linux-2.6
18 years ago[PATCH] htirq: refactor so we only have one function that writes to the chip
Eric W. Biederman [Thu, 9 Nov 2006 01:44:57 +0000 (17:44 -0800)] 
[PATCH] htirq: refactor so we only have one function that writes to the chip

This refactoring actually optimizes the code a little by caching the value
that we think the device is programmed with instead of reading it back from
the hardware.  Which simplifies the code a little and should speed things up a
bit.

This patch introduces the concept of a ht_irq_msg and modifies the
architecture read/write routines to update this code.

There is a minor consistency fix here as well as x86_64 forgot to initialize
the htirq as masked.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Andi Kleen <ak@suse.de>
Acked-by: Bryan O'Sullivan <bos@pathscale.com>
Cc: <olson@pathscale.com>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ipmi_si_intf.c: fix "&& 0xff" typos
Alexey Dobriyan [Thu, 9 Nov 2006 01:44:56 +0000 (17:44 -0800)] 
[PATCH] ipmi_si_intf.c: fix "&& 0xff" typos

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Corey Minyard <minyard@acm.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] IPMI: retry messages on certain error returns
Corey Minyard [Thu, 9 Nov 2006 01:44:55 +0000 (17:44 -0800)] 
[PATCH] IPMI: retry messages on certain error returns

Some more errors from the IPMI send message command are retryable, but are not
being retried by the IPMI code.  Make sure they get retried.

Signed-off-by: Corey Minyard <minyard@acm.org>
Cc: Frederic Lelievre <Frederic.Lelievre@ca.kontron.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] IPMI: Clean up the waiting message queue properly on unload
Corey Minyard [Thu, 9 Nov 2006 01:44:52 +0000 (17:44 -0800)] 
[PATCH] IPMI: Clean up the waiting message queue properly on unload

A wrong function was being used to free a list; this fixes the problem.
Otherwise, an oops at unload time was possible.  But not likely, since you
can't have any users when you unload the modules and it is very hard to get
messages into this queue without users.

Signed-off-by: Corey Minyard <minyard@acm.org>
Cc: Patrick Schoeller <Patrick.Schoeller@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] sysctl: Undeprecate sys_sysctl
Eric W. Biederman [Thu, 9 Nov 2006 01:44:51 +0000 (17:44 -0800)] 
[PATCH] sysctl: Undeprecate sys_sysctl

The basic issue is that despite have been deprecated and warned about as a
very bad thing in the man pages since its inception there are a few real
users of sys_sysctl.  It was my assumption that because sysctl had been
deprecated for all of 2.6 there would be no user space users by this point,
so I initially gave sys_sysctl a very short deprecation period.

Now that I know there are a few real users the only sane way to proceed
with deprecation is to push the time limit out to a year or two work and
work with distributions that have big testing pools like fedora core to
find these last remaining users.

Which means that the sys_sysctl interface needs to be maintained in the
meantime.

Since I have provided a technical measure that allows us to add new sysctl
entries without reserving more binary numbers I believe that is enough to
fix the sys_sysctl binary interface maintenance problems, because there is
no longer a need to change the binary interface at all.

Since the sys_sysctl implementation needs to stay around for a while and
the worst of the maintenance issues that caused us to occasionally break
the ABI have been addressed I don't see any advantage in continuing with
the removal of sys_sysctl.

So instead of merely increasing the deprecation period this patch removes
the deprecation of sys_sysctl and modifies the kernel to compile the code
in by default.

With committing to maintain sys_sysctl we get all of the advantages of a
fast interface for anything that needs it.  Currently sys_sysctl is about
5x faster than /proc/sys, for the same string data.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ia64: select ACPI_NUMA if ACPI
KAMEZAWA Hiroyuki [Thu, 9 Nov 2006 01:44:50 +0000 (17:44 -0800)] 
[PATCH] ia64: select ACPI_NUMA if ACPI

When ACPI && NUMA, pxm_to_node is used and it exists in drivers/acpi/numa.c

Tony said:

  The patch makes sense ...  if you pick both of "ACPI" and "NUMA", then you
  need (and should automatically be given) ACPI_NUMA too.

  The only open question is whether there is a better way of getting there.
  Perhaps with less configuration options in the first place?  We are heading
  towards a future where so many systems will be NUMA that there would seem to
  be little benefit in keeping ACPI_NUMA separate from ACPI ...  but perhaps
  we aren't quite there yet.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujtisu.com>
Cc: Len Brown <lenb@kernel.org>
Acked-by: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] kretprobe: fix kretprobe-booster to save regs and set status
Masami Hiramatsu [Thu, 9 Nov 2006 01:44:49 +0000 (17:44 -0800)] 
[PATCH] kretprobe: fix kretprobe-booster to save regs and set status

There are two bugs in the kretprobe-booster.

1) It doesn't make room for gs registers.

2) It doesn't change status of the current kprobe.  This status will
   effect the fault handling.

This patch fixes these bugs and, additionally, saves skipped registers for
compatibility with the original kretprobe.

Signed-off-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: do not freeze md threads for suspend
Rafael J. Wysocki [Thu, 9 Nov 2006 01:44:48 +0000 (17:44 -0800)] 
[PATCH] md: do not freeze md threads for suspend

If there's a swap file on a software RAID, it should be possible to use this
file for saving the swsusp's suspend image.  Also, this file should be
available to the memory management subsystem when memory is being freed before
the suspend image is created.

For the above reasons it seems that md_threads should not be frozen during the
suspend and the appended patch makes this happen, but then there is the
question if they don't cause any data to be written to disks after the suspend
image has been created, provided that all filesystems are frozen at that time.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: fix sizing problem with raid5-reshape and CONFIG_LBD=n
NeilBrown [Thu, 9 Nov 2006 01:44:48 +0000 (17:44 -0800)] 
[PATCH] md: fix sizing problem with raid5-reshape and CONFIG_LBD=n

I forgot to has the size-in-blocks to (loff_t) before shifting up to a
size-in-bytes.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: change ONLINE/OFFLINE events to a single CHANGE event
NeilBrown [Thu, 9 Nov 2006 01:44:47 +0000 (17:44 -0800)] 
[PATCH] md: change ONLINE/OFFLINE events to a single CHANGE event

It turns out that CHANGE is preferred to ONLINE/OFFLINE for various reasons
(not least of which being that udev understands it already).

