]> err.no Git - linux-2.6/log
linux-2.6
17 years agoRDMA/cma: Fix initialization of next_port
Sean Hefty [Tue, 5 Jun 2007 16:57:31 +0000 (09:57 -0700)]
RDMA/cma: Fix initialization of next_port

next_port should be between sysctl_local_port_range[0] and [1].
However, it is initially set to a random value with get_random_bytes().
If the value is negative when treated as a signed integer, next_port
can end up outside the expected range because of the result of the %
operator being negative.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoIB/mlx4: Fix zeroing of rnr_retry value in ib_modify_qp()
Jack Morgenstein [Wed, 6 Jun 2007 16:35:04 +0000 (19:35 +0300)]
IB/mlx4: Fix zeroing of rnr_retry value in ib_modify_qp()

The code in __mlx4_ib_modify_qp() overwrites context->params1 after
the RNR retry parameter is ORed in, which results in the RNR retry
parameter always being set to 0.  Fix this by moving where we OR in
the value to later in the function, after the initial assignment of
context->params1.

Found by the Mellanox firmware group.

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agomlx4_core: Don't set MTT address in dMPT entries with PA set
Jack Morgenstein [Fri, 8 Jun 2007 06:24:38 +0000 (23:24 -0700)]
mlx4_core: Don't set MTT address in dMPT entries with PA set

If a dMPT entry has the PA flag (direct physical address) set, then
the (unused) MTT base address field has to be set to 0.

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agomlx4_core: Check firmware command interface revision
Roland Dreier [Fri, 8 Jun 2007 06:24:36 +0000 (23:24 -0700)]
mlx4_core: Check firmware command interface revision

HCA firmware with incompatible changes to the FW commmand interface is
coming soon.  Add a check of the interface revision during
initialization and bail out if the firmware advertises a revision that
the driver doesn't know about.  This will avoid strange failures later
if the driver goes on using the wrong interface revision.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoIB/mthca, mlx4_core: Fix typo in comment
Roland Dreier [Mon, 4 Jun 2007 02:47:10 +0000 (19:47 -0700)]
IB/mthca, mlx4_core: Fix typo in comment

s/signifant/significant/

Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agomlx4_core: Free catastrophic error MSI-X interrupt with correct dev_id
Roland Dreier [Sat, 2 Jun 2007 14:16:02 +0000 (07:16 -0700)]
mlx4_core: Free catastrophic error MSI-X interrupt with correct dev_id

We need to pass the same dev_id to free_irq() and request_irq().  When
using MSI-X, the MLX4_EQ_CATAS interrupt uses a different dev_id from
the other interrupts.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agomlx4_core: Initialize ctx_list and ctx_lock earlier
Roland Dreier [Thu, 7 Jun 2007 18:51:58 +0000 (11:51 -0700)]
mlx4_core: Initialize ctx_list and ctx_lock earlier

We may call mlx4_dispatch_event() before mlx4_register_device() is
called for a device, because for example a catastrophic error happens
immediately after we enable interrupts.  Therefore priv->ctx_list and
priv->ctx_lock need to be initialized earlier.

This bug was actually exposed by the MSI-X bug that returned IRQ numbers
to drivers in reverse order, so that the first FW command
interrupt looked to mlx4 like a catastrophic error.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agomlx4_core: Fix CQ context layout
Eli Cohen [Wed, 30 May 2007 10:14:31 +0000 (13:14 +0300)]
mlx4_core: Fix CQ context layout

The reserved6 field should be 64 bits, not just 16 bits.  Without
this, the structure does not match the hardware layout on 32-bit
architectures: the db_rec_addr field ends up at offset 52 instead of
offset 56.  The bug slipped by because the alignment of __be64 members
ends up putting it in the right place on x86-64.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Thu, 7 Jun 2007 16:35:54 +0000 (09:35 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [VIDEO] sunxvr500fb: Fix pseudo_palette array size
  [VIDEO] sunxvr2500fb: Fix pseudo_palette array size
  [VIDEO] ffb: The pseudo_palette is only 16 elements long
  [VIDEO]: Fix section mismatch warning in promcon.
  [ATA]: Back out bogus (SPARC64 && !PCI) Kconfig depends.
  [SPARC64]: Fill in gaps in non-PCI dma_*() NOP implementation.
  [SPARC64]: Fix {mc,smt}_capable().
  [SPARC64]: Make core and sibling groups equal on UltraSPARC-IV.
  [SPARC64]: Proper multi-core scheduling support.
  [SPARC64]: Provide mmu statistics via sysfs.
  [SPARC64]: Fix service channel hypervisor function names.
  [SPARC64]: Export basic cpu properties via sysfs.
  [SPARC64]: Move topology init code into new file, sysfs.c

17 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Thu, 7 Jun 2007 15:54:55 +0000 (08:54 -0700)]
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] Fix building of COFF zImages
  [POWERPC] spufs: Fix error handling in spufs_fill_dir()
  [POWERPC] Add table of contents to booting-without-of.txt
  [POWERPC] spufs: Don't yield nosched context
  [POWERPC] Fix typo in booting-without-of-txt section numbering
  [POWERPC] scc_sio: Fix link failure
  [POWERPC] cbe_cpufreq: Limit frequency via cpufreq notifier chain
  [POWERPC] Fix pci_setup_phb_io_dynamic for pci_iomap
  [POWERPC] spufs scheduler: Fix wakeup races
  [POWERPC] spufs: Synchronize pte invalidation vs ps close
  [POWERPC] spufs: Free mm if spufs_fill_dir() failed
  [POWERPC] spufs: Fix gang destroy leaks
  [POWERPC] spufs: Hook up spufs_release_mem
  [POWERPC] spufs: Refuse to load the module when not running on cell
  [POWERPC] pasemi: Fix iommu + 64K PAGE_SIZE bug

17 years agoRestrict clearing TIF_SIGPENDING
Roland McGrath [Wed, 6 Jun 2007 10:59:00 +0000 (03:59 -0700)]
Restrict clearing TIF_SIGPENDING

This patch should get a few birds.  It prevents sigaction calls from
clearing TIF_SIGPENDING in other threads, which could leak -ERESTART*.
And It fixes ptrace_stop not to clear it, which done at the syscall exit
stop could leak -ERESTART*.  It probably removes the harm from signalfd,
at least assuming it never calls dequeue_signal on kernel threads that
might have used block_all_signals.

