openwrt/staging/blogic.git
18 years agoV4L/DVB (4674): Use NULL instead of 0 for ptrs
Randy Dunlap [Thu, 28 Sep 2006 17:03:26 +0000 (14:03 -0300)]
V4L/DVB (4674): Use NULL instead of 0 for ptrs

Use NULL instead of 0 for pointer value, eliminate sparse warnings.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4673): Mark the two newer ioctls as experimental
Mauro Carvalho Chehab [Thu, 28 Sep 2006 16:48:26 +0000 (13:48 -0300)]
V4L/DVB (4673): Mark the two newer ioctls as experimental

VIDIOC_ENUM_FRAMESIZES and VIDIOC_ENUM_FRAMEINTERVALS ioctls are meant
to be used to provide better support for webcams. Currently, it is not yet
used on kernel drivers.
Better to keep it marked as experimental, until we have several kernel drivers
supporting those features.

Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4672): Frame format enumeration (1/2)
Laurent Pinchart [Thu, 28 Sep 2006 16:42:05 +0000 (13:42 -0300)]
V4L/DVB (4672): Frame format enumeration (1/2)

Add VIDIOC_ENUM_FRAMESIZES and VIDIOC_ENUM_FRAMEINTERVALS ioctls to enumerate
supported frame sizes and frame intervals.

Signed-off-by: Martin Rubli <martin.rubli@epfl.ch>
Signed-off-by: Laurent Pinchart <laurent.pinchart@skynet.be>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4671): Support for SAA7134-based AVerTV Hybrid A16AR
Petr Baudis [Tue, 26 Sep 2006 19:53:53 +0000 (16:53 -0300)]
V4L/DVB (4671): Support for SAA7134-based AVerTV Hybrid A16AR

This adds support for a hybrid PAL/DVB/FM card.  Unfortunately I tested
only the DVB since I don't have any proper antenna available and I can
receive even the DVB just barely so; I can hear noise in the FM part but I
couldn't catch any station, then again I don't have an FM antenna either.
The PAL/FM and IR control data are based on what I harvested on the 'net.
Perhaps I or someone else will fix them if they turn out to be wrong.

Signed-off-by: Petr Baudis <pasky@ucw.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4670): Allow RC5 codes 64 - 127 in ir-kbd-i2c.c
David Hardeman [Tue, 26 Sep 2006 19:39:00 +0000 (16:39 -0300)]
V4L/DVB (4670): Allow RC5 codes 64 - 127 in ir-kbd-i2c.c

The RC5 coding has for a long time supported commands 64-127 in addition
to 0-63.  This is controlled by the second bit of the RC5 packet (see
The attached patch modifies ir-kbd-i2c.c to allow for commands 64-127,
tested with a PVR350 card in combination with a programmable remote.

Signed-off-by: David Hardeman <david@hardeman.nu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4669): Cx88: autodetect Club3D Zap TV2100 by subsystem id 12ab:2300
Michael Krufky [Mon, 25 Sep 2006 17:14:24 +0000 (14:14 -0300)]
V4L/DVB (4669): Cx88: autodetect Club3D Zap TV2100 by subsystem id 12ab:2300

The Club3D Zap TV2100 has been reported to be a clone of the Yuan PG300 and
KWorld/VStream XPert DVB-T with cx22702

Signed-off-by: Michael Krufky <mkrufky@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4668): Cx88: rename mpeg capability flags from CX88_BOARD_FOO to CX88_MPEG_FOO
Michael Krufky [Mon, 25 Sep 2006 17:09:10 +0000 (14:09 -0300)]
V4L/DVB (4668): Cx88: rename mpeg capability flags from CX88_BOARD_FOO to CX88_MPEG_FOO

The flags for mpeg capabilities are sub-optimally named as
CX88_BOARD_DVB and CX88_BOARD_BLACKBIRD, which creates some confusion.
This patch renames the above to CX88_MPEG_DVB and CX88_MPEG_BLACKBIRD.

Signed-off-by: Michael Krufky <mkrufky@linuxtv.org>
Signed-off-by: Trent Piepho <xyzzy@speakeasy.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4667): Changed cx88_board .dvb and .register to an enum.
Steven Toth [Mon, 25 Sep 2006 15:43:45 +0000 (12:43 -0300)]
V4L/DVB (4667): Changed cx88_board .dvb and .register to an enum.

Some basic cleanup in preperation for a future patch where
the cx88-mpeg functions have to deal with the port being
used by multiple frontends in (mpeg2 hw encoder and dvb demod).

Signed-off-by: Steven Toth <stoth@hauppauge.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4666): Ensure the WM8775 driver is loaded generically for any board.
Steven Toth [Mon, 25 Sep 2006 15:43:42 +0000 (12:43 -0300)]
V4L/DVB (4666): Ensure the WM8775 driver is loaded generically for any board.

A generic change to cards to allow any board to specify whether
it needs the wm8775 module loaded (by the core) or not.

Signed-off-by: Steven Toth <stoth@hauppauge.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4665): Add frontend structure callback for bus acquisition.
Steven Toth [Mon, 25 Sep 2006 15:41:53 +0000 (12:41 -0300)]
V4L/DVB (4665): Add frontend structure callback for bus acquisition.

This patch enables generic bus arbitration callbacks enabling
dvbcore frontend_open and frontend_release to pass 'acquire'
and 'release' hardware messages back into the DVB bridge frameworks.
Frameworks like cx88 can then implement single bus multiple demod
card sharing features, which would prohibit two frontends from attempting
to use a single transport bus at the same time.

Signed-off-by: Steven Toth <stoth@hauppauge.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4664): Pvrusb2: Don't use videodev.h; use v4l2-dev.h in its place
Mike Isely [Sun, 24 Sep 2006 02:47:50 +0000 (23:47 -0300)]
V4L/DVB (4664): Pvrusb2: Don't use videodev.h; use v4l2-dev.h in its place

The function prototype needed in pvrusb2-v4l2.c has been moved to
v4l2-dev.h.  Track that change.

