pandora-kernel.git
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Thu, 12 Jan 2006 01:06:31 +0000 (17:06 -0800)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

18 years ago[PATCH] new tty buffering access fix
Paul Fulghum [Wed, 11 Jan 2006 23:51:48 +0000 (17:51 -0600)]
[PATCH] new tty buffering access fix

Fix typos in new tty buffering that incorrectly
access and update buffers in pending queue.

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[SPARC64] arch/sparc64/Kconfig: fix HUGETLB_PAGE_SIZE_64K dependencies
Adrian Bunk [Wed, 11 Jan 2006 23:55:23 +0000 (15:55 -0800)]
[SPARC64] arch/sparc64/Kconfig: fix HUGETLB_PAGE_SIZE_64K dependencies

This patch fixes a typo in the dependencies of HUGETLB_PAGE_SIZE_64K.

It might be more logical to rename the HUGETLB_PAGE_SIZE_*K
dependencies to HUGETLB_PAGE_SIZE_*KB, but let's fix this bug first.

This bug was reported by Jean-Luc Leger <reiga@dspnet.fr.eu.org>.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agoFix mutex_trylock() copy-and-paste bug (x86, x86-64, generic mutex-dec.h)
Linus Torvalds [Wed, 11 Jan 2006 23:50:47 +0000 (15:50 -0800)]
Fix mutex_trylock() copy-and-paste bug (x86, x86-64, generic mutex-dec.h)

Noticed by Arjan originally on x86-64, then Ingo on x86, and finally me
grepping for it in the generic version.

Bad parenthesis nesting.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc-merge
Linus Torvalds [Wed, 11 Jan 2006 16:16:57 +0000 (08:16 -0800)]
Merge git://git./linux/kernel/git/paulus/powerpc-merge

18 years ago[PATCH] fix/simplify mutex debugging code
David Woodhouse [Wed, 11 Jan 2006 14:41:26 +0000 (14:41 +0000)]
[PATCH] fix/simplify mutex debugging code

Let's switch mutex_debug_check_no_locks_freed() to take (addr, len) as
arguments instead, since all its callers were just calculating the 'to'
address for themselves anyway... (and sometimes doing so badly).

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] powerpc: Don't build crash.c for PPC32
Michael Ellerman [Wed, 11 Jan 2006 05:41:29 +0000 (16:41 +1100)]
[PATCH] powerpc: Don't build crash.c for PPC32

arch/powerpc/kernel/crash.c isn't safe for PPC32 (yet?), so don't build it.

Built with CONFIG_KEXEC=y for pmac32_defconfig, pseries_defconfig,
and g5_defconfig.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[MUTEX]: linux/mutex.h needs linux/linkage.h too
David S. Miller [Wed, 11 Jan 2006 08:15:16 +0000 (00:15 -0800)]
[MUTEX]: linux/mutex.h needs linux/linkage.h too

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/kyle/parisc-2.6
Linus Torvalds [Wed, 11 Jan 2006 05:31:24 +0000 (21:31 -0800)]
Merge /pub/scm/linux/kernel/git/kyle/parisc-2.6

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Wed, 11 Jan 2006 05:29:53 +0000 (21:29 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

18 years ago[PATCH] powerpc: Fix clean_files in arch/powerpc/boot Makefile
Kumar Gala [Wed, 11 Jan 2006 04:36:37 +0000 (22:36 -0600)]
[PATCH] powerpc: Fix clean_files in arch/powerpc/boot Makefile

clean-files was being set twice rather than being appended to.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agopowerpc/32: Fix compile error caused by pud_t/pgt_t confusion
Paul Mackerras [Wed, 11 Jan 2006 05:27:21 +0000 (16:27 +1100)]
powerpc/32: Fix compile error caused by pud_t/pgt_t confusion

PPC32 is still using asm-generic/4level-fixup.h, but asm-powerpc/page.h
was defining pud_t and pgd_t.  Depending on the order in which files
got included, this could result in a compilation error.  Tweak the ifdef
so that page.h doesn't try to define pud_t on ppc32 (which uses 2-level
page tables).

Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agox86: fix "make install" target
Linus Torvalds [Wed, 11 Jan 2006 05:09:19 +0000 (21:09 -0800)]
x86: fix "make install" target

Removing the dependency on the boot image build was good, but it also
meant that the $< expansion by make needed to be done explicitly.

Noted by Stephen Hemminger.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[IPV6]: Fix modular build with netfilter enabled.
David S. Miller [Wed, 11 Jan 2006 05:02:21 +0000 (21:02 -0800)]
[IPV6]: Fix modular build with netfilter enabled.

Also, drop __exit marker from ipv6_netfilter_fini() as this
can be invoked from inet6_init() error handling paths.

