GitHub/exynos8895/android_kernel_samsung_universal8895.git
18 years ago[IA64] kprobe clears qp bits for special instructions
bibo,mao [Tue, 12 Dec 2006 20:04:42 +0000 (12:04 -0800)]
[IA64] kprobe clears qp bits for special instructions

On IA64 there exists some special instructions which
always need to be executed regradless of qp bits, such
as com.crel.unc, tbit.trel.unc etc.
This patch clears qp bits when inserting kprobe trap code
and disables probepoint on slot 1 for these special
instructions.

Signed-off-by: bibo,mao <bibo.mao@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] enable trap code on slot 1
Tony Luck [Tue, 14 Nov 2006 17:33:38 +0000 (09:33 -0800)]
[IA64] enable trap code on slot 1

Because slot 1 of one instr bundle crosses border of two consecutive
8-bytes, kprobe on slot 1 is disabled. This patch enables kprobe on
slot1, it only replaces higher 8-bytes of the instruction bundle and
changes the exception code to ignore the low 12 bits of the break
number (which is across the border in the lower 8-bytes of the bundle).

For those instructions which must execute regardless qp bits,
kprobe on slot 1 is still disabled.

Signed-off-by: bibo,mao <bibo.mao@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] Take defensive stance on ia64_pal_get_brand_info()
Tony Luck [Tue, 12 Dec 2006 19:56:36 +0000 (11:56 -0800)]
[IA64] Take defensive stance on ia64_pal_get_brand_info()

Stephane thought he saw a problem here (but was just confused
by the return value from ia64_pal_get_brand_info()).  But we
should be more defensive here in case an prototype PAL for
a future processor doesn't implement this PAL call.

Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] fix possible XPC deadlock when disconnecting
Dean Nelson [Wed, 22 Nov 2006 14:25:00 +0000 (08:25 -0600)]
[IA64] fix possible XPC deadlock when disconnecting

This patch eliminates a potential deadlock that is possible when XPC
disconnects a channel to a partition that has gone down. This deadlock will
occur if at least one of the kthreads created by XPC for the purpose of making
callouts to the channel's registerer is detained in the registerer and will
not be returning back to XPC until some registerer request occurs on the now
downed partition. The potential for a deadlock is removed by ensuring that
there always is a kthread available to make the channel disconnecting callout
to the registerer.

Signed-off-by: Dean Nelson <dcn@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] - Reduce overhead of FP exception logging messages
Jack Steiner [Wed, 22 Nov 2006 15:55:08 +0000 (09:55 -0600)]
[IA64] - Reduce overhead of FP exception logging messages

Improve the scalability of the fpswa code that rate-limits
logging of messages.

There are 2 distinctly different problems in this code.

1) If prctl is used to disable logging, last_time is never
   updated. The result is that fpu_swa_count is zeroed out on
   EVERY fp fault. This causes a very very hot cache line.
   The fix reduces the wallclock time of a 1024p FP exception test
   from 28734 sec to 19 sec!!!

2) On VERY large systems, excessive messages are logged because
   multiple cpus can each reset or increment fpu_swa_count at
   about the same time. The result is that hundreds of messages
   are logged each second. The fixes reduces the logging rate
   to ~1 per second.

Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] fix arch/ia64/mm/contig.c:235: warning: unused variable `nid'
Tony Luck [Tue, 12 Dec 2006 19:18:55 +0000 (11:18 -0800)]
[IA64] fix arch/ia64/mm/contig.c:235: warning: unused variable `nid'

This warning only shows up with CONFIG_VIRTUAL_MEM_MAP=y and
CONFIG_FLATMEM=y.

There is only one caller left for register_active_ranges() from the
contig.c code ... so it doesn't need to pick up the node number, the
node number is always zero.

Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] s/termios/ktermios/ in simserial.c
Tony Luck [Tue, 12 Dec 2006 18:47:36 +0000 (10:47 -0800)]
[IA64] s/termios/ktermios/ in simserial.c

This got missed in 606d099cdd1080bbb50ea50dc52d98252f8f10a1

Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] kexec/kdump: tidy up declaration of relocate_new_kernel_t
Horms [Tue, 12 Dec 2006 09:06:13 +0000 (18:06 +0900)]
[IA64] kexec/kdump: tidy up declaration of relocate_new_kernel_t

* Make NORET_TYPE and ATTRIB_NORET in line with the
  declaration for other architectures
* Add parameter names

Signed-Off-By: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] Kexec/Kdump: honour non-zero crashkernel offset.
Horms [Tue, 12 Dec 2006 09:08:10 +0000 (18:08 +0900)]
[IA64] Kexec/Kdump: honour non-zero crashkernel offset.

There seems to be a value in both allowing the kernel to determine
the base offset of the crashkernel automatically and allowing
users's to sepcify it.

The old behaviour on ia64, which is still the current behaviour on
most architectures is for the user to always specify the address.
Recently ia64 was changed so that it is always automatically determined.

With this patch the kernel automatically determines the offset if
the supplied value is 0, otherwise it uses the value provided.

This should probably be backed by a documentation change.

Signed-Off-By: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] CONFIG_KEXEC/CONFIG_CRASH_DUMP permutations
Horms [Tue, 12 Dec 2006 08:49:03 +0000 (17:49 +0900)]
[IA64] CONFIG_KEXEC/CONFIG_CRASH_DUMP permutations

Actually, on reflection I think that there is a good case for
keeping the options separate. I am thinking particularly of people
who want a very small crashdump kernel and thus don't want to compile
in kexec.

The patch below should fix things up so that all valid combinations of
KEXEC, CRASH_DUMP and VMCORE compile cleanly - VMCORE depends on
CRASH_DUMP which is why I said valid combinations. In a nutshell
it just untangles unrelated code and switches around a few defines.

Please note that it creats a new file, arch/ia64/kernel/crash_dump.c
This is in keeping with the i386 implementation.

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[IA64] Do not call SN_SAL_SET_CPU_NUMBER twice on cpu 0
Jay Lan [Mon, 11 Dec 2006 23:39:23 +0000 (15:39 -0800)]
[IA64] Do not call SN_SAL_SET_CPU_NUMBER twice on cpu 0

This is an SN specific patch.

Architectually, cpu_init is always called twice on cpu 0
and thus resulted in two SN_SAL_SET_CPU_NUMBER calls.

This was harmless in production kernel; however, it can
cause problem on booting up a crashdump kernel at Altix.

Here is the patch that detects the second sn_cpu_init
call and skips the second call to SN_SAL_SET_CPU_NUMBER.