So remove the recently added KOBJ_OFFLINE (no-one is likely to care anyway)
and change the ONLINE to a CHANGE event

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Neil Brown <neilb@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Tigran has moved
Tigran Aivazian [Thu, 9 Nov 2006 01:44:46 +0000 (17:44 -0800)] 
[PATCH] Tigran has moved

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] drivers/telephony/ixj: fix an array overrun
Adrian Bunk [Thu, 9 Nov 2006 01:44:45 +0000 (17:44 -0800)] 
[PATCH] drivers/telephony/ixj: fix an array overrun

The Coverity checker noted that in
drivers/telephony/ixj.c:ixj_build_filter_cadence(), filter_en[4] or
filter_en[5] could be written to.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: raid1: fix waiting for io on suspend
Jonathan E Brassow [Thu, 9 Nov 2006 01:44:44 +0000 (17:44 -0800)] 
[PATCH] dm: raid1: fix waiting for io on suspend

All device-mapper targets must complete outstanding I/O before suspending.
The mirror target generates I/O in its recovery phase and fails to wait for
it.  It needs to be tracked so we can ensure that it has completed before we
suspend.

[akpm@osdl.org: cleanup]
Signed-off-by: Jonathan E Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: <dm-devel@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: multipath: fix rr_add_path order
Jonathan E Brassow [Thu, 9 Nov 2006 01:44:43 +0000 (17:44 -0800)] 
[PATCH] dm: multipath: fix rr_add_path order

When adding paths to the round-robin path selector, their order gets inverted,
which is not desirable.

Fix by replacing list_add() with list_add_tail().

Signed-off-by: Jonathan E Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: <dm-devel@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: suspend: fix error path
Alasdair G Kergon [Thu, 9 Nov 2006 01:44:43 +0000 (17:44 -0800)] 
[PATCH] dm: suspend: fix error path

If the device is already suspended, just return the error and skip the code
that would incorrectly wipe md->suspended_bdev.

(This isn't currently a problem because existing code avoids calling this
function if the device is already suspended.)

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: <dm-devel@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: fix find_device race
Alasdair G Kergon [Thu, 9 Nov 2006 01:44:42 +0000 (17:44 -0800)] 
[PATCH] dm: fix find_device race

There is a race between dev_create() and find_device().

If the mdptr has not yet been stored against a device, find_device() needs to
behave as though no device was found.  It already returns NULL, but there is a
dm_put() missing: it must drop the reference dm_get_md() took.

The bug was introduced by dm-fix-mapped-device-ref-counting.patch.

It manifests itself if another dm ioctl attempts to reference a newly-created
device while the device creation ioctl is still running.  The consequence is
that the device cannot be removed until the machine is rebooted.  Certain udev
configurations can lead to this happening.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: <dm-devel@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] i386: Force data segment to be 4K aligned
Vivek Goyal [Thu, 9 Nov 2006 01:44:41 +0000 (17:44 -0800)] 
[PATCH] i386: Force data segment to be 4K aligned

o Currently there is no specific alignment restriction in linker script
  and in some cases it can be placed non 4K aligned addresses. This fails
  kexec which checks that segment to be loaded is page aligned.

o I guess, it does not harm data segment to be 4K aligned.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] nfsd4: fix open-create permissions
J. Bruce Fields [Thu, 9 Nov 2006 01:44:40 +0000 (17:44 -0800)] 
[PATCH] nfsd4: fix open-create permissions

In the case where an open creates the file, we shouldn't be rechecking
permissions to open the file; the open succeeds regardless of what the new
file's mode bits say.

This patch fixes the problem, but only by introducing yet another parameter
to nfsd_create_v3.  This is ugly.  This will be fixed by later patches.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Acked-by: Neil Brown <neilb@suse.de>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] nfsd4: reindent do_open_lookup()
J. Bruce Fields [Thu, 9 Nov 2006 01:44:39 +0000 (17:44 -0800)] 
[PATCH] nfsd4: reindent do_open_lookup()

Minor rearrangement, cleanup of do_open_lookup().  No change in behavior.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Acked-by: Neil Brown <neilb@suse.de>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] A minor fix for set_mb() in Documentation/memory-barriers.txt
Oleg Nesterov [Thu, 9 Nov 2006 01:44:38 +0000 (17:44 -0800)] 
[PATCH] A minor fix for set_mb() in Documentation/memory-barriers.txt

set_mb() is used by set_current_state() which needs mb(), not wmb().  I
think it would be right to assume that set_mb() implies mb(), all arches
seem to do just this.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Regression in 2.6.19-rc microcode driver
Arjan van de Ven [Thu, 9 Nov 2006 01:44:37 +0000 (17:44 -0800)] 
[PATCH] Regression in 2.6.19-rc microcode driver

If the microcode driver is built in (rather than module) there are some,
ehm, interesting effects happening due to the new "call out to userspace"
behavior that is introduced..  and which runs too early.  The result is a
boot hang; which is really nasty.

The patch below is a minimally safe patch to fix this regression for 2.6.19
by just not requesting actual microcode updates during early boot.  (That
is a good idea in general anyway)

The "real" fix is a lot more complex given the entire cpu hotplug scenario
(during cpu hotplug you normally need to load the microcode as well); but
the interactions for that are just really messy at this point; this fix at
least makes it work and avoids a full detangle of hotplug.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge merom:v2.6/linux
Linus Torvalds [Wed, 8 Nov 2006 18:45:37 +0000 (10:45 -0800)] 
Merge merom:v2.6/linux

* merom:v2.6/linux:
  x86-64: write IO APIC irq routing entries in correct order
  x86-64: clean up io-apic accesses

18 years agox86-64: write IO APIC irq routing entries in correct order
Linus Torvalds [Wed, 8 Nov 2006 18:27:54 +0000 (10:27 -0800)] 
x86-64: write IO APIC irq routing entries in correct order

This is the x86-64 version of f9dadfa71bc594df09044da61d1c72701121d802
that did the same thing on i386.