Based upon a report from Stephen Hemminger.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PATCH] powerpc: Remove redundant setting of htab_address
Michael Ellerman [Wed, 11 Jan 2006 04:27:25 +0000 (15:27 +1100)]
[PATCH] powerpc: Remove redundant setting of htab_address

iSeries doesn't need to set the htab_address explicitly, htab_initialize()
will do it for us later.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: Add ibm,pft-size to iSeries device tree
Michael Ellerman [Wed, 11 Jan 2006 04:27:24 +0000 (15:27 +1100)]
[PATCH] powerpc: Add ibm,pft-size to iSeries device tree

To make iSeries just a teensy bit less special, create ibm,pft-size properties
in the iSeries device tree. We can then rely on htab_dt_scan_pftsize() to set
ppc64_pft_size for us.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: Add some missing .gitignore's
Kumar Gala [Wed, 11 Jan 2006 04:12:23 +0000 (22:12 -0600)]
[PATCH] powerpc: Add some missing .gitignore's

ignore generated files under arch/powerpc

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: Updated Kconfig and Makefiles for 83xx support
Kumar Gala [Wed, 11 Jan 2006 03:43:56 +0000 (21:43 -0600)]
[PATCH] powerpc: Updated Kconfig and Makefiles for 83xx support

Updated Kconfig & Makefiles in prep for adding support for the Freescale
MPC83xx family of processors to arch/powerpc.  Moved around some config
options that are more globally applicable to other PowerPC processors.
Added a temporary config option (83xx) to match existing arch/ppc support
for the MPC83xx line.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] spufs: Suspend/restore MFC DMA operations at SPU context switch.
Geoff Levand [Wed, 11 Jan 2006 03:47:37 +0000 (19:47 -0800)]
[PATCH] spufs: Suspend/restore MFC DMA operations at SPU context switch.

The SPE Book IV indicates that MFC DMA operations must be
suspended and restored on SPU context switch (in Step 8).

This patch adds that operation, which is missing from the
current spufs implementation.

Signed-off-by: Masato Noguchi <Masato.Noguchi@jp.sony.com>
Signed-off-by: Geoff Levand <geoffrey.levand@am.sony.com>
Acked-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: fix for compile problem in kdump code when SMP disabled
Haren Myneni [Wed, 11 Jan 2006 03:25:25 +0000 (19:25 -0800)]
[PATCH] powerpc: fix for compile problem in kdump code when SMP disabled

This patch fixes the compilation error (shown below) when CONFIG_SMP=n.
    arch/powerpc/kernel/crash.c: In function `crash_kexec_prepare_cpus':
    arch/powerpc/kernel/crash.c:236: error: implicit declaration of
    function `smp_release_cpus'

Signed-off-by: Haren Myneni <haren@us.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agopowerpc: Fix compile error when CONFIG_PROC_VMCORE is not defined
Paul Mackerras [Wed, 11 Jan 2006 04:30:07 +0000 (15:30 +1100)]
powerpc: Fix compile error when CONFIG_PROC_VMCORE is not defined

We were getting elfcorehdr_addr undefined in this case.

Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: fix up iSeries console after TTY layer buffering revamp
Stephen Rothwell [Wed, 11 Jan 2006 02:51:22 +0000 (13:51 +1100)]
[PATCH] powerpc: fix up iSeries console after TTY layer buffering revamp

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc/64: per cpu data optimisations
Anton Blanchard [Wed, 11 Jan 2006 02:16:44 +0000 (13:16 +1100)]
[PATCH] powerpc/64: per cpu data optimisations

The current ppc64 per cpu data implementation is quite slow. eg:

        lhz 11,18(13)           /* smp_processor_id() */
        ld 9,.LC63-.LCTOC1(30)  /* per_cpu__variable_name */
        ld 8,.LC61-.LCTOC1(30)  /* __per_cpu_offset */
        sldi 11,11,3            /* form index into __per_cpu_offset */
        mr 10,9
        ldx 9,11,8              /* __per_cpu_offset[smp_processor_id()] */
        ldx 0,10,9              /* load per cpu data */

5 loads for something that is supposed to be fast, pretty awful. One
reason for the large number of loads is that we have to synthesize 2
64bit constants (per_cpu__variable_name and __per_cpu_offset).

By putting __per_cpu_offset into the paca we can avoid the 2 loads
associated with it:

        ld 11,56(13)            /* paca->data_offset */
        ld 9,.LC59-.LCTOC1(30)  /* per_cpu__variable_name */
        ldx 0,9,11              /* load per cpu data

Longer term we can should be able to do even better than 3 loads.
If per_cpu__variable_name wasnt a 64bit constant and paca->data_offset
was in a register we could cut it down to one load. A suggestion from
Rusty is to use gcc's __thread extension here. In order to do this we
would need to free up r13 (the __thread register and where the paca
currently is). So far Ive had a few unsuccessful attempts at doing that :)

The patch also allocates per cpu memory node local on NUMA machines.
This patch from Rusty has been sitting in my queue _forever_ but stalled
when I hit the compiler bug. Sorry about that.

Finally I also only allocate per cpu data for possible cpus, which comes
straight out of the x86-64 port. On a pseries kernel (with NR_CPUS == 128)
and 4 possible cpus we see some nice gains:

             total       used       free     shared    buffers cached
Mem:       4012228     212860    3799368          0          0 162424

             total       used       free     shared    buffers cached
Mem:       4016200     212984    3803216          0          0 162424

A saving of 3.75MB. Quite nice for smaller machines. Note: we now have
to be careful of per cpu users that touch data for !possible cpus.