Signed-off-by: Roland McGrath <roland@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agocheckpatch.pl: should be executable
Andy Whitcroft [Thu, 7 Jun 2007 15:16:40 +0000 (16:16 +0100)]
checkpatch.pl: should be executable

scripts/checkpatch.pl should be executable, make it so.

Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Joel Schopp <jschopp@austin.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[POWERPC] Fix building of COFF zImages
Paul Mackerras [Thu, 7 Jun 2007 12:21:31 +0000 (22:21 +1000)]
[POWERPC] Fix building of COFF zImages

The COFF zImage (for booting oldworld powermacs) wasn't being built
correctly because the procedure descriptor in crt0.S for the zImage
entry point wasn't declared as .globl, and therefore wasn't getting
pulled in from wrapper.a by the linker.  This adds the necessary
.globl statement.

Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Fix error handling in spufs_fill_dir()
Sebastian Siewior [Wed, 6 Jun 2007 04:03:58 +0000 (14:03 +1000)]
[POWERPC] spufs: Fix error handling in spufs_fill_dir()

The error path in spufs_fill_dir() is broken. If d_alloc_name() or
spufs_new_file() fails, spufs_prune_dir() is getting called. At this time
dir->inode is not set and a NULL pointer is dereferenced by mutex_lock().
This bugfix replaces spufs_prune_dir() with a shorter version that does
not touch dir->inode but simply removes all children.

Signed-off-by: Sebastian Siewior <bigeasy@linux.vnet.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Acked-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Add table of contents to booting-without-of.txt
Stuart Yoder [Tue, 5 Jun 2007 18:29:14 +0000 (04:29 +1000)]
[POWERPC] Add table of contents to booting-without-of.txt

Add table of contents.

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Don't yield nosched context
Christoph Hellwig [Tue, 5 Jun 2007 01:25:59 +0000 (11:25 +1000)]
[POWERPC] spufs: Don't yield nosched context

Nosched context sould never be scheduled out, thus we must not
deactivate them in spu_yield ever.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix typo in booting-without-of-txt section numbering
Stuart Yoder [Mon, 4 Jun 2007 21:16:15 +0000 (07:16 +1000)]
[POWERPC] Fix typo in booting-without-of-txt section numbering

Fix typo in section numbering.

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] scc_sio: Fix link failure
Christoph Hellwig [Mon, 4 Jun 2007 18:14:41 +0000 (04:14 +1000)]
[POWERPC] scc_sio: Fix link failure

scc_sio.o should only be built if the txx9 serial driver is actually
built into the kernel.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
--
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] cbe_cpufreq: Limit frequency via cpufreq notifier chain
Thomas Renninger [Mon, 4 Jun 2007 18:14:40 +0000 (04:14 +1000)]
[POWERPC] cbe_cpufreq: Limit frequency via cpufreq notifier chain

... and get rid of cpufreq_set_policy call that caused a build
failure due interfering commits.

Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Christian Krafft <krafft@de.ibm.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix pci_setup_phb_io_dynamic for pci_iomap
Arnd Bergmann [Mon, 4 Jun 2007 18:14:39 +0000 (04:14 +1000)]
[POWERPC] Fix pci_setup_phb_io_dynamic for pci_iomap

We had a problem on a system with only dynamically allocated
PCI buses (using of_pci_phb_driver) in combination with libata.
This setup ended up having no "primary" phb, which means
that pci_io_base never got initialized and all IO port
numbers are 64 bit numbers, which is larger than the
PIO_MASK limit.

Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs scheduler: Fix wakeup races
Christoph Hellwig [Mon, 4 Jun 2007 13:26:51 +0000 (23:26 +1000)]
[POWERPC] spufs scheduler: Fix wakeup races

Fix the race between checking for contexts on the runqueue and actually
waking them in spu_deactive and spu_yield.

The guts of spu_reschedule are split into a new helper called
grab_runnable_context which shows if there is a runnable thread below
a specified priority and if yes removes if from the runqueue and uses
it.  This function is used by the new __spu_deactivate hepler shared
by preemption and spu_yield to grab a new context before deactivating
a specified priority and if yes removes if from the runqueue and uses
it.  This function is used by the new __spu_deactivate hepler shared
by preemption and spu_yield to grab a new context before deactivating
the old one.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Synchronize pte invalidation vs ps close
Christoph Hellwig [Mon, 4 Jun 2007 13:26:51 +0000 (23:26 +1000)]
[POWERPC] spufs: Synchronize pte invalidation vs ps close

Make sure the mapping_lock also protects access to the various address_space
pointers used for tearing down the ptes on a spu context switch.

Because unmap_mapping_range can sleep we need to turn mapping_lock from
a spinlock into a sleeping mutex.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Free mm if spufs_fill_dir() failed
Sebastian Siewior [Mon, 4 Jun 2007 13:26:51 +0000 (23:26 +1000)]
[POWERPC] spufs: Free mm if spufs_fill_dir() failed

In case spufs_fill_dir() fails only put_spu_context()
gets called for cleanup and the acquired mm_struct never gets freed.

Signed-off-by: Sebastian Siewior <bigeasy@linux.vnet.ibm.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Fix gang destroy leaks
Jeremy Kerr [Mon, 4 Jun 2007 13:26:51 +0000 (23:26 +1000)]
[POWERPC] spufs: Fix gang destroy leaks

Previously, closing a SPE gang that still has contexts would trigger
a WARN_ON, and leak the allocated gang.