Since the "mask" bit is in the low word, when we write a new entry, we
need to write the high word first, before we potentially unmask it.

The exception is when we actually want to mask the interrupt, in which
case we want to write the low word first to make sure that the high word
doesn't change while the interrupt routing is still active.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agox86-64: clean up io-apic accesses
Linus Torvalds [Wed, 8 Nov 2006 18:23:03 +0000 (10:23 -0800)] 
x86-64: clean up io-apic accesses

This is just commit 130fe05dbc0114609cfef9815c0c5580b42decfa ported to
x86-64, for all the same reasons.  It cleans up the IO-APIC accesses in
order to then fix the ordering issues.

We move the accessor functions (that were only used by io_apic.c) out of
a header file, and use proper memory-mapped accesses rather than making
up our own "volatile" pointers.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoRevert "[PATCH] i386: Add MMCFG resources to i386 too"
Linus Torvalds [Wed, 8 Nov 2006 18:09:28 +0000 (10:09 -0800)] 
Revert "[PATCH] i386: Add MMCFG resources to i386 too"

This reverts commit de09bddb9d6f96785be470c832b881e6d72d589f.  It tried
to reserve the MMCONFIG mmio memory ranges, but since the MMCONFIG
information is broken and often bogus (which is why we don't dare use it
most of the time _anyway_), it does more harm than good.

Cc: Jeff Chua <jeff.chua.linux@gmail.com>
Cc: Adrian Bunk <bunk@stusta.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Wed, 8 Nov 2006 16:05:59 +0000 (08:05 -0800)] 
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [libata] sata_via: fix obvious typo

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Wed, 8 Nov 2006 15:34:11 +0000 (07:34 -0800)] 
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [DECNET]: Endianess fixes (try #2)
  [TG3]: Fix array overrun in tg3_read_partno().
  [NET]: Set truesize in pskb_copy
  [NETPOLL]: Compute checksum properly in netpoll_send_udp().
  [PKT_SCHED] sch_htb: Use hlist_del_init().
  [TCP]: Don't use highmem in tcp hash size calculation.
  [NET]: kconfig, correct traffic shaper

18 years ago[libata] sata_via: fix obvious typo
Jeff Garzik [Wed, 8 Nov 2006 12:46:02 +0000 (07:46 -0500)] 
[libata] sata_via: fix obvious typo

Spotted by Martin Devera.

Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years agoLinux 2.6.19-rc5 v2.6.19-rc5
Linus Torvalds [Wed, 8 Nov 2006 02:24:20 +0000 (18:24 -0800)] 
Linux 2.6.19-rc5

Ok, things are clearly starting to calm down.. Finally.

18 years agoMerge branch 'fixes_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbgla...
Linus Torvalds [Wed, 8 Nov 2006 02:22:31 +0000 (18:22 -0800)] 
Merge branch 'fixes_for_linus' of git://git./linux/kernel/git/jbglaw/vax-linux

* 'fixes_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbglaw/vax-linux:
  Update for the srm_env driver.

18 years ago[DECNET]: Endianess fixes (try #2)
Steven Whitehouse [Tue, 7 Nov 2006 23:09:17 +0000 (15:09 -0800)] 
[DECNET]: Endianess fixes (try #2)

Here are some fixes to endianess problems spotted by Al Viro.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: Fix array overrun in tg3_read_partno().
Michael Chan [Tue, 7 Nov 2006 22:57:51 +0000 (14:57 -0800)] 
[TG3]: Fix array overrun in tg3_read_partno().

Use proper upper limits for the loops and check for all error
conditions.

The problem was noticed by Adrian Bunk.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: Set truesize in pskb_copy
Herbert Xu [Tue, 7 Nov 2006 22:57:15 +0000 (14:57 -0800)] 
[NET]: Set truesize in pskb_copy

Since pskb_copy tacks on the non-linear bits from the original
skb, it needs to count them in the truesize field of the new skb.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETPOLL]: Compute checksum properly in netpoll_send_udp().
Chris Lalancette [Tue, 7 Nov 2006 22:56:19 +0000 (14:56 -0800)] 
[NETPOLL]: Compute checksum properly in netpoll_send_udp().

Signed-off-by: Chris Lalancette <clalance@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PKT_SCHED] sch_htb: Use hlist_del_init().
Stephen Hemminger [Tue, 7 Nov 2006 22:54:46 +0000 (14:54 -0800)] 
[PKT_SCHED] sch_htb: Use hlist_del_init().

Otherwise we can hit paths that (legally) do multiple deletes on the
same node and OOPS with the HLIST poison values there instead of
NULL.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TCP]: Don't use highmem in tcp hash size calculation.
John Heffner [Tue, 7 Nov 2006 07:10:51 +0000 (23:10 -0800)] 
[TCP]: Don't use highmem in tcp hash size calculation.

This patch removes consideration of high memory when determining TCP
hash table sizes.  Taking into account high memory results in tcp_mem
values that are too large.

Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: kconfig, correct traffic shaper
Jiri Slaby [Mon, 6 Nov 2006 22:34:48 +0000 (14:34 -0800)] 
[NET]: kconfig, correct traffic shaper

As Patrick McHardy <kaber@trash.net> suggested, Traffic Shaper is now
obsolete and alternative to it is no longer CBQ, since its problems with
virtual devices, alter Kconfig text to reflect this -- put a link to the
traffic schedulers as a whole.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Tue, 7 Nov 2006 22:55:40 +0000 (14:55 -0800)] 
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 3927/1: Allow show_mem() to work with holes in memory map.
  [ARM] 3926/1: make timer led handle HZ != 100
  [ARM] 3923/1: S3C24XX: update s3c2410_defconfig with new drivers
  [ARM] 3922/1: S3C24XX: update s3c2410_defconfig to 2.6.19-rc4
  [ARM] 3921/1: S3C24XX: remove bast_defconfig
  [ARM] 3920/1: S3C24XX: Remove smdk2410_defconfig
  [ARM] 3919/1: Fixed definition of some PXA270 CIF related registers
  [ARM] 3918/1: ixp4xx irq-chip rework
  [ARM] 3912/1: Make PXA270 advertise HWCAP_IWMMXT capability
  [ARM] 3915/1: S3C2412: Add s3c2410_gpio_getirq() to general gpio.c
  [ARM] 3917/1: Fix dmabounce symbol exports

18 years agoUpdate for the srm_env driver.
Jan-Benedict Glaw [Tue, 7 Nov 2006 22:50:37 +0000 (23:50 +0100)] 
Update for the srm_env driver.

This patch contains a fix for a bug introduced more than a year ago
(not setting *eof) and updates whitespace a bit.

Signed-off-by: Jan-Benedict Glaw <jbglaw@lug-owl.de>
18 years ago[ARM] 3927/1: Allow show_mem() to work with holes in memory map.
Ray Lehtiniemi [Tue, 7 Nov 2006 02:19:15 +0000 (03:19 +0100)] 
[ARM] 3927/1: Allow show_mem() to work with holes in memory map.

show_mem() was not correctly handling holes in the memory
map.  It was treating the freed sections of the map as
though they contained valid struct page entries.  This
could cause incorrect debugging output or even a kernel
panic.

This patch keeps the struct meminfo around after system
initialization so that show_mem() can use it when
scanning memory.  show_mem() now walks over each bank
of each online node, rather than assuming that each node
contains a single contiguous bank.

Signed-off-by: Ray Lehtiniemi <rayl@mail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
18 years ago[ARM] 3926/1: make timer led handle HZ != 100
David Brownell [Mon, 6 Nov 2006 18:29:16 +0000 (19:29 +0100)] 
[ARM] 3926/1: make timer led handle HZ != 100

The timer LED is unusable at HZ=large, since it's got
a hard-wired value of 100 ticks per cycle; when HZ=1024
(for example) it's essentially always-on.  This patch
just makes that be HZ ticks per cycle.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
18 years ago[PATCH] e1000: Fix regression: garbled stats and irq allocation during swsusp
Auke Kok [Mon, 6 Nov 2006 16:57:12 +0000 (08:57 -0800)] 
[PATCH] e1000: Fix regression: garbled stats and irq allocation during swsusp

e1000: Fix suspend/resume powerup and irq allocation

From: Auke Kok <auke-jan.h.kok@intel.com>

After 7.0.33/2.6.16, e1000 suspend/resume left the user with an enabled
device showing garbled statistics and undetermined irq allocation state,
where `ifconfig eth0 down` would display `trying to free already freed irq`.

Explicitly free and allocate irq as well as powerup the PHY during resume
fixes when needed.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years ago[PATCH] b44: change comment about irq mask register
Johannes Berg [Mon, 6 Nov 2006 22:17:20 +0000 (23:17 +0100)] 
[PATCH] b44: change comment about irq mask register

Through some experimentation with the similarly built bcm43xx I came to
the conclusion that if the hw/firmware sets a bit in the interrupt
register, an interrupt will only be raised if that bit is included in
the interrupt mask. Hence, the interrupt mask is more like an interrupt
control mask.

This patch changes the comment to reflect that.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 7 Nov 2006 03:53:12 +0000 (19:53 -0800)] 
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] Fix EV64120 and Ocelot builds by providing a plat_timer_setup().
  [MIPS] EV64120: Fix PCI interrupt allocation.
  [MIPS] Make irq number allocator generally available for fixing EV64120.
  [MIPS] EV64120: Fix timer initialization for HZ != 100.
  [MIPS] Ocelot 3: Fix MAC address detection after platform_device conversion.
  [MIPS] Ocelot C: Fix MAC address detection after platform_device conversion.
  [MIPS] SB1: On bootup only flush cache on local CPU.
  [MIPS] Ocelot 3: Fix large number of warnings.
  [MIPS] Ocelot C: Fix mapping of ioport address range.
  [MIPS] Ocelot C: Fix warning about missmatching format string.
  [MIPS] Ocelot C: fix eth registration after conversion to platform_device
  [MIPS] Ocelot C: Fix large number of warnings.

18 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/v4l-dvb
Linus Torvalds [Tue, 7 Nov 2006 03:52:29 +0000 (19:52 -0800)] 
Merge branch 'master' of git://git./linux/kernel/git/mchehab/v4l-dvb

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/v4l-dvb:
  V4L/DVB (4751): Fix DBV_FE_CUSTOMISE for card drivers compiled into kernel
  V4L/DVB (4784): [saa7146_i2c] short_delay mode fixed for fast machines
  V4L/DVB (4770): Fix mode switch of Compro Videomate T300
  V4L/DVB (4787): Budget-ci: Inversion setting fixed for Technotrend 1500 T
  V4L/DVB (4786): Pvrusb2: use NULL instead of 0
  V4L/DVB (4785): Budget-ci: Change DEBIADDR_IR to a safer default
  V4L/DVB (4752): DVB: Add DVB_FE_CUSTOMISE support for MT2060

18 years ago[MIPS] Fix EV64120 and Ocelot builds by providing a plat_timer_setup().
Ralf Baechle [Mon, 6 Nov 2006 18:17:35 +0000 (18:17 +0000)] 
[MIPS] Fix EV64120 and Ocelot builds by providing a plat_timer_setup().