Signed-Off-By: Jay Lan <jlan@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
18 years ago[PATCH] netpoll: fix netpoll lockup
Ingo Molnar [Tue, 12 Dec 2006 16:20:42 +0000 (17:20 +0100)]
[PATCH] netpoll: fix netpoll lockup

current -git doesnt boot on my laptop due to netpoll not unlocking the
tx lock in the else branch.

booted this up on my laptop with lockdep enabled and there are no
locking complaints and it works fine.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6
Linus Torvalds [Tue, 12 Dec 2006 16:14:46 +0000 (08:14 -0800)]
Merge /pub/scm/linux/kernel/git/lethal/sh-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6: (29 commits)
  sh: Fixup SH-2 BUG() trap handling.
  sh: Use early_param() for earlyprintk parsing.
  sh: Fix .empty_zero_page alignment for PAGE_SIZE > 4096.
  sh: Fixup .data.page_aligned.
  sh: Hook up SH7722 scif ipr interrupts.
  sh: Fixup sh_bios() trap handling.
  sh: SH-MobileR SH7722 CPU support.
  sh: Fixup dma_cache_sync() callers.
  sh: Convert remaining remap_area_pages() users to ioremap_page_range().
  sh: Fixup kernel_execve() for syscall cleanups.
  sh: Fix get_wchan().
  sh: BUG() handling through trapa vector.
  rtc: rtc-sh: alarm support.
  rtc: rtc-sh: fix rtc for out-by-one for the month.
  sh: Kill off unused SE7619 I/O ops.
  serial: sh-sci: Shut up various sci_rxd_in() gcc4 warnings.
  sh: Split out atomic ops logically.
  sh: Fix Solution Engine 7619 build.
  sh: Trivial build fixes for SH-2 support.
  sh: IPR IRQ updates for SH7619/SH7206.
  ...

18 years ago[PATCH] net, 8139too.c: fix netpoll deadlock
Ingo Molnar [Tue, 12 Dec 2006 12:49:35 +0000 (13:49 +0100)]
[PATCH] net, 8139too.c: fix netpoll deadlock

fix deadlock in the 8139too driver: poll handlers should never forcibly
enable local interrupts, because they might be used by netpoll/printk
from IRQ context.

  =================================
  [ INFO: inconsistent lock state ]
  2.6.19 #11
  ---------------------------------
  inconsistent {softirq-on-W} -> {in-softirq-W} usage.
  swapper/1 [HC0[0]:SC1[1]:HE1:SE0] takes:
   (&npinfo->poll_lock){-+..}, at: [<c0350a41>] net_rx_action+0x64/0x1de
  {softirq-on-W} state was registered at:
    [<c0134c86>] mark_lock+0x5b/0x39c
    [<c0135012>] mark_held_locks+0x4b/0x68
    [<c01351e9>] trace_hardirqs_on+0x115/0x139
    [<c02879e6>] rtl8139_poll+0x3d7/0x3f4
    [<c035c85d>] netpoll_poll+0x82/0x32f
    [<c035c775>] netpoll_send_skb+0xc9/0x12f
    [<c035cdcc>] netpoll_send_udp+0x253/0x25b
    [<c0288463>] write_msg+0x40/0x65
    [<c011cead>] __call_console_drivers+0x45/0x51
    [<c011cf16>] _call_console_drivers+0x5d/0x61
    [<c011d4fb>] release_console_sem+0x11f/0x1d8
    [<c011d7d7>] register_console+0x1ac/0x1b3
    [<c02883f8>] init_netconsole+0x55/0x67
    [<c010040c>] init+0x9a/0x24e
    [<c01049cf>] kernel_thread_helper+0x7/0x10
    [<ffffffff>] 0xffffffff
  irq event stamp: 819992
  hardirqs last  enabled at (819992): [<c0350a16>] net_rx_action+0x39/0x1de
  hardirqs last disabled at (819991): [<c0350b1e>] net_rx_action+0x141/0x1de
  softirqs last  enabled at (817552): [<c01214e4>] __do_softirq+0xa3/0xa8
  softirqs last disabled at (819987): [<c0106051>] do_softirq+0x5b/0xc9

  other info that might help us debug this:
  no locks held by swapper/1.

  stack backtrace:
   [<c0104d88>] dump_trace+0x63/0x1e8
   [<c0104f26>] show_trace_log_lvl+0x19/0x2e
   [<c010532d>] show_trace+0x12/0x14
   [<c0105343>] dump_stack+0x14/0x16
   [<c0134980>] print_usage_bug+0x23c/0x246
   [<c0134d33>] mark_lock+0x108/0x39c
   [<c01356a7>] __lock_acquire+0x361/0x9ed
   [<c0136018>] lock_acquire+0x56/0x72
   [<c03aff1f>] _spin_lock+0x35/0x42
   [<c0350a41>] net_rx_action+0x64/0x1de
   [<c0121493>] __do_softirq+0x52/0xa8
   [<c0106051>] do_softirq+0x5b/0xc9
   [<c0121338>] irq_exit+0x3c/0x48
   [<c0106163>] do_IRQ+0xa4/0xbd
   [<c01047c6>] common_interrupt+0x2e/0x34
   [<c011db92>] vprintk+0x2c0/0x309
   [<c011dbf6>] printk+0x1b/0x1d
   [<c01003f2>] init+0x80/0x24e
   [<c01049cf>] kernel_thread_helper+0x7/0x10
   =======================

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] lockdep: fix seqlock_init()
Ingo Molnar [Tue, 12 Dec 2006 11:10:28 +0000 (12:10 +0100)]
[PATCH] lockdep: fix seqlock_init()

seqlock_init() needs to use spin_lock_init() for dynamic locks, so that
lockdep is notified about the presence of a new lock.

(this is a fallout of the recent networking merge, which started using
the so-far unused seqlock_init() API.)

This fix solves the following lockdep-internal warning on current -git:

 INFO: trying to register non-static key.
 the code is fine but needs lockdep annotation.
 turning off the locking correctness validator.
     __lock_acquire+0x10c/0x9f9
     lock_acquire+0x56/0x72
     _spin_lock+0x35/0x42
     neigh_destroy+0x9d/0x12e
     neigh_periodic_timer+0x10a/0x15c
     run_timer_softirq+0x126/0x18e
     __do_softirq+0x6b/0xe6
     do_softirq+0x64/0xd2
     ksoftirqd+0x82/0x138

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/mchehab/v4l-dvb
Linus Torvalds [Tue, 12 Dec 2006 16:01:50 +0000 (08:01 -0800)]
Merge branch 'master' of /linux/kernel/git/mchehab/v4l-dvb

* 'master' of master.kernel.org:/pub/scm/linux/kernel/git/mchehab/v4l-dvb:
  V4L/DVB (4954): Fix: On ia64, i2c adap->inb/adap->outb are wrongly evaluated

18 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shaggy...
Linus Torvalds [Tue, 12 Dec 2006 15:45:48 +0000 (07:45 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/shaggy/jfs-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shaggy/jfs-2.6:
  JFS: Fix conflicting superblock flags

18 years agoV4L/DVB (4954): Fix: On ia64, i2c adap->inb/adap->outb are wrongly evaluated
Mauro Carvalho Chehab [Mon, 11 Dec 2006 08:37:14 +0000 (05:37 -0300)]
V4L/DVB (4954): Fix: On ia64, i2c adap->inb/adap->outb are wrongly evaluated

i2c defines two callbacks (inb/outb). On ia64, since it defines also two macros
with those names, it causes the following errors:
drivers/media/video/usbvision/usbvision-i2c.c:64:39: macro "outb" passed 4 arguments, but takes just 2
drivers/media/video/usbvision/usbvision-i2c.c: In function `try_write_address':
drivers/media/video/usbvision/usbvision-i2c.c:64: warning: assignment makes integer from pointer without a cast
drivers/media/video/usbvision/usbvision-i2c.c:89:38: macro "inb" passed 4 arguments, but takes just 1
drivers/media/video/usbvision/usbvision-i2c.c: In function `try_read_address':
drivers/media/video/usbvision/usbvision-i2c.c:89: warning: assignment makes integer from pointer without a cast
drivers/media/video/usbvision/usbvision-i2c.c:85: warning: unused variable `buf'
drivers/media/video/usbvision/usbvision-i2c.c:173:53: macro "inb" passed 4 arguments, but takes just 1
drivers/media/video/usbvision/usbvision-i2c.c: In function `usb_xfer':
drivers/media/video/usbvision/usbvision-i2c.c:173: warning: assignment makes integer from pointer without a cast
drivers/media/video/usbvision/usbvision-i2c.c:179:54: macro "outb" passed 4 arguments, but takes just 2
drivers/media/video/usbvision/usbvision-i2c.c:179: warning: assignment makes integer from pointer without a cast
thanks to Andrew Morton for pointing this.

Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years ago[PATCH] remove unnecessary blk_queue_bounce in SG_IO
FUJITA Tomonori [Tue, 12 Dec 2006 09:26:55 +0000 (10:26 +0100)]
[PATCH] remove unnecessary blk_queue_bounce in SG_IO

When I converted the original patch, I left unnecessary blk_queue_bounce in
SG_IO.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
18 years ago[PATCH] fix SG_IO bio leak
FUJITA Tomonori [Mon, 11 Dec 2006 09:01:34 +0000 (10:01 +0100)]
[PATCH] fix SG_IO bio leak

This patch fixes bio leaks in SG_IO. rq->bio can be changed after io
completion, so we need to reset rq->bio before calling blk_rq_unmap_user()

http://marc.theaimsgroup.com/?l=linux-kernel&m=116570666807983&w=2

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
18 years ago[PATCH] remove blk_queue_activity_fn
Boaz Harrosh [Tue, 5 Dec 2006 09:19:14 +0000 (10:19 +0100)]
[PATCH] remove blk_queue_activity_fn

While working on bidi support at struct request level
I have found that blk_queue_activity_fn is actually never used.
The only user is in ide-probe.c with this code:

/* enable led activity for disk drives only */
if (drive->media == ide_disk && hwif->led_act)
blk_queue_activity_fn(q, hwif->led_act, drive);

And led_act is never initialized anywhere.
(Looking back at older kernels it was used in the PPC arch, but was removed around 2.6.18)
Unless it is all for future use off course.
(this patch is against linux-2.6-block.git as off 2006/12/4)

Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 12 Dec 2006 02:35:17 +0000 (18:35 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (32 commits)
  [NETPOLL]: Fix local_bh_enable() warning.
  [IPVS]: Make ip_vs_sync.c <= 80col wide.
  [IPVS]: Use msleep_interruptable() instead of ssleep() aka msleep()
  [HAMRADIO]: Fix baycom_epp.c compile failure.
  [DCCP]: Whitespace cleanups
  [DCCP] ccid3: Fixup some type conversions related to rtts
  [DCCP] ccid3: BUG-FIX - conversion errors
  [DCCP] ccid3: Reorder packet history source file
  [DCCP] ccid3: Reorder packet history header file
  [DCCP] ccid3: Make debug output consistent
  [DCCP] ccid3: Perform history operations only after packet has been sent
  [DCCP] ccid3: TX history - remove unused field
  [DCCP] ccid3: Shift window counter computation
  [DCCP] ccid3: Sanity-check RTT samples
  [DCCP] ccid3: Initialise RTT values
  [DCCP] ccid: Deprecate ccid_hc_tx_insert_options
  [DCCP]: Warn when discarding packet due to internal errors
  [DCCP]: Only deliver to the CCID rx side in charge
  [DCCP]: Simplify TFRC calculation
  [DCCP]: Debug timeval operations
  ...

18 years agoMerge branch 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32
Linus Torvalds [Tue, 12 Dec 2006 02:28:59 +0000 (18:28 -0800)]
Merge branch 'for-linus' of git://atmel.no/~hskinnemoen/linux/kernel/avr32

* 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32:
  [AVR32] Add missing #include <linux/param.h> to delay.c
  [AVR32] Pass dev parameter to dma_cache_sync()
  [AVR32] Implement intc_get_pending()
  [AVR32] Don't include <asm/delay.h>
  [AVR32] Put the chip in "stop" mode when halting the system
  [AVR32] Set flow handler for external interrupts
  [AVR32] Remove unused file
  [AVR32] Remove mii_phy_addr and eth_addr from eth_platform_data
  [AVR32] Move ethernet tag parsing to board-specific code
  [AVR32] Add macb1 platform_device
  [AVR32] Portmux API update

18 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Tue, 12 Dec 2006 02:24:58 +0000 (18:24 -0800)]
Merge git://git./linux/kernel/git/paulus/powerpc

* git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (36 commits)
  [POWERPC] Generic BUG for powerpc
  [PPC] Fix compile failure do to introduction of PHY_POLL
  [POWERPC] Only export __mtdcr/__mfdcr if CONFIG_PPC_DCR is set
  [POWERPC] Remove old dcr.S
  [POWERPC] Fix SPU coredump code for max_fdset removal
  [POWERPC] Fix irq routing on some 32-bit PowerMacs
  [POWERPC] ps3: Add vuart support
  [POWERPC] Support ibm,dynamic-reconfiguration-memory nodes
  [POWERPC] dont allow pSeries_probe to succeed without initialising MMU
  [POWERPC] micro optimise pSeries_probe
  [POWERPC] Add SPURR SPR to sysfs
  [POWERPC] Add DSCR SPR to sysfs
  [POWERPC] Fix 440SPe CPU table entry
  [POWERPC] Add support for FP emulation for the e300c2 core
  [POWERPC] of_device_register: propagate device_create_file return code
  [POWERPC] Fix mmap of PCI resource with hack for X
  [POWERPC] iSeries: head_64.o needs to depend on lparmap.s
  [POWERPC] cbe_thermal: Fix initialization of sysfs attribute_group
  [POWERPC] Remove QE header files from lite5200.c
  [POWERPC] of_platform_make_bus_id(): make `magic' int
  ...

18 years ago[MIPS] paging_init(): use highend_pfn/highstart_pfn
Franck Bui-Huu [Wed, 6 Dec 2006 15:48:28 +0000 (16:48 +0100)]
[MIPS] paging_init(): use highend_pfn/highstart_pfn

This patch makes paging_init() use highend_pfn/highstart_pfn globals.

It removes the need of 'high' local which was needed only by HIGHMEM config.

More important perhaps, it fixes a bug when HIGHMEM is set but there's
actually no physical highmem (highend_pfn = 0)

Signed-off-by: Franck Bui-Huu <fbuihuu@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Malta: Resurrect MTD support for onboard flash.
Ralf Baechle [Mon, 11 Dec 2006 15:33:30 +0000 (15:33 +0000)]
[MIPS] Malta: Resurrect MTD support for onboard flash.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Discard .exit.text and .exit.data at runtime.
Ralf Baechle [Mon, 11 Dec 2006 11:54:52 +0000 (11:54 +0000)]
[MIPS] Discard .exit.text and .exit.data at runtime.

While the recent cset 86384d544157db23879064cde36061cdcafc6794 did improve
things it didn't resolve all the problems.  So bite the bullet and discard
.exit.text and .exit.data at runtime.  Which of course sucks because it
bloats binaries with code that will never ever be used but it's the only
thing that will work reliable as demonstrated by the function sd_major() in
drivers/scsi/sd.c.

