linux-2.6
16 years ago[SUNRPC]: Remove SPIN_LOCK_UNLOCKED
Jiri Slaby [Thu, 22 Nov 2007 11:40:22 +0000 (19:40 +0800)] 
[SUNRPC]: Remove SPIN_LOCK_UNLOCKED

SPIN_LOCK_UNLOCKED is deprecated, use DEFINE_SPINLOCK instead

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
16 years ago[SUNRPC]: Make xprtsock.c:xs_setup_{udp,tcp}() static
Adrian Bunk [Thu, 22 Nov 2007 11:38:25 +0000 (19:38 +0800)] 
[SUNRPC]: Make xprtsock.c:xs_setup_{udp,tcp}() static

xs_setup_{udp,tcp}() can now become static.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
16 years ago[PFKEY]: Sending an SADB_GET responds with an SADB_GET
Charles Hardin [Thu, 22 Nov 2007 11:35:15 +0000 (19:35 +0800)] 
[PFKEY]: Sending an SADB_GET responds with an SADB_GET

From: Charles Hardin <chardin@2wire.com>

Kernel needs to respond to an SADB_GET with the same message type to
conform to the RFC 2367 Section 3.1.5

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
16 years ago[IRDA]: Compilation for CONFIG_INET=n case
Pavel Emelyanov [Thu, 22 Nov 2007 11:15:56 +0000 (19:15 +0800)] 
[IRDA]: Compilation for CONFIG_INET=n case

Found this occasionally.

The CONFIG_INET=n is hardly ever set, but if it is the
irlan_eth_send_gratuitous_arp() compilation should produce a
warning about unused variable in_dev.

Too pedantic? :)

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
16 years agommc: Avoid re-using minor numbers before the original device is closed.
David Woodhouse [Wed, 21 Nov 2007 17:45:12 +0000 (18:45 +0100)] 
mmc: Avoid re-using minor numbers before the original device is closed.

Move the code which marks the minor number as free to mmc_blk_put() so
that it happens on the final close() (or removal), instead of doing it
at removal even when the device is still logically open.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
16 years agotifm_sd: handle non-power-of-2 block sizes
Alex Dubov [Wed, 14 Nov 2007 12:55:36 +0000 (23:55 +1100)] 
tifm_sd: handle non-power-of-2 block sizes

It is possible to handle arbitrary block sizes with tifm card reader by
conditionally switching to PIO in case such block has to be delivered. At
the beginning of each request, DMA is either disabled (non-power-of-2 block
size) or set to load time user preference.

Signed-off-by: Alex Dubov <oakad@yahoo.com>
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
16 years agommc_block: check card state after write
Pierre Ossman [Fri, 2 Nov 2007 17:21:13 +0000 (18:21 +0100)] 
mmc_block: check card state after write

Some cards have been reported to signal that they're ready prematurely.
Checking both the busy bit and card state solves the issue.

Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
16 years agoBlackfin arch: fix bug when enable uart1 with uart0 disabled => no initial console
Sonic Zhang [Wed, 21 Nov 2007 09:04:41 +0000 (17:04 +0800)] 
Blackfin arch: fix bug when enable uart1 with uart0 disabled => no initial console

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: split apart dump_bfin_regs and merge/remove show_regs from process...
Mike Frysinger [Wed, 21 Nov 2007 08:46:49 +0000 (16:46 +0800)] 
Blackfin arch: split apart dump_bfin_regs and merge/remove show_regs from process.c, which was largely duplicated

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: use common __INIT/__FINIT defines rather than setting the .section...
Mike Frysinger [Wed, 21 Nov 2007 08:45:08 +0000 (16:45 +0800)] 
Blackfin arch: use common __INIT/__FINIT defines rather than setting the .section ourselves to .init.text

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: fix bug when sending signals with the wrong PC, cause gdb get confused
Robin Getz [Wed, 21 Nov 2007 08:38:05 +0000 (16:38 +0800)] 
Blackfin arch: fix bug when sending signals with the wrong PC, cause gdb get confused

We need to send signals with the proper PC, or gdb gets
confused, and lots of tests fail. This should fix that.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: Ensure we printk out strings with the proper loglevel
Robin Getz [Wed, 21 Nov 2007 08:35:57 +0000 (16:35 +0800)] 
Blackfin arch: Ensure we printk out strings with the proper loglevel

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: Need to specify ax with the .init.text section,
Bernd Schmidt [Wed, 21 Nov 2007 08:33:47 +0000 (16:33 +0800)] 
Blackfin arch: Need to specify ax with the .init.text section,

Need to specify "ax" with the .init.text section, otherwise the linker will
make unique .init.text.1 .. .init.text.3 sections to cope with the flags

16 years agoBlackfin arch: Update Kconfig to latest Blackfin silicon datasheets
Robin Getz [Wed, 21 Nov 2007 08:29:20 +0000 (16:29 +0800)] 
Blackfin arch: Update Kconfig to latest Blackfin silicon datasheets

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: update defconfig files
Mike Frysinger [Wed, 21 Nov 2007 08:23:51 +0000 (16:23 +0800)] 
Blackfin arch: update defconfig files

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: Fix typo, and add ENDPROC - no functional changes
Mike Frysinger [Wed, 21 Nov 2007 08:15:48 +0000 (16:15 +0800)] 
Blackfin arch: Fix typo, and add ENDPROC - no functional changes

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: convert READY to DMA_READY as it causes build errors in common sound...
Mike Frysinger [Wed, 21 Nov 2007 08:17:11 +0000 (16:17 +0800)] 
Blackfin arch: convert READY to DMA_READY as it causes build errors in common sound code otherwise

