pandora-kernel.git
17 years agoi2c: Fix typo in SMBus Write Word Data description
Mike Frysinger [Tue, 13 Feb 2007 21:08:59 +0000 (22:08 +0100)]
i2c: Fix typo in SMBus Write Word Data description

Write data, don't read it.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
17 years agoi2c-piix4: Add support for the ATI SB600
Jean Delvare [Tue, 13 Feb 2007 21:08:59 +0000 (22:08 +0100)]
i2c-piix4: Add support for the ATI SB600

Add support for the ATI SB600 SMBus controller.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
17 years agoi2c-nforce2: Drop unused reference to pci_dev
Jean Delvare [Tue, 13 Feb 2007 21:08:58 +0000 (22:08 +0100)]
i2c-nforce2: Drop unused reference to pci_dev

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Hans-Frieder Vogt <hfvogt@gmx.net>
17 years agoi2c/vt8231: Remove superfluous initialization
David Brownell [Tue, 13 Feb 2007 21:08:58 +0000 (22:08 +0100)]
i2c/vt8231: Remove superfluous initialization

Remove a superfluous initialization from the vt8231 hwmon driver; the
i2c core does this, and the source field will be vanishing soon.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
17 years agoi2c-ali1563: Fix device initialization
Jean Delvare [Tue, 13 Feb 2007 21:08:57 +0000 (22:08 +0100)]
i2c-ali1563: Fix device initialization

The i2c-ali1563 initialization looks quite broken to me:
* If the I/O space isn't enabled, we forcibly set 3 bits in
  the PCI configuration space instead of just the one enabling
  the I/O space.
* After that we pretend to check if the write worked, but we
  don't actually read the new value from the register.
* It's probably not a good idea to enable the I/O space if no
  base address has been set.

So I propose the following changes to that part of the driver:
* Merge ali1563_enable() into ali1563_setup().
* Check the base address before the I/O space enabled bit.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Rudolf Marek <r.marek@assembler.cz>
17 years agoi2c-ali1563: Improve the status messages
Jean Delvare [Tue, 13 Feb 2007 21:08:57 +0000 (22:08 +0100)]
i2c-ali1563: Improve the status messages

Improve the status messages printed by the i2c-ali1563 driver.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Rudolf Marek <r.marek@assembler.cz>
17 years ago[POWERPC] cell: pm_rtas_activat_signals routine cleanup
Maynard Johnson [Tue, 13 Feb 2007 21:02:03 +0000 (22:02 +0100)]
[POWERPC] cell: pm_rtas_activat_signals routine cleanup

The code was setting up the debug bus for group 21 when profiling on the
event PPU CYCLES.  The debug bus is not actually used by the hardware
performance counters when counting PPU CYCLES.  Setting up the debug bus
for PPU CYCLES causes signal routing conflicts on the debug bus when
profiling PPU cycles and another PPU event.  This patch fixes the code to
only setup the debug bus to route the performance signals for the non
PPU CYCLE events.

Signed-off-by: Maynard Johnson <mpjohn@us.ibm.com>
Signed-off-by: Carl Love <carll@us.ibm.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] cell: PPU Oprofile cleanup patch
Carl Love [Tue, 13 Feb 2007 21:02:02 +0000 (22:02 +0100)]
[POWERPC] cell: PPU Oprofile cleanup patch

This is a clean up patch that includes the following changes:

 -Some comments were added to clarify the code based on feedback
  from the community.
 -The write_pm_cntrl() and set_count_mode() were passed a
  structure element from a global variable.  The argument was
  removed so the functions now just operate on the global directly.
 -The set_pm_event() function call in the cell_virtual_cntr()
  routine was moved to a for-loop before the for_each_cpu loop

Signed-off-by: Carl Love <carll@us.ibm.com>
Signed-off-by: Maynard Johnson <mpjohn@us.ibm.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[IPSEC]: Fix the address family to refer encap_family
Kazunori MIYAZAWA [Tue, 13 Feb 2007 20:57:16 +0000 (12:57 -0800)]
[IPSEC]: Fix the address family to refer encap_family

Fix the address family to refer encap_family
when comparing with a kernel generated xfrm_state

Signed-off-by: Kazunori MIYAZAWA <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPSEC]: changing API of xfrm6_tunnel_register
Kazunori MIYAZAWA [Tue, 13 Feb 2007 20:55:55 +0000 (12:55 -0800)]
[IPSEC]: changing API of xfrm6_tunnel_register

This patch changes xfrm6_tunnel register and deregister
interface to prepare for solving the conflict of device
tunnels with inter address family IPsec tunnel.
There is no device which conflicts with IPv4 over IPv6
IPsec tunnel.

Signed-off-by: Kazunori MIYAZAWA <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[POWERPC] spufs: avoid accessing kernel memory through mmapped /mem node
Masato Noguchi [Tue, 13 Feb 2007 20:54:30 +0000 (21:54 +0100)]
[POWERPC] spufs: avoid accessing kernel memory through mmapped /mem node

I found an exploit in current kernel.
Currently, there is no range check about mmapping "/mem" node in
spufs. Thus, an application can access privilege memory region.

In case this kernel already worked on a public server, I send this
information only here.
If there are such servers in somewhere, please replace it, ASAP.

Signed-off-by: Masato Noguchi <Masato.Noguchi@jp.sony.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spu sched: static timeslicing for SCHED_RR contexts
Christoph Hellwig [Tue, 13 Feb 2007 20:54:29 +0000 (21:54 +0100)]
[POWERPC] spu sched: static timeslicing for SCHED_RR contexts

For SCHED_RR tasks we can do some really trivial timeslicing.  Basically
we fire up a time for every scheduler tick that searches for a higher
or same priority thread that is on the runqueue and if there is one
context switches to it.  Because we can't lock spus from timer context
we actually run this from a delayed runqueue instead of a timer.