Gcc may compile sd_major() using a jump table which it will put into
.rodata.  If it also inlines sd_major's function body into exit_sd() which
gcc > 3.4.x does.  If CONFIG_BLK_DEV_SD has been set to y we would like ld
to discard exit_sd's code at link time.  However sd_major happens to
contain a switch statement which gcc will compile using a jump table in
.rodata on the architectures I checked.  So, when ld later discards
.exit.text only the jump table in .rodata with its stale references to
the discard .exit.text will be left which any no antique ld will honor
with a link error.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] IP27: Don't drag <asm/sn/arch.h> into topology.h.
Ralf Baechle [Mon, 11 Dec 2006 04:17:30 +0000 (04:17 +0000)]
[MIPS] IP27: Don't drag <asm/sn/arch.h> into topology.h.

Another way that old SGI types were getting dragged into generic code.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] IP27: Move definition of nic_t to its sole user.
Ralf Baechle [Mon, 11 Dec 2006 04:07:40 +0000 (04:07 +0000)]
[MIPS] IP27: Move definition of nic_t to its sole user.

This also fixes the duplicate definition of nic_t in the s2io driver.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] IP27: Don't include <asm/sn/arch.h>.
Ralf Baechle [Mon, 11 Dec 2006 04:00:37 +0000 (04:00 +0000)]
[MIPS] IP27: Don't include <asm/sn/arch.h>.

Nothing <asm/sn/arch.h> defines is used.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] compat.h uses struct pt_regs so needs to include ptrace.h.
Ralf Baechle [Mon, 11 Dec 2006 00:43:58 +0000 (00:43 +0000)]
[MIPS] compat.h uses struct pt_regs so needs to include ptrace.h.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[NETPOLL]: Fix local_bh_enable() warning.
Andrew Morton [Tue, 12 Dec 2006 01:24:46 +0000 (17:24 -0800)]
[NETPOLL]: Fix local_bh_enable() warning.

During boot we get:

netconsole: device eth0 not up yet, forcing it
e1000: eth0: e1000_watchdog: NIC Link is Up 100 Mbps Full Duplex
WARNING (!__warned) at kernel/softirq.c:137 local_bh_enable()

Call Trace:
 [<ffffffff80235baf>] local_bh_enable+0x41/0xa3
 [<ffffffff8045ab8e>] netpoll_send_skb+0x116/0x144
 [<ffffffff8045b1ee>] netpoll_send_udp+0x263/0x271
 [<ffffffff803d41ec>] write_msg+0x42/0x5e
 [<ffffffff80230c9b>] __call_console_drivers+0x5f/0x70
 [<ffffffff80230d19>] _call_console_drivers+0x6d/0x71
 [<ffffffff802313f0>] release_console_sem+0x148/0x1ec
 [<ffffffff802316ce>] register_console+0x1b1/0x1ba
 [<ffffffff803d4178>] init_netconsole+0x54/0x68
 [<ffffffff802071ae>] init+0x152/0x308
 [<ffffffff804dac8b>] _spin_unlock_irq+0x14/0x30
 [<ffffffff8022c15e>] schedule_tail+0x43/0x9f
 [<ffffffff8020a758>] child_rip+0xa/0x12

Herbert sayeth:

  Normally networking isn't invoked with interrupts turned off, but I
  suppose we don't have a choice here.  This is unique being a place where you
  can get called with BH on, off, or IRQs off.

  Given that this is only used for printk, the easiest solution is probably
  just to disable local IRQs instead of BH.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agosh: Fixup SH-2 BUG() trap handling.
Yoshinori Sato [Tue, 12 Dec 2006 00:11:45 +0000 (09:11 +0900)]
sh: Fixup SH-2 BUG() trap handling.

This adds in support for the BUG() trap on SH-2.

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Use early_param() for earlyprintk parsing.
Paul Mundt [Tue, 12 Dec 2006 00:00:47 +0000 (09:00 +0900)]
sh: Use early_param() for earlyprintk parsing.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fix .empty_zero_page alignment for PAGE_SIZE > 4096.
Paul Mundt [Mon, 11 Dec 2006 23:53:29 +0000 (08:53 +0900)]
sh: Fix .empty_zero_page alignment for PAGE_SIZE > 4096.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fixup .data.page_aligned.
Paul Mundt [Mon, 11 Dec 2006 23:50:36 +0000 (08:50 +0900)]
sh: Fixup .data.page_aligned.

This had a bogus .data.idt reference, fix it up..

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Hook up SH7722 scif ipr interrupts.
Paul Mundt [Mon, 11 Dec 2006 23:49:06 +0000 (08:49 +0900)]
sh: Hook up SH7722 scif ipr interrupts.

Add the SCIF IRQs to the IPR table for SH7722.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fixup sh_bios() trap handling.
Paul Mundt [Mon, 11 Dec 2006 11:29:28 +0000 (20:29 +0900)]
sh: Fixup sh_bios() trap handling.

This was inadvertently broken when the entry.S code split up,
restore the missing branch and get subsequent traps working
under debug again. This manifested itself as a lockup when
attempting to reload the VBR base.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: SH-MobileR SH7722 CPU support.
Paul Mundt [Mon, 11 Dec 2006 11:28:03 +0000 (20:28 +0900)]
sh: SH-MobileR SH7722 CPU support.

This adds CPU support for the SH7722.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fixup dma_cache_sync() callers.
Paul Mundt [Sat, 9 Dec 2006 00:17:01 +0000 (09:17 +0900)]
sh: Fixup dma_cache_sync() callers.

This now takes a struct device, update all of the callers.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Convert remaining remap_area_pages() users to ioremap_page_range().
Paul Mundt [Sat, 9 Dec 2006 00:16:12 +0000 (09:16 +0900)]
sh: Convert remaining remap_area_pages() users to ioremap_page_range().

A couple of these were missed.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fixup kernel_execve() for syscall cleanups.
Paul Mundt [Sat, 9 Dec 2006 00:14:35 +0000 (09:14 +0900)]
sh: Fixup kernel_execve() for syscall cleanups.

SH-2 and SH-2A need to use a different syscall base for the trapa
vector than the other parts, so fixup the logic in the kernel_execve()
case.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fix get_wchan().
Paul Mundt [Fri, 8 Dec 2006 08:46:29 +0000 (17:46 +0900)]
sh: Fix get_wchan().

Some time ago the schedule frame size changed and we failed to reflect
this in get_wchan() at the time. This first popped up as a problem on
SH7751R where schedule_frame ended up being unaligned and generating
an unaligned trap. This fixes it up again..

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: BUG() handling through trapa vector.
Paul Mundt [Fri, 8 Dec 2006 08:41:43 +0000 (17:41 +0900)]
sh: BUG() handling through trapa vector.

Previously we haven't been doing anything with verbose BUG() reporting,
and we've been relying on the oops path for handling BUG()'s, which is
rather sub-optimal.

This switches BUG handling to use a fixed trapa vector (#0x3e) where we
construct a small bug frame post trapa instruction to get the context
right. This also makes it trivial to wire up a DIE_BUG for the atomic
die chain, which we couldn't really do before.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agortc: rtc-sh: alarm support.
Jamie Lenehan [Fri, 8 Dec 2006 06:26:15 +0000 (15:26 +0900)]
rtc: rtc-sh: alarm support.