Signed-off-by: Mike Isely <isely@pobox.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4663): Pvrusb2: Get rid of private global context array brain damage
Mike Isely [Sun, 24 Sep 2006 01:30:50 +0000 (22:30 -0300)]
V4L/DVB (4663): Pvrusb2: Get rid of private global context array brain damage

A previous attempt to deal with the upcoming loss of
video_set_drvdata() and video_get_drvdata() resulted in logic which
causes a circular locking dependency - also known as a deadlock.  This
changeset attacks the problem in a different manner, using a technique
that no longer requires the problematic mutex (or that private global
array either).

Signed-off-by: Mike Isely <isely@pobox.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4662): Pvrusb2: Implement VIDIOC_INT_[G|S]_REGISTER
Mike Isely [Sun, 24 Sep 2006 01:26:52 +0000 (22:26 -0300)]
V4L/DVB (4662): Pvrusb2: Implement VIDIOC_INT_[G|S]_REGISTER

Implement VIDIOC_INT_SET_REGISTER and VIDIOC_INT_GET_REGISTER for the
pvrusb2 driver.  This is a debugging aid which will not be enabled
unless CONFIG_VIDEO_ADV_DEBUG has been enabled.

Signed-off-by: Mike Isely <isely@pobox.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4661): Pvrusb2: improve 24XXX config option description
Mike Isely [Sun, 24 Sep 2006 01:19:29 +0000 (22:19 -0300)]
V4L/DVB (4661): Pvrusb2: improve 24XXX config option description

The CONFIG_VIDEO_PVRUSB2_24XXX is not nearly as "experimental" as the
description suggests.  So refine the description to better match reality.

Signed-off-by: Mike Isely <isely@pobox.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4660): Some cleanups at helper chips menu
Mauro Carvalho Chehab [Sat, 30 Sep 2006 09:54:14 +0000 (06:54 -0300)]
V4L/DVB (4660): Some cleanups at helper chips menu

Rearranged itens, so that decoders came first;
cx25840 is, in fact, an audio and video decoder;
Fixed some incorrect upercases;
Removed the word "chip" for some audio processors at item name;
Removed the word "driver" for some item names;

Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4658): Another fix because of dvb_attach
Patrick Boettcher [Wed, 20 Sep 2006 09:06:58 +0000 (06:06 -0300)]
V4L/DVB (4658): Another fix because of dvb_attach

Fixed mt2060-usage with dvb_attach

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4657): Power control of the device for dual board
Patrick Boettcher [Wed, 20 Sep 2006 09:06:11 +0000 (06:06 -0300)]
V4L/DVB (4657): Power control of the device for dual board

Corrected power control of the device for dual boards

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4656): Fixed dvb_attach for dib3000mc in dibusb
Patrick Boettcher [Wed, 20 Sep 2006 07:42:23 +0000 (04:42 -0300)]
V4L/DVB (4656): Fixed dvb_attach for dib3000mc in dibusb

When converting the dib3000mc-driver to dvb_attach I forgot to invert the check for NULL.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4655): Removed compilation warnings
Patrick Boettcher [Wed, 20 Sep 2006 07:40:14 +0000 (04:40 -0300)]
V4L/DVB (4655): Removed compilation warnings

Removed compilation warnings for unused statics and locals

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4654): Added module parameter force_lna_activation
Patrick Boettcher [Wed, 20 Sep 2006 07:37:18 +0000 (04:37 -0300)]
V4L/DVB (4654): Added module parameter force_lna_activation

Added a module parameter for force the activation of any LNA on a board.
Suggest by Steve Toth.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4654a): dib700m is not yet ready.
Patrick Boettcher [Tue, 26 Sep 2006 07:10:14 +0000 (04:10 -0300)]
V4L/DVB (4654a): dib700m is not yet ready.

Removing include for kernel 2.6.19

Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4653): Misc fixes for Nova-T 500
Patrick Boettcher [Tue, 19 Sep 2006 15:51:56 +0000 (12:51 -0300)]
V4L/DVB (4653): Misc fixes for Nova-T 500

- forward the clock to the slave undivided
- when sleeping the 3000 do not shutdown the clock

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4652): Misc fixes for DiB3000MC and Nova-T 500
Patrick Boettcher [Tue, 19 Sep 2006 15:51:53 +0000 (12:51 -0300)]
V4L/DVB (4652): Misc fixes for DiB3000MC and Nova-T 500

- make the timing frequency update work.
- fix AGC calibration for Nova-T 500

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4651): Adding another USB product ID for Nova-T 500
Patrick Boettcher [Tue, 19 Sep 2006 15:51:49 +0000 (12:51 -0300)]
V4L/DVB (4651): Adding another USB product ID for Nova-T 500

Adding another USB product ID for the Nova-T 500. Reported by Jose Alberto Reguero.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4650): Misc fixes for dib0700 download
Patrick Boettcher [Tue, 19 Sep 2006 15:51:46 +0000 (12:51 -0300)]
V4L/DVB (4650): Misc fixes for dib0700 download

Several fixes for dib0700-module. (Firmware error checking, dependency)

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4649): Merged VP702x support to dvb-usb multi input
Patrick Boettcher [Tue, 19 Sep 2006 15:51:43 +0000 (12:51 -0300)]
V4L/DVB (4649): Merged VP702x support to dvb-usb multi input

The fixed support for the VP70x (supports only StarBox2) is now in sync with
latest changes in the dvb-usb framework.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4648): Hauppauge Nova-T 500 support added
Patrick Boettcher [Tue, 19 Sep 2006 15:51:40 +0000 (12:51 -0300)]
V4L/DVB (4648): Hauppauge Nova-T 500 support added