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: add defines for the on-chip L1 ROM of BF54x
Mike Frysinger [Wed, 21 Nov 2007 08:14:03 +0000 (16:14 +0800)] 
Blackfin arch: add defines for the on-chip L1 ROM of BF54x

Should not need separate cplb entry though as we cover L1 with a 4 meg page

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: cplb and map header file cleanup
Mike Frysinger [Wed, 21 Nov 2007 08:12:12 +0000 (16:12 +0800)] 
Blackfin arch: cplb and map header file cleanup

 - remove duplicated defines for the BF561
 - generalize L2 support (so that it works for BF54x) and mark it executable
 - add support for reading/executing the Boot ROM sections
   (since it has data/functions we may need at runtime)
 - and fixup names for each map

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: cleanup the cplb declares
Mike Frysinger [Wed, 21 Nov 2007 07:55:45 +0000 (15:55 +0800)] 
Blackfin arch: cleanup the cplb declares

 - no need to declare their sizes in the common header
 - no need to tack on the section attribute as only the definition matters, not references

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: fix broken on BF52x, remove silly checks on processors for L1_SCRATCH...
Mike Frysinger [Wed, 21 Nov 2007 07:57:53 +0000 (15:57 +0800)] 
Blackfin arch: fix broken on BF52x, remove silly checks on processors for L1_SCRATCH defines

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: add support for working around anomaly 05000312
Mike Frysinger [Wed, 21 Nov 2007 08:08:58 +0000 (16:08 +0800)] 
Blackfin arch: add support for working around anomaly 05000312

Anomaly 05000312 - Errors When SSYNC, CSYNC, or Loads to LT, LB and LC Registers Are Interrupted:

DESCRIPTION:
When instruction cache is enabled, erroneous behavior may occur when any of the following instructions are interrupted:

. CSYNC
• SSYNC
• LCx =
• LTx = (only when LCx is non-zero)
• LBx = (only when LCx is non-zero)

When this problem occurs, a variety of incorrect things could happen, including an illegal instruction exception. Additional errors could
show up as an exception, a hardware error, or an instruction that is valid but different than the one that was expected.

WORKAROUND:
Place a cli before all SSYNC, CSYNC, "LCx =", "LTx =", and "LBx =" instructions to disable interrupts, and place an sti after each of these
instructions to re-enable interrupts. When these instructions are executed in code that is already non-interruptible, the problem will not
occur.

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: cleanup BF54x header file and add BF547 definition
Mike Frysinger [Wed, 21 Nov 2007 08:04:49 +0000 (16:04 +0800)] 
Blackfin arch: cleanup BF54x header file and add BF547 definition

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: fix building for BF542 processors which only have 1 TWI
Mike Frysinger [Wed, 21 Nov 2007 08:03:07 +0000 (16:03 +0800)] 
Blackfin arch: fix building for BF542 processors which only have 1 TWI

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: rename _return_from_exception to _bfin_return_from_exception and export it
Mike Frysinger [Wed, 21 Nov 2007 07:53:49 +0000 (15:53 +0800)] 
Blackfin arch: rename _return_from_exception to _bfin_return_from_exception and export it

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: move EXPORT_SYMBOL() to C files where the symbol is actually defined
Mike Frysinger [Wed, 21 Nov 2007 07:34:51 +0000 (15:34 +0800)] 
Blackfin arch: move EXPORT_SYMBOL() to C files where the symbol is actually defined

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: fix bug NOR Flash MTD mount fail
Sonic Zhang [Wed, 21 Nov 2007 15:49:52 +0000 (23:49 +0800)] 
Blackfin arch: fix bug NOR Flash MTD mount fail

Config EBIU flash mode properly. EBIU_MODE EBIU_FCTL
EBIU_MBSCTL register should be configurd to Flash
mode.

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years agoBlackfin arch: export symbol get_dma_curr_desc_ptr for driver usage
Cliff Cai [Wed, 21 Nov 2007 07:22:58 +0000 (15:22 +0800)] 
Blackfin arch: export symbol get_dma_curr_desc_ptr for driver usage

Signed-off-by: Cliff Cai <cliff.cai@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
16 years ago[IPVS]: Fix compiler warning about unused register_ip_vs_protocol
Pavel Emelyanov [Wed, 21 Nov 2007 01:44:01 +0000 (17:44 -0800)] 
[IPVS]: Fix compiler warning about unused register_ip_vs_protocol

This is silly, but I have turned the CONFIG_IP_VS to m,
to check the compilation of one (recently sent) fix
and set all the CONFIG_IP_VS_PROTO_XXX options to n to
speed up the compilation.

In this configuration the compiler warns me about

  CC [M]  net/ipv4/ipvs/ip_vs_proto.o
net/ipv4/ipvs/ip_vs_proto.c:49: warning: 'register_ip_vs_protocol' defined but not used

Indeed. With no protocols selected there are no
calls to this function - all are compiled out with
ifdefs.

Maybe the best fix would be to surround this call with
ifdef-s or tune the Kconfig dependences, but I think that
marking this register function as __used is enough. No?

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[ARP]: Fix arp reply when sender ip 0
Jonas Danielsson [Wed, 21 Nov 2007 01:38:16 +0000 (17:38 -0800)] 
[ARP]: Fix arp reply when sender ip 0

Fix arp reply when received arp probe with sender ip 0.

Send arp reply with target ip address 0.0.0.0 and target hardware
address set to hardware address of requester. Previously sent reply
with target ip address and target hardware address set to same as
source fields.