This adds alarm support for the RTC_ALM_SET, RTC_ALM_READ,
RTC_WKALM_SET and RTC_WKALM_RD operations to rtc-sh.

The only unusual part is the handling of the alarm interrupt. If you
clear the alarm flag (AF) while the time in the RTC still matches the
time in the alarm registers than AF is immediately re-set, and if the
alarm interrupt (AIE) is still enabled then it re-triggers. I was
originally getting around 20k+ interrupts generated during the second
when the RTC and alarm registers matches.

The solution I've used is to clear AIE when the alarm goes off and
then use the carry interrupt to re-enabled it. The carry interrupt
will check AF and re-enabled AIE if it's clear. If AF is not clear
it'll clear it and then the check will be repeated next carry
interrupt. This a bit in rtc structure that indicates that it's
waiting to have AIE re-enabled so it doesn't turn it on when it
wasn't enabled anyway.

Signed-off-by: Jamie Lenehan <lenehan@twibble.org>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agortc: rtc-sh: fix rtc for out-by-one for the month.
Jamie Lenehan [Fri, 8 Dec 2006 05:49:30 +0000 (14:49 +0900)]
rtc: rtc-sh: fix rtc for out-by-one for the month.

The RMONCNT register, which holds the month in the RTC, takes a value
between 1 and 12 while the tm_mon field in the time structures takes
a value between 0 and 11. This wasn't being taken into account in
rtc-sh resulting in the month being out by one.

eg, on my board during boot the RTC is set to:

  RTC is set to Thu Jul 01 09:00:00 1999

but "hwclock -r" immediately after logging in was showing:

  Sun Aug  1 09:01:43 1999  0.000000 seconds

Signed-off-by: Jamie Lenehan <lenehan@twibble.org>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Kill off unused SE7619 I/O ops.
Yoshinori Sato [Fri, 8 Dec 2006 05:47:02 +0000 (14:47 +0900)]
sh: Kill off unused SE7619 I/O ops.

This can use the generic routines, so kill off the board-specific ones.

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agoserial: sh-sci: Shut up various sci_rxd_in() gcc4 warnings.
Paul Mundt [Fri, 8 Dec 2006 05:26:19 +0000 (14:26 +0900)]
serial: sh-sci: Shut up various sci_rxd_in() gcc4 warnings.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Split out atomic ops logically.
Paul Mundt [Thu, 7 Dec 2006 11:33:38 +0000 (20:33 +0900)]
sh: Split out atomic ops logically.

We have a few different ways to do the atomic operations, so split
them out in to different headers rather than bloating atomic.h.
Kernelspace gUSA will take this up to a third implementation.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Fix Solution Engine 7619 build.
Yoshinori Sato [Thu, 7 Dec 2006 10:59:35 +0000 (19:59 +0900)]
sh: Fix Solution Engine 7619 build.

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Trivial build fixes for SH-2 support.
Yoshinori Sato [Thu, 7 Dec 2006 09:07:27 +0000 (18:07 +0900)]
sh: Trivial build fixes for SH-2 support.

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: IPR IRQ updates for SH7619/SH7206.
Yoshinori Sato [Thu, 7 Dec 2006 09:01:23 +0000 (18:01 +0900)]
sh: IPR IRQ updates for SH7619/SH7206.

This updates the SH7619 and SH7206 code for the IPR IRQ changes.

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: gcc4 symbol export fixups.
Stuart Menefy [Thu, 7 Dec 2006 08:48:52 +0000 (17:48 +0900)]
sh: gcc4 symbol export fixups.

gcc 4 for sh changes the names of some compiler intrinsic functions
and adds some additional ones. This patch adds the new ones, and
fixes up various module symbol resolution issues.

Signed-off-by: Stuart Menefy <stuart.menefy@st.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: landisk board build fixes.
Paul Mundt [Thu, 7 Dec 2006 08:27:18 +0000 (17:27 +0900)]
sh: landisk board build fixes.

Get the landisk board building again..

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agortc: rtc-sh: fix for period rtc interrupts.
Jamie Lenehan [Thu, 7 Dec 2006 08:23:50 +0000 (17:23 +0900)]
rtc: rtc-sh: fix for period rtc interrupts.

When testing the per second interrupt support (RTC_UIE_ON/RTC_UIE_OFF)
of the new RTC system it would die in sh_rtc_interrupt due to a null
ptr dereference. The following gets it working correctly.

Signed-off-by: Jamie Lenehan <lenehan@twibble.org>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: register rtc resources for sh775x.
Jamie Lenehan [Thu, 7 Dec 2006 08:23:07 +0000 (17:23 +0900)]
sh: register rtc resources for sh775x.

Register the RTC resources for the sh775x subtype so that the new
generic RTC support in drivers/rtc/rtc-sh.c will work.

Signed-off-by: Jamie Lenehan <lenehan@twibble.org>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: SH-2 defconfig updates.
Yoshinori Sato [Thu, 7 Dec 2006 08:09:46 +0000 (17:09 +0900)]
sh: SH-2 defconfig updates.

This adds a new defconfig for SE7619 and updates SE7206.

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Add uImage and S-rec generation support.
Paul Mundt [Thu, 7 Dec 2006 08:00:32 +0000 (17:00 +0900)]
sh: Add uImage and S-rec generation support.

Add a couple of new targets, both for uImage and S-rec generation.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: push-switch fixups for work_struct API damage.
Paul Mundt [Thu, 7 Dec 2006 04:20:58 +0000 (13:20 +0900)]
sh: push-switch fixups for work_struct API damage.

INIT_WORK() dropped the data arg, so now we have to stash an extra
pointer and backpedal instead.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Shut up csum_ipv6_magic() warnings.
Paul Mundt [Thu, 7 Dec 2006 03:43:52 +0000 (12:43 +0900)]
sh: Shut up csum_ipv6_magic() warnings.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years agosh: Reworked swap cache entry encoding for SH-X2 MMU.
Paul Mundt [Thu, 7 Dec 2006 03:43:06 +0000 (12:43 +0900)]
sh: Reworked swap cache entry encoding for SH-X2 MMU.

In the 64-bit PTE case there's no point in restricting the encoding
to the low bits of the PTE, we can instead bump all of this up to
the high 32 bits and extend PTE_FILE_MAX_BITS to 32, adopting the
same convention used by x86 PAE.

There's a minor discrepency between the number of bits used for the
swap type encoding between 32 and 64-bit PTEs, but this is unlikely
to cause any problem given the extended offset.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
18 years ago[IPVS]: Make ip_vs_sync.c <= 80col wide.
Simon Horman [Mon, 11 Dec 2006 06:36:18 +0000 (22:36 -0800)]
[IPVS]: Make ip_vs_sync.c <= 80col wide.

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPVS]: Use msleep_interruptable() instead of ssleep() aka msleep()
Simon Horman [Mon, 11 Dec 2006 06:35:24 +0000 (22:35 -0800)]
[IPVS]: Use msleep_interruptable() instead of ssleep() aka msleep()

Dean Manners notices that when an IPVS synchonisation daemons are
started the system load slowly climbs up to 1. This seems to be related
to the call to ssleep(1) (aka msleep(1000) in the main loop. Replacing
this with a call to msleep_interruptable() seems to make the problem go
away. Though I'm not sure that it is correct.