This changeset finalizes the support of the Hauppauge Nova-T 500 (Dual DVB-T).
It adds correct AGC setting for the 3000P, correct firmware download state
detection.
Additionally it fixes the mt2060-driver to be able to be used with dvb_attach.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4647): Added module for DiB0700 based devices
Patrick Boettcher [Tue, 19 Sep 2006 15:51:37 +0000 (12:51 -0300)]
V4L/DVB (4647): Added module for DiB0700 based devices

Added module for DiB0700 based USB devices. This module is preliminary and
untested (because of a lack of test devices) but should work.
Further commits will be necessary to make it work properly.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4646): Misc. changes, DiB3000MC, MT2060
Patrick Boettcher [Tue, 19 Sep 2006 15:51:33 +0000 (12:51 -0300)]
V4L/DVB (4646): Misc. changes, DiB3000MC, MT2060

Changed the attach-function of the dib3000mc-driver to return only one
frontend. In case of multiple dib3000-chips on one board, one has to call the
i2c-enumeration manually before.
Added a field to Microtune 2060 config to output the clock to other
tuners/device on a board.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4645): Added new file for multiple input rewrite
Patrick Boettcher [Tue, 19 Sep 2006 15:51:30 +0000 (12:51 -0300)]
V4L/DVB (4645): Added new file for multiple input rewrite

Forgot to add usb-urb.c which is new after adding support for multiple inputs

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4644): Multi-input fix for dtt200u
Patrick Boettcher [Tue, 19 Sep 2006 15:51:27 +0000 (12:51 -0300)]
V4L/DVB (4644): Multi-input fix for dtt200u

Small fix for multi input for dtt200u module.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agoV4L/DVB (4643): Multi-input patch for DVB-USB device
Patrick Boettcher [Sat, 30 Sep 2006 09:53:48 +0000 (06:53 -0300)]
V4L/DVB (4643): Multi-input patch for DVB-USB device

This patch is the first commit of the Multiple Input Patch for the DVB-USB frame
work.
It changes the DVB-USB-device to be able to have more than one streaming input
(e.g. multiple DVB-T sources) on one device. This is a necessary feature for
the upcoming DiB7700 driven devices.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
18 years agox86: Fix booting with "no387 nofxsr"
Linus Torvalds [Tue, 3 Oct 2006 16:45:46 +0000 (09:45 -0700)]
x86: Fix booting with "no387 nofxsr"

Jesper Juhl reported that testing the software math-emulation by forcing
"no387" doesn't work on modern CPU's.

The reason was two-fold:
 - you also need to pass in "nofxsr" to make sure that we not only don't
   touch the old i387 legacy hardware, it also needs to disable the
   modern XMM/FXSR sequences
 - "nofxsr" didn't actually clear the capability bits immediately,
   leaving the early boot sequence still using FXSR until we got to
   the identify_cpu() stage.

This fixes the "nofxsr" flag to take effect immediately on the boot CPU.

Debugging by Randy Dunlap

Acked-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Jesper Juhl <jesper.juhl@gmail.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Tue, 3 Oct 2006 16:14:00 +0000 (09:14 -0700)]
Merge branch 'devel' of /home/rmk/linux-2.6-arm

* 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 3848/1: pxafb: Add option of fixing video modes and spitz QVGA mode support
  [ARM] 3880/1: remove the last trace of iop31x support
  [ARM] 3879/1: ep93xx: instantiate platform devices for ep93xx ethernet
  [ARM] 3809/3: get rid of 4 megabyte kernel image size limit
  [ARM] Fix XIP_KERNEL build error in arch/arm/mm/mmu.c
  [ARM] 3874/1: Remove leftover usage of asm/timeofday.h

18 years agoMerge branch 'devel' of master.kernel.org:/home/rmk/linux-2.6-serial
Linus Torvalds [Tue, 3 Oct 2006 16:13:29 +0000 (09:13 -0700)]
Merge branch 'devel' of /home/rmk/linux-2.6-serial

* 'devel' of master.kernel.org:/home/rmk/linux-2.6-serial: (21 commits)
  [SERIAL] add PNP IDs for FPI based touchscreens
  [SERIAL] Magic SysRq SAK does nothing on serial consoles
  [SERIAL] tickle NMI watchdog on serial output.
  [SERIAL] Fix oops when removing suspended serial port
  [SERIAL] Fix resume handling bug
  [SERIAL] Remove wrong asm/serial.h inclusions
  [SERIAL] CONFIG_PM=n slim: drivers/serial/8250_pci.c
  [SERIAL] OMAP1510 serial fix for 115200 baud
  [SERIAL] returning proper error from serial core driver
  [SERIAL] Make uart_line_info() correctly tell MMIO from I/O port
  [SERIAL] suspend/resume handlers don't have level arg anymore
  [SERIAL] 8250 resourse management fixes
  [SERIAL] serial_cs: Add quirk for brainboxes 2-port RS232 card
  [SERIAL] serial_cs: handle Nokia multi->single port bodge via config quirk
  [SERIAL] serial_cs: add configuration quirk
  [SERIAL] serial_cs: Convert Oxford 950 / Possio GCC wakeup quirk
  [SERIAL] serial_cs: convert IBM post-init handling to a quirk
  [SERIAL] serial_cs: allow wildcarded quirks
  [SERIAL] serial_cs: convert multi-port table to quirk table
  [SERIAL] serial_cs: Use clean up multiport card detection
  ...

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6
Linus Torvalds [Tue, 3 Oct 2006 15:54:28 +0000 (08:54 -0700)]
Merge /pub/scm/linux/kernel/git/lethal/sh-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6:
  sh: Fixup __raw_read_trylock().
  sh: Kill off remaining config.h references.
  sh: Initial gitignore list
  sh: build fixes for defconfigs.
  sh: Kill off more dead headers.
  sh: Set pclk default for SH7705.
  sh: defconfig updates.

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh64-2.6
Linus Torvalds [Tue, 3 Oct 2006 15:53:46 +0000 (08:53 -0700)]
Merge /pub/scm/linux/kernel/git/lethal/sh64-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh64-2.6:
  sh64: Remove me from sh64 maintainers.
  sh64: Update cayman defconfig.