A nice optimization would be to skip the actual priority bitmap search
when there are less contexts than physical spus available.  To implement
this I need a so far unpublished patch from Andre, and it will be added
after we have that patch in.

Note that right now we only do the time slicing for SCHED_RR tasks.
The code would work for SCHED_OTHER tasks aswell, but their prio
value is defered from the one the PPU thread has at time of spu_run,
and using this for spu scheduling decisions would make the code very
unfair.  SCHED_OTHER support will be enabled once we the spu scheduler
knows how to calculcate cpu_context.prio (very soon)

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spu sched: use DECLARE_BITMAP
Christoph Hellwig [Tue, 13 Feb 2007 20:54:28 +0000 (21:54 +0100)]
[POWERPC] spu sched: use DECLARE_BITMAP

use DECLARE_BITMAP in the spu scheduler instead of reimplementing it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spu sched: forced preemption at execution
Christoph Hellwig [Tue, 13 Feb 2007 20:54:27 +0000 (21:54 +0100)]
[POWERPC] spu sched: forced preemption at execution

If we start a spu context with realtime priority we want it to run
immediately and not wait until some other lower priority thread has
finished.  Try to find a suitable victim and use it's spu in this
case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spu sched: update some comments
Christoph Hellwig [Tue, 13 Feb 2007 20:54:26 +0000 (21:54 +0100)]
[POWERPC] spu sched: update some comments

Give spu_yield a kerneldoc comment and remove the old comment
documenting spu_activate, spu_deactive and spu_yield as all of them
now have descriptive kerneldoc comments of their own.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spu sched: simplity spu_remove_from_active_list
Christoph Hellwig [Tue, 13 Feb 2007 20:54:25 +0000 (21:54 +0100)]
[POWERPC] spu sched: simplity spu_remove_from_active_list

If we call spu_remove_from_active_list that spu is always guaranteed
to be on the active list and in runnable state, so we can simply
do a list_del to remove it and unconditionally take the was_active
codepath.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: optimize spu_run
Christoph Hellwig [Tue, 13 Feb 2007 20:54:24 +0000 (21:54 +0100)]
[POWERPC] spufs: optimize spu_run

There is no need to directly wake up contexts in spu_activate when
called from spu_run, so add a flag to surpress this wakeup.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: runqueue simplification
Christoph Hellwig [Tue, 13 Feb 2007 20:54:23 +0000 (21:54 +0100)]
[POWERPC] spufs: runqueue simplification

This is the biggest patch in this series, and it reworks the guts of
the spu scheduler runqueue mechanism:

 - instead of embedding a waitqueue in the runqueue there is now a
   simple doubly-linked list, the actual wakeups happen by reusing
   the stop_wq in the spu context (maybe we should rename it one day)
 - spu_free and spu_prio_wakeup are merged into a single spu_reschedule
   function
 - various functionality is split out into small helpers, and kerneldoc
   comments are added in various places to document what's going on.
 - spu_activate is rewritten into a tight loop by removing test for
   various impossible conditions and using the infrastructure in this
   patch.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: move prio to spu_context
Christoph Hellwig [Tue, 13 Feb 2007 20:54:22 +0000 (21:54 +0100)]
[POWERPC] spufs: move prio to spu_context

It doesn't make any sense to have a priority field in the physical spu
structure.  Move it into the spu context instead.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: state_mutex cleanup
Christoph Hellwig [Tue, 13 Feb 2007 20:54:21 +0000 (21:54 +0100)]
[POWERPC] spufs: state_mutex cleanup

Various cleanups in code surrounding the state semaphore:

 - inline spu_acquire/spu_release
 - cleanup spu_acquire_* and add kerneldoc comments to these functions
 - remove spu_release_exclusive and replace it with spu_release

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[IPSEC]: make sit use the xfrm4_tunnel_register
Kazunori MIYAZAWA [Tue, 13 Feb 2007 20:55:25 +0000 (12:55 -0800)]
[IPSEC]: make sit use the xfrm4_tunnel_register

This patch makes sit use xfrm4_tunnel_register instead of
inet_add_protocol. It solves conflict of sit device with
inter address family IPsec tunnel.

Signed-off-by: Kazunori MIYAZAWA <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPSEC]: Changing API of xfrm4_tunnel_register.
Kazunori MIYAZAWA [Tue, 13 Feb 2007 20:54:47 +0000 (12:54 -0800)]
[IPSEC]: Changing API of xfrm4_tunnel_register.

This patch changes xfrm4_tunnel register and deregister
interface to prepare for solving the conflict of device
tunnels with inter address family IPsec tunnel.

Signed-off-by: Kazunori MIYAZAWA <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[POWERPC] spufs: simplify state_mutex
Christoph Hellwig [Tue, 13 Feb 2007 20:36:50 +0000 (21:36 +0100)]
[POWERPC] spufs: simplify state_mutex

The r/w semaphore to lock the spus was overkill and can be replaced
with a mutex to make it faster, simpler and easier to debug.  It also
helps to allow making most spufs interruptible in future patches.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: sched.c cleanups
Christoph Hellwig [Tue, 13 Feb 2007 20:36:49 +0000 (21:36 +0100)]
[POWERPC] spufs: sched.c cleanups

Various cleanups to sched.c that don't change the global control flow:

 - add kerneldoc comments to various functions
 - add spu_ prefixes to various functions
 - add/remove context from the runqueue in bind/unbind_context as
   it's part of the logical operation
 - add a call to put_active_spu to spu_unbind_contex as it's logically
   part of the unbind operation

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: bind_context sets SPU_STATE_RUNNABLE
Christoph Hellwig [Tue, 13 Feb 2007 20:36:48 +0000 (21:36 +0100)]
[POWERPC] spufs: bind_context sets SPU_STATE_RUNNABLE