Signed-off-by: Jonas Danielsson <the.sator@gmail.com>
Acked-by: Alexey Kuznetov <kuznet@ms2.inr.ac.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV6] TCPMD5: Fix deleting key operation.
YOSHIFUJI Hideaki [Wed, 21 Nov 2007 01:31:23 +0000 (17:31 -0800)] 
[IPV6] TCPMD5: Fix deleting key operation.

Due to the bug, refcnt for md5sig pool was leaked when
an user try to delete a key if we have more than one key.
In addition to the leakage, we returned incorrect return
result value for userspace.

This fix should close Bug #9418, reported by <ming-baini@163.com>.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV6] TCPMD5: Check return value of tcp_alloc_md5sig_pool().
YOSHIFUJI Hideaki [Wed, 21 Nov 2007 01:30:56 +0000 (17:30 -0800)] 
[IPV6] TCPMD5: Check return value of tcp_alloc_md5sig_pool().

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV4] TCPMD5: Use memmove() instead of memcpy() because we have overlaps.
YOSHIFUJI Hideaki [Wed, 21 Nov 2007 01:30:31 +0000 (17:30 -0800)] 
[IPV4] TCPMD5: Use memmove() instead of memcpy() because we have overlaps.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV4] TCPMD5: Omit redundant NULL check for kfree() argument.
YOSHIFUJI Hideaki [Wed, 21 Nov 2007 01:30:06 +0000 (17:30 -0800)] 
[IPV4] TCPMD5: Omit redundant NULL check for kfree() argument.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'fixes-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville...
David S. Miller [Wed, 21 Nov 2007 01:24:29 +0000 (17:24 -0800)] 
Merge branch 'fixes-davem' of git://git./linux/kernel/git/linville/wireless-2.6

16 years agoPull bugzilla-9327 into release branch
Len Brown [Wed, 21 Nov 2007 01:07:49 +0000 (20:07 -0500)] 
Pull bugzilla-9327 into release branch

Conflicts:

drivers/acpi/ec.c

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: EC: Workaround for optimized controllers (version 3)
Alexey Starikovskiy [Wed, 21 Nov 2007 00:23:32 +0000 (03:23 +0300)] 
ACPI: EC: Workaround for optimized controllers (version 3)

Some controllers fail to send confirmation GPE after address or data write.
Detect this and don't expect such confirmation in future.
This is a generalization of previous workaround
(66c5f4e7367b0085652931b2f3366de29e7ff5ec), which did only read address.

Reference: http://bugzilla.kernel.org/show_bug.cgi?id=9327

Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de>
Tested-by: Romano Giannetti <romano.giannetti@gmail.com>
Tested-by: Mats Johannesson
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: EC: use printk_ratelimit(), add some DEBUG mode messages
Márton Németh [Wed, 21 Nov 2007 00:23:26 +0000 (03:23 +0300)] 
ACPI: EC: use printk_ratelimit(), add some DEBUG mode messages

Sometimes it is usefull to see raw protocol dump.
Uncomment '#define DEBUG' at the beginning of file to make EC
really verbose.

Signed-off-by: Márton Németh <nm127@freemail.hu>
Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoRevert "ACPI: EC: Workaround for optimized controllers"
Len Brown [Wed, 21 Nov 2007 00:59:08 +0000 (19:59 -0500)] 
Revert "ACPI: EC: Workaround for optimized controllers"

This reverts commit f2d68935ba08cf80f151bbdb5628381184e4a498.

16 years ago[CIFS] Fix potential data corruption when writing out cached dirty pages
Jeff Layton [Tue, 20 Nov 2007 23:19:03 +0000 (23:19 +0000)] 
[CIFS] Fix potential data corruption when writing out cached dirty pages

Fix RedHat bug 329431

The idea here is separate "conscious" from "unconscious" flushes.
Conscious flushes are those due to a fsync() or close(). Unconscious
ones are flushes that occur as a side effect of some other operation or
due to memory pressure.

Currently, when an error occurs during an unconscious flush (ENOSPC or
EIO), we toss out the page and don't preserve that error to report to
the user when a conscious flush occurs. If after the unconscious flush,
there are no more dirty pages for the inode, the conscious flush will
simply return success even though there were previous errors when writing
out pages. This can lead to data corruption.

The easiest way to reproduce this is to mount up a CIFS share that's
very close to being full or where the user is very close to quota. mv
a file to the share that's slightly larger than the quota allows. The
writes will all succeed (since they go to pagecache). The mv will do a
setattr to set the new file's attributes. This calls
filemap_write_and_wait,
which will return an error since all of the pages can't be written out.
Then later, when the flush and release ops occur, there are no more
dirty pages in pagecache for the file and those operations return 0. mv
then assumes that the file was written out correctly and deletes the
original.

CIFS already has a write_behind_rc variable where it stores the results
from earlier flushes, but that value is only reported in cifs_close.
Since the VFS ignores the return value from the release operation, this
isn't helpful. We should be reporting this error during the flush
operation.

This patch does the following:

1) changes cifs_fsync to use filemap_write_and_wait and cifs_flush and also
sync to check its return code. If it returns successful, they then check
the value of write_behind_rc to see if an earlier flush had reported any
errors. If so, they return that error and clear write_behind_rc.

2) sets write_behind_rc in a few other places where pages are written
out as a side effect of other operations and the code waits on them.

3) changes cifs_setattr to only call filemap_write_and_wait for
ATTR_SIZE changes.

4) makes cifs_writepages accurately distinguish between EIO and ENOSPC
errors when writing out pages.