This change fixes the problem by using the gang's reference counts to
destroy the gang instead. The gangs will persist until their last
reference (be it context or open file handle) is gone.

Also, avoid using statements with side-effects in a WARN_ON().

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Hook up spufs_release_mem
Christoph Hellwig [Mon, 4 Jun 2007 13:26:51 +0000 (23:26 +1000)]
[POWERPC] spufs: Hook up spufs_release_mem

Currently spufs_mem_release and the mem file doesn't have any release
method hooked up, leading to leaks everytime is used.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] spufs: Refuse to load the module when not running on cell
Arnd Bergmann [Mon, 4 Jun 2007 13:26:51 +0000 (23:26 +1000)]
[POWERPC] spufs: Refuse to load the module when not running on cell

As noticed by David Woodhouse, it's currently possible to mount
spufs on any machine, which means that it actually will get
mounted by fedora.
This refuses to load the module on platforms that have no
support for SPUs.

Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] pasemi: Fix iommu + 64K PAGE_SIZE bug
Olof Johansson [Mon, 4 Jun 2007 00:49:05 +0000 (10:49 +1000)]
[POWERPC] pasemi: Fix iommu + 64K PAGE_SIZE bug

The powerpc iommu code was refactored by Linas back in the 2.6.20 time
frame to map 4K pages from the generic code, but I had forgotten to go
back and fix my platform driver before submitting it.

Signed-off-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[VIDEO] sunxvr500fb: Fix pseudo_palette array size
Antonino Daplas [Tue, 5 Jun 2007 20:15:26 +0000 (13:15 -0700)]
[VIDEO] sunxvr500fb: Fix pseudo_palette array size

- the pseudo_palette is only 16 elements long.
- the pseudo_palette has only 16 elements. Do not write if regno (the array
  index) is more than 15.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[VIDEO] sunxvr2500fb: Fix pseudo_palette array size
Antonino Daplas [Tue, 5 Jun 2007 20:15:01 +0000 (13:15 -0700)]
[VIDEO] sunxvr2500fb: Fix pseudo_palette array size

- the pseudo_palette is only 16 elements long.
- the pseudo_palette has only 16 elements. Do not write if regno (the array
  index) is more than 15.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[VIDEO] ffb: The pseudo_palette is only 16 elements long
Antonino Daplas [Tue, 5 Jun 2007 20:14:33 +0000 (13:14 -0700)]
[VIDEO] ffb: The pseudo_palette is only 16 elements long

The pseudo_palette is only 16 elements long.

Signed-off-by: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[VIDEO]: Fix section mismatch warning in promcon.
Sam Ravnborg [Tue, 5 Jun 2007 06:52:45 +0000 (23:52 -0700)]
[VIDEO]: Fix section mismatch warning in promcon.

Fix the following warnings in promcon:
WARNING: o-sparc64/drivers/video/console/built-in.o(.text+0x480): Section mismatch: reference to .init.data:promfont_unitable (between 'promcon_init_unimap' and 'promcon_init')
WARNING: o-sparc64/drivers/video/console/built-in.o(.text+0x488): Section mismatch: reference to .init.data:promfont_unitable (between 'promcon_init_unimap' and 'promcon_init')
WARNING: o-sparc64/drivers/video/console/built-in.o(.text+0x48c): Section mismatch: reference to .init.data:promfont_unicount (between 'promcon_init_unimap' and 'promcon_init')
WARNING: o-sparc64/drivers/video/console/built-in.o(.text+0x490): Section mismatch: reference to .init.data:promfont_unicount (between 'promcon_init_unimap' and 'promcon_init')

The warnings happens because the function: promcon_init_unimap()
references promfont_unitable and promfont_unicount which are marked
__initdata by the conmakehash command in the drivers/video/console/Makefile

Fix the warning by removing the __initdata marker on the two variables.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ATA]: Back out bogus (SPARC64 && !PCI) Kconfig depends.
David S. Miller [Tue, 5 Jun 2007 06:33:04 +0000 (23:33 -0700)]
[ATA]: Back out bogus (SPARC64 && !PCI) Kconfig depends.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Fill in gaps in non-PCI dma_*() NOP implementation.
David S. Miller [Tue, 5 Jun 2007 06:32:23 +0000 (23:32 -0700)]
[SPARC64]: Fill in gaps in non-PCI dma_*() NOP implementation.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Fix {mc,smt}_capable().
David S. Miller [Tue, 5 Jun 2007 04:48:33 +0000 (21:48 -0700)]
[SPARC64]: Fix {mc,smt}_capable().

It's not just sun4v hypervisor platforms that should return true
for this, sun4u with UltraSPARC-IV should return true too.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Make core and sibling groups equal on UltraSPARC-IV.
David S. Miller [Tue, 5 Jun 2007 04:35:18 +0000 (21:35 -0700)]
[SPARC64]: Make core and sibling groups equal on UltraSPARC-IV.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Proper multi-core scheduling support.
David S. Miller [Tue, 5 Jun 2007 00:01:39 +0000 (17:01 -0700)]
[SPARC64]: Proper multi-core scheduling support.

The scheduling domain hierarchy is:

   all cpus -->
      cpus that share an instruction cache -->
          cpus that share an integer execution unit

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Provide mmu statistics via sysfs.
David Miller [Mon, 4 Jun 2007 06:38:09 +0000 (23:38 -0700)]
[SPARC64]: Provide mmu statistics via sysfs.

If the system supports hypervisor based statistics, allow them to
be fetched, enabled, and disabled via sysfs.

Enable and disable via the boolean:

/sys/devices/systems/cpu/cpuN/mmustat_enable

Statistic values are provided under:

/sys/devices/systems/cpu/cpuN/mmu_status/

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Fix service channel hypervisor function names.
David Miller [Mon, 4 Jun 2007 02:07:32 +0000 (19:07 -0700)]
[SPARC64]: Fix service channel hypervisor function names.

sed 's/scv/svc/'

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Export basic cpu properties via sysfs.
David S. Miller [Sun, 3 Jun 2007 03:46:36 +0000 (20:46 -0700)]
[SPARC64]: Export basic cpu properties via sysfs.