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Tue, 3 Oct 2006 15:53:26 +0000 (08:53 -0700)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Update defconfig.
  [SPARC64]: Do not include compat.h from asm-sparc64/signal.h any more.
  [SPARC64]: Move signal compat bits to new header file.

18 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Tue, 3 Oct 2006 15:52:26 +0000 (08:52 -0700)]
Merge git://git./linux/kernel/git/paulus/powerpc

* git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (29 commits)
  [POWERPC] Fix rheap alignment problem
  [POWERPC] Use check_legacy_ioport() for ISAPnP
  [POWERPC] Avoid NULL pointer in gpio1_interrupt
  [POWERPC] Enable generic rtc hook for the MPC8349 mITX
  [POWERPC] Add powerpc get/set_rtc_time interface to new generic rtc class
  [POWERPC] Create a "wrapper" script and use it in arch/powerpc/boot
  [POWERPC] fix spin lock nesting in hvc_iseries
  [POWERPC] EEH failure to mark pci slot as frozen.
  [POWERPC] update powerpc defconfig files after libata kconfig breakage
  [POWERPC] enable sysrq in pmac32_defconfig
  [POWERPC] UPIO_TSI cleanup
  [POWERPC] rewrite mkprep and mkbugboot in sane C
  [POWERPC] maple/pci iomem annotations
  [POWERPC] powerpc oprofile __user annotations
  [POWERPC] cell spufs iomem annotations
  [POWERPC] NULL noise removal: spufs
  [POWERPC] ppc math-emu needs -fno-builtin-fabs for math.c and fabs.c
  [POWERPC] update mpc8349_itx_defconfig and remove some debug settings
  [POWERPC] Always call cede in pseries dedicated idle loop
  [POWERPC] Fix loop logic in irq_alloc_virt()
  ...

18 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild
Linus Torvalds [Tue, 3 Oct 2006 15:51:38 +0000 (08:51 -0700)]
Merge git://git./linux/kernel/git/sam/kbuild

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild:
  kbuild: trivial documentation fixes
  kconfig: fix saving alternate kconfig file in parent dir
  kbuild: make modpost processing configurable
  kconfig/menuconfig: do not let ncurses clutter screen on exit
  kconfig/lxdialog: clear long menu lines
  kbuild: do not build mconf & lxdialog unless needed
  kconfig/lxdialog: fix make mrproper
  kconfig/lxdialog: support resize
  kconfig/lxdialog: let <ESC><ESC> behave as expected
  kconfig/menuconfig: lxdialog is now built-in
  kconfig/lxdialog: add a new theme bluetitle which is now default
  kconfig/lxdialog: add support for color themes and add blackbg theme
  kconfig/lxdialog: refactor color support

18 years ago[PATCH] pr_debug: check pr_debug() arguments
Zach Brown [Tue, 3 Oct 2006 08:16:15 +0000 (01:16 -0700)]
[PATCH] pr_debug: check pr_debug() arguments

check pr_debug() arguments

When DEBUG isn't defined pr_debug() is defined away as an empty macro.  By
throwing away the arguments we allow completely incorrect code to build.

Instead let's make it an empty inline which checks arguments and mark it so gcc
can check the format specification.

This results in a seemingly insignificant code size increase.  A x86-64
allyesconfig:

   text    data     bss     dec     hex filename
25354768        7191098 4854720 37400586        23ab00a vmlinux.before
25354945        7191138 4854720 37400803        23ab0e3 vmlinux

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] isdn: more pr_debug() fixes
Andrew Morton [Tue, 3 Oct 2006 08:16:15 +0000 (01:16 -0700)]
[PATCH] isdn: more pr_debug() fixes

drivers/isdn/sc/event.c: In function 'indicate_status':
drivers/isdn/sc/event.c:49: error: 'events' undeclared (first use in this function)
drivers/isdn/sc/event.c:49: error: (Each undeclared identifier is reported only once
drivers/isdn/sc/event.c:49: error: for each function it appears in.)
drivers/isdn/sc/event.c:49: warning: format '%d' expects type 'int', but argument 4 has type 'ulong'

drivers/isdn/sc/interrupt.c: In function 'interrupt_handler':
drivers/isdn/sc/interrupt.c:97: warning: format '%x' expects type 'unsigned int', but argument 4 has type 'long unsigned int'

drivers/isdn/sc/timer.c: In function 'check_reset':
drivers/isdn/sc/timer.c:80: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'

Acked-by: Karsten Keil <kkeil@suse.de>
Cc: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] isdn-debug-build-fix
Andrew Morton [Tue, 3 Oct 2006 08:16:14 +0000 (01:16 -0700)]
[PATCH] isdn-debug-build-fix

When pr_debug() is enabled:

drivers/isdn/sc/command.c: In function 'command':
drivers/isdn/sc/command.c:107: error: 'commands' undeclared (first use in this function)
drivers/isdn/sc/command.c:107: error: (Each undeclared identifier is reported only once
drivers/isdn/sc/command.c:107: error: for each function it appears in.)
drivers/isdn/sc/command.c:122: warning: format '%x' expects type 'unsigned int', but argument 3 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'dial':
drivers/isdn/sc/command.c:199: warning: format '%d' expects type 'int', but argument 4 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'answer':
drivers/isdn/sc/command.c:221: warning: format '%s' expects type 'char *', but argument 3 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'hangup':
drivers/isdn/sc/command.c:244: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'setl2':
drivers/isdn/sc/command.c:265: error: 'l2protos' undeclared (first use in this function)
drivers/isdn/sc/command.c: In function 'setl3':
drivers/isdn/sc/command.c:297: error: 'l3protos' undeclared (first use in this function)
drivers/isdn/sc/command.c:297: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'acceptb':
drivers/isdn/sc/command.c:315: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'clreaz':
drivers/isdn/sc/command.c:330: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'
drivers/isdn/sc/command.c: In function 'seteaz':
drivers/isdn/sc/command.c:345: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'

Cc: Zach Brown <zach.brown@oracle.com>
Acked-by: Karsten Keil <kkeil@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] arm build fail: vfpsingle.c
Frederik Deweerdt [Tue, 3 Oct 2006 08:16:11 +0000 (01:16 -0700)]
[PATCH] arm build fail: vfpsingle.c

It looks like Zach Brown's patch pr_debug-check-pr_debug-arguments
worked as inteded. That is, it doesn't "allow completely incorrect code
to build." :).

