GitHub/MotorolaMobilityLLC/kernel-slsi.git
16 years agoatyfb: correct_chipset() can fail
Ville Syrjala [Thu, 24 Jul 2008 04:31:30 +0000 (21:31 -0700)]
atyfb: correct_chipset() can fail

Atari probe code relies on correct_chipset() failing if the device is not
a mach64 GX/CX.  aty_chips[] array would be indexed with -1 in that case.

Signed-off-by: Ville Syrjala <syrjala@sci.fi>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoatyfb: remove dead code
Olaf Hering [Thu, 24 Jul 2008 04:31:29 +0000 (21:31 -0700)]
atyfb: remove dead code

Remove dead code.  This will slightly change the behaviour of the driver
on systems that support backlight control.  Previously they would just
turn the backlight off using the backlight control but now the generic LCD
code will also turn off the LCD using the POWER_MANAGEMENT register.

Signed-off-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Ville Syrjala <syrjala@sci.fi>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdev: xoffset, yoffset and yres are unsigned
Ville Syrjala [Thu, 24 Jul 2008 04:31:28 +0000 (21:31 -0700)]
fbdev: xoffset, yoffset and yres are unsigned

The xoffset, yoffset and yres members of fb_var_screeninfo are __u32.
Make them unsigned in the code as well.

Signed-off-by: Ville Syrjala <syrjala@sci.fi>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdev: width and height are unsigned
Ville Syrjala [Thu, 24 Jul 2008 04:31:27 +0000 (21:31 -0700)]
fbdev: width and height are unsigned

The width and height members of fb_var_screeninfo are __u32.  The code
initializes them to -1 which seems wrong, and 0 seems like an equally good
default value.

Signed-off-by: Ville Syrjala <syrjala@sci.fi>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohgafb: convert to new platform driver API
Krzysztof Helt [Thu, 24 Jul 2008 04:31:26 +0000 (21:31 -0700)]
hgafb: convert to new platform driver API

Convert the hgafb driver to use new platform driver API.

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=9689

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Anton Vorontsov <avorontsov@ru.mvista.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovfb: only enable if explicitly requested when compiled in
Frans Pop [Thu, 24 Jul 2008 04:31:26 +0000 (21:31 -0700)]
vfb: only enable if explicitly requested when compiled in

The Kconfig help for the vfb driver says:
   Do NOT enable it for normal systems! To protect the innocent, it
   has to be enabled explicitly at boot time using the kernel option
   `video=vfb:'.

This change lets the code match the description.

Support for vfb:disable is kept for backwards compatibility; vfb:off works
because it is tested at a higher level.

Note: any undefined option (e.g. vfb:enable) will also enable this driver.

The relevant code has been unchanged since before the migration to
git (2.6.12).

This patch fixes bugzilla #9310 and was the root cause behind
http://lkml.org/lkml/2008/5/31/220.

Signed-off-by: Frans Pop <elendil@planet.nl>
Cc: Antonino A. Daplas <adaplas@gmail.com>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdev: SuperH Mobile LCDC Driver
Magnus Damm [Thu, 24 Jul 2008 04:31:24 +0000 (21:31 -0700)]
fbdev: SuperH Mobile LCDC Driver

This is the SuperH Mobile LCDC frame buffer driver V2, adding support for
the LCDC block found in SuperH Mobile processors.  The hardware supports
up to two LCD panels per LCDC block, and both RGB and SYS interfaces can
be used to hook up LCD panels/modules.

The device driver is a regular platform driver, so LCD configuration and
board specific hooks are passed to the driver using platform data.  LCD
modules using SYS interface often require special configuration using the
SYS bus, and to solve this cleanly the driver provides SYS interface
operations to the board code.

Tested on sh7723 and sh7722 processors with a SYS16A QVGA panel and WVGA
panels using RGB16 and RGB18 interfaces.

Signed-off-by: Magnus Damm <damm@igel.co.jp>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Reviewed-by: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotdfxfb: remove ypan checks done by a higher layer
Krzysztof Helt [Thu, 24 Jul 2008 04:31:24 +0000 (21:31 -0700)]
tdfxfb: remove ypan checks done by a higher layer

These checks and assignments are done by a higher layer so remove them
from the driver.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovga16fb: source code improvement
Krzysztof Helt [Thu, 24 Jul 2008 04:31:23 +0000 (21:31 -0700)]
vga16fb: source code improvement

Use constants and functions from the vga.h file.  Also add module
description.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotdfxfb: add mode_option module parameter
Krzysztof Helt [Thu, 24 Jul 2008 04:31:22 +0000 (21:31 -0700)]
tdfxfb: add mode_option module parameter

Small step toward unification of mode setting parameter.  This is required
to fix the Bugzilla's bug 9847

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: documentation update
Krzysztof Helt [Thu, 24 Jul 2008 04:31:22 +0000 (21:31 -0700)]
tridentfb: documentation update