This is the second edition of this patch, which replaces ssleep()
in the main loop for both the master and backup threads, as well
as some thread synchronisation code. The latter is just for thorougness
as it shouldn't be causing any problems.

Signed-Off-By: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[HAMRADIO]: Fix baycom_epp.c compile failure.
Ralf Baechle [Sun, 10 Dec 2006 21:46:45 +0000 (13:46 -0800)]
[HAMRADIO]: Fix baycom_epp.c compile failure.

Fix foobar in 15b1c0e822f578306332d4f4c449250db5c5dceb and
e8cc49bb0fdb9e18a99e6780073d1400ba2b0d1f patch series.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP]: Whitespace cleanups
Arnaldo Carvalho de Melo [Sun, 10 Dec 2006 18:01:18 +0000 (16:01 -0200)]
[DCCP]: Whitespace cleanups

That accumulated over the last months hackaton, shame on me for not
using git-apply whitespace helping hand, will do that from now on.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Fixup some type conversions related to rtts
Arnaldo Carvalho de Melo [Sun, 10 Dec 2006 17:39:29 +0000 (15:39 -0200)]
[DCCP] ccid3: Fixup some type conversions related to rtts

Spotted by David Miller when compiling on sparc64, I reproduced it here on
parisc64, that are the only platforms to define __kernel_suseconds_t as an
'int', all the others, x86_64 and x86 included typedef it as a 'long', but from
the definition of suseconds_t it should just be an 'int' on platforms where it
is >= 32bits, it would not require all the castings from suseconds_t to (int)
when printking variables of this type, that are not needed on parisc64 and
sparc64.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: BUG-FIX - conversion errors
Gerrit Renker [Sun, 10 Dec 2006 02:24:57 +0000 (00:24 -0200)]
[DCCP] ccid3: BUG-FIX - conversion errors

This fixes conversion errors which arose by not properly type-casting
from u32 to __u64. Fixed by explicitly casting each type which is not
__u64, or by performing operation after assignment.

The patch further adds missing debug information to track the current
value of X_recv.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Reorder packet history source file
Gerrit Renker [Sun, 10 Dec 2006 02:24:33 +0000 (00:24 -0200)]
[DCCP] ccid3: Reorder packet history source file

No code change at all.

This reorders the source file to follow the same order as the corresponding
header file.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Reorder packet history header file
Gerrit Renker [Sun, 10 Dec 2006 02:24:11 +0000 (00:24 -0200)]
[DCCP] ccid3: Reorder packet history header file

No code change at all.

To make the header file easier to read, the following ordering is established
among the declarations:
* hist_new
* hist_delete
* hist_entry_new
* hist_head
* hist_find_entry
* hist_add_entry
* hist_entry_delete
* hist_purge

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Make debug output consistent
Gerrit Renker [Sun, 10 Dec 2006 02:14:12 +0000 (00:14 -0200)]
[DCCP] ccid3: Make debug output consistent

This patch does not alter any algorithm, just the debug message format:

 * s#%s, sk=%p#%s(%p)#g

 * when a statename is present, it now uses %s(%p, state=%s)

 * when only function entry is debugged, it adds an `- entry'

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Perform history operations only after packet has been sent
Gerrit Renker [Sun, 10 Dec 2006 02:09:21 +0000 (00:09 -0200)]
[DCCP] ccid3: Perform history operations only after packet has been sent

 This migrates all packet history operations into the routine
 ccid3_hc_tx_packet_sent, thereby removing synchronization problems
 that occur when, as before, the operations are spread over multiple
 routines.
 The following minor simplifications are also applied:
  * several simplifications now follow from this change - several tests
    are now no longer required
  * removal of one unnecessary variable (dp)

Justification:

 Currently packet history operations span two different routines,
 one of which is likely to pass through several iterations of sleeping
 and awakening.
 The first routine, ccid3_hc_tx_send_packet, allocates an entry and
 sets a few fields. The remaining fields are filled in when the second
 routine (which is not within a sleeping context), ccid3_hc_tx_packet_sent,
 is called. This has several strong drawbacks:
  * it is not necessary to split history operations - all fields can be
    filled in by the second routine
  * the first routine is called multiple times, until a packet can be sent,
    and sleeps meanwhile - this causes a lot of difficulties with regard to
    keeping the list consistent
  * since both routines do not have a producer-consumer like synchronization,
    it is very difficult to maintain data across calls to these routines
  * the fact that the routines are called in different contexts (sleeping, not
    sleeping) adds further problems

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: TX history - remove unused field
Gerrit Renker [Sun, 10 Dec 2006 02:08:09 +0000 (00:08 -0200)]
[DCCP] ccid3: TX history - remove unused field

This removes the `dccphtx_ccval' field since it is nowhere used in the code and
in fact not necessary for the accounting.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Shift window counter computation
Gerrit Renker [Sun, 10 Dec 2006 02:07:37 +0000 (00:07 -0200)]
[DCCP] ccid3: Shift window counter computation

 This puts the window counter computation [RFC 4342, 8.1] into a separate
 function which is called whenever a new packet is ready for immediate
 transmission in ccid3_hc_tx_send_packet.

Justification:

 The window counter update was previously computed after the packet was sent. This has
 two drawbacks, both fixed by this patch:
   1) re-compute another timestamp almost directly after the packet was sent (expensive),
   2) the CCVal for the window counter is needed at the instant the packet is sent.

Further details:

 The initialisation of the window counter is left in the state NO_SENT, as before.
 The algorithm will do nothing if either RTT is initialised to 0 (which is ok) or if
 the RTT value remains below 4 microseconds (which is almost pathological).

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Sanity-check RTT samples
Gerrit Renker [Sun, 10 Dec 2006 02:06:32 +0000 (00:06 -0200)]
[DCCP] ccid3: Sanity-check RTT samples

CCID3 performance depends much on the accuracy of RTT samples.  If RTT
samples grow too large, performance can be catastrophically poor.

To limit the amount of possible damage in such cases, the patch
 * introduces an upper limit which identifies a maximum `sane' RTT value;
 * uses a macro to enforce this upper limit.

Using a macro was given preference, since it is necessary to identify the
calling function in the warning message. Since exceeding this threshold
identifies a critical condition, DCCP_CRIT is used and not DCCP_WARN.

Many thanks to Ian McDonald for collaboration on this issue.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Initialise RTT values
Gerrit Renker [Sun, 10 Dec 2006 02:06:01 +0000 (00:06 -0200)]
[DCCP] ccid3: Initialise RTT values

In both the sender and the receiver it is possible that the stored
RTT value is accessed before an actual RTT estimate has been computed.

This patch
 * initialises the sender RTT to 0
     - the sender always accesses the RTT in ccid3_hc_tx_packet_sent
     - the RTT is further needed for the window counter algorithm

 * replaces the receiver initialisation of 5msec with 0
     - which has the same effect and removes an `XXX'
     - the RTT value is needed in ccid3_hc_rx_packet_recv as rtt_prev

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid: Deprecate ccid_hc_tx_insert_options
Gerrit Renker [Sun, 10 Dec 2006 02:05:38 +0000 (00:05 -0200)]
[DCCP] ccid: Deprecate ccid_hc_tx_insert_options

 The function ccid3_hc_tx_insert_options only does a redundant no-op,
 as the operation

  DCCP_SKB_CB(skb)->dccpd_ccval = hctx->ccid3hctx_last_win_count;

 is already performed _unconditionally_ in ccid3_hc_tx_send_packet.

