GitHub/moto-9609/android_kernel_motorola_exynos9610.git
16 years agohfs: handle match_strdup failure
Jim Meyering [Tue, 29 Apr 2008 07:59:07 +0000 (00:59 -0700)]
hfs: handle match_strdup failure

fs/hfs/super.c (parse_options): Handle match_strdup failure, twice.

Signed-off-by: Jim Meyering <meyering@redhat.com>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoaffs: handle match_strdup failure
Jim Meyering [Tue, 29 Apr 2008 07:59:06 +0000 (00:59 -0700)]
affs: handle match_strdup failure

fs/affs/super.c (parse_options): Remove useless initialization.  Handle
match_strdup failure.

Signed-off-by: Jim Meyering <meyering@redhat.com>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs: remove unused fops from struct char_device_struct
Jiri Olsa [Tue, 29 Apr 2008 07:59:05 +0000 (00:59 -0700)]
fs: remove unused fops from struct char_device_struct

struct char_device_struct::fops is no longer used: remove it.

Signed-off-by: Jiri Olsa <olsajiri@gmail.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoipwireless: remove dead code
Jiri Kosina [Tue, 29 Apr 2008 07:59:04 +0000 (00:59 -0700)]
ipwireless: remove dead code

Remove unused leftovers of debugging functions. 2.6.25 material.

Reported-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix sparse warning in root.c
Harvey Harrison [Tue, 29 Apr 2008 07:59:03 +0000 (00:59 -0700)]
autofs4: fix sparse warning in root.c

fs/autofs4/root.c:536:23: warning: symbol 'ino' shadows an earlier one
fs/autofs4/root.c:510:22: originally declared here

There is no need to redeclare, we are at the end of the loop and in
the next iteration of the loop, ino will be reset.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Acked-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoremove mca_is_adapter_used()
Adrian Bunk [Tue, 29 Apr 2008 07:59:02 +0000 (00:59 -0700)]
remove mca_is_adapter_used()

Remove the no longer used mca_is_adapter_used().

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomake BINFMT_FLAT a bool
Adrian Bunk [Tue, 29 Apr 2008 07:59:02 +0000 (00:59 -0700)]
make BINFMT_FLAT a bool

I have not yet seen anyone saying he has a reasonable use case for using
BINFMT_FLAT modular on his embedded device.

Considering that fs/binfmt_flat.c even lacks a MODULE_LICENSE() I really doubt
there is any, and this patch therefore makes BINFMT_FLAT a bool.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Bryan Wu <cooloney.lkml@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomake fs/buffer.c:cont_expand_zero() static
Adrian Bunk [Tue, 29 Apr 2008 07:59:01 +0000 (00:59 -0700)]
make fs/buffer.c:cont_expand_zero() static

cont_expand_zero() can become static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoremove generic_commit_write()
Adrian Bunk [Tue, 29 Apr 2008 07:59:00 +0000 (00:59 -0700)]
remove generic_commit_write()

Remove the obsolete and no longer used generic_commit_write().

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/timerfd.c should #include <linux/syscalls.h>
Adrian Bunk [Tue, 29 Apr 2008 07:58:59 +0000 (00:58 -0700)]
fs/timerfd.c should #include <linux/syscalls.h>

Every file should include the headers containing the prototypes for its global
functions (in this case for sys_timerfd_*()).

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agolet LOG_BUF_SHIFT default to 17
Adrian Bunk [Tue, 29 Apr 2008 07:58:58 +0000 (00:58 -0700)]
let LOG_BUF_SHIFT default to 17

16 kB is often no longer enough for a normal boot of an UP system.

And even less when people e.g. use suspend.

17 seems to be a more reasonable default for current kernels on current
hardware (it's just the default, anyone who is memory limited can still lower
it).

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/aio.c: make 3 functions static
Adrian Bunk [Tue, 29 Apr 2008 07:58:57 +0000 (00:58 -0700)]
fs/aio.c: make 3 functions static

Make the following needlessly global functions static:

- __put_ioctx()
- lookup_ioctx()
- io_submit_one()

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Zach Brown <zach.brown@oracle.com>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/drop_caches.c: make 2 functions static
Adrian Bunk [Tue, 29 Apr 2008 07:58:57 +0000 (00:58 -0700)]
fs/drop_caches.c: make 2 functions static

Make the following needlessly global functions static:

- drop_pagecache()
- drop_slab()

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/fs-writeback.c: make 2 functions static
Adrian Bunk [Tue, 29 Apr 2008 07:58:56 +0000 (00:58 -0700)]
fs/fs-writeback.c: make 2 functions static

Make the following needlessly global functions static:

- writeback_acquire()
- writeback_release()

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomake vfs_ioctl() static
Adrian Bunk [Tue, 29 Apr 2008 07:58:55 +0000 (00:58 -0700)]
make vfs_ioctl() static

Make the needlessly global vfs_ioctl() static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomake __put_super() static
Adrian Bunk [Tue, 29 Apr 2008 07:58:54 +0000 (00:58 -0700)]
make __put_super() static

Make the needlessly global __put_super() static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/freevxfs/: proper externs
Adrian Bunk [Tue, 29 Apr 2008 07:58:54 +0000 (00:58 -0700)]
fs/freevxfs/: proper externs

Move the extern declarations of several structs to vxfs_extern.h

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/hfsplus/: proper externs
Adrian Bunk [Tue, 29 Apr 2008 07:58:52 +0000 (00:58 -0700)]
fs/hfsplus/: proper externs

Add proper extern declarations for two structs in fs/hfsplus/hfsplus_fs.h

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/ramfs/ extern cleanup
Adrian Bunk [Tue, 29 Apr 2008 07:58:51 +0000 (00:58 -0700)]
fs/ramfs/ extern cleanup

- internal.h shouldn't duplicate the extern declaration for
  ramfs_file_operations already in include/linux/ramfs.h
- file-mmu.c needs two #include's for seeing the extern declarations
  of it's global struct's

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocpu: fix section mismatch warning in reference to register_cpu_notifier
Sam Ravnborg [Tue, 29 Apr 2008 07:58:51 +0000 (00:58 -0700)]
cpu: fix section mismatch warning in reference to register_cpu_notifier

Fix following warnings:
WARNING: vmlinux.o(.text+0xc60): Section mismatch in reference from the function kvm_init() to the function .cpuinit.text:register_cpu_notifier()
WARNING: vmlinux.o(.text+0x33869a): Section mismatch in reference from the function xfs_icsb_init_counters() to the function .cpuinit.text:register_cpu_notifier()
WARNING: vmlinux.o(.text+0x5556a1): Section mismatch in reference from the function acpi_processor_install_hotplug_notify() to the function .cpuinit.text:register_cpu_notifier()
WARNING: vmlinux.o(.text+0xfe6b28): Section mismatch in reference from the function cpufreq_register_driver() to the function .cpuinit.text:register_cpu_notifier()

register_cpu_notifier() are only really defined when HOTPLUG_CPU is enabled.
So references to the function are OK.