Cache sizes, udelay_val, and clock_tick.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SPARC64]: Move topology init code into new file, sysfs.c
David S. Miller [Sat, 2 Jun 2007 21:41:44 +0000 (14:41 -0700)]
[SPARC64]: Move topology init code into new file, sysfs.c

Also, use per-cpu data for struct cpu.  Calling kmalloc for
each cpu in topology_init() is just plain clumsy.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoLinux 2.6.22-rc4 v2.6.22-rc4
Linus Torvalds [Tue, 5 Jun 2007 00:57:25 +0000 (17:57 -0700)]
Linux 2.6.22-rc4

17 years agoMerge git://git.infradead.org/mtd-2.6
Linus Torvalds [Tue, 5 Jun 2007 00:54:09 +0000 (17:54 -0700)]
Merge git://git.infradead.org/mtd-2.6

* git://git.infradead.org/mtd-2.6:
  [JFFS2] Fix obsoletion of metadata nodes in jffs2_add_tn_to_tree()
  [MTD] Fix error checking after get_mtd_device() in get_sb_mtd functions
  [JFFS2] Fix buffer length calculations in jffs2_get_inode_nodes()
  [JFFS2] Fix potential memory leak of dead xattrs on unmount.
  [JFFS2] Fix BUG() caused by failing to discard xattrs on deleted files.
  [MTD] generalise the handling of MTD-specific superblocks
  [MTD] [MAPS] don't force uclinux mtd map to be root dev

17 years agolibata: always use polling SETXFER
Tejun Heo [Sun, 27 May 2007 13:10:40 +0000 (15:10 +0200)]
libata: always use polling SETXFER

Several people have reported LITE-ON LTR-48246S detection failed
because SETXFER fails.  It seems the device raises IRQ too early after
SETXFER.  This is controller independent.  The same problem has been
reported for different controllers.

So, now we have pata_via where the controller raises IRQ before it's
ready after SETXFER and a device which does similar thing.  This patch
makes libata always execute SETXFER via polling.  As this only happens
during EH, performance impact is nil.  Setting ATA_TFLAG_POLLING is
also moved from issue hot path to ata_dev_set_xfermode() - the only
place where SETXFER can be issued.

Note that ATA_TFLAG_POLLING applies only to drivers which implement
SFF TF interface and use libata HSM.  More advanced controllers ignore
the flag.  This doesn't matter for this fix as SFF TF controllers are
the problematic ones.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Mon, 4 Jun 2007 20:31:39 +0000 (13:31 -0700)]
Merge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] JAZZ ESP and SUN ESP need SPI_ATTRS
  [SCSI] atari_NCR5380: update_timeout removal
  [SCSI] aacraid: fix shutdown handler to also disable interrupts.
  [SCSI] qla2xxx: fix timeout in qla2x00_down_timeout
  [SCSI] fix CONFIG_SCSI_WAIT_SCAN=m

17 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Mon, 4 Jun 2007 20:27:33 +0000 (13:27 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: reduce raciness when input handlers disconnect
  Input: ucb1x00 - do not access input_dev->private directly
  Input: logips2pp - fix typo in Kconfig
  Input: db9 - do not ignore dev2 module parameter

17 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 4 Jun 2007 20:26:49 +0000 (13:26 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [ICMP]: Fix icmp_errors_use_inbound_ifaddr sysctl
  [IPV4]: Fix "ipOutNoRoutes" counter error for TCP and UDP
  [NET] gso: Fix GSO feature mask in sk_setup_caps
  [TCP]: Fix GSO ignorance of pkts_acked arg (cong.cntrl modules)
  [NET]: Fix comparisons of unsigned < 0.
  [NET]: Make net watchdog timers 1 sec jiffy aligned.
  [ATM]: Fix warning.
  [TCP]: Use default 32768-61000 outgoing port range in all cases.
  [AF_UNIX]: Fix datagram connect race causing an OOPS.
  [TG3]: Fix link problem on Dell's onboard 5906.
  [AF_UNIX]: Make socket locking much less confusing.

17 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Mon, 4 Jun 2007 20:25:59 +0000 (13:25 -0700)]
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [ATA]: Don't allow to enable this for SPARC64 without PCI.
  [VIDEO]: XVR500 and XVR2500 require FB=y

17 years agovanishing ioctl handler debugging
Andrew Morton [Sun, 3 Jun 2007 20:50:41 +0000 (13:50 -0700)]
vanishing ioctl handler debugging

We've had several reoprts of the CPU jumping to 0x00000000 is do_ioctl().  I
assume that there's a race and someone is zeroing out the ioctl handler while
this CPU waits for the lock_kernel().

The patch adds code to detect this, then emits stuff which will hopefuly lead
us to the culprit.

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agomtrr atomicity fix
Andrew Morton [Sun, 3 Jun 2007 20:50:40 +0000 (13:50 -0700)]
mtrr atomicity fix

Rafael gets this on an SMP box with kernel preemption enabled, during
hibernation and restore (100% of the time):

Enabling non-boot CPUs ...
BUG: using smp_processor_id() in preemptible [00000001] code: bash/4514
caller is mtrr_save_state+0x9/0x40

Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoSPI dynamic busid generation bugfix
David Brownell [Sun, 3 Jun 2007 20:50:40 +0000 (13:50 -0700)]
SPI dynamic busid generation bugfix

Fix SPI dynamic bus ID assignment to start at 2^15-1 rather than a negative
number.  Valid bus ids are supposed to be positive, and are (now) stored in
an 's16' value.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[SCSI] JAZZ ESP and SUN ESP need SPI_ATTRS
David S. Miller [Mon, 4 Jun 2007 00:56:04 +0000 (17:56 -0700)]
[SCSI] JAZZ ESP and SUN ESP need SPI_ATTRS

Reported by Meelis Roos.