Make the tridentfb documentation closer to current state of the tridentfb
driver.  Fix also some formatting.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agouvesafb: change mode parameter to mode_option
Krzysztof Helt [Thu, 24 Jul 2008 04:31:21 +0000 (21:31 -0700)]
uvesafb: change mode parameter to mode_option

Make more drivers use the "mode_option" parameter.  This one is quite new
so drop the old "mode" parameter before someone starts using it seriously.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbcon: make logo_height a local variable
Krzysztof Helt [Thu, 24 Jul 2008 04:31:21 +0000 (21:31 -0700)]
fbcon: make logo_height a local variable

Make logo_height variable local in the only function it is used.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoatmel_lcdfb: FIFO underflow management
Nicolas Ferre [Thu, 24 Jul 2008 04:31:20 +0000 (21:31 -0700)]
atmel_lcdfb: FIFO underflow management

Manage atmel_lcdfb FIFO underflow

Resetting the LCD and DMA allows to fix screen shifting after a FIFO
underflow.  It follows reset sequence from errata "LCD Screen Shifting
After a Reset".

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Haavard Skinnemoen <hskinnemoen@atmel.com>
Cc: Andrew Victor <linux@maxim.org.za>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovga16fb: test virtual screen range before subtraction on unsigned
Roel Kluin [Thu, 24 Jul 2008 04:31:19 +0000 (21:31 -0700)]
vga16fb: test virtual screen range before subtraction on unsigned

dx and dy are u32's, so the test should occur before the subtraction

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Cc: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoatafb: test virtual screen range before subtraction on unsigned
Roel Kluin [Thu, 24 Jul 2008 04:31:18 +0000 (21:31 -0700)]
atafb: test virtual screen range before subtraction on unsigned

dx and dy are u32's, so the test should occur before the subtraction

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Cc: Tim Schmielau <tim@physik3.uni-rostock.de>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Antonino Daplas <adaplas@pol.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoamifb: test virtual screen range before subtraction on unsigned
Roel Kluin [Thu, 24 Jul 2008 04:31:18 +0000 (21:31 -0700)]
amifb: test virtual screen range before subtraction on unsigned

dx and dy are u32's, so the test should occur before the subtraction

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Cc: Antonino Daplas <adaplas@pol.net>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoaty128fb: test below 0 on unsigned pll->post_divider
Roel Kluin [Thu, 24 Jul 2008 04:31:17 +0000 (21:31 -0700)]
aty128fb: test below 0 on unsigned pll->post_divider

pll->post_divider is unsigned, so the test fails

Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agodrivers/video/aty/radeon_base.c: notify user if sysfs_create_bin_file() failed
Tony Breeds [Thu, 24 Jul 2008 04:31:16 +0000 (21:31 -0700)]
drivers/video/aty/radeon_base.c: notify user if sysfs_create_bin_file() failed

Current kernel builds warn about:
drivers/video/aty/radeon_base.c: In function 'radeonfb_pci_register':
drivers/video/aty/radeon_base.c:2334: warning: ignoring return value of 'sysfs_create_bin_file', declared with attribute warn_unused_result
drivers/video/aty/radeon_base.c:2336: warning: ignoring return value of 'sysfs_create_bin_file', declared with attribute warn_unused_result

Do minimal checking of these functions and issue a warning if either
fails.  They don't seem to be critical..

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Tony Breeds <tony@bakeyournoodle.com>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agosa1100fb: convert ctrlr_sem in a mutex
Matthias Kaehlcke [Thu, 24 Jul 2008 04:31:16 +0000 (21:31 -0700)]
sa1100fb: convert ctrlr_sem in a mutex

The semaphore ctrlr_sem is used as a mutex.  Convert it to the mutex API

Signed-off-by: Matthias Kaehlcke <matthias@kaehlcke.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopxafb: convert ctrlr_sem in a mutex
Matthias Kaehlcke [Thu, 24 Jul 2008 04:31:14 +0000 (21:31 -0700)]
pxafb: convert ctrlr_sem in a mutex

The semaphore ctrlr_sem is used as a mutex.  Convert it to the mutex API.

Signed-off-by: Matthias Kaehlcke <matthias@kaehlcke.net>
Cc: Daniel Mack <daniel@caiaq.de>
Cc: Eric Miao <eric.miao@marvell.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovideo/sis/: remove compat code
Adrian Bunk [Thu, 24 Jul 2008 04:31:12 +0000 (21:31 -0700)]
video/sis/: remove compat code

This patch removes compat code for older kernel versions.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: <thomas@winischhofer.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomdacon messing up default vc's - set default to vc13-16 again
Roland Kletzing [Thu, 24 Jul 2008 04:31:10 +0000 (21:31 -0700)]
mdacon messing up default vc's - set default to vc13-16 again

mdacon incorrectly detects MDA hardware on systems without such graphics card.

One may load this module by chance, for example when doing some systematical
module-testing, and if there is no Monochrome Display Adapter attached ,
module init renders vc1-16 completely unusable.