Annotate it with __ref so we do not get warnings from callers and do not get
warnings for the functions/data used by register_cpu_notifier().

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocpu: fix section mismatch warnings in *cpu_down
Sam Ravnborg [Tue, 29 Apr 2008 07:58:50 +0000 (00:58 -0700)]
cpu: fix section mismatch warnings in *cpu_down

Fix following warnings:
WARNING: vmlinux.o(.text+0x75c8d): Section mismatch in reference from the function take_cpu_down() to the variable .cpuinit.data:cpu_chain
WARNING: vmlinux.o(.text+0x75d2a): Section mismatch in reference from the function _cpu_down() to the variable .cpuinit.data:cpu_chain
WARNING: vmlinux.o(.text+0x75d4d): Section mismatch in reference from the function _cpu_down() to the variable .cpuinit.data:cpu_chain
WARNING: vmlinux.o(.text+0x75de4): Section mismatch in reference from the function _cpu_down() to the variable .cpuinit.data:cpu_chain
WARNING: vmlinux.o(.text+0x75e33): Section mismatch in reference from the function _cpu_down() to the variable .cpuinit.data:cpu_chain

cpu_down is only used from code surrounded by HOTPLUG_CPU so any references to
__cpuinit is OK.

Add a few __ref to tech modpost to ignore the references.

This is just papering over the fact that the cpu hotplug code is fragile with
respect to use of HOTPLUG_CPU and in many cases rely on __cpuinit to get rid
of code when HOTPLUG_CPU is not enabled.  For now this is the least invasive
change.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocpu: fix section mismatch warning in unregister_cpu_notifier
Sam Ravnborg [Tue, 29 Apr 2008 07:58:48 +0000 (00:58 -0700)]
cpu: fix section mismatch warning in unregister_cpu_notifier

Fix following warning:
WARNING: vmlinux.o(.text+0x75f4e): Section mismatch in reference from the function unregister_cpu_notifier() to the variable .cpuinit.data:cpu_chain

We know that unregister_cpu_notifier is using HOTPLUG_CPU
stuff - so ignore these references.
Annotating unregister_cpu_notifier had been another option
but this caused far more warnings since not all callers were
annotated __cpuinit.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocpu: fix section mismatch warnings in hotcpu_register
Sam Ravnborg [Tue, 29 Apr 2008 07:58:47 +0000 (00:58 -0700)]
cpu: fix section mismatch warnings in hotcpu_register

Fix following warnings:
WARNING: vmlinux.o(.data+0x5020): Section mismatch in reference from the variable cpu_vsyscall_notifier_nb.12876 to the function .cpuinit.text:cpu_vsyscall_notifier()
WARNING: vmlinux.o(.data+0x9ce0): Section mismatch in reference from the variable profile_cpu_callback_nb.17654 to the function .devinit.text:profile_cpu_callback()
WARNING: vmlinux.o(.data+0xd380): Section mismatch in reference from the variable workqueue_cpu_callback_nb.15004 to the function .devinit.text:workqueue_cpu_callback()
WARNING: vmlinux.o(.data+0x11d00): Section mismatch in reference from the variable relay_hotcpu_callback_nb.19626 to the function .cpuinit.text:relay_hotcpu_callback()
WARNING: vmlinux.o(.data+0x12970): Section mismatch in reference from the variable cpu_callback_nb.24694 to the function .devinit.text:cpu_callback()
WARNING: vmlinux.o(.data+0x3fee0): Section mismatch in reference from the variable percpu_counter_hotcpu_callback_nb.10903 to the function .cpuinit.text:percpu_counter_hotcpu_callback()
WARNING: vmlinux.o(.data+0x74ce0): Section mismatch in reference from the variable topology_cpu_callback_nb.12506 to the function .cpuinit.text:topology_cpu_callback()

Functions used as argument are by definition only used in HOTPLUG_CPU
situations so thay are annotated __cpuinit.  Annotate the static variable used
by hotcpu_register with __cpuinitdata to match this definition.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agobefs: fix sparse warning in linuxvfs.c
Harvey Harrison [Tue, 29 Apr 2008 07:58:44 +0000 (00:58 -0700)]
befs: fix sparse warning in linuxvfs.c

Use link as the variable name to avoid shadowing the arg.

fs/befs/linuxvfs.c:492:8: warning: symbol 'p' shadows an earlier one
fs/befs/linuxvfs.c:488:77: originally declared here

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "Sergey S. Kostyliov" <rathamahata@php4.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocoda: add static to functions in dir.c
Harvey Harrison [Tue, 29 Apr 2008 07:58:44 +0000 (00:58 -0700)]
coda: add static to functions in dir.c

coda_unlink, coda_rmdir, coda_readdir can all be static, the forward
declarations already were.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Jan Harkes <jaharkes@cs.cmu.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoadd RUSAGE_THREAD
Sripathi Kodi [Tue, 29 Apr 2008 07:58:42 +0000 (00:58 -0700)]
add RUSAGE_THREAD

Add the RUSAGE_THREAD option for the getrusage system call.  This is
essentially Roland's patch from http://lkml.org/lkml/2008/1/18/589, but the
line about RUSAGE_LWP line has been removed, as suggested by Ulrich and
Christoph.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Michael Kerrisk <mtk.manpages@googlemail.com>
Cc: Ulrich Drepper <drepper@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoadfs: work around bogus sparse warning
Harvey Harrison [Tue, 29 Apr 2008 07:58:41 +0000 (00:58 -0700)]
adfs: work around bogus sparse warning

fs/adfs/dir_f.c:126:4: warning: do-while statement is not a compound statement

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoTaint kernel after WARN_ON(condition)
Nur Hussein [Tue, 29 Apr 2008 07:58:39 +0000 (00:58 -0700)]
Taint kernel after WARN_ON(condition)

The kernel is sent to tainted within the warn_on_slowpath() function, and
whenever a warning occurs the new taint flag 'W' is set.  This is useful to
know if a warning occurred before a BUG by preserving the warning as a flag
in the taint state.

This does not work on architectures where WARN_ON has its own definition.
These archs are:
1. s390
2. superh
3. avr32
4. parisc

The maintainers of these architectures have been added in the Cc: list
in this email to alert them to the situation.

The documentation in oops-tracing.txt has been updated to include the
new flag.