Some simple testing indicates that the patch works as expected and that
it fixes the reproduceable known problem.

Acked-by: Dave Kleikamp <shaggy@austin.rr.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
16 years agoieee80211: Stop net_ratelimit/IEEE80211_DEBUG_DROP log pollution
Guillaume Chazarain [Mon, 19 Nov 2007 09:07:00 +0000 (10:07 +0100)] 
ieee80211: Stop net_ratelimit/IEEE80211_DEBUG_DROP log pollution

if (net_ratelimit())
IEEE80211_DEBUG_DROP(...)

can pollute the logs with messages like:

printk: 1 messages suppressed.
printk: 2 messages suppressed.
printk: 7 messages suppressed.

if debugging information is disabled. These messages are printed by
net_ratelimit(). Add a wrapper to net_ratelimit() that takes into account
the log level, so that net_ratelimit() is called only when we really want
to print something.

Signed-off-by: Guillaume Chazarain <guichaz@yahoo.fr>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: add missing space in error message
Bruno Randolf [Fri, 16 Nov 2007 08:04:01 +0000 (17:04 +0900)] 
mac80211: add missing space in error message

Signed-off-by: Bruno Randolf <bruno@thinktube.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years ago[POWERPC] 83xx: Update mpc8349emitx(gp) defconfig for USB
Grant Likely [Wed, 17 Oct 2007 23:41:00 +0000 (17:41 -0600)] 
[POWERPC] 83xx: Update mpc8349emitx(gp) defconfig for USB

USB support for the 8349itx got added a while back; but the defconfig
never got updated.  This patch adds the appropriate USB config options
to the defconfigs

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
CC: Scott Wood <scottwood@freescale.com>
CC: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
16 years ago[POWERPC] 83xx: Update mpc832x_rdb_defconfig to enable MMC-over-SPI
Anton Vorontsov [Tue, 13 Nov 2007 17:00:43 +0000 (20:00 +0300)] 
[POWERPC] 83xx: Update mpc832x_rdb_defconfig to enable MMC-over-SPI

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
16 years ago[POWERPC] 83xx: MPC832x RDB - remove spidev stub, use mmc_spi
Anton Vorontsov [Tue, 13 Nov 2007 17:00:11 +0000 (20:00 +0300)] 
[POWERPC] 83xx: MPC832x RDB - remove spidev stub, use mmc_spi

mmc_spi has hit the mainline, so we can start using it.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
16 years ago[POWERPC] 8xxx: MDS board RTC fixes
Kim Phillips [Thu, 8 Nov 2007 19:37:06 +0000 (13:37 -0600)] 
[POWERPC] 8xxx: MDS board RTC fixes

Now the rtc class ds1374 driver has been added, remove the old rtc
driver hookup code, add rtc node to device trees, and turn on the
new driver in the defconfigs.

Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
16 years agomac80211: fix allmulti/promisc behaviour
Johannes Berg [Fri, 16 Nov 2007 01:54:53 +0000 (02:54 +0100)] 
mac80211: fix allmulti/promisc behaviour

When an interface with promisc/allmulti bit is taken down,
the mac80211 state can become confused. This fixes it by
making mac80211 keep track of all *active* interfaces that
have the promisc/allmulti bit set in the sdata, we sync
the interface bit into sdata at set_multicast_list() time
so this works.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: fix ieee80211_set_multicast_list
Johannes Berg [Fri, 16 Nov 2007 00:49:11 +0000 (01:49 +0100)] 
mac80211: fix ieee80211_set_multicast_list

I recently experienced unexplainable behaviour with the b43
driver when I had broken firmware uploaded. The cause may have
been that promisc mode was not correctly enabled or disabled
and this bug may have been the cause.

Note how the values are compared later in the function so
just doing the & will result in the wrong thing being
compared and the test being false almost always.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years ago[ALSA] version 1.0.15
Jaroslav Kysela [Tue, 20 Nov 2007 19:16:43 +0000 (20:16 +0100)] 
[ALSA] version 1.0.15

Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] emu10k1 - Check value ranges in ctl callbacks
Takashi Iwai [Thu, 15 Nov 2007 12:16:02 +0000 (13:16 +0100)] 
[ALSA] emu10k1 - Check value ranges in ctl callbacks

Check value ranges in ctl callbacks properly.  This fixes the unexpected
crash due to wrong value assignment.
Also, remove invalid comments in the last patch.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] emu10k1: Add mixer controls parameter checking.
James Courtier-Dutton [Mon, 12 Nov 2007 14:55:19 +0000 (14:55 +0000)] 
[ALSA] emu10k1: Add mixer controls parameter checking.

Signed-off-by: James Courtier-Dutton <James@superbug.co.uk>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] fix private data pointer calculation in CS4270 driver
Timur Tabi [Wed, 14 Nov 2007 11:07:58 +0000 (12:07 +0100)] 
[ALSA] fix private data pointer calculation in CS4270 driver

Fix the calculation of the private_data pointer in the CS4270 driver.