The arm build fails with the following message:
  CC      arch/arm/vfp/vfpsingle.o
  arch/arm/vfp/vfpsingle.c: In function `__vfp_single_normaliseround':
  arch/arm/vfp/vfpsingle.c:201: error: `func' undeclared (first use in
  this function)
  arch/arm/vfp/vfpsingle.c:201: error: (Each undeclared identifier is
  reported only once
  arch/arm/vfp/vfpsingle.c:201: error: for each function it appears in.)
  make[1]: *** [arch/arm/vfp/vfpsingle.o] Error 1
  make: *** [arch/arm/vfp] Error 2

The following patch fixes the issue by using func only when DEBUG is
defined.

Signed-off-by: Frederik Deweerdt <frederik.deweerdt@gmail.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: trident: use size_t length modifier in pr_debug format arguments
Zach Brown [Tue, 3 Oct 2006 08:16:10 +0000 (01:16 -0700)]
[PATCH] pr_debug: trident: use size_t length modifier in pr_debug format arguments

trident: use size_t length modifier in pr_debug format arguments

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: ifb: replace missing comma to separate pr_debug arguments
Zach Brown [Tue, 3 Oct 2006 08:16:10 +0000 (01:16 -0700)]
[PATCH] pr_debug: ifb: replace missing comma to separate pr_debug arguments

ifb: replace missing comma to separate pr_debug arguments

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: dell_rbu: fix pr_debug argument warnings
Zach Brown [Tue, 3 Oct 2006 08:16:09 +0000 (01:16 -0700)]
[PATCH] pr_debug: dell_rbu: fix pr_debug argument warnings

dell_rbu: fix pr_debug argument warnings

Use size_t length modifier when outputting size_t and use %p instead of %lu for
'u8 *'.

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: tipar: repair nonexistant pr_debug argument use
Zach Brown [Tue, 3 Oct 2006 08:16:08 +0000 (01:16 -0700)]
[PATCH] pr_debug: tipar: repair nonexistant pr_debug argument use

tipar: repair nonexistant pr_debug argument use

I guessed what the pr_debug meant by 'data'.
Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: umem: repair nonexistant bh pr_debug reference
Zach Brown [Tue, 3 Oct 2006 08:16:07 +0000 (01:16 -0700)]
[PATCH] pr_debug: umem: repair nonexistant bh pr_debug reference

umem: repair nonexistant bh pr_debug reference

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: sysfs: use size_t length modifier in pr_debug format arguments
Zach Brown [Tue, 3 Oct 2006 08:16:06 +0000 (01:16 -0700)]
[PATCH] pr_debug: sysfs: use size_t length modifier in pr_debug format arguments

sysfs: use size_t length modifier in pr_debug format arguments

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: configfs: use size_t length modifier in pr_debug format argument
Zach Brown [Tue, 3 Oct 2006 08:16:05 +0000 (01:16 -0700)]
[PATCH] pr_debug: configfs: use size_t length modifier in pr_debug format argument

configfs: use size_t length modifier in pr_debug format argument

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Acked-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pr_debug: aio: use size_t length modifier in pr_debug format arguments
Zach Brown [Tue, 3 Oct 2006 08:16:04 +0000 (01:16 -0700)]
[PATCH] pr_debug: aio: use size_t length modifier in pr_debug format arguments

aio: use size_t length modifier in pr_debug format arguments

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: add error reporting to superblock write failure
NeilBrown [Tue, 3 Oct 2006 08:16:03 +0000 (01:16 -0700)]
[PATCH] md: add error reporting to superblock write failure

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: allow SET_BITMAP_FILE to work on 64bit kernel with 32bit userspace
Paul Clements [Tue, 3 Oct 2006 08:16:02 +0000 (01:16 -0700)]
[PATCH] md: allow SET_BITMAP_FILE to work on 64bit kernel with 32bit userspace

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: use ffz instead of find_first_set to convert multiplier to shift
Paul Clements [Tue, 3 Oct 2006 08:16:01 +0000 (01:16 -0700)]
[PATCH] md: use ffz instead of find_first_set to convert multiplier to shift

find_first_set doesn't find the least-significant bit on bigendian machines,
so it is really wrong to use it.

ffs is closer, but takes an 'int' and we have a 'unsigned long'.  So use
ffz(~X) to convert a chunksize into a chunkshift.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: remove 'experimental' classification from raid5 reshape
NeilBrown [Tue, 3 Oct 2006 08:16:00 +0000 (01:16 -0700)]
[PATCH] md: remove 'experimental' classification from raid5 reshape

I have had enough success reports not to believe that this is safe for 2.6.19.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: remove MAX_MD_DEVS which is an arbitrary limit
NeilBrown [Tue, 3 Oct 2006 08:15:59 +0000 (01:15 -0700)]
[PATCH] md: remove MAX_MD_DEVS which is an arbitrary limit

Once upon a time we needed to fixed limit to the number of md devices,
probably because we preallocated some array.  This need no longer exists, but
we still have an arbitrary limit.

So remove MAX_MD_DEVS and allow as many devices as we can fit into the 'minor'
part of a device number.