Signed-off-by: Nur Hussein <nurhussein@gmail.com>
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: "Randy.Dunlap" <rdunlap@xenotime.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Haavard Skinnemoen <hskinnemoen@atmel.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofs/coda: remove static inline forward declarations
Ilpo Järvinen [Tue, 29 Apr 2008 07:58:37 +0000 (00:58 -0700)]
fs/coda: remove static inline forward declarations

They're defined later on in the same file with bodies and nothing in
between needs them.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: Jan Harkes <jaharkes@cs.cmu.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoAvoid divides in BITS_TO_LONGS
Eric Dumazet [Tue, 29 Apr 2008 07:58:35 +0000 (00:58 -0700)]
Avoid divides in BITS_TO_LONGS

BITS_PER_LONG is a signed value (32 or 64)

DIV_ROUND_UP(nr, BITS_PER_LONG) performs signed arithmetic if "nr" is signed too.

Converting BITS_TO_LONGS(nr) to DIV_ROUND_UP(nr, BITS_PER_BYTE *
sizeof(long)) makes sure compiler can perform a right shift, even if "nr"
is a signed value, instead of an expensive integer divide.

Applying this patch saves 141 bytes on x86 when CONFIG_CC_OPTIMIZE_FOR_SIZE=y
and speedup bitmap operations.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoepoll: avoid kmemcheck warning
Davide Libenzi [Tue, 29 Apr 2008 07:58:34 +0000 (00:58 -0700)]
epoll: avoid kmemcheck warning

Epoll calls rb_set_parent(n, n) to initialize the rb-tree node, but
rb_set_parent() accesses node's pointer in its code.  This creates a
warning in kmemcheck (reported by Vegard Nossum) about an uninitialized
memory access.  The warning is harmless since the following rb-tree node
insert is going to overwrite the node data.  In any case I think it's
better to not have that happening at all, and fix it by simplifying the
code to get rid of a few lines that became superfluous after the previous
epoll changes.

Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Cc: Vegard Nossum <vegard.nossum@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomake /dev/kmem a config option
Arjan van de Ven [Tue, 29 Apr 2008 07:58:34 +0000 (00:58 -0700)]
make /dev/kmem a config option

Make /dev/kmem a config option; /dev/kmem is VERY rarely used, and when
used, it's generally for no good (rootkits tend to be the most common
users).  With this config option, users have the choice to disable
/dev/kmem, saving some size as well.

A patch to disable /dev/kmem has been in the Fedora and RHEL kernels for
4+ years now without any known problems or legit users of /dev/kmem.

[akpm@linux-foundation.org: make CONFIG_DEVKMEM default to y]
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopower: replace remaining __FUNCTION__ occurrences
Harvey Harrison [Tue, 29 Apr 2008 07:58:29 +0000 (00:58 -0700)]
power: replace remaining __FUNCTION__ occurrences

__FUNCTION__ is gcc-specific, use __func__

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Greg KH <greg@kroah.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopage allocator: explicitly retry hugepage allocations
Nishanth Aravamudan [Tue, 29 Apr 2008 07:58:26 +0000 (00:58 -0700)]
page allocator: explicitly retry hugepage allocations

Add __GFP_REPEAT to hugepage allocations.  Do so to not necessitate userspace
putting pressure on the VM by repeated echo's into /proc/sys/vm/nr_hugepages
to grow the pool.  With the previous patch to allow for large-order
__GFP_REPEAT attempts to loop for a bit (as opposed to indefinitely), this
increases the likelihood of getting hugepages when the system experiences (or
recently experienced) load.

Mel tested the patchset on an x86_32 laptop.  With the patches, it was easier
to use the proc interface to grow the hugepage pool.  The following is the
output of a script that grows the pool as much as possible running on
2.6.25-rc9.

Allocating hugepages test
-------------------------
Disabling OOM Killer for current test process
Starting page count: 0
Attempt 1: 57 pages Progress made with 57 pages
Attempt 2: 73 pages Progress made with 16 pages
Attempt 3: 74 pages Progress made with 1 pages
Attempt 4: 75 pages Progress made with 1 pages
Attempt 5: 77 pages Progress made with 2 pages

77 pages was the most it allocated but it took 5 attempts from userspace
to get it. With the 3 patches in this series applied,

Allocating hugepages test
-------------------------
Disabling OOM Killer for current test process
Starting page count: 0
Attempt 1: 75 pages Progress made with 75 pages
Attempt 2: 76 pages Progress made with 1 pages
Attempt 3: 79 pages Progress made with 3 pages

And 79 pages was the most it got. Your patches were able to allocate the
bulk of possible pages on the first attempt.

Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Tested-by: Mel Gorman <mel@csn.ul.ie>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopage allocator: smarter retry of costly-order allocations
Nishanth Aravamudan [Tue, 29 Apr 2008 07:58:25 +0000 (00:58 -0700)]
page allocator: smarter retry of costly-order allocations

Because of page order checks in __alloc_pages(), hugepage (and similarly
large order) allocations will not retry unless explicitly marked
__GFP_REPEAT. However, the current retry logic is nearly an infinite
loop (or until reclaim does no progress whatsoever). For these costly
allocations, that seems like overkill and could potentially never
terminate. Mel observed that allowing current __GFP_REPEAT semantics for
hugepage allocations essentially killed the system. I believe this is
because we may continue to reclaim small orders of pages all over, but
never have enough to satisfy the hugepage allocation request. This is
clearly only a problem for large order allocations, of which hugepages
are the most obvious (to me).

Modify try_to_free_pages() to indicate how many pages were reclaimed.
Use that information in __alloc_pages() to eventually fail a large
__GFP_REPEAT allocation when we've reclaimed an order of pages equal to
or greater than the allocation's order. This relies on lumpy reclaim
functioning as advertised. Due to fragmentation, lumpy reclaim may not
be able to free up the order needed in one invocation, so multiple
iterations may be requred. In other words, the more fragmented memory
is, the more retry attempts __GFP_REPEAT will make (particularly for
higher order allocations).

This changes the semantics of __GFP_REPEAT subtly, but *only* for
allocations > PAGE_ALLOC_COSTLY_ORDER. With this patch, for those size
allocations, we will try up to some point (at least 1<<order reclaimed
pages), rather than forever (which is the case for allocations <=
PAGE_ALLOC_COSTLY_ORDER).

This change improves the /proc/sys/vm/nr_hugepages interface with a
follow-on patch that makes pool allocations use __GFP_REPEAT. Rather
than administrators repeatedly echo'ing a particular value into the
sysctl, and forcing reclaim into action manually, this change allows for
the sysctl to attempt a reasonable effort itself. Similarly, dynamic
pool growth should be more successful under load, as lumpy reclaim can
try to free up pages, rather than failing right away.

Choosing to reclaim only up to the order of the requested allocation
strikes a balance between not failing hugepage allocations and returning
to the caller when it's unlikely to every succeed. Because of lumpy
reclaim, if we have freed the order requested, hopefully it has been in
big chunks and those chunks will allow our allocation to succeed. If
that isn't the case after freeing up the current order, I don't think it
is likely to succeed in the future, although it is possible given a
particular fragmentation pattern.

Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Tested-by: Mel Gorman <mel@csn.ul.ie>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: fix misleading __GFP_REPEAT related comments
Nishanth Aravamudan [Tue, 29 Apr 2008 07:58:23 +0000 (00:58 -0700)]
mm: fix misleading __GFP_REPEAT related comments

The definition and use of __GFP_REPEAT, __GFP_NOFAIL and __GFP_NORETRY in the
core VM have somewhat differing comments as to their actual semantics.
Annoyingly, the flags definition has inline and header comments, which might
be interpreted as not being equivalent.  Just add references to the header
comments in the inline ones so they don't go out of sync in the future.  In
their use in __alloc_pages() clarify that the current implementation treats
low-order allocations and __GFP_REPEAT allocations as distinct cases.

To clarify, the flags' semantics are:

__GFP_NORETRY means try no harder than one run through __alloc_pages

__GFP_REPEAT means __GFP_NOFAIL

__GFP_NOFAIL means repeat forever

order <= PAGE_ALLOC_COSTLY_ORDER means __GFP_NOFAIL

Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: fix usemap initialization
KAMEZAWA Hiroyuki [Tue, 29 Apr 2008 07:58:21 +0000 (00:58 -0700)]
mm: fix usemap initialization

usemap must be initialized only when pfn is within zone.  If not, it corrupts
memory.

And this patch also reduces the number of calls to set_pageblock_migratetype()
from
(pfn & (pageblock_nr_pages -1)
to
!(pfn & (pageblock_nr_pages-1)
it should be called once per pageblock.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Shi Weihua <shiwh@cn.fujitsu.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agodrivers: fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 23:50:04 +0000 (16:50 -0700)]
drivers: fix integer as NULL pointer warnings

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomedia: fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 23:50:03 +0000 (16:50 -0700)]
media: fix integer as NULL pointer warnings

drivers/media/video/v4l2-common.c:719:16: warning: Using plain integer as NULL pointer
drivers/media/video/au0828/au0828-dvb.c:122:19: warning: Using plain integer as NULL pointer
drivers/media/video/ivtv/ivtv-yuv.c:1101:22: warning: Using plain integer as NULL pointer
drivers/media/video/ivtv/ivtv-yuv.c:1102:23: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-audio.c:78:39: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-video-v4l.c:84:39: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-v4l2.c:1264:9: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-context.c:197:28: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-cx2584x-v4l.c:126:39: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-dvb.c:133:32: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-dvb.c:145:31: warning: Using plain integer as NULL pointer
drivers/media/video/pvrusb2/pvrusb2-dvb.c:177:55: warning: Using plain integer as NULL pointer
drivers/media/video/videobuf-core.c:100:9: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoscsi: fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 23:50:03 +0000 (16:50 -0700)]
scsi: fix integer as NULL pointer warnings

drivers/scsi/aic7xxx/aic7770_osm.c:53:58: warning: Using plain integer as NULL pointer
drivers/scsi/aic7xxx/aic7xxx_osm_pci.c:355:47: warning: Using plain integer as NULL pointer
drivers/scsi/aic7xxx/aic7xxx_osm_pci.c:372:55: warning: Using plain integer as NULL pointer
drivers/scsi/aha152x.c:997:28: warning: Using plain integer as NULL pointer
drivers/scsi/aha152x.c:1003:28: warning: Using plain integer as NULL pointer
drivers/scsi/aha152x.c:1165:46: warning: Using plain integer as NULL pointer
drivers/scsi/fdomain.c:1446:40: warning: Using plain integer as NULL pointer
drivers/scsi/sym53c8xx_2/sym_hipd.c:1650:51: warning: Using plain integer as NULL pointer
drivers/scsi/sym53c8xx_2/sym_hipd.c:3171:42: warning: Using plain integer as NULL pointer
drivers/scsi/sym53c8xx_2/sym_hipd.c:5732:52: warning: Using plain integer as NULL pointer
drivers/scsi/ncr53c8xx.c:8189:31: warning: Using plain integer as NULL pointer
drivers/scsi/ncr53c8xx.c:8225:34: warning: Using plain integer as NULL pointer
drivers/scsi/dpt_i2o.c:156:32: warning: Using plain integer as NULL pointer
drivers/scsi/ultrastor.c:954:42: warning: Using plain integer as NULL pointer
drivers/scsi/ultrastor.c:1104:18: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Tue, 29 Apr 2008 00:30:26 +0000 (17:30 -0700)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6: (35 commits)
  siimage: coding style cleanup (take 2)
  ide-cd: clean up cdrom_analyze_sense_data()
  ide-cd: fix test unsigned var < 0
  ide: add TSSTcorp CDDVDW SH-S202H to ivb_list[]
  piix: add Asus Eee 701 controller to short cable list
  ARM: always select HAVE_IDE
  remove the broken ETRAX_IDE driver
  ide: remove ->dma_prdtable field from ide_hwif_t
  ide: remove ->dma_vendor{1,3} fields from ide_hwif_t
  scc_pata: add ->dma_host_set and ->dma_start methods
  ide: skip "VLB sync" if host uses MMIO
  ide: add ide_pad_transfer() helper
  ide: remove ->INW and ->OUTW methods
  ide: use IDE I/O helpers directly in ide_tf_{load,read}()
  ns87415: add ->tf_read method
  scc_pata: add ->tf_{load,read} methods
  ide-h8300: add ->tf_{load,read} methods
  ide-cris: add ->tf_{load,read} methods
  ide: add ->tf_load and ->tf_read methods
  ide: move ide_tf_{load,read} to ide-iops.c
  ...

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild
Linus Torvalds [Tue, 29 Apr 2008 00:29:43 +0000 (17:29 -0700)]
Merge git://git./linux/kernel/git/sam/kbuild

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild:
  kconfig: add named choice group
  kconfig: fix choice dependency check
  kconifg: 'select' considered less evil
  dontdiff: ignore timeconst.h
  dontdiff: add modules.order
  kbuild: fix unportability in gen_initramfs_list.sh
  kbuild: fix help output to show correct arch
  kbuild: show defconfig subdirs in make help
  kconfig: reversed borderlines in inputbox