Only bind_context/unbind_context change the spu context state.  Thus
we can move all assignents of SPU_STATE_RUNNABLE into bind_context,
which parallels the unbind side aswell.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: remove superfluous SPU_STATE_SAVED assignments
Christoph Hellwig [Tue, 13 Feb 2007 20:36:47 +0000 (21:36 +0100)]
[POWERPC] spufs: remove superfluous SPU_STATE_SAVED assignments

unbind_context already sets the context state to SPU_STATE_SAVED, thus
the spu_deactivate callers don't need to do it again.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: remove empty last line in run.c
Christoph Hellwig [Tue, 13 Feb 2007 20:36:46 +0000 (21:36 +0100)]
[POWERPC] spufs: remove empty last line in run.c

Remove the empty last line in arch/powerpc/platforms/cell/spufs/run.c.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[POWERPC] spufs: remove SPU_CONTEXT_PREEMPT
Christoph Hellwig [Tue, 13 Feb 2007 20:36:45 +0000 (21:36 +0100)]
[POWERPC] spufs: remove SPU_CONTEXT_PREEMPT

Remove the SPU_CONTEXT_PREEMPT define.  It's unused and won't be used
in this form after the scheduler rework.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
17 years ago[TCP]: Prevent pseudo garbage in SYN's advertized window
Ilpo Järvinen [Tue, 13 Feb 2007 20:42:11 +0000 (12:42 -0800)]
[TCP]: Prevent pseudo garbage in SYN's advertized window

TCP may advertize up to 16-bits window in SYN packets (no window
scaling allowed). At the same time, TCP may have rcv_wnd
(32-bits) that does not fit to 16-bits without window scaling
resulting in pseudo garbage into advertized window from the
low-order bits of rcv_wnd. This can happen at least when
mss <= (1<<wscale) (see tcp_select_initial_window). This patch
fixes the handling of SYN advertized windows (compile tested
only).

In worst case (which is unlikely to occur though), the receiver
advertized window could be just couple of bytes. I'm not sure
that such situation would be handled very well at all by the
receiver!? Fortunately, the situation normalizes after the
first non-SYN ACK is received because it has the correct,
scaled window.

Alternatively, tcp_select_initial_window could be changed to
prevent too large rcv_wnd in the first place.

[ tcp_make_synack() has the same bug, and I've added a fix for
  that to this patch -DaveM ]

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET_SCHED]: sch_hfsc: replace ASSERT macro by WARN_ON
Patrick McHardy [Tue, 13 Feb 2007 20:36:57 +0000 (12:36 -0800)]
[NET_SCHED]: sch_hfsc: replace ASSERT macro by WARN_ON

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[BRIDGE] br_if: Fix oops in port_carrier_check
Jarek Poplawski [Tue, 13 Feb 2007 20:35:26 +0000 (12:35 -0800)]
[BRIDGE] br_if: Fix oops in port_carrier_check

Signed-off-by: Jarek Poplawski <jarkao2@o2.pl>
Acked-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: Clear GSO bits for TCP reset packet
Herbert Xu [Tue, 13 Feb 2007 20:32:58 +0000 (12:32 -0800)]
[NETFILTER]: Clear GSO bits for TCP reset packet

The TCP reset packet is copied from the original.  This
includes all the GSO bits which do not apply to the new
packet.  So we should clear those bits.

Spotted by Patrick McHardy.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Update copyright, version, and reldate.
Michael Chan [Tue, 13 Feb 2007 20:18:46 +0000 (12:18 -0800)]
[TG3]: Update copyright, version, and reldate.

Update version to 3.73.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Add some tx timeout debug messages.
Michael Chan [Tue, 13 Feb 2007 20:18:30 +0000 (12:18 -0800)]
[TG3]: Add some tx timeout debug messages.

Print the most useful information during tx timeout to help debug.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Use constant for PHY register 0x1e.
Michael Chan [Tue, 13 Feb 2007 20:18:15 +0000 (12:18 -0800)]
[TG3]: Use constant for PHY register 0x1e.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Power down 5704 serdes transceiver when shutting down.
Michael Chan [Tue, 13 Feb 2007 20:17:57 +0000 (12:17 -0800)]
[TG3]: Power down 5704 serdes transceiver when shutting down.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: 5906 doesn't need to switch to slower clock.
Michael Chan [Tue, 13 Feb 2007 20:17:38 +0000 (12:17 -0800)]
[TG3]: 5906 doesn't need to switch to slower clock.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: 5722/5756 don't need PHY jitter workaround.
Michael Chan [Tue, 13 Feb 2007 20:17:25 +0000 (12:17 -0800)]
[TG3]: 5722/5756 don't need PHY jitter workaround.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Use lower DMA watermark for 5703.
Michael Chan [Tue, 13 Feb 2007 20:17:03 +0000 (12:17 -0800)]
[TG3]: Use lower DMA watermark for 5703.

Set DMA read watermark to 4 on 5703 in PCIX mode.  This is needed to
prevent some tx timeouts.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Save MSI state before suspend.
Michael Chan [Tue, 13 Feb 2007 20:16:45 +0000 (12:16 -0800)]
[TG3]: Save MSI state before suspend.

This fixes the following problem:

http://bugzilla.kernel.org/show_bug.cgi?id=7969

The MSI state needs to be saved during suspend.  PCI state saved
during tg3_init_one() does not contain valid MSI state because
MSI hasn't been enabled.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoRevert "[PATCH] Fix d_path for lazy unmounts"
Linus Torvalds [Tue, 13 Feb 2007 20:08:18 +0000 (12:08 -0800)]
Revert "[PATCH] Fix d_path for lazy unmounts"

This reverts commit eb3dfb0cb1f4a44e2d0553f89514ce9f2a9fcaf1.