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] EV64120: Fix PCI interrupt allocation.
Ralf Baechle [Mon, 6 Nov 2006 18:05:08 +0000 (18:05 +0000)] 
[MIPS] EV64120: Fix PCI interrupt allocation.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Make irq number allocator generally available for fixing EV64120.
Ralf Baechle [Mon, 6 Nov 2006 17:41:06 +0000 (17:41 +0000)] 
[MIPS] Make irq number allocator generally available for fixing EV64120.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] EV64120: Fix timer initialization for HZ != 100.
Ralf Baechle [Mon, 6 Nov 2006 13:32:39 +0000 (13:32 +0000)] 
[MIPS] EV64120: Fix timer initialization for HZ != 100.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot 3: Fix MAC address detection after platform_device conversion.
Ralf Baechle [Sun, 5 Nov 2006 01:18:43 +0000 (01:18 +0000)] 
[MIPS] Ocelot 3: Fix MAC address detection after platform_device conversion.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot C: Fix MAC address detection after platform_device conversion.
Ralf Baechle [Sun, 5 Nov 2006 01:18:11 +0000 (01:18 +0000)] 
[MIPS] Ocelot C: Fix MAC address detection after platform_device conversion.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] SB1: On bootup only flush cache on local CPU.
Ralf Baechle [Sat, 4 Nov 2006 23:26:27 +0000 (23:26 +0000)] 
[MIPS] SB1: On bootup only flush cache on local CPU.

This fixes a warning on bootup warning in smp_call_function.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot 3: Fix large number of warnings.
Ralf Baechle [Sat, 4 Nov 2006 13:02:46 +0000 (13:02 +0000)] 
[MIPS] Ocelot 3: Fix large number of warnings.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot C: Fix mapping of ioport address range.
Ralf Baechle [Fri, 3 Nov 2006 18:06:33 +0000 (18:06 +0000)] 
[MIPS] Ocelot C: Fix mapping of ioport address range.

 o Fix warnings
 o 768MB worth of I/O ports were insane
 o 64-bit kernels don't need special handling because ioremap does the magic

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot C: Fix warning about missmatching format string.
Ralf Baechle [Fri, 3 Nov 2006 17:48:17 +0000 (17:48 +0000)] 
[MIPS] Ocelot C: Fix warning about missmatching format string.

  CC      arch/mips/momentum/ocelot_c/setup.o
arch/mips/momentum/ocelot_c/setup.c: In function 'momenco_time_init':
arch/mips/momentum/ocelot_c/setup.c:223: warning: format '%d' expects type 'int', but argument 2 has type 'long unsigned int'

Change data type to match format string; a 32-bit type better suits our
needs.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot C: fix eth registration after conversion to platform_device
Ralf Baechle [Fri, 3 Nov 2006 17:45:25 +0000 (17:45 +0000)] 
[MIPS] Ocelot C: fix eth registration after conversion to platform_device

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Ocelot C: Fix large number of warnings.
Ralf Baechle [Fri, 3 Nov 2006 17:42:47 +0000 (17:42 +0000)] 
[MIPS] Ocelot C: Fix large number of warnings.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes
Linus Torvalds [Mon, 6 Nov 2006 17:10:25 +0000 (09:10 -0800)] 
Merge git://git./linux/kernel/git/steve/gfs2-2.6-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes:
  [DLM] fix oops in kref_put when removing a lockspace
  [DLM] Fix kref_put oops
  [GFS2] Fix OOM error handling
  [GFS2] Fix incorrect fs sync behaviour.
  [GFS2] don't panic needlessly

18 years agoMerge branch 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32
Linus Torvalds [Mon, 6 Nov 2006 17:07:19 +0000 (09:07 -0800)] 
Merge branch 'for-linus' of git://atmel.no/~hskinnemoen/linux/kernel/avr32

* 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32:
  AVR32: Add missing return instruction in __raw_writesb
  AVR32: Wire up sys_epoll_pwait
  AVR32: Fix thinko in generic_find_next_zero_le_bit()
  AVR32: Get rid of board_early_init

18 years agoMerge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Mon, 6 Nov 2006 17:06:53 +0000 (09:06 -0800)] 
Merge branch 'for-linus' of git://git390.osdl.marist.edu/linux-2.6

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] IRQs too early enabled.
  [S390] revert add_active_range() usage patch.

18 years ago[DLM] fix oops in kref_put when removing a lockspace
Patrick Caulfield [Mon, 6 Nov 2006 08:53:28 +0000 (08:53 +0000)] 
[DLM] fix oops in kref_put when removing a lockspace

Now that the lockspace struct is freed when the last sysfs object is released
this patch prevents use of that lockspace by sysfs. We attempt to re-get the
lockspace from the lockspace list and fail the request if it has been removed.

Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
18 years ago[DLM] Fix kref_put oops
Patrick Caulfield [Thu, 2 Nov 2006 14:41:23 +0000 (14:41 +0000)] 
[DLM] Fix kref_put oops

This patch fixes the recounting on the lockspace kobject. Previously the lockspace was freed while userspace could have had a
reference to one of its sysfs files, causing an oops in kref_put.

Now the lockspace kfree is moved into the kobject release() function

Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
18 years ago[GFS2] Fix OOM error handling
Steven Whitehouse [Mon, 30 Oct 2006 21:59:08 +0000 (16:59 -0500)] 
[GFS2] Fix OOM error handling

Fix the OOM error handling in inode.c where it was possible for
a NULL pointer to be dereferenced.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
18 years ago[GFS2] Fix incorrect fs sync behaviour.
Steven Whitehouse [Wed, 1 Nov 2006 14:57:57 +0000 (09:57 -0500)] 
[GFS2] Fix incorrect fs sync behaviour.

This adds a sync_fs superblock operation for GFS2 and removes
the journal flush from write_super in favour of sync_fs where it
ought to be. This is more or less identical to the way in which ext3
does this.