16 years agodrivers: atm, char fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 21:13:20 +0000 (14:13 -0700)]
drivers: atm, char fix integer as NULL pointer warnings

drivers/atm/nicstar.c:418:25: warning: Using plain integer as NULL pointer
drivers/char/drm/r128_cce.c:820:25: warning: Using plain integer as NULL pointer
drivers/char/tty_io.c:1183:10: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 21:13:19 +0000 (14:13 -0700)]
mm: fix integer as NULL pointer warnings

mm/hugetlb.c:207:11: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agokernel: fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 21:13:19 +0000 (14:13 -0700)]
kernel: fix integer as NULL pointer warnings

kernel/cpuset.c:1268:52: warning: Using plain integer as NULL pointer
kernel/pid_namespace.c:95:24: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Reviewed-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoinit: fix integer as NULL pointer warnings
Harvey Harrison [Mon, 28 Apr 2008 21:13:14 +0000 (14:13 -0700)]
init: fix integer as NULL pointer warnings

init/do_mounts_rd.c:215:13: warning: Using plain integer as NULL pointer
init/do_mounts_md.c:136:45: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agosiimage: coding style cleanup (take 2)
Sergei Shtylyov [Mon, 28 Apr 2008 21:44:44 +0000 (23:44 +0200)]
siimage: coding style cleanup (take 2)

Fix 18 errors and several warnings given by checkpatch.pl:

- use of C99 // comments;

- trailing whitespace;

- 'switch' and 'case' not at the same indentation level;

- no space before the open parenthesis of the 'if' and 'switch' statements;

- space between function name and open parenthesis (though I have introduced
  such warnins in some places since the code looks prettier with the spaces);

- including <asm/io.h> instead of <linux/io.h>;

- line over 80 characters.

In addition to these changes, also do the following:

- make the arrays in sil_set_pio_mode() 'static', and make the arrays in
  sil_set_dma_mode() 'static const';

- change the string of the 'if' statements into the 'switch' statement in
  sil_pata_udma_filter();

- drop the needless '==' operators from the 'if' statements where a condition
  is a mere bit test;

- remove needless initializer for the 'tmp' variable in init_chipset_siimage();

- beautify groups of the variable initializers and assignment operators;

- add new line after variable definitions;

- remove new line between the comment and the statements it refers to;

- remove needless curly braces and parentheses;

- fix typos, capitalize acronyms, etc. in the comments...

Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-cd: clean up cdrom_analyze_sense_data()
Roel Kluin [Mon, 28 Apr 2008 21:44:43 +0000 (23:44 +0200)]
ide-cd: clean up cdrom_analyze_sense_data()

[bart: fix handling of bio_sectors(failed_command->bio) == 0]

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Cc: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-cd: fix test unsigned var < 0
Roel Kluin [Mon, 28 Apr 2008 21:44:43 +0000 (23:44 +0200)]
ide-cd: fix test unsigned var < 0

valid is unsigned and cannot be below 0.

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Cc: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add TSSTcorp CDDVDW SH-S202H to ivb_list[]
Alexander Smal [Mon, 28 Apr 2008 21:44:43 +0000 (23:44 +0200)]
ide: add TSSTcorp CDDVDW SH-S202H to ivb_list[]

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agopiix: add Asus Eee 701 controller to short cable list
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:43 +0000 (23:44 +0200)]
piix: add Asus Eee 701 controller to short cable list

Based on ata_piix patch by Dan McGee.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoARM: always select HAVE_IDE
Adrian Bunk [Mon, 28 Apr 2008 21:44:43 +0000 (23:44 +0200)]
ARM: always select HAVE_IDE

It's plain wrong for PCMCIA to select HAVE_IDE that implies e.g. the
availability of an asm/ide.h

It turns out this was done for ARM, and we can simply always select
HAVE_IDE on ARM instead of manually tracking which platforms might
possible have an IDE controller directly or indirectly.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Russell King <rmk+lkml@arm.linux.org.uk>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoremove the broken ETRAX_IDE driver
Adrian Bunk [Mon, 28 Apr 2008 21:44:42 +0000 (23:44 +0200)]
remove the broken ETRAX_IDE driver

ETRAX_IDE was marked as broken last year with the comment
"it doesn't even compile currently".

Remove it since it won't get fixed in the near future.

On Mon, Apr 14, 2008 at 02:50:19PM +0200, Mikael Starvik wrote:
> You can remove it for now and we will resubmit a new if/when we get around
> to fix it.

[bart: ported it over IDE tree]

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Mikael Starvik <mikael.starvik@axis.com>
Cc: Jesper Nilsson <Jesper.Nilsson@axis.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ->dma_prdtable field from ide_hwif_t
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:42 +0000 (23:44 +0200)]
ide: remove ->dma_prdtable field from ide_hwif_t

* Use 'hwif->dma_base + {4,8}' instead of hwif->dma_prdtable in
  {ide,scc}_dma_setup().

* Remove no longer needed ->dma_prdtable field from ide_hwif_t.

While at it:

* Use ATA_DMA_TABLE_OFS define.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ->dma_vendor{1,3} fields from ide_hwif_t
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:42 +0000 (23:44 +0200)]
ide: remove ->dma_vendor{1,3} fields from ide_hwif_t

* Use 'hwif->dma_base + {1,3}' instead of hwif->dma_vendor{1,3} in
  pdc202xx_new host driver.

* Remove no longer needed ->dma_vendor{1,3} fields from ide_hwif_t.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoscc_pata: add ->dma_host_set and ->dma_start methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:41 +0000 (23:44 +0200)]
scc_pata: add ->dma_host_set and ->dma_start methods

Add ->dma_host_set and ->dma_start methods (+ __scc_dma_end() helper)
so scc_ide_{in,out}b() can be used directly.

There should be no functional changes caused by this patch.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: skip "VLB sync" if host uses MMIO
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:41 +0000 (23:44 +0200)]
ide: skip "VLB sync" if host uses MMIO

* Skip "VLB sync" in ata_{in,out}put_data() if host uses MMIO.

* Use I/O ops directly in ata_vlb_sync() an drop no longer needed
  'ide_drive_t *drive' argument.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_pad_transfer() helper
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:41 +0000 (23:44 +0200)]
ide: add ide_pad_transfer() helper

* Add ide_pad_transfer() helper (which uses ->{in,out}put_data methods
  internally so the transfer is also padded to drive+host requirements)
  and use it instead of ide_atapi_{write_zeros,discard_data}().

* Remove no longer needed ide_atapi_{write_zeros,discard_data}().

Cc: Borislav Petkov <petkovbb@gmail.com>
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ->INW and ->OUTW methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:41 +0000 (23:44 +0200)]
ide: remove ->INW and ->OUTW methods

* Remove no longer used ->INW and ->OUTW methods.