It causes some strange Gnome problem with dbus-daemon getting stuck, so
we'll revert it until that problem is understood.

Reported by both walt and Greg KH, who both independently git-bisected
the problem to this commit.

Andreas is looking at it.

Reported-by: walt <wa1ter@myrealbox.com>
Reported-by: Greg KH <greg@kroah.com>
Acked-by: Andreas Gruenbacher <agruen@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch '83xx' into for_paulus
Kumar Gala [Tue, 13 Feb 2007 18:51:38 +0000 (12:51 -0600)]
Merge branch '83xx' into for_paulus

17 years ago[POWERPC] 83xx: Updated mpc834x_itx_defconfig
Timur Tabi [Fri, 9 Feb 2007 16:01:50 +0000 (10:01 -0600)]
[POWERPC] 83xx: Updated mpc834x_itx_defconfig

This patch updates the defconfig for the MPC8349E-mITX.  In addition to picking
up changes from recent kernels, disables support for e100 (which doesn't ship
with the system), turns off input devices, turns on some I2C support, turns
off HW monitoring (HW not yet supported), turns off OHCI USB (not used), turns
off USB gadget support (HW not yet supported), turns on DOS FS support, and
turns off kernel debugging.

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
17 years ago[POWERPC] 83xx: Add support for MPC8349E-mITX-GP
Timur Tabi [Fri, 9 Feb 2007 20:00:36 +0000 (14:00 -0600)]
[POWERPC] 83xx: Add support for MPC8349E-mITX-GP

This patch adds a defconfig and a DTS for the MPC8349E-mITX-GP, a variant of
the MPC8349E-mITX.

USB is disabled because the only USB port is not setup properly by
firmware/kernel

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
17 years agoMerge branch '85xx' into for_paulus
Kumar Gala [Tue, 13 Feb 2007 18:41:23 +0000 (12:41 -0600)]
Merge branch '85xx' into for_paulus

17 years ago[POWERPC] 85xx: Add a defconfig for the 8568 MDS
Andy Fleming [Fri, 9 Feb 2007 23:30:09 +0000 (17:30 -0600)]
[POWERPC] 85xx: Add a defconfig for the 8568 MDS

Add defconfig for the MPC8568 MDS reference board

Signed-off-by: Andrew Fleming <afleming@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
17 years ago[POWERPC] 85xx: Add support for the 8568 MDS board
Andy Fleming [Fri, 9 Feb 2007 23:28:31 +0000 (17:28 -0600)]
[POWERPC] 85xx: Add support for the 8568 MDS board

Add support for the MPC8568 MDS reference board

Signed-off-by: Andrew Fleming <afleming@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
17 years agoMerge branch 'master' of /home/trondmy/kernel/linux-2.6/
Trond Myklebust [Tue, 13 Feb 2007 06:43:25 +0000 (22:43 -0800)]
Merge branch 'master' of /home/trondmy/kernel/linux-2.6/

Conflicts:

net/sunrpc/auth_gss/gss_krb5_crypto.c
net/sunrpc/auth_gss/gss_spkm3_token.c
net/sunrpc/clnt.c

Merge with mainline and fix conflicts.

17 years agoNFS: disconnect before retrying NFSv4 requests over TCP
Chuck Lever [Tue, 6 Feb 2007 23:26:11 +0000 (18:26 -0500)]
NFS: disconnect before retrying NFSv4 requests over TCP

RFC3530 section 3.1.1 states an NFSv4 client MUST NOT send a request
twice on the same connection unless it is the NULL procedure.  Section
3.1.1 suggests that the client should disconnect and reconnect if it
wants to retry a request.

Implement this by adding an rpc_clnt flag that an ULP can use to
specify that the underlying transport should be disconnected on a
major timeout.  The NFSv4 client asserts this new flag, and requests
no retries after a minor retransmit timeout.

Note that disconnecting on a retransmit is in general not safe to do
if the RPC client does not reuse the TCP port number when reconnecting.

See http://bugzilla.linux-nfs.org/show_bug.cgi?id=6

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
17 years agoNFS: Don't use ClearPageUptodate() when writeback fails
Trond Myklebust [Tue, 6 Feb 2007 19:07:15 +0000 (11:07 -0800)]
NFS: Don't use ClearPageUptodate() when writeback fails

ClearPageUptodate() will just cause races here. What we really want to do
is to invalidate the page cache.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
17 years agoNFS: Cleanup - avoid rereading 'jiffies' more than once in the same routine
Trond Myklebust [Mon, 5 Feb 2007 22:44:22 +0000 (14:44 -0800)]
NFS: Cleanup - avoid rereading 'jiffies' more than once in the same routine

Micro-optimisations for nfs_fhget() and nfs_wcc_update_inode().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
17 years agoNFS: Fix a wraparound issue with nfsi->cache_change_attribute
Trond Myklebust [Mon, 5 Feb 2007 22:26:28 +0000 (14:26 -0800)]
NFS: Fix a wraparound issue with nfsi->cache_change_attribute

Fix wraparound issue with nfsi->cache_change_attribute. If it is found
to lie in the future, then update it to lie in the past. Patch based on
a suggestion by Neil Brown.

..and minor micro-optimisation: avoid reading 'jiffies' more than once in
nfs_update_inode().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
17 years agoMerge branch 'master' into 85xx
Kumar Gala [Tue, 13 Feb 2007 05:57:21 +0000 (23:57 -0600)]
Merge branch 'master' into 85xx

17 years agoMerge branch 'master' into 83xx
Kumar Gala [Tue, 13 Feb 2007 05:56:24 +0000 (23:56 -0600)]
Merge branch 'master' into 83xx

17 years agoMerge branch 'master' into for_paulus
Kumar Gala [Tue, 13 Feb 2007 05:52:26 +0000 (23:52 -0600)]
Merge branch 'master' into for_paulus

17 years ago[POWERPC] spufs: Fix bitrot of the SPU mmap facility
Benjamin Herrenschmidt [Tue, 13 Feb 2007 00:46:08 +0000 (11:46 +1100)]
[POWERPC] spufs: Fix bitrot of the SPU mmap facility

It looks like we've had some serious bitrot there mostly due to tracking
of address_space's of mmap'ed files getting out of sync with the actual
mmap code. The mfc, mss and psmap were not tracked properly and thus
not invalidated on context switches (oops !)