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years ago[SCSI] atari_NCR5380: update_timeout removal
Michael Schmitz [Sun, 3 Jun 2007 10:55:04 +0000 (12:55 +0200)]
[SCSI] atari_NCR5380: update_timeout removal

Atari SCSI driver fixes: remove update_timeout kludge

Signed-off-by: Michael Schmitz <schmitz@debian.org>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
17 years agoInput: reduce raciness when input handlers disconnect
Dmitry Torokhov [Mon, 4 Jun 2007 03:29:36 +0000 (23:29 -0400)]
Input: reduce raciness when input handlers disconnect

There is a race between input handler's release() and disconnect()
methods: when input handler disconnects it wakes up all regular
users and then process to walk user list to wake up async. users.
While disconnect() walks the list release() removes elements of
the same list causing oopses.

While this is not a substibute for proper locking we can reduce
odds of getting an oops if we wake up normal readers after walking
the list.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
17 years agoInput: ucb1x00 - do not access input_dev->private directly
Dmitry Torokhov [Fri, 11 May 2007 05:16:12 +0000 (01:16 -0400)]
Input: ucb1x00 - do not access input_dev->private directly

Use input_get_drvdata() and input_set_drvdata() helpers to do that.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Acked-by: Pavel Machek <pavel@ucw.cz>
17 years ago[ATA]: Don't allow to enable this for SPARC64 without PCI.
David S. Miller [Mon, 4 Jun 2007 00:39:56 +0000 (17:39 -0700)]
[ATA]: Don't allow to enable this for SPARC64 without PCI.

Based upon a report from Meelis Roos.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[VIDEO]: XVR500 and XVR2500 require FB=y
David S. Miller [Mon, 4 Jun 2007 00:35:24 +0000 (17:35 -0700)]
[VIDEO]: XVR500 and XVR2500 require FB=y

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ICMP]: Fix icmp_errors_use_inbound_ifaddr sysctl
Patrick McHardy [Fri, 1 Jun 2007 18:45:04 +0000 (11:45 -0700)]
[ICMP]: Fix icmp_errors_use_inbound_ifaddr sysctl

Currently when icmp_errors_use_inbound_ifaddr is set and an ICMP error is
sent after the packet passed through ip_output(), an address from the
outgoing interface is chosen as ICMP source address since skb->dev doesn't
point to the incoming interface anymore.

Fix this by doing an interface lookup on rt->dst.iif and using that device.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4]: Fix "ipOutNoRoutes" counter error for TCP and UDP
Wei Dong [Fri, 1 Jun 2007 05:49:28 +0000 (22:49 -0700)]
[IPV4]: Fix "ipOutNoRoutes" counter error for TCP and UDP

Signed-off-by: Wei Dong <weidong@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] gso: Fix GSO feature mask in sk_setup_caps
Herbert Xu [Fri, 1 Jun 2007 05:15:50 +0000 (22:15 -0700)]
[NET] gso: Fix GSO feature mask in sk_setup_caps

This isn't a bug just yet as only TCP uses sk_setup_caps for GSO.
However, if and when UDP or something else starts using it this is
likely to cause a problem if we forget to add software emulation
for it at the same time.

The problem is that right now we translate GSO emulation to the
bitmask NETIF_F_GSO_MASK, which includes every protocol, even
ones that we cannot emulate.

This patch makes it provide only the ones that we can emulate.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Fix GSO ignorance of pkts_acked arg (cong.cntrl modules)
Ilpo Järvinen [Fri, 1 Jun 2007 04:37:55 +0000 (21:37 -0700)]
[TCP]: Fix GSO ignorance of pkts_acked arg (cong.cntrl modules)

The code used to ignore GSO completely, passing either way too
small or zero pkts_acked when GSO skb or part of it got ACKed.
In addition, there is no need to calculate the value in the loop
but simple arithmetics after the loop is sufficient. There is
no need to handle SYN case specially because congestion control
modules are not yet initialized when FLAG_SYN_ACKED is set.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Fix comparisons of unsigned < 0.
Bill Nottingham [Fri, 1 Jun 2007 04:33:35 +0000 (21:33 -0700)]
[NET]: Fix comparisons of unsigned < 0.

Recent gcc versions emit warnings when unsigned variables are
compared < 0 or >= 0.

Signed-off-by: Bill Nottingham <notting@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Make net watchdog timers 1 sec jiffy aligned.
Venkatesh Pallipadi [Fri, 1 Jun 2007 04:28:44 +0000 (21:28 -0700)]
[NET]: Make net watchdog timers 1 sec jiffy aligned.

round_jiffies for net dev watchdog timer.

Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ATM]: Fix warning.
Jeff Garzik [Fri, 1 Jun 2007 04:26:23 +0000 (21:26 -0700)]
[ATM]: Fix warning.

The compiler warning

drivers/atm/firestream.c: In function â€˜top_off_fp’:
drivers/atm/firestream.c:1505: warning: cast to pointer from integer of different size

does indicate a bug, albeit a minor one.  Fixed, by using a 32-bit
temporary prior to the call to bus_to_virt().

The larger bug is still present:  the entire driver assumes that machine
pointers are 32-bit, as it stores pointers in 32-bit hardware registers.
This is obvious to anyone who knows the driver well, but for the casual
readers it is helpfully noted with FIXME.

Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Use default 32768-61000 outgoing port range in all cases.
Mark Glines [Thu, 31 May 2007 22:44:48 +0000 (15:44 -0700)]
[TCP]: Use default 32768-61000 outgoing port range in all cases.

This diff changes the default port range used for outgoing connections,
from "use 32768-61000 in most cases, but use N-4999 on small boxes
(where N is a multiple of 1024, depending on just *how* small the box
is)" to just "use 32768-61000 in all cases".