At this stage it might be worth making the NUMA and possible cpu
optimisations generic, but per cpu init is done so early we have to be
careful that all architectures have their possible map setup correctly.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: parallel port init fix
Michael Neuling [Wed, 11 Jan 2006 01:08:56 +0000 (12:08 +1100)]
[PATCH] powerpc: parallel port init fix

This stops parport from accessing nonexistent parallel ports.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: Make early debugging configurable via Kconfig
Michael Ellerman [Wed, 11 Jan 2006 00:54:09 +0000 (11:54 +1100)]
[PATCH] powerpc: Make early debugging configurable via Kconfig

This patch adds Kconfig entries to control the early debugging options,
currently in setup_64.c.

Doing this via Kconfig rather than #defines means you can have one source tree,
which is buildable for multiple platforms - and you can enable the correct
early debug option for each platform via .config.

I made udbg_early_init() a static inline because otherwise GCC is to daft to
optimise it away when debugging is off.

Now that we have udbg_init_rtas() we can make call_rtas_display_status* static.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: Early debugging support for iSeries
Michael Ellerman [Wed, 11 Jan 2006 00:54:08 +0000 (11:54 +1100)]
[PATCH] powerpc: Early debugging support for iSeries

Connect iSeries up to the standard early debugging infrastructure.

To actually use this you need to enable the iSeries early debugging
in setup_64.c. Then after the messages are logged hit Ctrl-x Ctrl-x on
your console to dump the Hypervisor console buffer.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] powerpc: remove remaining crash_notes variable from machine_kexec.c
Olaf Hering [Tue, 10 Jan 2006 21:31:53 +0000 (22:31 +0100)]
[PATCH] powerpc: remove remaining crash_notes variable from machine_kexec.c

remove remaining crash_notes definition to fix compile error

/dev/shm/linux-2.6/arch/powerpc/kernel/machine_kexec.c:21: error: conflicting types for `crash_notes'
/dev/shm/linux-2.6/include/linux/kexec.h:129: error: previous declaration of `crash_notes'

Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] PCI Hotplug/powerpc: module build break
linas [Tue, 10 Jan 2006 21:18:16 +0000 (15:18 -0600)]
[PATCH] PCI Hotplug/powerpc: module build break

The RPAPHP hoplug driver will not build as a module, because it calls
on a pcibios routine which is not exported. This exports the symbol.
Problem reported by Olaf Hering <olh@suse.de>

Signed-off-by: Linas Vepstas <linas@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years ago[PATCH] enable the RTC driver in ppc64_defconfig
Olaf Hering [Tue, 10 Jan 2006 20:57:40 +0000 (21:57 +0100)]
[PATCH] enable the RTC driver in ppc64_defconfig

Enable the RTC driver.

Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agoFix net/core/wireless.c link failure
Linus Torvalds [Wed, 11 Jan 2006 03:35:19 +0000 (19:35 -0800)]
Fix net/core/wireless.c link failure

It needs <linux/etherdevice.h> for compare_ether_addr()

18 years ago[PARISC] Use STABS_DEBUG macro from vmlinux.lds.h
Matthew Wilcox [Wed, 11 Jan 2006 01:48:07 +0000 (20:48 -0500)]
[PARISC] Use STABS_DEBUG macro from vmlinux.lds.h

Cleanup vmlinux.lds.S by using STABS_DEBUG macro from vmlinux.lds.h
instead of repeating the sections.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Fix Dino reporting on J2240
Matthew Wilcox [Wed, 11 Jan 2006 01:48:06 +0000 (20:48 -0500)]
[PARISC] Fix Dino reporting on J2240

Fix Dino reporting on J2240. This particular machine thought it
had a Cujo. Also add J2240 Dino chip to the hp_hardware_list.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] stifb: Remove obsolete MODULE_PARM()
Helge Deller [Wed, 11 Jan 2006 01:48:05 +0000 (20:48 -0500)]
[PARISC] stifb: Remove obsolete MODULE_PARM()

The bpp module parameter has been obsoleted in favour of
a setup string, so remove the MODULE_PARM.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] stifb: Fix framebuffer console at 32bpp
Helge Deller [Wed, 11 Jan 2006 01:48:04 +0000 (20:48 -0500)]
[PARISC] stifb: Fix framebuffer console at 32bpp

Fix stifb framebuffer console at 32bpp on a HCRX-24 card
by properly setting DIRECTCOLOR. Also a few nice cleanups
to the code.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] stifb: use F_EXTEND macro
Helge Deller [Wed, 11 Jan 2006 01:48:03 +0000 (20:48 -0500)]
[PARISC] stifb: use F_EXTEND macro

Use the F_EXTEND() macro instead of open coding it with an
#ifdef. Provides a nice cleanup.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Use C99 initializers in asm-parisc/processor.h
Alexey Dobriyan [Wed, 11 Jan 2006 01:48:02 +0000 (20:48 -0500)]
[PARISC] Use C99 initializers in asm-parisc/processor.h

Cleanup asm-parisc/processor.h to use C99 initializers in
INIT_THREAD().