Also remove some useless noise at init time (which reports MAX_MD_DEVS) and
remove MD_THREAD_NAME_MAX which hasn't been used for a while.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: fix duplicity of levels in md.txt
NeilBrown [Tue, 3 Oct 2006 08:15:58 +0000 (01:15 -0700)]
[PATCH] md: fix duplicity of levels in md.txt

md.txt has two sections describing the 'level' sysfs attribute, and some of
the text is out-of-date.  So make just one section, and make it right.

Cc: Christian Kujau <evil@g-house.de>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: make messages about resync/recovery etc more specific
NeilBrown [Tue, 3 Oct 2006 08:15:57 +0000 (01:15 -0700)]
[PATCH] md: make messages about resync/recovery etc more specific

It is possible to request a 'check' of an md/raid array where the whole array
is read and consistancies are reported.

This uses the same mechanisms as 'resync' and so reports in the kernel logs
that a resync is being started.  This understandably confuses/worries people.

Also the text in /proc/mdstat suggests a 'resync' is happen when it is just a
check.

This patch changes those messages to be more specific about what is happening.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: add a ->congested_fn function for raid5/6
NeilBrown [Tue, 3 Oct 2006 08:15:56 +0000 (01:15 -0700)]
[PATCH] md: add a ->congested_fn function for raid5/6

This is very different from other raid levels and all requests go through a
'stripe cache', and it has congestion management already.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: define ->congested_fn for raid1, raid10, and multipath
NeilBrown [Tue, 3 Oct 2006 08:15:54 +0000 (01:15 -0700)]
[PATCH] md: define ->congested_fn for raid1, raid10, and multipath

raid1, raid10 and multipath don't report their 'congested' status through
bdi_*_congested, but should.

This patch adds the appropriate functions which just check the 'congested'
status of all active members (with appropriate locking).

raid1 read_balance should be modified to prefer devices where
bdi_read_congested returns false.  Then we could use the '&' branch rather
than the '|' branch.  However that should would need some benchmarking first
to make sure it is actually a good idea.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: define backing_dev_info.congested_fn for raid0 and linear
NeilBrown [Tue, 3 Oct 2006 08:15:53 +0000 (01:15 -0700)]
[PATCH] md: define backing_dev_info.congested_fn for raid0 and linear

Each backing_dev needs to be able to report whether it is congested, either by
modulating BDI_*_congested in ->state, or by defining a ->congested_fn.
md/raid did neither of these.  This patch add a congested_fn which simply
checks all component devices to see if they are congested.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: Improve locking around error handling
NeilBrown [Tue, 3 Oct 2006 08:15:53 +0000 (01:15 -0700)]
[PATCH] md: Improve locking around error handling

The error handling routines don't use proper locking, and so two concurrent
errors could trigger a problem.

So:
  - use test-and-set and test-and-clear to synchonise
    the In_sync bits with the ->degraded count
  - use the spinlock to protect updates to the
    degraded count (could use an atomic_t but that
    would be a bigger change in code, and isn't
    really justified)
  - remove un-necessary locking in raid5

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: Remove working_disks from raid1 state data
NeilBrown [Tue, 3 Oct 2006 08:15:52 +0000 (01:15 -0700)]
[PATCH] md: Remove working_disks from raid1 state data

It is equivalent to conf->raid_disks - conf->mddev->degraded.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: Factor out part of raid1d into a separate function
NeilBrown [Tue, 3 Oct 2006 08:15:51 +0000 (01:15 -0700)]
[PATCH] md: Factor out part of raid1d into a separate function

raid1d has toooo many nested block, so take the fix_read_error functionality
out into a separate function.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: remove unnecessary variable x in stripe_to_pdidx()
Coywolf Qi Hunt [Tue, 3 Oct 2006 08:15:50 +0000 (01:15 -0700)]
[PATCH] md: remove unnecessary variable x in stripe_to_pdidx()

Signed-off-by: Coywolf Qi Hunt <qiyong@freeforge.net>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: new sysfs interface for setting bits in the write-intent-bitmap
Paul Clements [Tue, 3 Oct 2006 08:15:49 +0000 (01:15 -0700)]
[PATCH] md: new sysfs interface for setting bits in the write-intent-bitmap

Add a new sysfs interface that allows the bitmap of an array to be dirtied.
The interface is write-only, and is used as follows:

echo "1000" > /sys/block/md2/md/bitmap

(dirty the bit for chunk 1000 [offset 0] in the in-memory and on-disk
bitmaps of array md2)

echo "1000-2000" > /sys/block/md1/md/bitmap

(dirty the bits for chunks 1000-2000 in md1's bitmap)

This is useful, for example, in cluster environments where you may need to
combine two disjoint bitmaps into one (following a server failure, after a
secondary server has taken over the array).  By combining the bitmaps on
the two servers, a full resync can be avoided (This was discussed on the
list back on March 18, 2005, "[PATCH 1/2] md bitmap bug fixes" thread).

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: remove 'working_disks' from raid10 state
NeilBrown [Tue, 3 Oct 2006 08:15:48 +0000 (01:15 -0700)]
[PATCH] md: remove 'working_disks' from raid10 state

It isn't needed as mddev->degraded contains equivalent info.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: remove the working_disks and failed_disks from raid5 state data.
NeilBrown [Tue, 3 Oct 2006 08:15:47 +0000 (01:15 -0700)]
[PATCH] md: remove the working_disks and failed_disks from raid5 state data.

They are not needed.  conf->failed_disks is the same as mddev->degraded and
conf->working_disks is conf->raid_disks - mddev->degraded.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: replace magic numbers in sb_dirty with well defined bit flags
NeilBrown [Tue, 3 Oct 2006 08:15:46 +0000 (01:15 -0700)]
[PATCH] md: replace magic numbers in sb_dirty with well defined bit flags

Instead of magic numbers (0,1,2,3) in sb_dirty, we have
some flags instead:
MD_CHANGE_DEVS
   Some device state has changed requiring superblock update
   on all devices.
MD_CHANGE_CLEAN
   The array has transitions from 'clean' to 'dirty' or back,
   requiring a superblock update on active devices, but possibly
   not on spares
MD_CHANGE_PENDING
   A superblock update is underway.