This bug was pointed out by Russell Cattelan <cattelan@redhat.com>

Cc: Russell Cattelan <cattelan@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
18 years ago[GFS2] don't panic needlessly
Alexey Dobriyan [Fri, 27 Oct 2006 23:03:48 +0000 (03:03 +0400)] 
[GFS2] don't panic needlessly

First, SLAB_PANIC is unjustified. Second, all error propagating and backing out
is in place.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
18 years agoAVR32: Add missing return instruction in __raw_writesb
Haavard Skinnemoen [Mon, 6 Nov 2006 13:06:45 +0000 (14:06 +0100)] 
AVR32: Add missing return instruction in __raw_writesb

__raw_writesb ends with a conditional branch, which is obviously
wrong. It should return after the last loop terminates.

Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
18 years agoAVR32: Wire up sys_epoll_pwait
Haavard Skinnemoen [Mon, 6 Nov 2006 13:02:44 +0000 (14:02 +0100)] 
AVR32: Wire up sys_epoll_pwait

Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
18 years ago[S390] IRQs too early enabled.
Heiko Carstens [Mon, 6 Nov 2006 09:49:02 +0000 (10:49 +0100)] 
[S390] IRQs too early enabled.

setup_lowcore() calls ctl_set_bit() which returns withs interrupts
enabled. The setup arch code is not supposed to enable interrupts that
early. Therefore use the __ctl_set_bit() variant.
This fixes the not working lock dependency validator on non 64 bit
systems.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
18 years ago[S390] revert add_active_range() usage patch.
Heiko Carstens [Mon, 6 Nov 2006 09:49:00 +0000 (10:49 +0100)] 
[S390] revert add_active_range() usage patch.

Commit 7676bef9c183fd573822cac9992927ef596d584c breaks DCSS support on
s390. DCSS needs initialized struct pages to work. With the usage of
add_active_range() only the struct pages for physically present pages
are initialized.
This could be fixed if the DCSS driver would initiliaze the struct pages
itself, but this doesn't work too. This is because the mem_map array
does not include holes after the last present memory area and therefore
there is nothing that could be initialized.
To fix this and to avoid some dirty hacks revert this patch for now.
Will be added later when we move to a virtual mem_map.

Cc: Carsten Otte <cotte@de.ibm.com>
Cc: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
18 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Mon, 6 Nov 2006 09:46:43 +0000 (01:46 -0800)] 
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
  [PATCH] Kconfig: remove redundant NETDEVICES depends
  [PATCH] ehea: 64K page support fix
  [PATCH] ehea: Removed redundant define
  [PATCH] ehea: Nullpointer dereferencation fix
  [PATCH] bcm43xx: fix unexpected LED control values in BCM4303 sprom
  [PATCH] bcm43xx: Fix low-traffic netdev watchdog TX timeouts
  [PATCH] hostap_plx: fix CIS verification
  [PATCH] ieee80211: don't flood log with errors

18 years ago[PATCH] sunrpc: add missing spin_unlock
Akinobu Mita [Mon, 6 Nov 2006 07:52:13 +0000 (23:52 -0800)] 
[PATCH] sunrpc: add missing spin_unlock

auth_domain_put() forgot to unlock acquired spinlock.

Cc: Olaf Kirch <okir@monad.swb.de>
Cc: Andy Adamson <andros@citi.umich.edu>
Cc: J. Bruce Fields <bfields@citi.umich.edu>
Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Neil Brown <neilb@suse.de>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] sysctl: implement CTL_UNNUMBERED
Eric W. Biederman [Mon, 6 Nov 2006 07:52:13 +0000 (23:52 -0800)] 
[PATCH] sysctl: implement CTL_UNNUMBERED

This patch takes the CTL_UNNUMBERD concept from NFS and makes it available to
all new sysctl users.

At the same time the sysctl binary interface maintenance documentation is
updated to mention and to describe what is needed to successfully maintain the
sysctl binary interface.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] sysctl: allow a zero ctl_name in the middle of a sysctl table
Eric W. Biederman [Mon, 6 Nov 2006 07:52:12 +0000 (23:52 -0800)] 
[PATCH] sysctl: allow a zero ctl_name in the middle of a sysctl table

Since it is becoming clear that there are just enough users of the binary
sysctl interface that completely removing the binary interface from the kernel
will not be an option for foreseeable future, we need to find a way to address
the sysctl maintenance issues.

The basic problem is that sysctl requires one central authority to allocate
sysctl numbers, or else conflicts and ABI breakage occur.  The proc interface
to sysctl does not have that problem, as names are not densely allocated.

By not terminating a sysctl table until I have neither a ctl_name nor a
procname, it becomes simple to add sysctl entries that don't show up in the
binary sysctl interface.  Which allows people to avoid allocating a binary
sysctl value when not needed.

I have audited the kernel code and in my reading I have not found a single
sysctl table that wasn't terminated by a completely zero filled entry.  So
this change in behavior should not affect anything.

I think this mechanism eases the pain enough that combined with a little
disciple we can solve the reoccurring sysctl ABI breakage.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Improve the removed sysctl warnings
Eric W. Biederman [Mon, 6 Nov 2006 07:52:11 +0000 (23:52 -0800)] 
[PATCH] Improve the removed sysctl warnings

Don't warn about libpthread's access to kernel.version.  When it receives
-ENOSYS it will read /proc/sys/kernel/version.

If anything else shows up print the sysctl number string.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Cal Peake <cp@absolutedigital.net>
Cc: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] lockdep: fix delayacct locking bug
Peter Zijlstra [Mon, 6 Nov 2006 07:52:10 +0000 (23:52 -0800)] 
[PATCH] lockdep: fix delayacct locking bug

Make the delayacct lock irqsave; this avoids the possible deadlock where
an interrupt is taken while holding the delayacct lock which needs to
take the delayacct lock.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Shailabh Nagar <nagar@watson.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] isdn/gigaset: convert warning message
Tilman Schmidt [Mon, 6 Nov 2006 07:52:08 +0000 (23:52 -0800)] 
[PATCH] isdn/gigaset: convert warning message

Make the failed-to-allocate-skb warning a non-debug message.

Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Cc: Hansjoerg Lipp <hjlipp@web.de>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Kai Germaschewski <kai.germaschewski@gmx.de>
Cc: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix for LKDTM MEM_SWAPOUT crashpoint
Ankita Garg [Mon, 6 Nov 2006 07:52:07 +0000 (23:52 -0800)] 
[PATCH] Fix for LKDTM MEM_SWAPOUT crashpoint

The MEM_SWAPOUT crashpoint in LKDTM could be broken as some compilers
inline the call to shrink_page_list() and symbol lookup for this function
name fails.  Replacing it with the function shrink_inactive_list(), which
is the only function calling shrink_page_list().

Signed-off-by: Ankita Garg <ankita@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix the spurious unlock_cpu_hotplug false warnings
Gautham R Shenoy [Mon, 6 Nov 2006 07:52:04 +0000 (23:52 -0800)] 
[PATCH] Fix the spurious unlock_cpu_hotplug false warnings

Cpu-hotplug locking has a minor race case caused because of setting the
variable "recursive" to NULL *after* releasing the cpu_bitmask_lock in the
function unlock_cpu_hotplug,instead of doing so before releasing the
cpu_bitmask_lock.

This was the cause of most of the recent false spurious lock_cpu_unlock
warnings.

This should fix the problem reported by Martin Lorenz reported in
http://lkml.org/lkml/2006/10/29/127.

Thanks to Srinivasa DS for pointing it out.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoAVR32: Fix thinko in generic_find_next_zero_le_bit()
Haavard Skinnemoen [Tue, 31 Oct 2006 19:01:11 +0000 (20:01 +0100)] 
AVR32: Fix thinko in generic_find_next_zero_le_bit()

The existing implementation of this function seems to be looking for
a one although it should be looking for a zero. This causes trouble
for the ext2 filesystem, which tends to report -ENOSPC without this
patch.

Fix this by complementing each word before scanning.

Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
18 years agoAVR32: Get rid of board_early_init
Haavard Skinnemoen [Mon, 30 Oct 2006 08:31:27 +0000 (09:31 +0100)] 
AVR32: Get rid of board_early_init

board_early_init() is left over from some early prototyping work
where we had to initialize the SDRAM controller ourselves. This
depends on the kernel being loaded into static RAM, which just
isn't possible on any commercially available products today.

In order to run without a boot loader, we need to create a zImage
stub or have the debugger initialize the SDRAM for us (for really
low-level debugging)

Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
18 years agoMerge branch 'upstream-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/linvil...
Jeff Garzik [Mon, 6 Nov 2006 07:58:09 +0000 (02:58 -0500)] 
Merge branch 'upstream-fixes' of git://git./linux/kernel/git/linville/wireless-2.6 into upstream-fixes

18 years ago[PATCH] Kconfig: remove redundant NETDEVICES depends
Randy Dunlap [Thu, 2 Nov 2006 05:18:58 +0000 (21:18 -0800)] 
[PATCH] Kconfig: remove redundant NETDEVICES depends

drivers/net/Kconfig says:
# All the following symbols are dependent on NETDEVICES - do not repeat
# that for each of the symbols.

so remove duplicate 'depends' uses of NETDEVICES.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years ago[PATCH] ehea: 64K page support fix
Thomas Klein [Fri, 3 Nov 2006 16:48:23 +0000 (17:48 +0100)] 
[PATCH] ehea: 64K page support fix

This patch fixes 64k page support by using PAGE_MASK and appropriate pagesize defines in several places.

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years ago[PATCH] ehea: Removed redundant define
Thomas Klein [Fri, 3 Nov 2006 16:47:52 +0000 (17:47 +0100)] 
[PATCH] ehea: Removed redundant define

Removed define H_CB_ALIGNMENT which is already defined in include/asm-powerpc/hvcall.h

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years ago[PATCH] ehea: Nullpointer dereferencation fix
Thomas Klein [Fri, 3 Nov 2006 16:47:20 +0000 (17:47 +0100)] 
[PATCH] ehea: Nullpointer dereferencation fix

Fix: Must check for nullpointer before dereferencing it - not afterwards.

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 6 Nov 2006 03:10:04 +0000 (19:10 -0800)] 
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [NETLABEL]: Fix build failure.
  [IPV6]: Give sit driver an appropriate module alias.
  [IPV6]: Add ndisc_netdev_notifier unregister.
  [NET]: __alloc_pages() failures reported due to fragmentation
  [PKTGEN]: TCI endianness fixes
  [TG3]: Fix 2nd ifup failure on 5752M.
  [NETFILTER] bug: skb->protocol is already net-endian
  [NETFILTER] bug: nfulnl_msg_config_mode ->copy_range is 32bit
  [NETFILTER] bug: NFULA_CFG_QTHRESH uses 32bit
  [IPV6]: Fix ECN bug on big-endian
  [IPX]: Annotate and fix IPX checksum
  [IPX]: Trivial parts of endianness annotations

18 years ago[SPARC]: Fix robust futex syscalls and wire up migrate_pages.
David S. Miller [Mon, 6 Nov 2006 00:51:03 +0000 (16:51 -0800)] 
[SPARC]: Fix robust futex syscalls and wire up migrate_pages.

When I added the entries for the robust futex syscall entries, I
forgot to bump NR_SYSCALLS.  The current situation is error-prone
because NR_SYSCALLS lives in entry.S where the system call limit
checks are enforced.  Move the definition to asm/unistd.h in order to
make this mistake much more difficult to make.

And wire up sys_migrate_pages since the powerpc folks implemented the
compat wrapper for us.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETLABEL]: Fix build failure.
Paul Moore [Mon, 6 Nov 2006 00:44:06 +0000 (16:44 -0800)] 
[NETLABEL]: Fix build failure.