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] portman2x4 - Fix probe error
Takashi Iwai [Mon, 29 Oct 2007 10:14:31 +0000 (11:14 +0100)] 
[ALSA] portman2x4 - Fix probe error

Reported by Ingo Molnar,
when booting an allyesconfig bzImage kernel the bootup hangs in the
portman2x4 driver (on a box that does not have this hardware), at:
 Pid: 1, comm:              swapper
 EIP: 0060:[<c02f763c>] CPU: 0
 EIP is at parport_pc_read_status+0x4/0x8
  EFLAGS: 00000202    Not tainted  (2.6.23-rc9 #904)
 EAX: f7e57a7f EBX: 00000010 ECX: c2b808c0 EDX: 00000379
 ESI: f7cb8230 EDI: 00000010 EBP: f7cb8230 DS: 007b ES: 007b FS: 0000
 CR0: 8005003b CR2: fff9c000 CR3: 007ec000 CR4: 00000690
 DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
 DR6: ffff0ff0 DR7: 00000400
  [<c04613de>] portman_flush_input+0xde/0x12c
  [<c0461a24>] snd_portman_probe+0x368/0x484
  [<c02fbb8c>] __device_attach+0x0/0x8
  [<c02fce68>] platform_drv_probe+0xc/0x10
  [<c02fba6c>] driver_probe_device+0x74/0x194
  [<c0587174>] klist_next+0x38/0x70
  [<c02fbb8c>] __device_attach+0x0/0x8
  [<c02faea1>] bus_for_each_drv+0x35/0x68
  [<c02fbc22>] device_attach+0x72/0x78
the reason is due to an inconsistent error return code of 1 or 2, while
snd_portman_probe only realizes negative error codes.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] ca0106 - Fix write proc assignment
Takashi Iwai [Wed, 24 Oct 2007 16:02:17 +0000 (18:02 +0200)] 
[ALSA] ca0106 - Fix write proc assignment

The driver assigns the write proc callback to read wrongly.
Fixed now.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] s3c2443-ac97: compilation fix
Krzysztof Helt [Fri, 19 Oct 2007 06:23:00 +0000 (08:23 +0200)] 
[ALSA] s3c2443-ac97: compilation fix

The Samsung S3C24xx uses new architecture file layout in the post 2.6.23
kernel.  This patch fixes include path for the s3c2443-ac97.c.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[ALSA] hda-codec - Revert volume knob controls in STAC codecs
Takashi Iwai [Mon, 19 Nov 2007 10:56:26 +0000 (11:56 +0100)] 
[ALSA] hda-codec - Revert volume knob controls in STAC codecs

Volume knob controls with STAC codecs seem to cause problems with some
devices.  Volumes change very slowly or silent suddenly.  It's likely
due to conflict between the software and the hardware volume knob
setup.
Since we'll have a virtual master control in future, it's safer to
remove this control completely right now.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jaroslav Kysela <perex@perex.cz>
16 years ago[NETFILTER]: Fix kernel panic with REDIRECT target.
Evgeniy Polyakov [Tue, 20 Nov 2007 12:27:35 +0000 (04:27 -0800)] 
[NETFILTER]: Fix kernel panic with REDIRECT target.

When connection tracking entry (nf_conn) is about to copy itself it can
have some of its extension users (like nat) as being already freed and
thus not required to be copied.

Actually looking at this function I suspect it was copied from
nf_nat_setup_info() and thus bug was introduced.

Report and testing from David <david@unsolicited.net>.

[ Patrick McHardy states:

I now understand whats happening:

- new connection is allocated without helper
- connection is REDIRECTed to localhost
- nf_nat_setup_info adds NAT extension, but doesn't initialize it yet
- nf_conntrack_alter_reply performs a helper lookup based on the
   new tuple, finds the SIP helper and allocates a helper extension,
   causing reallocation because of too little space
- nf_nat_move_storage is called with the uninitialized nat extension

So your fix is entirely correct, thanks a lot :)  ]

Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[WIRELESS] WEXT: Fix userspace corruption on 64-bit.
David S. Miller [Tue, 20 Nov 2007 11:29:53 +0000 (03:29 -0800)] 
[WIRELESS] WEXT: Fix userspace corruption on 64-bit.

On 64-bit systems sizeof(struct ifreq) is 8 bytes larger than
sizeof(struct iwreq).

For GET calls, the wireless extension code copies back into userspace
using sizeof(struct ifreq) but userspace and elsewhere only allocates
a "struct iwreq".  Thus, this copy writes past the end of the iwreq
object and corrupts whatever sits after it in memory.

Fix the copy_to_user() length.

This particularly hurts the compat case because the wireless compat
code uses compat_alloc_userspace() and right after this allocated
buffer is the current bottom of the user stack, and that's what gets
overwritten by the copy_to_user() call.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[S390] cio: Register/unregister subchannels only from kslowcrw.
Cornelia Huck [Tue, 20 Nov 2007 10:13:41 +0000 (11:13 +0100)] 
[S390] cio: Register/unregister subchannels only from kslowcrw.

Make sure all subchannel handling is done on the slow path workqueue
so that we don't have races between an old subchannel unregistering
and a new subchannel with the same name registering.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Add missing die_notifier() call to die().
Heiko Carstens [Tue, 20 Nov 2007 10:13:40 +0000 (11:13 +0100)] 
[S390] Add missing die_notifier() call to die().

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Fix memory detection.
Heiko Carstens [Tue, 20 Nov 2007 10:13:39 +0000 (11:13 +0100)] 
[S390] Fix memory detection.

Before we're getting short on memory detection fixes here is the next
one: if neither sclp nor diag260 report the storage size the detection
loop will return immediately without detecting anything. Fix this by
breaking the detection loop only if the memory end is known.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Explicitly code allocpercpu calls in iucv
Christoph Lameter [Tue, 20 Nov 2007 10:13:38 +0000 (11:13 +0100)] 
[S390] Explicitly code allocpercpu calls in iucv

The iucv is the only user of the various functions that are used to bring
parts of cpus up and down. Its the only allocpercpu user that will do
I/O on per cpu objects (which is difficult to do with virtually mapped memory).
And its the only use of allocpercpu where a GFP_DMA allocation is done.