We wait for an update to complete by waiting for all flags to be clear.  A
flag can be set at any time, even during an update, without risk that the
change will be lost.

Stop exporting md_update_sb - isn't needed.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: factor out part of raid10d into a separate function.
NeilBrown [Tue, 3 Oct 2006 08:15:45 +0000 (01:15 -0700)]
[PATCH] md: factor out part of raid10d into a separate function.

raid10d has toooo many nested block, so take the fix_read_error functionality
out into a separate function.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: fix a comment that is wrong in raid5.h
NeilBrown [Tue, 3 Oct 2006 08:15:45 +0000 (01:15 -0700)]
[PATCH] md: fix a comment that is wrong in raid5.h

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] md: the scheduled removal of the START_ARRAY ioctl for md
Adrian Bunk [Tue, 3 Oct 2006 08:15:44 +0000 (01:15 -0700)]
[PATCH] md: the scheduled removal of the START_ARRAY ioctl for md

This patch contains the scheduled removal of the START_ARRAY ioctl for md.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm table: add target flush
Bryn Reeves [Tue, 3 Oct 2006 08:15:43 +0000 (01:15 -0700)]
[PATCH] dm table: add target flush

This patch adds support for a per-target dm_flush_fn method.  This is needed
to allow dm-loop to invalidate page cache mappings in response to BLKFLSBUF
ioctl commands.

Signed-off-by: Bryn Reeves <breeves@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: extract device limit setting
Bryn Reeves [Tue, 3 Oct 2006 08:15:42 +0000 (01:15 -0700)]
[PATCH] dm: extract device limit setting

Separate the setting of device I/O limits from dm_get_device().  dm-loop will
use this.

Signed-off-by: Bryn Reeves <breeves@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: use private biosets
Stefan Bader [Tue, 3 Oct 2006 08:15:41 +0000 (01:15 -0700)]
[PATCH] dm: use private biosets

I found a problem within device-mapper that occurs in low-mem situations.  It
was found using a mirror target but I think in theory it would hit any setup
that stacks device-mapper devices (like LVM on top of multipath).

Since device-mapper core uses the common fs_bioset in clone_bio(), and a
private, but still global, bio_set in split_bvec() it is possible that the
filesystem and the first level target successfully get bios but the lower
level target doesn't because there is no more memory and the pool was drained
by upper layers.  So the remapping will be stuck forever.  To solve this
device-mapper core needs to use a private bio_set for each device.

Signed-off-by: Stefan Bader <Stefan.Bader@de.ibm.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm crypt: use private biosets
Milan Broz [Tue, 3 Oct 2006 08:15:40 +0000 (01:15 -0700)]
[PATCH] dm crypt: use private biosets

In the low memory situation dm-crypt needs to use a private mempool of bios to
avoid blocking.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm crypt: move io to workqueue
Milan Broz [Tue, 3 Oct 2006 08:15:39 +0000 (01:15 -0700)]
[PATCH] dm crypt: move io to workqueue

This patch is designed to help dm-crypt comply with the
new constraints imposed by the following patch in -mm:
  md-dm-reduce-stack-usage-with-stacked-block-devices.patch

Under low memory the existing implementation relies upon waiting for I/O
submitted recursively to generic_make_request() completing before the original
generic_make_request() call can return.

This patch moves the I/O submission to a workqueue so the original
generic_make_request() can return immediately.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm crypt: restructure write processing
Milan Broz [Tue, 3 Oct 2006 08:15:38 +0000 (01:15 -0700)]
[PATCH] dm crypt: restructure write processing

Restructure the dm-crypt write processing in preparation for workqueue changes
in the next patches.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm crypt: restructure for workqueue change
Milan Broz [Tue, 3 Oct 2006 08:15:37 +0000 (01:15 -0700)]
[PATCH] dm crypt: restructure for workqueue change

Restructure part of the dm-crypt code in preparation for workqueue changes.

Use 'base_bio' or 'clone' variable names consistently throughout.  No
functional changes are included in this patch.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm crypt: add key msg
Milan Broz [Tue, 3 Oct 2006 08:15:37 +0000 (01:15 -0700)]
[PATCH] dm crypt: add key msg

Add the facility to wipe the encryption key from memory (for example while a
laptop is suspended) and reinstate it later (when the laptop gets resumed).

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm table: add target preresume
Milan Broz [Tue, 3 Oct 2006 08:15:36 +0000 (01:15 -0700)]
[PATCH] dm table: add target preresume

This patch adds a target preresume hook.

It is called before the targets are resumed and if it returns an error the
resume gets cancelled.

The crypt target will use this to indicate that it is unable to process I/O
because no encryption key has been supplied.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: add debug macro
Bryn Reeves [Tue, 3 Oct 2006 08:15:35 +0000 (01:15 -0700)]
[PATCH] dm: add debug macro

Add CONFIG_DM_DEBUG and DMDEBUG() macro.

Signed-off-by: Bryn Reeves <breeves@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: add uevent change event on resume
Hannes Reinecke [Tue, 3 Oct 2006 08:15:35 +0000 (01:15 -0700)]
[PATCH] dm: add uevent change event on resume

Device-mapper devices are not accessible until a 'resume' ioctl has been
issued.  For userspace to find out when this happens we need to generate an
uevent for udev to take appropriate action.

As discussed at OLS we should send 'change' events for 'resume'.  We can think
of no useful purpose served by also having 'suspend' events.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm mpath: use kzalloc
Micha³ Miros³aw [Tue, 3 Oct 2006 08:15:34 +0000 (01:15 -0700)]
[PATCH] dm mpath: use kzalloc

Use kzalloc() instead of kmalloc() + memset().