 Since there is further no current need for this function, it is removed
 entirely. Since furthermore, there is actually no present need for the
 entire interface function ccid_hc_tx_insert_options, it was decided to
 remove it also, to clean up the interface.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP]: Warn when discarding packet due to internal errors
Gerrit Renker [Sun, 10 Dec 2006 02:05:12 +0000 (00:05 -0200)]
[DCCP]: Warn when discarding packet due to internal errors

This adds a (debug) warning message which is triggered whenever a packet is
discarded due to send failure.

It also adds a conditional, so that an interruption during dccp_wait_for_ccid
is not treated as a `BUG': the rationale is that interruptions are external,
whereas bug warnings are concerned with the internals.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP]: Only deliver to the CCID rx side in charge
Gerrit Renker [Sun, 10 Dec 2006 02:04:43 +0000 (00:04 -0200)]
[DCCP]: Only deliver to the CCID rx side in charge

This is an optimisation to reduce CPU load. The received feedback is now
only directed to the active CCID component, without requiring processing
also by the inactive one.

As a consequence, a similar test in ccid3.c is now redundant and is
also removed.

Justification:

 Currently DCCP works as a unidirectional service, i.e. a listening server
 is not at the same time a connecting client.
 As far as I can see, several modifications are necessary until that
 becomes possible.
 At the present time, received feedback is both fed to the rx/tx CCID
 modules. In unidirectional service, only one of these is active at any
 one time.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP]: Simplify TFRC calculation
Gerrit Renker [Sun, 10 Dec 2006 02:04:16 +0000 (00:04 -0200)]
[DCCP]: Simplify TFRC calculation

In migrating towards using the newer functions scaled_div/scaled_div32
for TFRC computations mapped from floating-point onto integer arithmetic,
this completes the last stage of modifications.

In particular, the overflow case for computing X_calc is circumvented by
 * breaking the computation into two stages
 * the first stage, res = (s*1E6)/R, cannot overflow due to use of u64
 * in the second stage, res = (res*1E6)/f, overflow on u32 is avoided due
   to (i) returning UINT_MAX in this case (which is logically appropriate)
   and (ii) issuing a warning message into the system log (since very likely
   there is a problem somewhere else with the parameters)

Lastly, all such scaling operations are now exported into tfrc.h, since
actually this form of scaled computation is specific to TFRC and not to CCID3.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP]: Debug timeval operations
Gerrit Renker [Sun, 10 Dec 2006 02:03:51 +0000 (00:03 -0200)]
[DCCP]: Debug timeval operations

Problem:

 Most target types in the CCID3 code are u32, so subtle conversion errors
 can occur if signed time calculations yield negative results: the original
 values are lost in the conversion to unsigned, calculation errors go undetected.

 This patch therefore
   * sets all critical time types from unsigned to suseconds_t
   * avoids comparison between signed/unsigned via type-casting
   * provides ample warning messages in case time calculations are negative

 These warning messages can be removed at a later stage when the code
 has undergone more testing.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Simplify calculation for reverse lookup of p
Gerrit Renker [Sun, 10 Dec 2006 02:03:30 +0000 (00:03 -0200)]
[DCCP] ccid3: Simplify calculation for reverse lookup of p

 This simplifies the calculation of a value p for a given fval when the
 first loss interval is computed (RFC 3448, 6.3.1). It makes use of the
 two new functions scaled_div/scaled_div32 to provide overflow protection.

 Additionally, protection against divide-by-zero is extended - in this
 case the function will return the maximally possible value of p=100%.

Background:

 The maximum fval, f(100%), is approximately 244, i.e. the scaled value of fval
 should never exceed 244E6, which fits easily into u32. The problem is the scaling
 by 10^6, since additionally R(TT) is in microseconds.
 This is resolved by breaking the division into two stages: the first stage
 computes fval=(s*10^6)/R, stores that into u64; the second stage computes
 fval = (fval*10^6)/X_recv and complains if overflow is reached for u32.
 This case is safe since the TFRC reverse-lookup routine then returns p=100%.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Replace scaled division operations
Gerrit Renker [Sun, 10 Dec 2006 02:02:51 +0000 (00:02 -0200)]
[DCCP] ccid3: Replace scaled division operations

This replaces the remaining uses of usecs_div with scaled_div32, which
internally uses 64bit division and produces a warning on overflow.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Finer-grained resolution of sending rates
Gerrit Renker [Sun, 10 Dec 2006 02:02:12 +0000 (00:02 -0200)]
[DCCP] ccid3: Finer-grained resolution of sending rates

This patch
 * resolves a bug where packets smaller than 32/64 bytes resulted in sending rates of 0
 * supports all sending rates from 1/64 bytes/second up to 4Gbyte/second
 * simplifies the present overflow problems in calculations

Current sending rate X and the cached value X_recv of the receiver-estimated
sending rate are both scaled by 64 (2^6) in order to
 * cope with low sending rates (minimally 1 byte/second)
 * allow upgrading to use a packets-per-second implementation of CCID 3
 * avoid calculation errors due to integer arithmetic cut-off

The patch implements a revised strategy from
http://www.mail-archive.com/dccp@vger.kernel.org/msg01040.html

The only difference with regard to that strategy is that t_ipi is already
used in the calculation of the nofeedback timeout, which saves one division.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Fix two bugs in sending rate computation
Gerrit Renker [Sun, 10 Dec 2006 02:01:22 +0000 (00:01 -0200)]
[DCCP] ccid3: Fix two bugs in sending rate computation

This fixes
 1) a bug in the recomputation of the sending rate by the nofeedback
    timer when no feedback at all has so far been sent by the receiver:
    min_t was used instead of max_t, which is wrong (cf. RFC 3448, p. 10);

 2) an error in the computation of larger initial windows: instead of
    min(... max()) (cf. RFC 4342, 5.), the code had used max(... max()).

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Two optimisations for sending rate recomputation
Gerrit Renker [Sun, 10 Dec 2006 02:00:14 +0000 (00:00 -0200)]
[DCCP] ccid3: Two optimisations for sending rate recomputation

This performs two optimisations for the recomputation of the sending rate.

1) Currently the target sending rate X_calc is recalculated whenever
a) the nofeedback timer expires, or
b) a feedback packet is received.
   In the (a) case, recomputing X_calc is redundant, since

    * the parameters p and RTT do not change in between the
      reception of feedback packets;

    * the parameter X_recv is either modified from received
      feedback or via the nofeedback timer;

    * a test (`p == 0') in the nofeedback timer avoids using
      a stale/undefined value of X_calc if p was previously 0.

2) The nofeedback timer now only recomputes a timestamp when p == 0.
   This is according to step (4) of [RFC 3448, 4.3] and avoids
   unnecessarily determining a timestamp.

A debug statement about not updating X is also removed - it helps very
little in debugging and just clutters the logs.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP] ccid3: Check against too large p
Gerrit Renker [Sun, 10 Dec 2006 01:59:14 +0000 (23:59 -0200)]
[DCCP] ccid3: Check against too large p

This patch follows a suggestion by Ian McDonald and ensures that in
the current code the value of p can not exceed 100%.  Such a value is
illegal and would consequently cause a bug condition in tfrc_calc_x().