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] pdc_stable: More robust sysfs error checking
Thibaut VARENE [Wed, 11 Jan 2006 01:48:01 +0000 (20:48 -0500)]
[PARISC] pdc_stable: More robust sysfs error checking

pdc_stable 0.10:
As mentioned on LKML, pdc_stable wasn't checky enough on the return
values of some calls. This patch makes it more robust to errors when
registering objects in sysfs.

Signed-off-by: Thibaut VARENE <varenet@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Make PCI_HOST_ADDR and PCI_BUS_ADDR symmetrical
Grant Grundler [Wed, 11 Jan 2006 01:48:00 +0000 (20:48 -0500)]
[PARISC] Make PCI_HOST_ADDR and PCI_BUS_ADDR symmetrical

Change to asm-parisc/pci.h makes the define of PCI_HOST_ADDR symmetrical
with PCI_BUS_ADDR. Also add a comment about PA_VIEW and LMMIO/ELMMIO/GMMIO.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Fix parport_gsc by selecting PARPORT_NOT_PC
Guy Martin [Wed, 11 Jan 2006 01:47:59 +0000 (20:47 -0500)]
[PARISC] Fix parport_gsc by selecting PARPORT_NOT_PC

PARPORT_GSC requires selecting PARPORT_NOT_PC in order to
work properly.

Signed-off-by: Guy Martin <gmsoft@tuxicoman.be>
Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] OSS: Fix build of parisc harmony driver
Stuart Brady [Wed, 11 Jan 2006 01:47:58 +0000 (20:47 -0500)]
[PARISC] OSS: Fix build of parisc harmony driver

OSS Harmony got missed on the conversion of parisc_device.hpa to a
struct resource... fix its build.

Signed-off-by: Stuart Brady <sdb@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Fix BLK_BOUNCE_HIGH on parisc by initializing max_low_pfn
Grant Grundler [Wed, 11 Jan 2006 01:47:57 +0000 (20:47 -0500)]
[PARISC] Fix BLK_BOUNCE_HIGH on parisc by initializing max_low_pfn

max_low_pfn was not being set in arch/parisc/mm/init.c, causing severe
problems whenever anything tried to use BLK_BOUNCE_HIGH. Set it to
max_pfn like other similar architectures do.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Truncate overlapping PAT PDC reported ranges
Grant Grundler [Wed, 11 Jan 2006 01:47:56 +0000 (20:47 -0500)]
[PARISC] Truncate overlapping PAT PDC reported ranges

Deal with overlapping LBA MMIO resources,

rp3440 PDC BUG: PDC reports lmmio range for the last rope that overlaps
with the CPU HPA. Console output was:

...
Found devices:
1. Storm Peak Fast at 0xfffffffffe798000 [152] { 0, 0x0, 0x889, 0x00004 }
2. Storm Peak Fast at 0xfffffffffe799000 [153] { 0, 0x0, 0x889, 0x00004 }
...
FAILED: lba_fixup_bus() request for lmmio_space
[fffffffff0000000/fffffffffecffffe]

Output is now:

LBA: Truncating lmmio_space [fffffffff0000000/fffffffffecffffe] to
[fffffffff0000000,fffffffffe797fff]

My only concern with this patch is how C8000 (PAT PDC) will report
elmmio ranges when a gfx card is installed. I'll have to test this
another day.

Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: James Bottomley <jejb@parisc-linux.org>
Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Add __iomem to __raw_check_addr()
Alexey Dobriyan [Wed, 11 Jan 2006 01:47:56 +0000 (20:47 -0500)]
[PARISC] Add __iomem to __raw_check_addr()

Add __iomem to __raw_check_addr(), which nukes ~13809 sparse
warnings on allmodconfig.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Add some defines for HugeTLB pages
Helge Deller [Wed, 11 Jan 2006 01:47:55 +0000 (20:47 -0500)]
[PARISC] Add some defines for HugeTLB pages

Define some constants for HugeTLB pages, not that parisc-linux supports
it yet.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Introduce DINO_LOCAL_IRQS and use it for gsc_find_local_irq
Helge Deller [Wed, 11 Jan 2006 01:47:54 +0000 (20:47 -0500)]
[PARISC] Introduce DINO_LOCAL_IRQS and use it for gsc_find_local_irq

Fix dino by using DINO_LOCAL_IRQS as the limit for gsc_find_local_irq()
instead of the irq itself.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Fix GSC graphics cards with 64MB regions
Matthew Wilcox [Wed, 11 Jan 2006 01:47:53 +0000 (20:47 -0500)]
[PARISC] Fix GSC graphics cards with 64MB regions

Make knapps work with its 64MB gfx card.  I probably just broke another
machine in the process, but assuming 64MB when 64MB aligned is probably
safer than assuming 32MB all the time.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Fix and cleanup ioremap.c to work with 4level-fixup.h
Kyle McMartin [Wed, 11 Jan 2006 01:47:52 +0000 (20:47 -0500)]
[PARISC] Fix and cleanup ioremap.c to work with 4level-fixup.h

Fixup ioremap a bit. It seems to work on 32-bit kernels, but fails
miserably on the first ioremapped access on 64-bit kernels. Also, having
STI enabled causes it to fail. Probably because we're passing an ioremapped
region to a real-mode STI call...

Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Add __user annotation to eisa_eeprom.c
Alexey Dobriyan [Wed, 11 Jan 2006 01:47:51 +0000 (20:47 -0500)]
[PARISC] Add __user annotation to eisa_eeprom.c

Annotate eisa_eeprom_read() with __user.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Fix Cirrus 6832 Cardbus on RDI Tadpole PARISC Laptop
Helge Deller [Wed, 11 Jan 2006 01:47:50 +0000 (20:47 -0500)]
[PARISC] Fix Cirrus 6832 Cardbus on RDI Tadpole PARISC Laptop

Fix irq-off-by-one for Cirrus 6832 Cardbus on RDI Tadpole PARISC Laptop.
We just DECLARE_PCI_FIXUP_ENABLE as it is unlikely that this will be
found in any other parisc system.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Make local cache flushes take a void *
Matthew Wilcox [Wed, 11 Jan 2006 01:47:49 +0000 (20:47 -0500)]
[PARISC] Make local cache flushes take a void *

Make flush_data_cache_local, flush_instruction_cache_local and
flush_tlb_all_local take a void * so they don't have to be cast
when using on_each_cpu().  This becomes a problem when on_each_cpu
is a macro (as it is in current -mm).

Also move the prototype of flush_tlb_all_local into tlbflush.h and
remove its declaration from .c files.

Signed-off-by: Matthew Wilcox <willy@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years ago[PARISC] Add __read_mostly section for parisc
Helge Deller [Wed, 11 Jan 2006 01:35:03 +0000 (20:35 -0500)]
[PARISC] Add __read_mostly section for parisc

Flag a whole bunch of things as __read_mostly on parisc. Also flag a few
branches as unlikely() and cleanup a bit of code.

Signed-off-by: Helge Deller <deller@parisc-linux.org>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
18 years agofs/proc/vmcore.c: header included twice
Nicolas Kaiser [Wed, 11 Jan 2006 01:10:28 +0000 (02:10 +0100)]
fs/proc/vmcore.c: header included twice

Header included twice.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agofs/attr.c: header included twice
Nicolas Kaiser [Wed, 11 Jan 2006 01:08:57 +0000 (02:08 +0100)]
fs/attr.c: header included twice

Header included twice.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoasm-powerpc: header included twice
Nicolas Kaiser [Wed, 11 Jan 2006 01:07:34 +0000 (02:07 +0100)]
asm-powerpc: header included twice

Header included twice.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoxfs: header included twice
Nicolas Kaiser [Wed, 11 Jan 2006 01:06:20 +0000 (02:06 +0100)]
xfs: header included twice

Header included twice.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agonetfilter: headers included twice
Nicolas Kaiser [Wed, 11 Jan 2006 01:04:35 +0000 (02:04 +0100)]
netfilter: headers included twice

Headers included twice.

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agocpuset two little doc fixes
Paul Jackson [Wed, 11 Jan 2006 01:01:38 +0000 (02:01 +0100)]
cpuset two little doc fixes

Two little cpuset documentation fixes.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agodrivers/net/gianfar_sysfs.c: update email address of Kumar Gala
Adrian Bunk [Wed, 11 Jan 2006 01:00:10 +0000 (02:00 +0100)]
drivers/net/gianfar_sysfs.c: update email address of Kumar Gala

drivers/net/gianfar_sysfs.c still contained the old email address
of Kumar Gala.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoadd loglevel to printk in fs/afs/cmservice.c
Jesper Juhl [Wed, 11 Jan 2006 00:52:40 +0000 (01:52 +0100)]
add loglevel to printk in fs/afs/cmservice.c

This is a small patch that adds loglevel to a printk in
fs/afs/cmservice.c

Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agomissing printk loglevel and tiny tiny whitespace change in binfmt_elf()
Jesper Juhl [Wed, 11 Jan 2006 00:51:26 +0000 (01:51 +0100)]
missing printk loglevel and tiny tiny whitespace change in binfmt_elf()

Patch adds a mising printk loglevel (I think KERN_WARNING is appropriate
here) in fs/binfmt_elf.c, and while I was there I made some tiny tiny tiny
adjustments to whitespacing in the neighborhood.

Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoadd missing printk loglevel in mm/swapfile.c
Jesper Juhl [Wed, 11 Jan 2006 00:50:28 +0000 (01:50 +0100)]
add missing printk loglevel in mm/swapfile.c

in mm/swapfile.c a printk() is missing a loglevel. I believe the proper
loglevel for this situation is KERN_ERR, so that's what the patch below
sets -if you agree, please apply.

Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoTiny esthetic changes to Documentation/laptop-mode.txt
Jesper Juhl [Wed, 11 Jan 2006 00:48:08 +0000 (01:48 +0100)]
Tiny esthetic changes to Documentation/laptop-mode.txt

Here's a tiny patch making a few esthetic changes to
Documentation/laptop-mode.txt
To me this patch makes sense, but feel free to disagree, I don't feel
strongly about it at all.

It changes a single URL to its strictly correct form (directories should
end in /), and it makes the arguments to main in an included example
program follow convention and be named argc and argv.

Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoMAINTAINERS: sh: update the mailing list
Adrian Bunk [Wed, 11 Jan 2006 00:43:01 +0000 (01:43 +0100)]
MAINTAINERS: sh: update the mailing list

Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years agoext2: trivial indentation fix.
Luiz Fernando Capitulino [Wed, 11 Jan 2006 00:38:27 +0000 (01:38 +0100)]
ext2: trivial indentation fix.

This memset() line was indented with seven spaces, this patch fixes
it to use a tab instead. Yes, very trivial but it's the third time
I have to look at this line..

Signed-off-by: Luiz Capitulino <lcapitulino@mandriva.com.br>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
18 years ago[PATCH] mutex: trivial whitespace cleanups
Ingo Molnar [Tue, 10 Jan 2006 22:15:02 +0000 (23:15 +0100)]
[PATCH] mutex: trivial whitespace cleanups

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 10 Jan 2006 21:44:43 +0000 (13:44 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

18 years ago[PATCH] mark mutex_lock*() as might_sleep()
Ingo Molnar [Tue, 10 Jan 2006 21:10:36 +0000 (22:10 +0100)]
[PATCH] mark mutex_lock*() as might_sleep()

Mark mutex_lock() and mutex_lock_interruptible() as might_sleep()
functions.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fix i386 mutex fastpath on FRAME_POINTER && !DEBUG_MUTEXES
Ingo Molnar [Tue, 10 Jan 2006 21:07:44 +0000 (22:07 +0100)]
[PATCH] fix i386 mutex fastpath on FRAME_POINTER && !DEBUG_MUTEXES

Call the mutex slowpath more conservatively - e.g.  FRAME_POINTERS can
change the calling convention, in which case a direct branch to the
slowpath becomes illegal.  Bug found by Hugh Dickins.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] remove unnecessary asm/mutex.h from kernel/mutex-debug.c
Ingo Molnar [Tue, 10 Jan 2006 20:48:59 +0000 (15:48 -0500)]
[PATCH] remove unnecessary asm/mutex.h from kernel/mutex-debug.c

Remove unnecessary (and incorrect) inclusion of asm/mutex.h, pointed out
by David Howells.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[EBTABLES] Don't match tcp/udp source/destination port for IP fragments
Bart De Schuymer [Tue, 10 Jan 2006 21:12:22 +0000 (13:12 -0800)]
[EBTABLES] Don't match tcp/udp source/destination port for IP fragments

Signed-off-by: Bart De Schuymer <bdschuym@pandora.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IRDA] DONGLE_OLD: remove dependency on non-existing symbol
Adrian Bunk [Tue, 10 Jan 2006 21:11:44 +0000 (13:11 -0800)]
[IRDA] DONGLE_OLD: remove dependency on non-existing symbol

Jean-Luc Leger <reiga@dspnet.fr.eu.org> reported this alternative
dependency on a non-existing symbol.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IRDA]: kill drivers/net/irda/sir_core.c
Adrian Bunk [Tue, 10 Jan 2006 21:10:02 +0000 (13:10 -0800)]
[IRDA]: kill drivers/net/irda/sir_core.c

EXPORT_SYMBOL's do nowadays belong to the files where the actual
functions are.

Moving the module_init/module_exit to the file with the actual functions
has the advantage of saving a few bytes due to the removal of two
functions.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Jean Tourrilhes <jt@hpl.hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM]: Remove unneeded kmalloc() return value casts + tiny whitespace cleanup
Jesper Juhl [Tue, 10 Jan 2006 21:09:01 +0000 (13:09 -0800)]
[ATM]: Remove unneeded kmalloc() return value casts + tiny whitespace cleanup

Small cleanups for drivers/atm/zatm.c
 Get rid of unneeded cast of kmalloc() return value.
 Small whitespace/CodingStyle/formatting cleanup (since I was in there anyway).

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: Remove unneeded kmalloc() return value casts
Jesper Juhl [Tue, 10 Jan 2006 21:08:21 +0000 (13:08 -0800)]
[NET]: Remove unneeded kmalloc() return value casts

Get rid of needless casting of kmalloc() return value in net/

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[RXRPC]: Decrease number of pointer derefs in connection.c
Jesper Juhl [Tue, 10 Jan 2006 21:07:44 +0000 (13:07 -0800)]
[RXRPC]: Decrease number of pointer derefs in connection.c

Decrease the number of pointer derefs in net/rxrpc/connection.c

Benefits of the patch:
 - Fewer pointer dereferences should make the code slightly faster.
 - Size of generated code is smaller
 - improved readability

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[AF_NETLINK]: Fix DoS in netlink_rcv_skb()
Martin Murray [Tue, 10 Jan 2006 21:02:29 +0000 (13:02 -0800)]
[AF_NETLINK]: Fix DoS in netlink_rcv_skb()

From: Martin Murray <murrayma@citi.umich.edu>

Sanity check nlmsg_len during netlink_rcv_skb.  An nlmsg_len == 0 can
cause infinite loop in kernel, effectively DoSing machine.  Noted by
Matin Murray.

Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: Fix timeout sysctls on big-endian 64bit architectures
Patrick McHardy [Tue, 10 Jan 2006 01:48:09 +0000 (17:48 -0800)]
[NETFILTER]: Fix timeout sysctls on big-endian 64bit architectures

The connection tracking timeout variables are unsigned long, but
proc_dointvec_jiffies is used with sizeof(unsigned int) in the sysctl
tables. Since there is no proc_doulongvec_jiffies function, change the
timeout variables to unsigned int.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: Remove unused function from NAT protocol helpers
Patrick McHardy [Tue, 10 Jan 2006 00:44:36 +0000 (16:44 -0800)]
[NETFILTER]: Remove unused function from NAT protocol helpers

->print and ->print_range are not used (and apparently never were).

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: Fix return value confusion in PPTP NAT helper
Patrick McHardy [Tue, 10 Jan 2006 00:44:18 +0000 (16:44 -0800)]
[NETFILTER]: Fix return value confusion in PPTP NAT helper

ip_nat_mangle_tcp_packet doesn't return NF_* values but 0/1 for
failure/success.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: Fix another crash in ip_nat_pptp
Patrick McHardy [Tue, 10 Jan 2006 00:44:00 +0000 (16:44 -0800)]
[NETFILTER]: Fix another crash in ip_nat_pptp

The PPTP NAT helper calculates the offset at which the packet needs
to be mangled as difference between two pointers to the header. With
non-linear skbs however the pointers may point to two seperate buffers
on the stack and the calculation results in a wrong offset beeing
used.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: Fix crash in ip_nat_pptp
Patrick McHardy [Tue, 10 Jan 2006 00:43:43 +0000 (16:43 -0800)]
[NETFILTER]: Fix crash in ip_nat_pptp

When an inbound PPTP_IN_CALL_REQUEST packet is received the
PPTP NAT helper uses a NULL pointer in pointer arithmentic to
calculate the offset in the packet which needs to be mangled
and corrupts random memory or crashes.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: net/ipv[46]/netfilter.c cleanups
Patrick McHardy [Tue, 10 Jan 2006 00:43:13 +0000 (16:43 -0800)]
[NETFILTER]: net/ipv[46]/netfilter.c cleanups

Don't wrap entire file in #ifdef CONFIG_NETFILTER, remove a few
unneccessary includes.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: Change memcmp(,,ETH_ALEN) to compare_ether_addr()
Kris Katterjohn [Tue, 10 Jan 2006 00:01:43 +0000 (16:01 -0800)]
[NET]: Change memcmp(,,ETH_ALEN) to compare_ether_addr()

This changes some memcmp(one,two,ETH_ALEN) to compare_ether_addr(one,two).

Signed-off-by: Kris Katterjohn <kjak@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[INET]: congestion and af_ops can be const
Stephen Hemminger [Mon, 9 Jan 2006 23:56:09 +0000 (15:56 -0800)]
[INET]: congestion and af_ops can be const

The congestion ops and af_ops in the inet_connection_sock
can be const.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PATCH] moxa serial: add proper capability check
Alan Cox [Mon, 9 Jan 2006 14:35:28 +0000 (09:35 -0500)]
[PATCH] moxa serial: add proper capability check

This requires the proper capabilities for the moxa bios update ioctl's.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: save reg a5 on context change
Greg Ungerer [Tue, 10 Jan 2006 07:02:57 +0000 (17:02 +1000)]
[PATCH] m68knommu: save reg a5 on context change

Fix a5 register corruption when processing user space signals handlers.
We need to save a5 through each contenxt change.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: fix find_next_zero_bit in bitops.h
Greg Ungerer [Tue, 10 Jan 2006 07:00:39 +0000 (17:00 +1000)]
[PATCH] m68knommu: fix find_next_zero_bit in bitops.h

We're starting a number of big applications (memory footprint app.
1MByte) on our Arcturus uC5272.  Therefore memory fragmentation is a
real pain for us.  We've switched to uClinux-2.4.27-uc1 and found that
page_alloc2 fragments the memory heavily.

Digging into it we found a bug in the find_next_zero_bit function in the
m68knommu/bitops.h file.  if the size isn't a multiple of 32 than the
upper bits of the last word to be searched should be masked.  But the
functions masks the lower bits of the last word because it uses a right
shift instead of a left shift operator.

Patch submitted by Sascha Smejkal <s.smejkal@centersystems.at>

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] uclinux: delay binfmt_flat trace
Greg Ungerer [Tue, 10 Jan 2006 06:59:37 +0000 (16:59 +1000)]
[PATCH] uclinux: delay binfmt_flat trace

Modify the initial trace output (which is based on flags in the binary
header) so that it is not done until after the magic number check.  This
may well not be a flat format binary, so the flags could be invalid.
(Prime example, running a script).