I also removed the various file->f_mapping = inode->i_mapping;
assignments that were done in the other open() routines since that
is already done for us by __dentry_open.

One improvement we might want to do later is to assign the various
ctx-> fields at mmap time instead of file open/close time so that we
don't call unmap_mapping_range() on thing that have not been mmap'ed

Finally, I added some smp_wmb's after assigning the ctx-> fields to make
sure they are visible to other CPUs. I don't think this is really
necessary as I suspect locking in the fs layer will make that happen
anyway but better safe than sorry.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] powerpc: Remove SPU struct pages for PS3
Benjamin Herrenschmidt [Tue, 13 Feb 2007 00:46:07 +0000 (11:46 +1100)]
[POWERPC] powerpc: Remove SPU struct pages for PS3

Struct page are no longer needed for SPUs, so let's not create them
on PS3 anymore.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: remove need for struct page for SPEs
Benjamin Herrenschmidt [Tue, 13 Feb 2007 00:46:06 +0000 (11:46 +1100)]
[POWERPC] spufs: remove need for struct page for SPEs

This patch removes the need for struct page for SPE local store
and registers from spufs. It also makes the locking much more
obvious and no longer relying on the truncate logic black magic
for protecting against races between unmap_mapping_range() and
new pages faulted in. It does so by switching to a nopfn() handler
and using the new vm_insert_pfn() to setup the PTEs itself while
holding a lock on the SPE.

The nice thing is that this patch actually removes a lot more code
than it adds :-)

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Small cleanup of EFIKA platform
Sylvain Munaut [Mon, 12 Feb 2007 22:13:24 +0000 (23:13 +0100)]
[POWERPC] Small cleanup of EFIKA platform

The efika platform used three files efika-pci.c efika-setup.c and
a 2 line efika.h to link the two. The total of code in those is
really not much and therefore, I think they're better merged
in a single file.

There is absolutely _no_code_change_ at all, just merged the files.

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix unbalanced of_node_{get,put} in efika-setup.c
Sylvain Munaut [Mon, 12 Feb 2007 22:13:23 +0000 (23:13 +0100)]
[POWERPC] Fix unbalanced of_node_{get,put} in efika-setup.c

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Use common 52xx of_platform probe code for EFIKA
Sylvain Munaut [Mon, 12 Feb 2007 22:13:22 +0000 (23:13 +0100)]
[POWERPC] Use common 52xx of_platform probe code for EFIKA

Now that the device tree has the good properties, we can
remove all the efika_init code by a single call to common code.

While we're modifying that file, a few whitespaces/alignement/typo
fixes are made (nothing significant).

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Restore 'proper' link order in platform
Sylvain Munaut [Mon, 12 Feb 2007 22:13:21 +0000 (23:13 +0100)]
[POWERPC] Restore 'proper' link order in platform

The 52xx was put before CHRP to allow EFIKA to be recognized
properly. Now the efika tree is fixed up in prom_init so
no need for this ugly hack. So we restore the 'normal'
order.

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Add device tree fixups for the EFIKA
Sylvain Munaut [Mon, 12 Feb 2007 22:13:20 +0000 (23:13 +0100)]
[POWERPC] Add device tree fixups for the EFIKA

We make the efika device tree compliant with the defined bindings
(at least compliant enough). This is mostly done by mangling
the device_type and compatible properties, but also adding
some missing bits.

Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] mpc5200 device tree bindings refinement
Grant Likely [Mon, 12 Feb 2007 20:36:54 +0000 (13:36 -0700)]
[POWERPC] mpc5200 device tree bindings refinement

Much needed refinement of mpc5200 device tree binding specifications.

Short list:
- drop mpc52xx designator; only two supported chips exist, 5200 and 5200b.
  It's premature to refer to them as '52xx'.
- Specify optional 'model' and 'revision' properties in the soc5200 node
- Specify reqiured 'cell-index' property to identify between multiple SOC
  devices of the same type.  (Useful for arbitrating shared register access)
- Specify optional 'port-number' property for adjusting the logical serial
  port assignments.
- Specify optional 'has-wdt' property for gpt0 node.
- Add system-frequency property to soc5200 node

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fixup mp5200 drivers to match device tree changes
Grant Likely [Mon, 12 Feb 2007 20:36:55 +0000 (13:36 -0700)]
[POWERPC] Fixup mp5200 drivers to match device tree changes

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] correct a prototype
Stephen Rothwell [Mon, 12 Feb 2007 11:10:48 +0000 (22:10 +1100)]
[POWERPC] correct a prototype

This rids us of a warning.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Wire up sys_getcpu
Stephen Rothwell [Mon, 12 Feb 2007 11:09:18 +0000 (22:09 +1100)]
[POWERPC] Wire up sys_getcpu

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] ppc: Add support for AMCC Taishan 440GX eval board
Stefan Roese [Mon, 12 Feb 2007 10:29:04 +0000 (11:29 +0100)]
[POWERPC] ppc: Add support for AMCC Taishan 440GX eval board

This patch adds support for the AMCC Taishan PPC440GX evaluation
board.

This is still an arch/ppc port. I'm aware that the move of
4xx to arch/powerpc is making good progress right now. So this
patch is mainly intended to make the Taishan support available
for the community right now.

Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix vDSO page count calculation
Benjamin Herrenschmidt [Mon, 12 Feb 2007 02:31:08 +0000 (13:31 +1100)]
[POWERPC] Fix vDSO page count calculation

The recent vDSO consolidation patches broke powerpc due to a mistake
in the definition of MAXPAGES constants. This fixes it by moving to
a dynamically allocated array of pages instead as I don't like much
hard coded size limits. Also move the vdso initialisation to an initcall
since it doesn't really need to be done -that- early.

Applogies for not catching the breakage earlier, Roland _did_ CC me on
his patches a while ago, I got busy with other things and forgot to test
them.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Correct AC Power: in /proc/pmu/info on ibook1
Olaf Hering [Sat, 10 Feb 2007 20:40:00 +0000 (21:40 +0100)]
[POWERPC] Correct AC Power: in /proc/pmu/info on ibook1

/proc/pmu/info contains AC Power: 0 when booting without battery.
Force AC Power, it will be updated whenever the battery state changes.

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Mark winbond IDE PCI resources with start 0 as unassigned
Olaf Hering [Sat, 10 Feb 2007 20:38:37 +0000 (21:38 +0100)]
[POWERPC] Mark winbond IDE PCI resources with start 0 as unassigned

libata calls pci_request_regions to claim PCI BAR 0 - 5
pci_request_regions fails if one of the regions cant be claimed.
bar 5 has start == 0,  __request_resource will fail.

Tested on a p630 in SMP mode with pata_sl82c105

 00:03.1 IDE interface: Symphony Labs SL82c105 (rev 05) (prog-if 8f [Master SecP SecO PriP PriO])
         Control: I/O+ Mem- BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr+ Stepping- SERR+ FastB2B-
         Status: Cap- 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
         Interrupt: pin A routed to IRQ 86
         Region 0: I/O ports at 3fd3000f000 [size=8]
         Region 1: I/O ports at 3fd3000f010 [size=4]
         Region 2: I/O ports at 3fd3000f020 [size=8]
         Region 3: I/O ports at 3fd3000f030 [size=4]
         Region 4: I/O ports at 3fd3000f040 [size=16]
         Region 5: I/O ports at 3fd30000000 [size=16]
 00: ad 10 05 01 41 01 80 02 05 8f 01 01 08 48 80 00
 10: 01 f0 00 00 11 f0 00 00 21 f0 00 00 31 f0 00 00
 20: 41 f0 00 00 01 00 00 00 00 00 00 00 00 00 00 00
 30: 00 00 00 00 00 00 00 00 00 00 00 00 56 01 02 28
 40: b3 08 ff 00 09 09 00 00 09 09 00 00 09 09 00 00
 50: 09 09 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 60: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 70: 00 00 00 00 00 00 00 00 00 00 00 00 ff 00 00 00
 80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 90: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 a0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 c0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

 name             "ide"
 linux,phandle    00d5cdc0 (14011840)
 assigned-addresses 81001910 00000000 0000f000 00000000 00000008 81001914
                  00000000 0000f010 00000000 00000004 81001918 00000000
                  0000f020 00000000 00000008 8100191c 00000000 0000f030
                  00000000 00000004 81001920 00000000 0000f040 00000000
                  00000010 81001924 00000000 00000000 00000000 00000010
 interrupts       00000003
 built-in
 #size-cells      00000000
 #address-cells   00000001
 device_type      "ide"
 reg              00001900 00000000 00000000 00000000 00000000
                  41001910 00000000 00000000 00000000 00000008
                  41001914 00000000 00000000 00000000 00000004
                  41001918 00000000 00000000 00000000 00000008
                  4100191c 00000000 00000000 00000000 00000004
                  41001920 00000000 00000000 00000000 00000010
                  41001924 00000000 00000000 00000000 00000010
 compatible       "pci10ad,105"
                  "pciclass,01018f"
 ibm,fw-slot-number 00000000
 fast-back-to-back
 devsel-speed     00000001
 max-latency      00000028 (40)
 min-grant        00000002
 class-code       0001018f (65935)
 revision-id      00000005
 device-id        00000105 (261)
 vendor-id        000010ad (4269)
 ibm,loc-code     "U0.1-P1/Q6"

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] move variables in drivers/macintosh to bss
Olaf Hering [Sat, 10 Feb 2007 20:35:12 +0000 (21:35 +0100)]
[POWERPC] move variables in drivers/macintosh to bss

Move all the initialized variables to bss.
Mark a version string as const.

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Virtual DMA support for floppy driver for new powerpc architecture
Pavel Fedin [Mon, 29 Jan 2007 12:13:03 +0000 (15:13 +0300)]
[POWERPC] Virtual DMA support for floppy driver for new powerpc architecture

During ppc64+ppc merge virtual DMA code for floppy driver was not
ported.  This patch restores virtual DMA support for floppy in new
powerpc target.

It is necessary at least on Pegasos and AmigaOne machines for the
floppy drive to function.  ISA DMA controller works incorrectly there
due to its addressing limitations.

Virtual DMA mode is activated by floppy=nodma option passed to the
kernel (or module).  There's no automatic switch like on i386.

Signed-off-by: Pavel Fedin <sonic_amiga@rambler.ru>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Celleb: improve MMU hashtable locking
Akira Iguchi [Fri, 9 Feb 2007 07:53:59 +0000 (16:53 +0900)]
[POWERPC] Celleb: improve MMU hashtable locking

Disabling IRQ is required only in invalidation.  This changes
"spin_lock_irqsave" to "spin_lock" in other ops.