I don't believe there are any drawbacks to this change, and it keeps
outgoing connection ports farther away from the mess of
IANA-registered ports.

Signed-off-by: Mark Glines <mark@glines.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[AF_UNIX]: Fix datagram connect race causing an OOPS.
David S. Miller [Thu, 31 May 2007 22:19:20 +0000 (15:19 -0700)]
[AF_UNIX]: Fix datagram connect race causing an OOPS.

Based upon an excellent bug report and initial patch by
Frederik Deweerdt.

The UNIX datagram connect code blindly dereferences other->sk_socket
via the call down to the security_unix_may_send() function.

Without locking 'other' that pointer can go NULL via unix_release_sock()
which does sock_orphan() which also marks the socket SOCK_DEAD.

So we have to lock both 'sk' and 'other' yet avoid all kinds of
potential deadlocks (connect to self is OK for datagram sockets and it
is possible for two datagram sockets to perform a simultaneous connect
to each other).  So what we do is have a "double lock" function similar
to how we handle this situation in other areas of the kernel.  We take
the lock of the socket pointer with the smallest address first in
order to avoid ABBA style deadlocks.

Once we have them both locked, we check to see if SOCK_DEAD is set
for 'other' and if so, drop everything and retry the lookup.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TG3]: Fix link problem on Dell's onboard 5906.
Michael Chan [Thu, 31 May 2007 21:49:51 +0000 (14:49 -0700)]
[TG3]: Fix link problem on Dell's onboard 5906.

The bug is caused by code that always set
(TG3_FLAG_USE_MI_INTERRUPT | TG3_FLAG_USE_LINKCHG_REG) on all Dell's
onboard devices.  With these 2 flags set, the link status is polled
by tg3_timer() and will only work when the PHY is set up to interrupt
the MAC on link changes.  This breaks 5906 because the 5906 PHY does
not support TG3_FLAG_USE_MI_INTERRUPT the same as other PHYs.

For correctness, only Dell's onboard 5701 needs these 2 flags to be
set.  This change will fix the 5906 problem and will change other
Dell devices except 5700 and 5701 to use the more efficient
interrupt-driven link changes.

Update version to 3.77.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[AF_UNIX]: Make socket locking much less confusing.
David S. Miller [Thu, 31 May 2007 20:24:26 +0000 (13:24 -0700)]
[AF_UNIX]: Make socket locking much less confusing.

The unix_state_*() locking macros imply that there is some
rwlock kind of thing going on, but the implementation is
actually a spinlock which makes the code more confusing than
it needs to be.

So use plain unix_state_lock and unix_state_unlock.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Sun, 3 Jun 2007 19:41:05 +0000 (12:41 -0700)]
Merge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
  NET: add MAINTAINERS entry for ucc_geth driver
  myri10ge: report link up/down in standard ethtool way
  NetXen: Removal of extra free_irq call
  Update tulip maintainer email address
  smc91x: sh solution engine fixes.
  e1000: disable polling before registering netdevice
  network drivers: eliminate unneeded kill_vid code
  atl1: eliminate unneeded kill_vid code
  8139cp: fix VLAN unregistration
  sky2: Fix VLAN unregistration
  VLAN: kill_vid is only useful for VLAN filtering devices
  qla3xxx: device doesnt do hardware checksumming.

17 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Sun, 3 Jun 2007 19:36:56 +0000 (12:36 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  [WATCHDOG] clean-up watchdog documentation
  [WATCHDOG] ks8695_wdt.c - new KS8695 watchdog driver

17 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Sun, 3 Jun 2007 19:29:15 +0000 (12:29 -0700)]
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] Fix zImage.coff generation for 32-bit pmac
  [POWERPC] Fix compile breakage for IBM/AMCC 4xx arch/ppc platforms
  [POWERPC] Don't allow PMAC_APM_EMU for 64-bit
  [POWERPC] Compare irq numbers with NO_IRQ not IRQ_NONE
  [POWERPC] Fix return from pte_alloc_one() in out-of-memory case
  [POWERPC] Fix compile warning in pseries xics code
  [POWERPC] Don't use HOSTCFLAGS in BOOTCFLAGS
  [POWERPC] Create a zImage for legacy iSeries
  [POWERPC] pasemi idle uses hard_smp_processor_id
  [POWERPC] ps3/interrupt.c uses get_hard_smp_processor_id
  [POWERPC] Fix possible access to free pages
  [POWERPC] Fix compiler/assembler flags for Ebony platform boot files
  [POWERPC] Fix ppc32 single-stepping out of syscalls
  [POWERPC] Update documentation for of_find_node_by_type()

17 years agoNET: add MAINTAINERS entry for ucc_geth driver
Li Yang [Fri, 25 May 2007 05:54:02 +0000 (13:54 +0800)]
NET: add MAINTAINERS entry for ucc_geth driver

Signed-off-by: Li Yang <leoli@freescale.com>
Acked-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agomyri10ge: report link up/down in standard ethtool way
Brice Goglin [Wed, 30 May 2007 19:13:59 +0000 (21:13 +0200)]
myri10ge: report link up/down in standard ethtool way

Report link up/down in standard ethtool way

Signed-off-by: Brice Goglin <brice@myri.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoNetXen: Removal of extra free_irq call
Mithlesh Thukral [Fri, 1 Jun 2007 11:13:08 +0000 (04:13 -0700)]
NetXen: Removal of extra free_irq call

NetXen: Removal of redundant free_irq
This patch removes a redundant free_irq() call from remove() routine.
This will also eliminate a warning during unload of driver.

Signed-by: Mithlesh Thukral <mithlesh@netxen.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoUpdate tulip maintainer email address
Valerie Henson [Wed, 30 May 2007 12:08:54 +0000 (06:08 -0600)]
Update tulip maintainer email address

I've quit Intel and gone into business as a Linux consultant.  Update
my email address in MAINTAINERS.