While at it:

* scc_pata.c: scc_ide_{out,in}w() is called only in scc_tf_{load,read}()
  so inline it there.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: use IDE I/O helpers directly in ide_tf_{load,read}()
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:41 +0000 (23:44 +0200)]
ide: use IDE I/O helpers directly in ide_tf_{load,read}()

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agons87415: add ->tf_read method
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:40 +0000 (23:44 +0200)]
ns87415: add ->tf_read method

Add ->tf_read method so out{b,w}(), in{b,w}() and superio_ide_inb()
can be used directly.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoscc_pata: add ->tf_{load,read} methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:40 +0000 (23:44 +0200)]
scc_pata: add ->tf_{load,read} methods

Add ->tf_{load,read} methods so scc_ide_{outb,outw,inb,inw}()
can be used directly.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-h8300: add ->tf_{load,read} methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:40 +0000 (23:44 +0200)]
ide-h8300: add ->tf_{load,read} methods

Add ->tf_{load,read} methods so outb()/inb() and mm_outw()/mm_inw()
can be used directly.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-cris: add ->tf_{load,read} methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:40 +0000 (23:44 +0200)]
ide-cris: add ->tf_{load,read} methods

Add ->tf_{load,read} methods so cris_ide_{outb,outw,inb,inw}()
can be used directly.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ->tf_load and ->tf_read methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:40 +0000 (23:44 +0200)]
ide: add ->tf_load and ->tf_read methods

* Add ->tf_load and ->tf_read methods to ide_hwif_t and set the default
  methods in default_hwif_transport().

* Use ->tf_{load,read} instead o calling ide_tf_{load,read}() directly.

* Make ide_tf_{load,read}() static.

There should be no functional changes caused by this patch.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: move ide_tf_{load,read} to ide-iops.c
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:39 +0000 (23:44 +0200)]
ide: move ide_tf_{load,read} to ide-iops.c

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: factor out debugging code from ide_tf_load()
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:39 +0000 (23:44 +0200)]
ide: factor out debugging code from ide_tf_load()

Factor out debugging code from ide_tf_load() to ide_tf_dump() helper
and update ide_tf_load() users accordingly.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add ide_execute_pkt_cmd() helper
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:39 +0000 (23:44 +0200)]
ide: add ide_execute_pkt_cmd() helper

Add ide_execute_pkt_cmd() helper for executing PACKET command,
then convert ATAPI device drivers to use it.

As a nice side-effect this fixes ide-{floppy,tape,scsi} w.r.t.
ide_lock taking (ide-cd was OK).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-{floppy,tape,scsi}: 400ns delay is required after executing the command
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:39 +0000 (23:44 +0200)]
ide-{floppy,tape,scsi}: 400ns delay is required after executing the command

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: always use ->OUTBSYNC method for executing commands
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:38 +0000 (23:44 +0200)]
ide: always use ->OUTBSYNC method for executing commands

Always use ->OUTBSYNC method for executing commands so the posting is done
if needed (this affects only pmac and scc_pata host drivers at the moment).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agosiimage: remove proc_reports_siimage()
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:38 +0000 (23:44 +0200)]
siimage: remove proc_reports_siimage()

* proc_reports_siimage() is now only called by init_chipset_siimage()
  so inline it there.

* Use array instead of switch statement for reporting clock modes.

There should be no functional changes caused by this patch.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agosiimage: add sil_* I/O ops
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:38 +0000 (23:44 +0200)]
siimage: add sil_* I/O ops

Add sil_iowrite{8,16,32}() and sil_ioread{8,16}() helpers, then use them to
merge code accessing configuration registers through PCI and MMIO together.

[ because of this SATA initialization bits from setup_mmio_siimage() are
  moved to init_chipset_siimage() ]

This also cuts code size a bit:

   text    data     bss     dec     hex filename
   4437     164       0    4601    11f9 drivers/ide/pci/siimage.o.before
   3979     164       0    4143    102f drivers/ide/pci/siimage.o.after

While at it:

* Use I/O ops directly instead of using ->IN{B,W} and ->OUT{B,W}.

* Fixup CodingStyle in setup_mmio_siimage().

* Rename 'tmpbyte' variable to 'tmp' in init_chipset_siimage().

There should be no functional changes caused by this patch.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agosiimage: do clocking register posting earlier in setup_mmio_siimage()
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:38 +0000 (23:44 +0200)]
siimage: do clocking register posting earlier in setup_mmio_siimage()

Do clocking register posting earlier in setup_mmio_siimage()
to match code in init_chipset_siimage().

This is a preparation for the next patch which merges PCI and MMIO
code paths together.

Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: remove ->INS{W,L} and ->OUTS{W,L} methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:37 +0000 (23:44 +0200)]
ide: remove ->INS{W,L} and ->OUTS{W,L} methods

* Use ins{w,l}()/outs{w,l}() and __ide_mm_ins{w,l}()/__ide_mm_outs{w,l}()
  directly in ata_{in,out}put_data() (by using IDE_HFLAG_MMIO host flag to
  decide which I/O ops are required).

* Remove no longer needed ->INS{W,L} and ->OUTS{W,L} methods (ide-h8300,
  au1xxx-ide and scc_pata implement their own ->{in,out}put_data methods).

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: add IDE_HFLAG_MMIO host flag (take 2)
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:37 +0000 (23:44 +0200)]
ide: add IDE_HFLAG_MMIO host flag (take 2)

* Add IDE_HFLAG_MMIO host flag and set it for hosts which use
  default_hwif_mmiops().

v2:
* Fix kernel panic in pmac host driver (',' should be '|').

  Thanks to Kamalesh for reporting it + testing the fix
  and to Andrew for hinting me about the source of the issue.

Cc: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide-h8300: add ->{in,out}put_data methods (take 2)
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:37 +0000 (23:44 +0200)]
ide-h8300: add ->{in,out}put_data methods (take 2)

v2:

* Update ->{in,out}_data methods to take 'struct request *rq' argument.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoau1xxx-ide: add ->{in,out}put_data methods (take 2)
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:37 +0000 (23:44 +0200)]
au1xxx-ide: add ->{in,out}put_data methods (take 2)

v2:

* Update ->{in,out}_data methods to take 'struct request *rq' argument.

There should be no functional changes caused by this patch.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoscc_pata: add ->{in,out}put_data methods (take 2)
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:36 +0000 (23:44 +0200)]
scc_pata: add ->{in,out}put_data methods (take 2)

v2:

* Update ->{in,out}_data methods to take 'struct request *rq' argument
  (thanks to Stephen Rothwell for catching it).

There should be no functional changes caused by this patch.

Cc: Kou Ishizaki <kou.ishizaki@toshiba.co.jp>
Cc: Akira Iguchi <akira2.iguchi@toshiba.co.jp>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: merge ->atapi_*put_bytes and ->ata_*put_data methods
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:36 +0000 (23:44 +0200)]
ide: merge ->atapi_*put_bytes and ->ata_*put_data methods

* Merge ->atapi_{in,out}put_bytes and ->ata_{in,out}put_data methods
  into new ->{in,out}put_data methods which take number of bytes to
  transfer as an argument and always do padding.

While at it:

* Use 'hwif' or 'drive->hwif' instead of 'HWIF(drive)'.