The receiver case is also tested, and a warning message is added.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP]: Update documentation
Ian McDonald [Sun, 10 Dec 2006 01:58:10 +0000 (23:58 -0200)]
[DCCP]: Update documentation

As Eddie Kohler points out the RFC is Proposed Standard not experimental.
Also removed documentation about deprecated socket option.

Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[DCCP]: Remove timeo from output.c
Ian McDonald [Sun, 10 Dec 2006 01:56:09 +0000 (23:56 -0200)]
[DCCP]: Remove timeo from output.c

It simplifies waiting for the CCID module to signal that a packet
is ready to be sent.  Other simplifications flow on from this such as
removing constants.

As a result of this EAGAIN is not returned any more by dccp_wait_for_ccid
(which would otherwise lead to unnecessarily discarding the packet in
dccp_write_xmit).

Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[NETPOLL]: Make sure TX lock is taken with BH disabled.
Andrew Morton [Sat, 9 Dec 2006 22:01:49 +0000 (14:01 -0800)]
[NETPOLL]: Make sure TX lock is taken with BH disabled.

Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM]: Ignore generated file pca200e_ecd.bin2
Ralf Baechle [Sat, 9 Dec 2006 21:59:25 +0000 (13:59 -0800)]
[ATM]: Ignore generated file pca200e_ecd.bin2

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: Fix WAN routers kconfig dependency.
Randy Dunlap [Sat, 9 Dec 2006 21:58:42 +0000 (13:58 -0800)]
[NET]: Fix WAN routers kconfig dependency.

Currently WAN router drivers can be built in-kernel while the
register/unregister_wan_device interfaces are built as modules.
This causes:

drivers/built-in.o: In function `cycx_init':
cycx_main.c:(.init.text+0x5c4b): undefined reference to `register_wan_device'
drivers/built-in.o: In function `cycx_exit':
cycx_main.c:(.exit.text+0x560): undefined reference to `unregister_wan_device'
make: *** [.tmp_vmlinux1] Error 1

The problem is caused by tristate -> bool conversion (y or m => y),
so convert WAN_ROUTER_DRIVERS to a tristate so that the correct
dependency is preserved.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[CRYPTO] sha512: Fix sha384 block size
Herbert Xu [Fri, 8 Dec 2006 02:59:52 +0000 (13:59 +1100)]
[CRYPTO] sha512: Fix sha384 block size

The SHA384 block size should be 128 bytes, not 96 bytes.  This was
spotted by Andrew Donofrio.

Fortunately the block size isn't actually used anywhere so this typo
has had no real impact.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[PATCH] Fix typo in 'EXPERIMENTAL' in CC_STACKPROTECTOR on x86_64
Brice Goglin [Mon, 11 Dec 2006 19:14:15 +0000 (20:14 +0100)]
[PATCH] Fix typo in 'EXPERIMENTAL' in CC_STACKPROTECTOR on x86_64

Fix typo in 'EXPERIMENTAL' in config CC_STACKPROTECTOR in arch/x86_64/Kconfig.

Signed-off-by: Brice Goglin <brice@myri.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] LOG2: Make powerpc's __ilog2_u64() take a 64-bit argument
David Howells [Mon, 11 Dec 2006 13:16:05 +0000 (13:16 +0000)]
[PATCH] LOG2: Make powerpc's __ilog2_u64() take a 64-bit argument

Make powerpc's __ilog2_u64() take a 64-bit argument.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc
Linus Torvalds [Mon, 11 Dec 2006 20:26:03 +0000 (12:26 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/drzeus/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc:
  AT91 MMC update for 2.6.19
  mmc: Change SDHCI iomem error to a warning
  mmc: fix "prev->state: 2 != TASK_RUNNING??" problem on SD/MMC  card removal
  AT91 MMC 5 : Minor cleanups
  AT91 MMC 4 : Interrupt handler cleanup
  AT91 MMC 3 : Move global mci_clk variable
  AT91 MMC 2 : Use platform resources
  AT91 MMC 1: Pass host structure.

18 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Mon, 11 Dec 2006 20:22:58 +0000 (12:22 -0800)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
  [PATCH] smc91x: Kill off excessive versatile hooks.
  [PATCH] myri10ge: update driver version to 1.1.0
  [PATCH] myri10ge: fix big_bytes in case of vlan frames
  [PATCH] myri10ge: Full vlan frame in small_bytes
  [PATCH] myri10ge: drop contiguous skb routines
  [PATCH] myri10ge: switch to page-based skb
  [PATCH] myri10ge: add page-based skb routines
  [PATCH] myri10ge: indentation cleanups
  [PATCH] chelsio: working NAPI
  [PATCH] MACB: Use __raw register access
  [PATCH] MACB: Use struct delayed_work instead of struct work_struct
  [PATCH] ucc_geth: Initialize mdio_lock.
  [PATCH] ucc_geth: compilation error fixes

18 years agoMake sure we populate the initroot filesystem late enough
Linus Torvalds [Mon, 11 Dec 2006 20:12:04 +0000 (12:12 -0800)]
Make sure we populate the initroot filesystem late enough

We should not initialize rootfs before all the core initializers have
run.  So do it as a separate stage just before starting the regular
driver initializers.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMake SLES9 "get_kernel_version" work on the kernel binary again
Linus Torvalds [Mon, 11 Dec 2006 17:28:46 +0000 (09:28 -0800)]
Make SLES9 "get_kernel_version" work on the kernel binary again

As reported by Andy Whitcroft, at least the SLES9 initrd build process
depends on getting the kernel version from the kernel binary.  It does
that by simply trawling the binary and looking for the signature of the
"linux_banner" string (the string "Linux version " to be exact. Which
is really broken in itself, but whatever..)

That got broken when the string was changed to allow /proc/version to
change the UTS release information dynamically, and "get_kernel_version"
thus returned "%s" (see commit a2ee8649ba6d71416712e798276bf7c40b64e6e5:
"[PATCH] Fix linux banner utsname information").

This just restores "linux_banner" as a static string, which should fix
the version finding.  And /proc/version simply uses a different string.

To avoid wasting even that miniscule amount of memory, the early boot
string should really be marked __initdata, but that just causes the same
bug in SLES9 to re-appear, since it will then find other occurrences of
"Linux version " first.

Cc: Andy Whitcroft <apw@shadowen.org>
Acked-by: Herbert Poetzl <herbert@13thfloor.at>
Cc: Andi Kleen <ak@suse.de>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Steve Fox <drfickle@us.ibm.com>
Acked-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] smc91x: Kill off excessive versatile hooks.
Paul Mundt [Mon, 11 Dec 2006 10:30:06 +0000 (19:30 +0900)]
[PATCH] smc91x: Kill off excessive versatile hooks.

This looks like a result of too many auto-merges. The
CONFIG_ARCH_VERSATILE case was handled a total of 6 times.
This kills 5 of them.

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

 drivers/net/smc91x.h |   90 ---------------------------------------------------
 1 file changed, 90 deletions(-)
Signed-off-by: Jeff Garzik <jeff@garzik.org>
18 years ago[PATCH] myri10ge: update driver version to 1.1.0
Brice Goglin [Mon, 11 Dec 2006 10:27:55 +0000 (11:27 +0100)]
[PATCH] myri10ge: update driver version to 1.1.0

Update driver version to 1.1.0.

Signed-off-by: Brice Goglin <brice@myri.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>