Changes prompted by patches from Stuart Hughs.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: set irq priority/level different for each ColdFire serial port
Greg Ungerer [Tue, 10 Jan 2006 06:59:04 +0000 (16:59 +1000)]
[PATCH] m68knommu: set irq priority/level different for each ColdFire serial port

Set the hardware interrupt priority to a different value for each
attached ColdFire serial port.  According to the CPU documentation you
should not use the same combination of level/priority on more than one
device.  People have reported odd serial port behavior with them set the
same.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: allow configure of FEC for M520x CPU family
Greg Ungerer [Tue, 10 Jan 2006 06:58:16 +0000 (16:58 +1000)]
[PATCH] m68knommu: allow configure of FEC for M520x CPU family

Allow the ColdFire FEC ethernet driver to be enabled on the M520x CPU
family.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: fix ram length of m5208evb board
Greg Ungerer [Tue, 10 Jan 2006 06:43:39 +0000 (16:43 +1000)]
[PATCH] m68knommu: fix ram length of m5208evb board

Adjust length of M5208EVB ram define.  It should size up to 32MB after
adding in the dBUG reserved 128k.

Problem pointed out be Milton Miller <miltonm@bga.com>.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: fix a5 reg corruption in signal handlers
Greg Ungerer [Tue, 10 Jan 2006 06:42:59 +0000 (16:42 +1000)]
[PATCH] m68knommu: fix a5 reg corruption in signal handlers

This is a patch adapted from a posting by Andrea Tarani which was
pointed out to me by Bernardo Innocenti.  Thanks to both of them for
their help and patience.

The original posting is here:
  http://mailman.uclinux.org/pipermail/uclinux-dev/2005-July/033543.html

The problem first manifest itself as busybox ping terminating with an
"Illegal instruction".  I reduced this to a test case and found that
variable size arrays allocated on the stack could lead to stacks not
aligned on 32 bit boundaries.  For the Coldfire this proved fatal.

Having been pointed out this patch by Bernardo, I applied it and it
fixed the first test case.  I then went back to busybox's ping.  This
still failed with "Illegal instruction", but in a different way.  Before
it depended on the size allocated for the ping buffer, now it happened
every time.  I also found it depended on optimisation level (gcc-3.4.0)
-Os was okay but not -O2.

After a lot of looking, it turned out that register a5 was being
corrupted by the signal handler (after applying the patch).  I re-worked
the patch a bit to save/restore a5 and now all seems well.

Patch submitted by Stuart Hughs <stuarth@freescale.com>

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: fix mangled 'truct' in ptrace.c
Greg Ungerer [Tue, 10 Jan 2006 06:42:18 +0000 (16:42 +1000)]
[PATCH] m68knommu: fix mangled 'truct' in ptrace.c

Fix broken "truct" -> "struct" in arch_ptrace() parameter list.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] m68knommu: don't set gcc optimizer flags
Greg Ungerer [Tue, 10 Jan 2006 06:40:21 +0000 (16:40 +1000)]
[PATCH] m68knommu: don't set gcc optimizer flags

Don't specify compiler optimization flags in the m68knommu Makefile.
Let the top level Makefile/config set it.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] h8300: remove MAGIC_ROM_PTR from memory.c
Greg Ungerer [Tue, 10 Jan 2006 06:39:35 +0000 (16:39 +1000)]
[PATCH] h8300: remove MAGIC_ROM_PTR from memory.c

Remove obsolete MAGIC_ROM_PTR code from h8300 architecture.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] h8300: remove MAGIC_ROM_PTR from k8300_ksyms.c
Greg Ungerer [Tue, 10 Jan 2006 06:39:30 +0000 (16:39 +1000)]
[PATCH] h8300: remove MAGIC_ROM_PTR from k8300_ksyms.c

Remove obsolete MAGIC_ROM_PTR code from h8300 architecture.

Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 10 Jan 2006 17:00:55 +0000 (09:00 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

18 years agoFix rpc shutdown event condition bug
Linus Torvalds [Tue, 10 Jan 2006 16:56:39 +0000 (08:56 -0800)]
Fix rpc shutdown event condition bug

We want to wait for the cl_users to go down to zero, not for it to stay
positive.  Quoth Trond (who wasn't even the author, but acked the wrong
version): "Argh! I need to increase my daily caffeine dosages."

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] rcu: fix hotplug-cpu ->donelist leak
Oleg Nesterov [Tue, 10 Jan 2006 14:24:53 +0000 (17:24 +0300)]
[PATCH] rcu: fix hotplug-cpu ->donelist leak

Pointed out by Srivatsa Vaddagiri <vatsa@in.ibm.com>.

rcu_do_batch() stops after processing maxbatch callbacks
on ->donelist leaving rcu_tasklet in TASKLET_STATE_SCHED
state.

If CPU_DEAD event happens remaining ->donelist entries are
lost, rcu_offline_cpu() kills this tasklet.

With this patch ->donelist migrates along with ->curlist
and ->nxtlist to the current cpu.

Compile tested.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Paul E. McKenney <paulmck@us.ibm.com>
Cc: Srivatsa Vaddagiri <vatsa@in.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] rcu: join rcu_ctrlblk and rcu_state
Oleg Nesterov [Tue, 10 Jan 2006 13:48:02 +0000 (16:48 +0300)]
[PATCH] rcu: join rcu_ctrlblk and rcu_state

This patch moves rcu_state into the rcu_ctrlblk. I think there
are no reasons why we should have 2 different variables to control
rcu state. Every user of rcu_state has also "rcu_ctrlblk *rcp" in
the parameter list.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Paul E. McKenney <paulmck@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>