Signed-off-by: Kou Ishizaki <kou.ishizaki@toshiba.co.jp>
Signed-off-by: Akira Iguchi <akira2.iguchi@toshiba.co.jp>
Acked-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[XFRM]: Fix IPv4 tunnel mode decapsulation with IPV6=n
Patrick McHardy [Tue, 13 Feb 2007 04:27:10 +0000 (20:27 -0800)]
[XFRM]: Fix IPv4 tunnel mode decapsulation with IPV6=n

Add missing break when CONFIG_IPV6=n.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] HASHTABLES: Use appropriate seed for caluculating ehash index.
YOSHIFUJI Hideaki [Tue, 13 Feb 2007 04:26:39 +0000 (20:26 -0800)]
[IPV6] HASHTABLES: Use appropriate seed for caluculating ehash index.

Tetsuo Handa <handat@pm.nttdata.co.jp> told me that connect(2) with TCPv6
socket almost always took a few minutes to return when we did not have any
ports available in the range of net.ipv4.ip_local_port_range.

The reason was that we used incorrect seed for calculating index of
hash when we check established sockets in __inet6_check_established().

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[POWERPC] Remove ibm4{xx,4x}.h from arch/powerpc
David Gibson [Tue, 6 Feb 2007 01:48:31 +0000 (12:48 +1100)]
[POWERPC] Remove ibm4{xx,4x}.h from arch/powerpc

ARCH=powerpc should not use the ghastly un-multiplatformable tangle of
includes that starts with asm-ppc/ibm4xx.h.  This patch removes a
compile-breaking include of it from head_44x.S.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] 85xx: Don't write reserved values to MAS1[TSIZE]
Becky Bruce [Mon, 12 Feb 2007 23:43:46 +0000 (17:43 -0600)]
[POWERPC] 85xx: Don't write reserved values to MAS1[TSIZE]

Some of the current tlbwe instructions early on in head_fsl_booke.S take
advantage of unarchitected behavior that allows the writing of reserved
values to the TSIZE field.  This patch corrects that, as well as an error
where an uninitialized (by linux) value was written into a MAS register and
used for a tlbwe.

Correct this for both arch/ppc and arch/powerpc.

Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
17 years agoMerge branch 'master' into 83xx
Kumar Gala [Tue, 13 Feb 2007 03:28:39 +0000 (21:28 -0600)]
Merge branch 'master' into 83xx

17 years agoMerge branch 'master' into 85xx
Kumar Gala [Tue, 13 Feb 2007 03:27:10 +0000 (21:27 -0600)]
Merge branch 'master' into 85xx

17 years agoMerge branch 'master' into for_paulus
Kumar Gala [Tue, 13 Feb 2007 03:17:37 +0000 (21:17 -0600)]
Merge branch 'master' into for_paulus

17 years agoMerge branch 'for_paulus' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerpc
Paul Mackerras [Tue, 13 Feb 2007 02:28:00 +0000 (13:28 +1100)]
Merge branch 'for_paulus' of /linux/kernel/git/galak/powerpc

17 years agoIB/mthca: Always fill MTTs from CPU
Michael S. Tsirkin [Sat, 10 Feb 2007 21:14:25 +0000 (23:14 +0200)]
IB/mthca: Always fill MTTs from CPU

Speed up memory registration by filling in MTTs directly when the CPU
can write directly to the whole table (all mem-free cards, and to
Tavor mode on 64-bit systems with the patch I posted earlier).  This
reduces the number of FW commands needed to register an MR by at least
a factor of 2 and speeds up memory registration significantly.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoIB/mthca: Merge MR and FMR space on 64-bit systems
Michael S. Tsirkin [Sat, 10 Feb 2007 21:13:12 +0000 (23:13 +0200)]
IB/mthca: Merge MR and FMR space on 64-bit systems

For Tavor, we currently reserve separate MPT and MTT space for FMRs to
avoid abusing the vmalloc space on 32 bit kernels. No such problem
exists on 64 bit kernels so let's not do it there.

This way we have a shared pool for MR and FMR resources, used on
demand.  This will also make it possible to write MTTs for regular
regions directly from driver.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoIB/mthca: Fix access to MTT and MPT tables on non-cache-coherent CPUs
Michael S. Tsirkin [Sat, 10 Feb 2007 21:15:08 +0000 (23:15 +0200)]
IB/mthca: Fix access to MTT and MPT tables on non-cache-coherent CPUs

We allocate the MTT table with alloc_pages() and then do pci_map_sg(),
so we must call pci_dma_sync_sg() after the CPU writes to the MTT
table.  This works since the device will never write MTTs on mem-free
HCAs, once we get rid of the use of the WRITE_MTT firmware command.
This change is needed to make that work, and is an improvement for
now, since it gives FMRs a chance at working.

For MPTs, both the device and CPU might write there, so we must
allocate DMA coherent memory for these.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoIB/mthca: Give reserved MTTs a separate cache line
Michael S. Tsirkin [Sat, 10 Feb 2007 21:17:26 +0000 (23:17 +0200)]
IB/mthca: Give reserved MTTs a separate cache line

MTTs are allocated in non-cache-coherent memory, so we must give
reserved MTTs their own cache line, to prevent both device and
CPU from writing into the same cache line at the same time.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoIB/mthca: Fix reserved MTTs calculation on mem-free HCAs
Michael S. Tsirkin [Sat, 10 Feb 2007 21:17:26 +0000 (23:17 +0200)]
IB/mthca: Fix reserved MTTs calculation on mem-free HCAs

The reserved_mtts field has different meaning in Tavor and Arbel, so
we are wasting mtt entries on memfree. Fix the Arbel case to match
Tavor semantics.

Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoRDMA/cxgb3: Add driver for Chelsio T3 RNIC
Steve Wise [Tue, 13 Feb 2007 00:16:18 +0000 (16:16 -0800)]
RDMA/cxgb3: Add driver for Chelsio T3 RNIC

Add an RDMA/iWARP driver for the Chelsio T3 1GbE and 10GbE adapters.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Mon, 12 Feb 2007 23:35:30 +0000 (15:35 -0800)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC]: Re-export saved_command_line to modules.
  [SPARC64]: Increase command line size to 2048 like other arches.
  [SPARC64]: We do not need ZONE_DMA.