Signed-off-by: Valerie Henson <val@nmt.edu>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosmc91x: sh solution engine fixes.
Paul Mundt [Fri, 1 Jun 2007 08:36:48 +0000 (17:36 +0900)]
smc91x: sh solution engine fixes.

The current smc91x I/O routines ifdef the solution engine subtypes
individually, which is rather bogus, as they can simply use
CONFIG_SOLUTION_ENGINE instead. This fixes it for some of the other
solution engines that weren't included in the ifdef list (SH7206
specifically).

There are also inb/outb definitions which are totally bogus (missing
brackets in _both_ cases, SMC_CAN_USE_8BIT == 0, and even better, they
try to use a 16-bit access to fake 8-bit access). Kill that nonsense off
completely.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
--

 drivers/net/smc91x.h |    5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoe1000: disable polling before registering netdevice
Auke Kok [Fri, 1 Jun 2007 17:22:39 +0000 (10:22 -0700)]
e1000: disable polling before registering netdevice

To assure the symmetry of poll enable/disable in up/down, we should
initialize the netdevice to be poll_disabled at load time. Doing
this after register_netdevice leaves us open to another race, so
lets move all the netif_* calls above register_netdevice so the
stack starts out how we expect it to be.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Doug Chapman <doug.chapman@hp.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agonetwork drivers: eliminate unneeded kill_vid code
Stephen Hemminger [Fri, 1 Jun 2007 16:44:01 +0000 (09:44 -0700)]
network drivers: eliminate unneeded kill_vid code

Many drivers had code that did kill_vid, but they weren't doing vlan
filtering. With new API the stub is unneeded unless device sets
NETIF_F_HW_VLAN_FILTER.

Bad habit: I couldn't resist fixing a couple of nearby style things
in acenic, and forcedeth.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoatl1: eliminate unneeded kill_vid code
Stephen Hemminger [Fri, 1 Jun 2007 16:44:00 +0000 (09:44 -0700)]
atl1: eliminate unneeded kill_vid code

This driver has unneeded stubs for VLAN filtering.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years ago8139cp: fix VLAN unregistration
Stephen Hemminger [Fri, 1 Jun 2007 16:43:59 +0000 (09:43 -0700)]
8139cp: fix VLAN unregistration

The 8139cp driver did VLAN unregistration incorrectly.
It disables VLAN completely when the first VID is unregistered. It
should instead disable VLAN when the group is unregistered by calling
cp_vlan_rx_register with a NULL grp.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: Fix VLAN unregistration
Stephen Hemminger [Fri, 1 Jun 2007 16:43:58 +0000 (09:43 -0700)]
sky2: Fix VLAN unregistration

Fix sky2 disabling VLAN completely when the first vid is unregistered.

sky2 disables VLAN completely when the first VID is unregistered. It
should instead disable VLAN when the group is unregistered by calling
sky2_vlan_rx_register with grp = NULL.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
 drivers/net/sky2.c |   25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoVLAN: kill_vid is only useful for VLAN filtering devices
Stephen Hemminger [Fri, 1 Jun 2007 16:43:57 +0000 (09:43 -0700)]
VLAN: kill_vid is only useful for VLAN filtering devices

The interface for network device VLAN extension was confusing.
The kill_vid function is only really useful for devices that do
hardware filtering. Devices that only do VLAN receiption without
filtering were being forced to provide the hook, and there were
bugs in those devices.

Many drivers had kill_vid routine that called vlan_group_set_device, with
NULL, but that is done already.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoqla3xxx: device doesnt do hardware checksumming.
Stephen Hemminger [Wed, 30 May 2007 21:23:17 +0000 (14:23 -0700)]
qla3xxx: device doesnt do hardware checksumming.

Reading the code for ql_hw_csum_setup(), it is obvious that
this driver is broken for IPV6. The driver sets the NETIF_F_HW_SUM
flag, but the code for checksum setup only deals with IPV4.

Compile tested only, no hardware available.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years ago[POWERPC] Fix zImage.coff generation for 32-bit pmac
Milton Miller [Wed, 30 May 2007 15:29:01 +0000 (01:29 +1000)]
[POWERPC] Fix zImage.coff generation for 32-bit pmac

Commit 9da82a6dee9db4cd5ae7a74ab4f51afb52b6efb9 inadvertently
removed the platform override for zImage.coff to be generated
with pmaccoff.   Rather than add a special makefile rule,
change the platform for which the wrapper platform uses
the special rules.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix compile breakage for IBM/AMCC 4xx arch/ppc platforms
Stefan Roese [Sat, 2 Jun 2007 09:30:20 +0000 (19:30 +1000)]
[POWERPC] Fix compile breakage for IBM/AMCC 4xx arch/ppc platforms

The IBM/AMCC 405 platforms don't compile anymore in the current
kernel version.  This fixes the compile breakage.

Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Don't allow PMAC_APM_EMU for 64-bit
Johannes Berg [Sat, 2 Jun 2007 09:13:44 +0000 (19:13 +1000)]
[POWERPC] Don't allow PMAC_APM_EMU for 64-bit

In b302887854d6f0c6f9fc3f1080535e7c1bd53134 I switched the apm emulation
code to use the generic code but accidentally dropped the PPC32
dependency from the configuration symbol. This adds it back.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Compare irq numbers with NO_IRQ not IRQ_NONE
Michael Ellerman [Fri, 1 Jun 2007 07:23:26 +0000 (17:23 +1000)]
[POWERPC] Compare irq numbers with NO_IRQ not IRQ_NONE

There is a thinko in the irq code, it uses IRQ_NONE to indicate no irq,
whereas it should be using NO_IRQ.  IRQ_NONE is returned from irq
handlers to say "not handled".

As it happens they currently have the same value (0), so this is just for
future proof-ness.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix return from pte_alloc_one() in out-of-memory case
Akinobu Mita [Tue, 29 May 2007 10:46:51 +0000 (20:46 +1000)]
[POWERPC] Fix return from pte_alloc_one() in out-of-memory case

pte_alloc_one() is expected to return NULL if out of memory.
But it returns virt_to_page(NULL), which is not NULL.
This fixes it.

Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Akinobu Mita <mita@fixstars.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix compile warning in pseries xics code
Michael Neuling [Tue, 29 May 2007 07:01:52 +0000 (17:01 +1000)]
[POWERPC] Fix compile warning in pseries xics code

In 616883df78bd4b3fcdb6ddc39bd3d4cb902bfa32 request_irq was marked as
__must_check so we must... er... check it.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Don't use HOSTCFLAGS in BOOTCFLAGS
David Gibson [Tue, 29 May 2007 05:37:12 +0000 (15:37 +1000)]
[POWERPC] Don't use HOSTCFLAGS in BOOTCFLAGS

In the bootwrapper code for powerpc, we include HOSTCFLAGS into the
BOOTCFLAGS used for building the zImage wrapper code.  Since the
wrapper code is not host code, this makes no sense.  This patch
removes the use of HOSTCFLAGS here, instead including directly into
BOOTCFLAGS those flags from the normal kernel CFLAGS which also make
sense in the bootwrapper code.

In particular, this makes the bootwrapper use -msoft-float, preventing
the compiler from generating floating point instructions.  Previously,
under some circumstances the compiler could generate floating point
instructions in the bootwrapper which would cause exceptions on
embedded CPUS which don't have floating point support.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Acked-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Create a zImage for legacy iSeries
Stephen Rothwell [Mon, 28 May 2007 06:12:59 +0000 (16:12 +1000)]
[POWERPC] Create a zImage for legacy iSeries

This zImage is really just the stripped vmlinux, but it means that there
is one less special case for iSeries and also that the zImages will be
built for a combined kernel build that happens to include iSeries.

This zImage boots fine on legacy iSeries.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] pasemi idle uses hard_smp_processor_id
Stephen Rothwell [Mon, 28 May 2007 00:20:45 +0000 (10:20 +1000)]
[POWERPC] pasemi idle uses hard_smp_processor_id

and so needs to include asm/smp.h so a UP build works.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] ps3/interrupt.c uses get_hard_smp_processor_id
Stephen Rothwell [Mon, 28 May 2007 00:19:08 +0000 (10:19 +1000)]
[POWERPC] ps3/interrupt.c uses get_hard_smp_processor_id

and so needs to include asm/smp.h for a UP build to work.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Geoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix possible access to free pages
Benjamin Herrenschmidt [Sun, 27 May 2007 05:18:22 +0000 (15:18 +1000)]
[POWERPC] Fix possible access to free pages

I think we have a subtle race on ppc64 with the tlb batching.  The
common code expects tlb_flush() to actually flush any pending TLB
batch.  It does that because it delays all page freeing until after
tlb_flush() is called, in order to ensure no stale reference to
those pages exist in any TLB, thus causing potential access to
the freed pages.

However, our tlb_flush only triggers the RCU for freeing page
table pages, it does not currently trigger a flush of a pending
TLB/hash batch, which is, I think, an error.  This fixes it.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix compiler/assembler flags for Ebony platform boot files
David Gibson [Fri, 25 May 2007 03:19:17 +0000 (13:19 +1000)]
[POWERPC] Fix compiler/assembler flags for Ebony platform boot files

The recent addition of assembler flags for 44x.c and ebony.c in the
bootwrapper to make them compile on certain toolchains was not correct
and could break other platforms.  This patch switches to using a
compiler flag instead, which implies the appropriate assembler flag,
and also stops the compiler itself generating instructions which are
invalid for the platform in question.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Acked-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix ppc32 single-stepping out of syscalls
Benjamin Herrenschmidt [Thu, 24 May 2007 05:41:04 +0000 (15:41 +1000)]
[POWERPC] Fix ppc32 single-stepping out of syscalls

The ppc32 kernel didn't properly set/clear the TIF_SINGLESTEP
flag, causing return from syscalls to not SIGTRAP, thus executing
one more instruction before stopping again.

This fixes it.  The ptrace code is a bit of a mess, and is overdue
for at least a -proper- 32/64 bits split and possibly more cleanups
but this minimum fix should be ok for 2.6.22

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Update documentation for of_find_node_by_type()
Michael Ellerman [Wed, 23 May 2007 08:08:13 +0000 (18:08 +1000)]
[POWERPC] Update documentation for of_find_node_by_type()

The documentation for of_find_node_by_type() incorrectly refers to the
"name" parameter - it should be "type".

Also the behaviour when from == NULL is not really documented, fix that.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years agoPull osi-now into release branch
Len Brown [Sat, 2 Jun 2007 05:02:09 +0000 (01:02 -0400)]
Pull osi-now into release branch

17 years agoPull now into release branch
Len Brown [Sat, 2 Jun 2007 04:48:48 +0000 (00:48 -0400)]
Pull now into release branch

17 years agoACPICA: Support for external package objects as method arguments
Bob Moore [Tue, 3 Apr 2007 23:59:37 +0000 (19:59 -0400)]
ACPICA: Support for external package objects as method arguments

Implemented support to allow Package objects to be passed as
method arguments to the acpi_evaluate_object interface. Previously,
this would return an AE_NOT_IMPLEMENTED exception.

Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
17 years agoACPI: Section mismatch ... acpi_map_pxm_to_node
Luck, Tony [Thu, 24 May 2007 20:57:40 +0000 (13:57 -0700)]
ACPI: Section mismatch ... acpi_map_pxm_to_node

Last of the "Section mismatch" errors from ia64 builds! acpi_map_pxm_to_node()
is defined with attribute __cpuinit, but is called by "normal" kernel functions
acpi_getnode() and acpi_map_cpu2node().

Commit f363d16fbb9374c0bd7f2757d412c287169094c9 moved the data structures on
which this routine operates from __cpuinitdata to regular memory, so this
routine can also move out of init space.

Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>