Remove the allocpercpu calls from iucv and code the allocation and freeing
manually. After this patch it is possible to remove a large part of
the allocpercpu API.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Dont overwrite lowcores on smp_send_stop().
Heiko Carstens [Tue, 20 Nov 2007 10:13:37 +0000 (11:13 +0100)] 
[S390] Dont overwrite lowcores on smp_send_stop().

Don't perform a sigp store-status-at-address on smp_send_stop().
It will overwrite the lowcores of other cpus and destroys valueable
debug informations.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Optimize storage key handling for anonymous pages
Christian Borntraeger [Tue, 20 Nov 2007 10:13:36 +0000 (11:13 +0100)] 
[S390] Optimize storage key handling for anonymous pages

page_mkclean used to call page_clear_dirty for every given page. This
is different to all other architectures, where the dirty bit in the
PTEs is only resetted, if page_mapping() returns a non-NULL pointer.
We can move the page_test_dirty/page_clear_dirty sequence into the
2nd if to avoid unnecessary iske/sske sequences, which are expensive.

This change also helps kvm for s390 as the host must transfer the
dirty bit into the guest status bits. By moving the page_clear_dirty
operation into the 2nd if, the vm will only call page_clear_dirty
for pages where it walks the mapping anyway. There it calls
ptep_clear_flush for writable ptes, so we can transfer the dirty bit
to the guest.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Fix kernel preemption.
Heiko Carstens [Tue, 20 Nov 2007 10:13:35 +0000 (11:13 +0100)] 
[S390] Fix kernel preemption.

When returning from IRQ handling and TIF_NEED_RESCHED is set we must
call preempt_schedule_irq() instead of schedule().
Otherwise the BKL might be unlocked in schedule() and therfore
everything that relies on the BKL is broken.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] appldata: remove unused binary sysctls.
Heiko Carstens [Tue, 20 Nov 2007 10:13:34 +0000 (11:13 +0100)] 
[S390] appldata: remove unused binary sysctls.

Remove binary sysctls that never worked due to missing strategy functions.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Gerald Schaefer <geraldsc@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] cmm: remove unused binary sysctls.
Heiko Carstens [Tue, 20 Nov 2007 10:13:33 +0000 (11:13 +0100)] 
[S390] cmm: remove unused binary sysctls.

Remove binary sysctls that never worked due to missing strategy functions.

Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] Fix irq tracing and lockdep_sys_exit calls.
Heiko Carstens [Tue, 20 Nov 2007 10:13:32 +0000 (11:13 +0100)] 
[S390] Fix irq tracing and lockdep_sys_exit calls.

Current support for TRACE_IRQFLAGS and lockdep_sys_exit is broken.
IRQ flag tracing is broken for program checks. Even worse is that
the newly introduced calls to lockdep_sys_exit are in the critical
section code which is not supposed to call any C functions. In
addition the checks if locks are still held are also done when
returning to kernel code which is broken as well.
Fix all this by disabling interrupts and machine checks at the
exit paths and then do the appropriate checks and calls.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] magic sysrq: check for in_atomic before doing an console_unblank
Christian Borntraeger [Tue, 20 Nov 2007 10:13:31 +0000 (11:13 +0100)] 
[S390] magic sysrq: check for in_atomic before doing an console_unblank

When doing an magic sysrq reboot on s390 the following bug message
appears:

SysRq : Resetting
BUG: sleeping function called from invalid context at include/asm/semaphore.h:61

in_atomic():1, irqs_disabled():0
07000000004002a8 000000000fe6bc48 0000000000000002 0000000000000000
       000000000fe6bce8 000000000fe6bc60 000000000fe6bc60 000000000012a79a
       0000000000000000 07000000004002a8 0000000000000006 0000000000000000
       0000000000000000 000000000fe6bc48 000000000000000d 000000000fe6bcb8
       00000000004000c8 0000000000103234 000000000fe6bc48 000000000fe6bc90
Call Trace:
(¬<00000000001031b2>| show_trace+0x12e/0x148)
 ¬<000000000011ffca>| __might_sleep+0x10a/0x118
 ¬<0000000000129fba>| acquire_console_sem+0x92/0xf4
 ¬<000000000012a2ca>| console_unblank+0xc2/0xc8
 ¬<0000000000107bb4>| machine_restart+0x54/0x6c
 ¬<000000000028e806>| sysrq_handle_reboot+0x26/0x30
 ¬<000000000028e52a>| __handle_sysrq+0xa6/0x180
 ¬<0000000000140134>| run_workqueue+0xcc/0x18c
 ¬<000000000014029a>| worker_thread+0xa6/0x108
 ¬<00000000001458e4>| kthread+0x64/0x9c
 ¬<0000000000106f0e>| kernel_thread_starter+0x6/0xc
 ¬<0000000000106f08>| kernel_thread_starter+0x0/0xc

The only reason for doing a console_unblank on s390 is to flush the
log buffer. We have to check for in_atomic before doing a
console_unblank as the console is otherwise filled with an unrelated
bug message.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[S390] cio: change device sense procedure to work with pav aliases
Peter Oberparleiter [Tue, 20 Nov 2007 10:13:30 +0000 (11:13 +0100)] 
[S390] cio: change device sense procedure to work with pav aliases

Modify the sense id channel program to allow device sensing of pav
alias devices which belong to a base device with ungrouped paths.

Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
16 years ago[IRDA]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:48:30 +0000 (23:48 -0800)] 
[IRDA]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[SUNRPC]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:48:08 +0000 (23:48 -0800)] 
[SUNRPC]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[SCTP]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:47:47 +0000 (23:47 -0800)] 
[SCTP]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV6]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:47:25 +0000 (23:47 -0800)] 
[IPV6]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[BRIDGE]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:46:55 +0000 (23:46 -0800)] 
[BRIDGE]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPV4]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:46:29 +0000 (23:46 -0800)] 
[IPV4]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[DCCP]: Add missing "space"
Joe Perches [Tue, 20 Nov 2007 07:46:02 +0000 (23:46 -0800)] 
[DCCP]: Add missing "space"

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[POWERPC] Fix 8xx build breakage due to _tlbie changes
Benjamin Herrenschmidt [Tue, 20 Nov 2007 07:32:12 +0000 (18:32 +1100)] 
[POWERPC] Fix 8xx build breakage due to _tlbie changes

My changes to _tlbie to fix 4xx unfortunately broke 8xx build in a
couple of places.  This fixes it.

Spotted by Olof Johansson.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Vitaly Bordug <vitb@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years agoMerge branch 'for-2.6.24' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerp...
Paul Mackerras [Tue, 20 Nov 2007 07:40:23 +0000 (18:40 +1100)] 
Merge branch 'for-2.6.24' of /linux/kernel/git/galak/powerpc into merge

16 years ago[TCP]: Problem bug with sysctl_tcp_congestion_control function
Sam Jansen [Tue, 20 Nov 2007 07:28:21 +0000 (23:28 -0800)] 
[TCP]: Problem bug with sysctl_tcp_congestion_control function

From: "Sam Jansen" <sjansen@google.com>

sysctl_tcp_congestion_control seems to have a bug that prevents it
from actually calling the tcp_set_default_congestion_control
function. This is not so apparent because it does not return an error
and generally the /proc interface is used to configure the default TCP
congestion control algorithm.  This is present in 2.6.18 onwards and
probably earlier, though I have not inspected 2.6.15--2.6.17.

sysctl_tcp_congestion_control calls sysctl_string and expects a successful
return code of 0. In such a case it actually sets the congestion control
algorithm with tcp_set_default_congestion_control. Otherwise, it returns the
value returned by sysctl_string. This was correct in 2.6.14, as sysctl_string
returned 0 on success. However, sysctl_string was updated to return 1 on
success around about 2.6.15 and sysctl_tcp_congestion_control was not updated.
Even though sysctl_tcp_congestion_control returns 1, do_sysctl_strategy
converts this return code to '0', so the caller never notices the error.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[TCP] MTUprobe: fix potential sk_send_head corruption
Ilpo Järvinen [Tue, 20 Nov 2007 07:24:09 +0000 (23:24 -0800)] 
[TCP] MTUprobe: fix potential sk_send_head corruption

When the abstraction functions got added, conversion here was
made incorrectly. As a result, the skb may end up pointing
to skb which got included to the probe skb and then was freed.
For it to trigger, however, skb_transmit must fail sending as
well.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[PKTGEN]: Fix double unlock of xfrm_state->lock
Pavel Emelyanov [Tue, 20 Nov 2007 06:51:24 +0000 (22:51 -0800)] 
[PKTGEN]: Fix double unlock of xfrm_state->lock

The pktgen_output_ipsec() function can unlock this lock twice
due to merged error and plain paths. Remove one of the calls
to spin_unlock.

Other possible solution would be to place "return 0" right
after the first unlock, but at this place the err is known
to be 0, so these solutions are the same except for this one
makes the code shorter.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoPull fluff into release branch
Len Brown [Tue, 20 Nov 2007 06:21:47 +0000 (01:21 -0500)] 
Pull fluff into release branch

Conflicts:

drivers/acpi/ec.c

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoPull video-2.6.24 into release branch
Len Brown [Tue, 20 Nov 2007 06:20:57 +0000 (01:20 -0500)] 
Pull video-2.6.24 into release branch

16 years agoPull thinkpad-2.6.24 into release branch
Len Brown [Tue, 20 Nov 2007 06:20:42 +0000 (01:20 -0500)] 
Pull thinkpad-2.6.24 into release branch

16 years agoPull thermal into release branch
Len Brown [Tue, 20 Nov 2007 06:20:31 +0000 (01:20 -0500)] 
Pull thermal into release branch

16 years agoPull procfs-default into release branch
Len Brown [Tue, 20 Nov 2007 06:20:00 +0000 (01:20 -0500)] 
Pull procfs-default into release branch

Conflicts:

drivers/acpi/sbs.c

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoPull cpuidle into release branch
Len Brown [Tue, 20 Nov 2007 06:18:37 +0000 (01:18 -0500)] 
Pull cpuidle into release branch

16 years agoPull bugzilla-9327 into release branch
Len Brown [Tue, 20 Nov 2007 06:18:19 +0000 (01:18 -0500)] 
Pull bugzilla-9327 into release branch

16 years agoPull bugzilla-9262 into release branch
Len Brown [Tue, 20 Nov 2007 06:18:07 +0000 (01:18 -0500)] 
Pull bugzilla-9262 into release branch

16 years agoPull bugzilla-9153 into release branch
Len Brown [Tue, 20 Nov 2007 06:17:54 +0000 (01:17 -0500)] 
Pull bugzilla-9153 into release branch

16 years agoPull battery into release branch
Len Brown [Tue, 20 Nov 2007 06:17:42 +0000 (01:17 -0500)] 
Pull battery into release branch