There should be no functional changes caused by this patch (all users
of ->ata_{in,out}put_data methods were using multiply-of-4 word counts).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agofalconide/q40ide: add ->atapi_*put_bytes and ->ata_*put_data methods (take 2)
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:36 +0000 (23:44 +0200)]
falconide/q40ide: add ->atapi_*put_bytes and ->ata_*put_data methods (take 2)

* Add ->atapi_{in,out}put_bytes and ->ata_{in,out}put_data methods to
  falconide and q40ide host drivers (->ata_* methods are implemented on
  top of ->atapi_* methods so they also do byte-swapping now).

* Cleanup atapi_{in,out}put_bytes().

v2:
* Add 'struct request *rq' argument to ->ata_{in,out}put_data methods
  and don't byte-swap disk fs requests (we shouldn't un-swap fs requests
  because fs itself is stored byte-swapped on the disk) - this is how
  things were done before the patch (ideally device mapper should be
  used instead but it would break existing setups and would have some
  performance impact).

Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Michael Schmitz <schmitz@debian.org>
Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Richard Zidlicky <rz@linux-m68k.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoide: fix au1xxx-ide breakage
Bartlomiej Zolnierkiewicz [Mon, 28 Apr 2008 21:44:35 +0000 (23:44 +0200)]
ide: fix au1xxx-ide breakage

On Monday 28 April 2008, Sergei Shtylyov wrote:
> Hello, I wrote:
>
> > Fix these warnings emitted when compiling drivers/ide/mips/au1xxx-ide.c:
>
> > include/asm/mach-au1x00/au1xxx_ide.h:137: warning: 'auide_tune_drive' declared
> > `static' but never defined
> > include/asm/mach-au1x00/au1xxx_ide.h:138: warning: 'auide_tune_chipset' declared
> >  `static' but never defined
>
> > by wiping out the whole "function prototyping" section from the header file
> > <asm-mips/mach-au1x00/au1xxx_ide.h> as it mostly declared functions that are
> > already dead in the IDE driver; move the only useful prototype into the driver.
> >
> > Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
>
> > ---
> > I'm not sure thru which tree this should go -- probably thru Linux/MIPS one...
>
> > Bart, au1xxx-ide-fix-mwdma-support.patch will probably need to be updated to
> > remove that added prototype since it won't be needed anymore...
>
>     Which you haven't done either in that patch or in
> au1xxx-ide-use-init_dma-method.patch. So, face the consequences:
>
> drivers/ide/mips/au1xxx-ide.c:456: error: conflicting types for 'auide_ddma_init'
> drivers/ide/mips/au1xxx-ide.c:51: error: previous declaration of
> 'auide_ddma_init' was here
> drivers/ide/mips/au1xxx-ide.c:456: error: conflicting types for 'auide_ddma_init'
> drivers/ide/mips/au1xxx-ide.c:51: error: previous declaration of
> 'auide_ddma_init' was here
> drivers/ide/mips/au1xxx-ide.c:51: warning: 'auide_ddma_init' used but never
> defined

Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agoptrace: conditionalize compat_ptrace_request
Roland McGrath [Mon, 28 Apr 2008 20:57:19 +0000 (13:57 -0700)]
ptrace: conditionalize compat_ptrace_request

My recent additions to compat_ptrace_request made it mandatory
for CONFIG_COMPAT arch's to define copy_siginfo_from_user32.
This broke some builds, though they all really should get cleaned
up in that way.

Since all the arch's that actually call compat_ptrace_request have
now been cleaned up to use the generic compat_sys_ptrace, we can
avoid the build problems on the crufty arch's by changing the
conditionals on the definition.

Signed-off-by: Roland McGrath <roland@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/penberg...
Linus Torvalds [Mon, 28 Apr 2008 21:08:56 +0000 (14:08 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/penberg/slab-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/slab-2.6:
  slub: pack objects denser
  slub: Calculate min_objects based on number of processors.
  slub: Drop DEFAULT_MAX_ORDER / DEFAULT_MIN_OBJECTS
  slub: Simplify any_slab_object checks
  slub: Make the order configurable for each slab cache
  slub: Drop fallback to page allocator method
  slub: Fallback to minimal order during slab page allocation
  slub: Update statistics handling for variable order slabs
  slub: Add kmem_cache_order_objects struct
  slub: for_each_object must be passed the number of objects in a slab
  slub: Store max number of objects in the page struct.
  slub: Dump list of objects not freed on kmem_cache_close()
  slub: free_list() cleanup
  slub: improve kmem_cache_destroy() error message
  slob: fix bug - when slob allocates "struct kmem_cache", it does not force alignment.

16 years agokconfig: add named choice group
Roman Zippel [Fri, 29 Feb 2008 04:11:50 +0000 (05:11 +0100)]
kconfig: add named choice group

As choice dependency are now fully checked, it's quite easy to add support
for named choices. This lifts the restriction that a choice value can only
appear once, although it still has to be within the same group,
but multiple choices can be joined by giving them a name.
While at it I cleaned up a little the choice type logic to simplify it a
bit.

Signed-off-by: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agokconfig: fix choice dependency check
Roman Zippel [Fri, 29 Feb 2008 04:10:24 +0000 (05:10 +0100)]
kconfig: fix choice dependency check

Properly check the dependency of choices as a group.
Also fix that sym_check_deps() correctly terminates the dependency loop
error check (otherwise it would continue printing the dependency chain).

Signed-off-by: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agokconifg: 'select' considered less evil
Matthew Wilcox [Sat, 19 Apr 2008 20:45:11 +0000 (14:45 -0600)]
kconifg: 'select' considered less evil

While select should be used with care, it is not actually evil.

Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agodontdiff: ignore timeconst.h
Ben Dooks [Fri, 18 Apr 2008 15:28:35 +0000 (16:28 +0100)]
dontdiff: ignore timeconst.h

Ignore the autobuilt kernel/timeconst.h when
using diff on an built kernel tree.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agodontdiff: add modules.order
Ben Dooks [Fri, 18 Apr 2008 15:18:16 +0000 (16:18 +0100)]
dontdiff: add modules.order

Add modules.order to the list of files that
shoud be ignored when using diff on a built
kernel tree.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agokbuild: fix unportability in gen_initramfs_list.sh
Felix Fietkau [Wed, 2 Apr 2008 12:50:05 +0000 (14:50 +0200)]
kbuild: fix unportability in gen_initramfs_list.sh

On a Mac OS X machine the output of ls -l is different from a standard
Linux machine. Use readlink instead of parsing a hardcoded field number
from the ls output.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agox86_64 vDSO: use initdata
Roland McGrath [Mon, 28 Apr 2008 01:45:38 +0000 (18:45 -0700)]
x86_64 vDSO: use initdata

The 64-bit vDSO image is in a special ".vdso" section for no reason
I can determine.  Furthermore, the location of the vdso_end symbol
includes some wrongly-calculated padding space in the image, which
is then (correctly) rounded to page size, resulting in an extra page
of zeros in the image mapped in to user processes.

