block layer.  Make libata
      set queue alignment correctly for ATAPI devices and drop broken
      sg mangling from ata_sg_setup().
    * Use request->raw_data_len for ATAPI transfer chunk size.
    * Killed qc->raw_nbytes.
    * Separated out killing qc->n_iter.

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
16 years agoblock: clear drain buffer if draining for write command
Tejun Heo [Tue, 19 Feb 2008 10:36:55 +0000 (11:36 +0100)]
block: clear drain buffer if draining for write command

Clear drain buffer before chaining if the command in question is a
write.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
16 years agoblock: implement request_queue->dma_drain_needed
Tejun Heo [Tue, 19 Feb 2008 10:36:53 +0000 (11:36 +0100)]
block: implement request_queue->dma_drain_needed

Draining shouldn't be done for commands where overflow may indicate
data integrity issues.  Add dma_drain_needed callback to
request_queue.  Drain buffer is appened iff this function returns
non-zero.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
16 years agoblock: add request->raw_data_len
Tejun Heo [Tue, 19 Feb 2008 10:36:35 +0000 (11:36 +0100)]
block: add request->raw_data_len

With padding and draining moved into it, block layer now may extend
requests as directed by queue parameters, so now a request has two
sizes - the original request size and the extended size which matches
the size of area pointed to by bios and later by sgs.  The latter size
is what lower layers are primarily interested in when allocating,
filling up DMA tables and setting up the controller.

Both padding and draining extend the data area to accomodate
controller characteristics.  As any controller which speaks SCSI can
handle underflows, feeding larger data area is safe.

So, this patch makes the primary data length field, request->data_len,
indicate the size of full data area and add a separate length field,
request->raw_data_len, for the unmodified request size.  The latter is
used to report to higher layer (userland) and where the original
request size should be fed to the controller or device.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
16 years agoblock: update bio according to DMA alignment padding
Tejun Heo [Tue, 19 Feb 2008 10:35:38 +0000 (11:35 +0100)]
block: update bio according to DMA alignment padding

DMA start address and transfer size alignment for PC requests are
achieved using bio_copy_user() instead of bio_map_user().  This works
because bio_copy_user() always uses full pages and block DMA alignment
isn't allowed to go over PAGE_SIZE.

However, the implementation didn't update the last bio of the request
to make this padding visible to lower layers.  This patch makes
blk_rq_map_user() extend the last bio such that it includes the
padding area and the size of area pointed to by the request is
properly aligned.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
16 years agolibata: update ATAPI overflow draining
Tejun Heo [Tue, 19 Feb 2008 10:35:37 +0000 (11:35 +0100)]
libata: update ATAPI overflow draining

For misc ATAPI commands which transfer variable length data to the
host, overflow can occur due to application or hardware bug.  Such
overflows can be ignored safely as long as overflow data is properly
drained.  libata HSM implementation has this implemented in
__atapi_pio_bytes() and recently updated for 2.6.24-rc but it requires
further improvements.  Improve drain logic such that...

* Report overflow errors using ehi desc mechanism instead of printing
  directly.

* Properly calculate the number of bytes to be drained considering
  actual number of consumed bytes for partial draining.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Acked-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>