16 years agoACPI: fix two IRQ8 issues in IOAPIC mode
Shaohua Li [Sat, 17 Nov 2007 06:05:28 +0000 (01:05 -0500)] 
ACPI: fix two IRQ8 issues in IOAPIC mode

Use mp_irqs[] to get PNP device's interrupt polarity and trigger.
There are two reasons to do this:
1. BIOS bug for PNP interrupt
2. BIOS explictly does override
mp_irqs[] should cover all the cases.

http://bugzilla.kernel.org/show_bug.cgi?id=5243
http://bugzilla.kernel.org/show_bug.cgi?id=7679
http://bugzilla.kernel.org/show_bug.cgi?id=9153

[lenb: fixed !IOAPIC and 64-bit !SMP builds]

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years ago[IPVS]: Move remaining sysctl handlers over to CTL_UNNUMBERED
Simon Horman [Tue, 20 Nov 2007 05:51:13 +0000 (21:51 -0800)] 
[IPVS]: Move remaining sysctl handlers over to CTL_UNNUMBERED

Switch the remaining IPVS sysctl entries over to to use CTL_UNNUMBERED,
I stronly doubt that anyone is using the sys_sysctl interface to
these variables.

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[IPVS]: Fix sysctl warnings about missing strategy in schedulers
Simon Horman [Tue, 20 Nov 2007 05:50:21 +0000 (21:50 -0800)] 
[IPVS]: Fix sysctl warnings about missing strategy in schedulers

sysctl table check failed: /net/ipv4/vs/lblc_expiration .3.5.21.19 Missing strategy
[...]
sysctl table check failed: /net/ipv4/vs/lblcr_expiration .3.5.21.20 Missing strategy

Switch these entried over to use CTL_UNNUMBERED as clearly
the sys_syscal portion wasn't working.

This is along the same lines as Christian Borntraeger's patch that fixes
up entries with no stratergy in net/ipv4/ipvs/ip_vs_ctl.c

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[POWERPC] Fix device tree interrupt map for Freescale ULI1575 boards
Kumar Gala [Tue, 20 Nov 2007 05:36:23 +0000 (23:36 -0600)] 
[POWERPC] Fix device tree interrupt map for Freescale ULI1575 boards

The interrupt map for the PCI PHB that had the ULI1575 was not correct
on the boards that have it.

* 8544 DS:
   - Fix interrupt mask
   - Be explicit about use of INTA for on chip peripherals

* 8572 DS/8641 HPCN:
   - Fix interrupt mask
   - Expand interrupt map for PCI slots to cover all functions
   - Be explicit about use of INTA for on chip peripherals

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
16 years ago[IPVS]: Fix sysctl warnings about missing strategy
Christian Borntraeger [Tue, 20 Nov 2007 05:49:25 +0000 (21:49 -0800)] 
[IPVS]: Fix sysctl warnings about missing strategy

Running the latest git code I get the following messages during boot:
sysctl table check failed: /net/ipv4/vs/drop_entry .3.5.21.4 Missing strategy
[...]
sysctl table check failed: /net/ipv4/vs/drop_packet .3.5.21.5 Missing strategy
[...]
sysctl table check failed: /net/ipv4/vs/secure_tcp .3.5.21.6 Missing strategy
[...]
sysctl table check failed: /net/ipv4/vs/sync_threshold .3.5.21.24 Missing strategy

I removed the binary sysctl handler for those messages and also removed
the definitions in ip_vs.h. The alternative would be to implement a
proper strategy handler, but syscall sysctl is deprecated.

There are other sysctl definitions that are commented out or work with
the default sysctl_data strategy. I did not touch these.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[POWERPC] Fix possible division by zero in scaled time accounting
Michael Neuling [Tue, 20 Nov 2007 04:18:40 +0000 (15:18 +1100)] 
[POWERPC] Fix possible division by zero in scaled time accounting

If we get no user time and no system time allocated since the last
account_system_vtime, the system to user time ratio estimate can end
up dividing by zero.

This was causing a problem noticed by Balbir Singh.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years ago[POWERPC] spufs: Fix context destroy vs /spu readdir race
Jeremy Kerr [Fri, 16 Nov 2007 02:32:23 +0000 (13:32 +1100)] 
[POWERPC] spufs: Fix context destroy vs /spu readdir race

We can currently cause an oops by repeatedly creating and destroying
contexts, while doing getdents() calls on the "/spu" directory.

This is due to the context's top-level dentry remaining hashed while
the context is being destroyed.

Fix this by unhashing the context's dentry with the
dentry->d_inode->i_mutex held. This way, we'll hit the check for
d_unhashed in dentry_readdir, and won't be included in the
list of subdirs for /spu.

test: spufs-testsuite:tests/01-spu_create/07-destroy-vs-readdir-race

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years ago[POWERPC] Fix RTAS os-term usage on kernel panic
Linas Vepstas [Tue, 20 Nov 2007 01:28:15 +0000 (12:28 +1100)] 
[POWERPC] Fix RTAS os-term usage on kernel panic

The rtas_os_term() routine was being called at the wrong time.
The actual rtas call "os-term" will not ever return, and so
calling it from the panic notifier is too early.  Instead,
call it from the machine_reset() call.

This splits the rtas_os_term() routine into two: one part to capture
the kernel panic message, invoked during the panic notifier, and
another part that is invoked during machine_reset().

Prior to this patch, the os-term call was never being made,
because panic_timeout was always non-zero.  Calling os-term
helps keep the hypervisor happy!  We have to keep the hypervisor
happy to avoid service, dump and error reporting problems.

Signed-off-by: Linas Vepstas <linas@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>