This changes it to put the vdso.so image into normal initdata as we
have always done for the 32-bit vDSO images.  The extra padding is
gone, so the user VMA is one page instead of two.  The image that
was already copied around at boot time is now in initdata, so we
recover that wasted space after boot.

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Mon, 28 Apr 2008 20:47:21 +0000 (13:47 -0700)]
Merge git://git./linux/kernel/git/davej/cpufreq

* git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] Make acpi-cpufreq more robust against BIOS freq changes behind our back.
  [CPUFREQ] change cpu freq tables to per_cpu variables
  [CPUFREQ] fix show_trans_table
  [CPUFREQ] Warn when cpufreq_register_notifier called before pure initcalls
  [CPUFREQ] Refactor locking in cpufreq_add_dev
  [CPUFREQ] more CodingStyle
  [CPUFREQ] CodingStyle
  [CPUFREQ] Slightly shorten the error paths of cpufreq_suspend/cpufreq_resume

16 years agokbuild: fix help output to show correct arch
Andres Salomon [Sat, 26 Apr 2008 02:34:58 +0000 (22:34 -0400)]
kbuild: fix help output to show correct arch

Signed-off-by: Andres Salomon <dilinger@debian.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agomm/memory_hotplug.c must #include "internal.h"
Adrian Bunk [Mon, 28 Apr 2008 17:40:08 +0000 (20:40 +0300)]
mm/memory_hotplug.c must #include "internal.h"

This patch fixes the following compile error caused by commit
04753278769f3b6c3b79a080edb52f21d83bf6e2 ("memory hotplug: register
section/node id to free"):

    CC      mm/memory_hotplug.o
  /home/bunk/linux/kernel-2.6/git/linux-2.6/mm/memory_hotplug.c: In function ‘put_page_bootmem’:
  /home/bunk/linux/kernel-2.6/git/linux-2.6/mm/memory_hotplug.c:82: error: implicit declaration of function ‘__free_pages_bootmem’
  /home/bunk/linux/kernel-2.6/git/linux-2.6/mm/memory_hotplug.c: At top level:
  /home/bunk/linux/kernel-2.6/git/linux-2.6/mm/memory_hotplug.c:87: warning: no previous prototype for ‘register_page_bootmem_info_section’
  make[2]: *** [mm/memory_hotplug.o] Error 1

[ Andrew: "Argh.  The -mm-only memory-hotplug-add-removable-to-sysfs-
  to-show-memblock-removability.patch debugging patch adds that include
  so nobody hit this before. ]

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agokbuild: show defconfig subdirs in make help
Segher Boessenkool [Sun, 6 Apr 2008 20:16:07 +0000 (22:16 +0200)]
kbuild: show defconfig subdirs in make help

PowerPC will start moving board defconfigs into subarch-specific
subdirs soon.  "make help" currently does not look in subdirs to
find the defconfigs to show.  This is partially a good thing,
since there are way too many defconfigs for one list.

This patch makes the main "make help" display something like

  help-40x         - Show 40x-specific targets
  help-44x         - Show 44x-specific targets
  help-boards      - Show all of the above

and wires up stuff so those new help-* commands actually work.

[sam: fixed it up to display x86 defconfigs too]
Cc: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agokconfig: reversed borderlines in inputbox
Roel Kluin [Thu, 20 Mar 2008 20:30:32 +0000 (21:30 +0100)]
kconfig: reversed borderlines in inputbox

Fix reversal of dlg.border.atr and dlg.dialog.atr for draw_box()
Makes the inputbox look like expected

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
16 years agohrtimer: raise softirq unlocked to avoid circular lock dependency
Thomas Gleixner [Mon, 28 Apr 2008 07:23:24 +0000 (09:23 +0200)]
hrtimer: raise softirq unlocked to avoid circular lock dependency

The scheduler hrtimer bits in 2.6.25 introduced a circular lock
dependency in a rare code path:

=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.25-sched-devel.git-x86-latest.git #19
-------------------------------------------------------
X/2980 is trying to acquire lock:
 (&rq->rq_lock_key#2){++..}, at: [<ffffffff80230146>] task_rq_lock+0x56/0xa0

but task is already holding lock:
 (&cpu_base->lock){++..}, at: [<ffffffff80257ae1>] lock_hrtimer_base+0x31/0x60

which lock already depends on the new lock.

The scenario which leads to this is:

posix-timer signal is delivered
 -> posix-timer is rearmed
    timer is already expired in hrtimer_enqueue()
     -> softirq is raised

To prevent this we need to move the raise of the softirq out of the
base->lock protected code path.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@kernel.org
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
16 years ago[CPUFREQ] Make acpi-cpufreq more robust against BIOS freq changes behind our back.
Venkatesh Pallipadi [Mon, 28 Apr 2008 19:13:43 +0000 (15:13 -0400)]
[CPUFREQ] Make acpi-cpufreq more robust against BIOS freq changes behind our back.

We checked the hardware freq with OS cached freq value in get_cur_freqon_cpu().

Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Dave Jones <davej@redhat.com>
16 years ago[CPUFREQ] change cpu freq tables to per_cpu variables
Mike Travis [Wed, 5 Mar 2008 16:31:29 +0000 (08:31 -0800)]
[CPUFREQ] change cpu freq tables to per_cpu variables

Change cpufreq tables from arrays to per_cpu variables in
drivers/acpi/processor_thermal.c

Based on git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git

Cc: Len Brown <len.brown@intel.com>
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Dave Jones <davej@codemonkey.org.uk>
16 years ago[CPUFREQ] fix show_trans_table
Cesar Eduardo Barros [Sat, 16 Feb 2008 10:41:25 +0000 (08:41 -0200)]
[CPUFREQ] fix show_trans_table

Fix show_trans_table when it overflows PAGE_SIZE.

* Not all snprintf calls were protected against being passed a negative
length.
* When show_trans_table overflows, len might be > PAGE_SIZE. In that case,
returns PAGE_SIZE.

Signed-off-by: Cesar Eduardo Barros <cesarb@cesarb.net>
Signed-off-by: Dave Jones <davej@codemonkey.org.uk>
16 years ago[CPUFREQ] Warn when cpufreq_register_notifier called before pure initcalls
Cesar Eduardo Barros [Sat, 16 Feb 2008 10:41:24 +0000 (08:41 -0200)]
[CPUFREQ] Warn when cpufreq_register_notifier called before pure initcalls

If cpufreq_register_notifier is called before pure initcalls,
init_cpufreq_transition_notifier_list will overwrite whatever it did,
causing notifiers to be ignored.

Print some noise to the kernel log if that happens.

Signed-off-by: Cesar Eduardo Barros <cesarb@cesarb.net>
Signed-off-by: Dave Jones <davej@codemonkey.org.uk>