Signed-off-by: Micha³ Miros³aw <mirq-linux@rere.qmqm.pl>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm mpath: tidy ctr
Micha³ Miros³aw [Tue, 3 Oct 2006 08:15:33 +0000 (01:15 -0700)]
[PATCH] dm mpath: tidy ctr

After initialising m->ti, there's no need to pass it in subsequent calls to
static functions used for parsing parameters.

Signed-off-by: Micha³ Miros³aw <mirq-linux@rere.qmqm.pl>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm mirror: remove trailing space from table
Jonathan Brassow [Tue, 3 Oct 2006 08:15:32 +0000 (01:15 -0700)]
[PATCH] dm mirror: remove trailing space from table

Remove trailing space from 'dmsetup table' output.

Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: fix freeing pending exception
Alasdair G Kergon [Tue, 3 Oct 2006 08:15:31 +0000 (01:15 -0700)]
[PATCH] dm snapshot: fix freeing pending exception

If a snapshot became invalid while there are outstanding pending_exceptions,
when pending_complete() processes each one it forgets to remove the
corresponding exception from its exception table before freeing it.

Fix this by moving the 'out:' label up one statement so that
remove_exception() is always called.  Then __invalidate_exception() no longer
needs to call it and its 'pe' argument become superfluous.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: tidy pe ref counting
Alasdair G Kergon [Tue, 3 Oct 2006 08:15:30 +0000 (01:15 -0700)]
[PATCH] dm snapshot: tidy pe ref counting

Rename sibling_count to ref_count and introduce get and put functions.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: add workqueue
Alasdair G Kergon [Tue, 3 Oct 2006 08:15:30 +0000 (01:15 -0700)]
[PATCH] dm snapshot: add workqueue

Add a workqueue so that I/O can be queued up to be flushed from a separate
thread (e.g.  if local interrupts are disabled).

A new per-snapshot spinlock pe_lock is introduced to protect queued_bios.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: tidy pending_complete
Alasdair G Kergon [Tue, 3 Oct 2006 08:15:29 +0000 (01:15 -0700)]
[PATCH] dm snapshot: tidy pending_complete

This patch rearranges the pending_complete() code so that the functional
changes in subsequent patches are clearer.

By consolidating the error and the non-error paths, we can move
error_snapshot_bios() and __flush_bios() in line.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: tidy snapshot_map
Alasdair G Kergon [Tue, 3 Oct 2006 08:15:28 +0000 (01:15 -0700)]
[PATCH] dm snapshot: tidy snapshot_map

This patch rearranges the snapshot_map code so that the functional changes in
subsequent patches are clearer.

The only functional change is to replace the existing read lock with a write
lock which the next patch needs.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: fix metadata writing when suspending
Mark McLoughlin [Tue, 3 Oct 2006 08:15:27 +0000 (01:15 -0700)]
[PATCH] dm snapshot: fix metadata writing when suspending

When suspending a device-mapper device, dm_suspend() sleeps until all
necessary I/O is completed.  This state is triggered by a callback from
persistent_commit().  But some I/O can still be issued *after* the callback
(to prepare the next metadata area for use if the current one is full).  This
patch delays the callback until after that I/O is complete.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: make read and write exception functions void
Mark McLoughlin [Tue, 3 Oct 2006 08:15:26 +0000 (01:15 -0700)]
[PATCH] dm snapshot: make read and write exception functions void

read_exception() and write_exception() only return an error if supplied with
an out-of-range index.  If this ever happens it's the result of a bug in the
calling code so we handle this with an assertion and remove the error handling
in the callers.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: fix metadata error handling
Mark McLoughlin [Tue, 3 Oct 2006 08:15:25 +0000 (01:15 -0700)]
[PATCH] dm snapshot: fix metadata error handling

Fix the error handling when store.read_metadata is called: the error should be
returned immediately.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: allow zero chunk_size
Mark McLoughlin [Tue, 3 Oct 2006 08:15:25 +0000 (01:15 -0700)]
[PATCH] dm snapshot: allow zero chunk_size

The chunk size of snapshots cannot be changed so it is redundant to require it
as a parameter when activating an existing snapshot.  Allow a value of zero in
this case and ignore it.  For a new snapshot, use a default value if zero is
specified.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm snapshot: fix invalidation ENOMEM
Milan Broz [Tue, 3 Oct 2006 08:15:24 +0000 (01:15 -0700)]
[PATCH] dm snapshot: fix invalidation ENOMEM

Fix ENOMEM error sign.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: fix alloc_dev error path
Ishai Rabinovitz [Tue, 3 Oct 2006 08:15:22 +0000 (01:15 -0700)]
[PATCH] dm: fix alloc_dev error path

While reading the code I found a bug in the error path in alloc_dev in dm.c

When blk_alloc_queue fails there is no call to free_minor.

This patch fixes the problem.

Signed-off-by: Ishai Rabinovitz <ishai@mellanox.co.il>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: support ioctls on mapped devices: fix with fake file
Milan Broz [Tue, 3 Oct 2006 08:15:22 +0000 (01:15 -0700)]
[PATCH] dm: support ioctls on mapped devices: fix with fake file

The new ioctl code passes the wrong file pointer to the underlying device.
No file pointer is available so make a temporary fake one.

ioctl_by_bdev() does set_fs(KERNEL_DS) so it's for ioctls originating
within the kernel and unsuitable here.  We are processing ioctls that
originated in userspace and mapping them to different devices.  Fixing the
existing callers that pass a NULL file struct and consolidating the
fake_file users are separate matters to solve in later patches.

Signed-off-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] dm: export blkdev_driver_ioctl
Alasdair G Kergon [Tue, 3 Oct 2006 08:15:21 +0000 (01:15 -0700)]
[PATCH] dm: export blkdev_driver_ioctl

Export blkdev_driver_ioctl for device-mapper.

If we get as far as the device-mapper ioctl handler, we know the ioctl is not
a standard block layer BLK* one, so we don't need to check for them a second
time and can call blkdev_driver_ioctl() directly.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>