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 12 Feb 2007 23:34:17 +0000 (15:34 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (25 commits)
  [XFRM]: Fix OOPSes in xfrm_audit_log().
  [TCP]: cleanup of htcp (resend)
  [TCP]: Use read mostly for CUBIC parameters.
  [NETFILTER]: nf_conntrack_tcp: make sysctl variables static
  [NETFILTER]: ip6t_mh: drop piggyback payload packet on MH packets
  [NETFILTER]: Fix whitespace errors
  [NETFILTER]: Kconfig: improve dependency handling
  [NETFILTER]: xt_mac/xt_CLASSIFY: use IPv6 hook names for IPv6 registration
  [NETFILTER]: nf_conntrack: change nf_conntrack_l[34]proto_unregister to void
  [NETFILTER]: nf_conntrack: properly use RCU for nf_conntrack_destroyed callback
  [NETFILTER]: ip_conntrack: properly use RCU for ip_conntrack_destroyed callback
  [NETFILTER]: nf_conntrack: fix invalid conntrack statistics RCU assumption
  [NETFILTER]: ip_conntrack: fix invalid conntrack statistics RCU assumption
  [NETFILTER]: nf_conntrack: properly use RCU API for nf_ct_protos/nf_ct_l3protos arrays
  [NETFILTER]: ip_conntrack: properly use RCU API for ip_ct_protos array
  [NETFILTER]: nf_nat: properly use RCU API for nf_nat_protos array
  [NETFILTER]: ip_nat: properly use RCU API for ip_nat_protos array
  [NETFILTER]: nf_log: minor cleanups
  [NETFILTER]: nf_log: switch logger registration/unregistration to mutex
  [NETFILTER]: nf_log: make nf_log_unregister_pf return void
  ...

17 years ago[SPARC]: Re-export saved_command_line to modules.
David S. Miller [Mon, 12 Feb 2007 23:10:56 +0000 (15:10 -0800)]
[SPARC]: Re-export saved_command_line to modules.

This reverts some bogosity from the dynamic command-line
changes made on sparc32 and sparc64.

Drivers such as drivers/sbus/char/openprom.c reference
saved_command_line, and can be modular.

The boot_command_line is __initdata, yet the dynamic command-line
changes add modular exports of that symbol, obviously wrong.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Increase command line size to 2048 like other arches.
David S. Miller [Mon, 12 Feb 2007 19:01:21 +0000 (11:01 -0800)]
[SPARC64]: Increase command line size to 2048 like other arches.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: We do not need ZONE_DMA.
David S. Miller [Mon, 12 Feb 2007 08:13:31 +0000 (00:13 -0800)]
[SPARC64]: We do not need ZONE_DMA.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agodon't use 'localversion*' files twice
Linus Torvalds [Mon, 12 Feb 2007 23:05:15 +0000 (15:05 -0800)]
don't use 'localversion*' files twice

Since we look in both source and object directories for localversion*
files, we accidentally ended up getting them twice.  Use 'sort -u' to
avoid that.

Reported-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[XFRM]: Fix OOPSes in xfrm_audit_log().
David S. Miller [Mon, 12 Feb 2007 21:53:54 +0000 (13:53 -0800)]
[XFRM]: Fix OOPSes in xfrm_audit_log().

Make sure that this function is called correctly, and
add BUG() checking to ensure the arguments are sane.

Based upon a patch by Joy Latten.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[WATCHDOG] timers cleanup
Jiri Slaby [Thu, 8 Feb 2007 17:39:36 +0000 (18:39 +0100)]
[WATCHDOG] timers cleanup

- Use timer macros to set function and data members and to modify
  expiration time.
- Use DEFINE_TIMER for single (platform dependent) watchdog timers and
  do not init them at run-time in these cases.
- del_timer_sync is common in most cases -- we want to wait for timer
  function if it's still running.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Steve Hill <steve@navaho.co.uk>
Cc: Heiko Ronsdorf <hero@ihg.uni-duisburg.de>
Cc: Fernando Fuganti <fuganti@conectiva.com.br>
Cc: Gergely Madarasz <gorgo@itc.hu>
Cc: Ken Hollis <khollis@bitgate.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[TCP]: cleanup of htcp (resend)
Stephen Hemminger [Mon, 12 Feb 2007 21:34:03 +0000 (13:34 -0800)]
[TCP]: cleanup of htcp (resend)

Minor non-invasive cleanups:
 * white space around operators and line wrapping
 * use const
 * use __read_mostly

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Use read mostly for CUBIC parameters.
Stephen Hemminger [Mon, 12 Feb 2007 21:15:20 +0000 (13:15 -0800)]
[TCP]: Use read mostly for CUBIC parameters.

These module parameters should be in the read mostly area to avoid
cache pollution.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: nf_conntrack_tcp: make sysctl variables static
Patrick McHardy [Mon, 12 Feb 2007 19:16:58 +0000 (11:16 -0800)]
[NETFILTER]: nf_conntrack_tcp: make sysctl variables static

sysctls are registered by the protocol module itself since 2.6.19, no need
to have them visible to others.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: ip6t_mh: drop piggyback payload packet on MH packets
Masahide NAKAMURA [Mon, 12 Feb 2007 19:16:17 +0000 (11:16 -0800)]
[NETFILTER]: ip6t_mh: drop piggyback payload packet on MH packets

Regarding RFC3775, MH payload proto field should be IPPROTO_NONE. Otherwise
it must be discarded (and the receiver should send ICMP error).

We assume filter should drop such piggyback everytime to disallow slipping
through firewall rules, even the final receiver will discard it.

Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>