> the build with the attached .config failed, make ends with:
> ...
> : undefined reference to `cipso_v4_sock_getattr'
> net/built-in.o: In function `netlbl_socket_getattr':

 ...

It looks like I was stupid and made NetLabel depend on CONFIG_NET and not
CONFIG_INET, the patch below should fix this by making NetLabel depend on
CONFIG_INET and CONFIG_SECURITY.  Please review and apply for 2.6.19.

Signed-off-by: Paul Moore <paul.moore@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: Give sit driver an appropriate module alias.
Patrick McHardy [Sun, 5 Nov 2006 23:47:04 +0000 (15:47 -0800)] 
[IPV6]: Give sit driver an appropriate module alias.

It would be nice to keep things working even with this built as a
module, it took me some time to realize my IPv6 tunnel was broken
because of the missing sit module. This module alias fixes things
until distributions have added an appropriate alias to modprobe.conf.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: Add ndisc_netdev_notifier unregister.
Dmitry Mishin [Sat, 4 Nov 2006 00:08:19 +0000 (16:08 -0800)] 
[IPV6]: Add ndisc_netdev_notifier unregister.

If inet6_init() fails later than ndisc_init() call, or IPv6 module is
unloaded, ndisc_netdev_notifier call remains in the list and will follows in
oops later.

Signed-off-by: Dmitry Mishin <dim@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: __alloc_pages() failures reported due to fragmentation
Larry Woodman [Sat, 4 Nov 2006 00:05:45 +0000 (16:05 -0800)] 
[NET]: __alloc_pages() failures reported due to fragmentation

We have seen a couple of __alloc_pages() failures due to
fragmentation, there is plenty of free memory but no large order pages
available.  I think the problem is in sock_alloc_send_pskb(), the
gfp_mask includes __GFP_REPEAT but its never used/passed to the page
allocator.  Shouldnt the gfp_mask be passed to alloc_skb() ?

Signed-off-by: Larry Woodman <lwoodman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PKTGEN]: TCI endianness fixes
Al Viro [Fri, 3 Nov 2006 11:49:56 +0000 (03:49 -0800)] 
[PKTGEN]: TCI endianness fixes

open-coded variant there works only for little-endian

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: Fix 2nd ifup failure on 5752M.
Michael Chan [Fri, 3 Nov 2006 09:01:03 +0000 (01:01 -0800)] 
[TG3]: Fix 2nd ifup failure on 5752M.

This fixes a bug reported in:

http://bugzilla.kernel.org/show_bug.cgi?id=7438

tg3_close() turns off the PHY if WoL and ASF are both disabled.  On
the next tg3_open(), some devices such as the 5752M will not be
brought up correctly without a PHY reset early in the reset sequence.
The PHY clock is needed for some internal MAC blocks to function
correctly.

This problem is fixed by always resetting the PHY early in
tg3_reset_hw() when it is called from tg3_open() or tg3_resume().
tg3_setup_phy() can then be called later in the sequence without the
reset_phy parameter set to 1, since the PHY reset is already done.

Update version to 3.68.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER] bug: skb->protocol is already net-endian
Al Viro [Fri, 3 Nov 2006 08:59:17 +0000 (00:59 -0800)] 
[NETFILTER] bug: skb->protocol is already net-endian

htons() is not needed (and no, it's not misspelled ntohs() -
userland expects net-endian here).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER] bug: nfulnl_msg_config_mode ->copy_range is 32bit
Al Viro [Fri, 3 Nov 2006 08:58:41 +0000 (00:58 -0800)] 
[NETFILTER] bug: nfulnl_msg_config_mode ->copy_range is 32bit

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER] bug: NFULA_CFG_QTHRESH uses 32bit
Al Viro [Fri, 3 Nov 2006 08:58:17 +0000 (00:58 -0800)] 
[NETFILTER] bug: NFULA_CFG_QTHRESH uses 32bit

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: Fix ECN bug on big-endian
Al Viro [Fri, 3 Nov 2006 08:55:35 +0000 (00:55 -0800)] 
[IPV6]: Fix ECN bug on big-endian

__constant_htons(2<<4) is not a replacement for
htonl(2<<20).

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPX]: Annotate and fix IPX checksum
Al Viro [Fri, 3 Nov 2006 08:28:23 +0000 (00:28 -0800)] 
[IPX]: Annotate and fix IPX checksum

Calculation of IPX checksum got buggered about 2.4.0.  The old variant
mangled the packet; that got fixed, but calculation itself got buggered.
Restored the correct logics, fixed a subtle breakage we used to have even
back then: if the sum is 0 mod 0xffff, we want to return 0, not 0xffff.
The latter has special meaning for IPX (cheksum disabled).  Observation
(and obvious fix) nicked from history of FreeBSD ipx_cksum.c...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPX]: Trivial parts of endianness annotations
Al Viro [Fri, 3 Nov 2006 08:27:06 +0000 (00:27 -0800)] 
[IPX]: Trivial parts of endianness annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agoMake sure "user->sigpending" count is in sync
Linus Torvalds [Sat, 4 Nov 2006 21:03:00 +0000 (13:03 -0800)] 
Make sure "user->sigpending" count is in sync

The previous commit (45c18b0bb579b5c1b89f8c99f1b6ffa4c586ba08, aka "Fix
unlikely (but possible) race condition on task->user access") fixed a
potential oops due to __sigqueue_alloc() getting its "user" pointer out
of sync with switch_user(), and accessing a user pointer that had been
de-allocated on another CPU.

It still left another (much less serious) problem, where a concurrent
__sigqueue_alloc and swich_user could cause sigqueue_alloc to do signal
pending reference counting for a _different_ user than the one it then
actually ended up using.  No oops, but we'd end up with the wrong signal
accounting.

Another case of Oleg's eagle-eyes picking up the problem.

This is trivially fixed by just making sure we load whichever "user"
structure we decide to use (it doesn't matter _which_ one we pick, we
just need to pick one) just once.

Acked-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>