I and others have run into this more than once.  see [Bug 224522 - modprobe
mdacon freezes machine -> https://bugzilla.novell.com/show_bug.cgi?id=224522 ]
for example

Apparently  proper MDA detection seems to be broken for a long time - seems to
be related to those #ifdef TEST_MDA_B statements added by Edward Betts.

this commit back in 2002 made things even worse :
http://git.kernel.org/?p=linux/kernel/git/tglx/history.git;a=commit;h=c72757b49c88914433244757fb4967fc63546685

It changed default vc allocation from 13-16 to 1-16 for no apparent reason
(!?) , and with that (and without X), mdacon grabs the vc you`re currently
sitting on and locks you out.

this is from Kconfig :
>config MDA_CONSOLE
>        depends on !M68K && !PARISC && ISA
>        tristate "MDA text console (dual-headed) (EXPERIMENTAL)"
>        ---help---
>          Say Y here if you have an old MDA or monochrome Hercules graphics
>          adapter in your system acting as a second head ( = video card). You
>          will then be able to use two monitors with your Linux system. Do not
>          say Y here if your MDA card is the primary card in your system; the
>          normal VGA driver will handle it.

As we can see mdacon is just meant as an additional driver for dual-head
setup, and since kernel 2.4.36 still defaults to vc13-16 , setting the default
back to that value again shouldn`t do any harm.

Hereby i'm reverting that change, setting default back to to vc13-16 again.

Besides the fact that mdacon may be rarely or never be used these days and
could perhaps put to trash anyway (pre-dinosaur hardware!), indeed this is not
a real solution, but at least it removes the unfortunate side-effect of
messing up the vc you`re working on.

Signed-off-by: Roland Kletzing <devzero@web.de>
Cc: James Simmons <jsimmons@infradead.org>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Tim Schmielau <tim@physik3.uni-rostock.de>
Cc: Jan Engelhardt <jengelh@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: remove warning message that cyblafb driver should be used
Krzysztof Helt [Thu, 24 Jul 2008 04:31:09 +0000 (21:31 -0700)]
tridentfb: remove warning message that cyblafb driver should be used

The tridentfb driver should handle now all chipsets handled by the cyblafb
driver.  Remove the message which claims that support will be removed.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: add imageblit acceleration for Blade3D family
Krzysztof Helt [Thu, 24 Jul 2008 04:31:08 +0000 (21:31 -0700)]
tridentfb: add imageblit acceleration for Blade3D family

Add imageblit acceleration for the Blade3D family of cores.  The code is
based on code from the cyblafb driver.

It is a step toward assimilating back the cyblafb driver into the
tridentfb driver.  The cyblafb driver handles a subfamily of the Trident
Blade3d cores.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: Blade3D clock fixes
Krzysztof Helt [Thu, 24 Jul 2008 04:31:08 +0000 (21:31 -0700)]
tridentfb: Blade3D clock fixes

This patch fixes following problems:
- does not allow the m parameter to reach 0 as
  it locks the graphics core (power cycle needed)
- for the newer chips (with new clock registers)
  does not allow of n / m ratio below 4 as it gives
  unstable image on the Blade3D core
- extend shift parameter (k) range  to 2 for the newer
  chips to cope with the n /m >= 4 limit at low resolution
  (bandwidth) modes
- prefer modes with higher n / m ratio (higher k values)

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: y-panning fixes
Krzysztof Helt [Thu, 24 Jul 2008 04:31:07 +0000 (21:31 -0700)]
tridentfb: y-panning fixes

The Trident cards uses only 20-bit address of screen start in double
words.  This allows addressing for only 4MB of video memory so check this.

Also remove some redundant checks and assignments.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: fix 224 color logo at 8 bpp
Krzysztof Helt [Thu, 24 Jul 2008 04:31:06 +0000 (21:31 -0700)]
tridentfb: fix 224 color logo at 8 bpp

Fix depth setting for 8 bpp mode.  The nice 224 color logo is not
displayed in 8 bpp depth without this fix.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: fix console freeze when switching from X11
Krzysztof Helt [Thu, 24 Jul 2008 04:31:06 +0000 (21:31 -0700)]
tridentfb: fix console freeze when switching from X11

This patch fixes two problems when acceleration is enabled:

 - console switch from the Xorg locks up the computer
   because the Xorg code locks some registers and disables
   the mmio mode, so reenable these in the
   tridentfb_set_par() and enable_mmio()

 - blacklist the Image975 chipset from setting PCI burst
   mode. This helps with random lock ups of the
   framebuffer on this chip. The same fix is probably
   needed for the Xorg as well.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: source code improvements
Krzysztof Helt [Thu, 24 Jul 2008 04:31:05 +0000 (21:31 -0700)]
tridentfb: source code improvements

This patch contains general source code improvments:
 - more simple functions are inline
 - removes some meaningless output and the VERSION
   string as it is no use
 - eng_par is moved into the tridentfb_par
 - removed small section of code for CyberBladeXPAi1
   which is maybe right for only one resolution
   and refresh rate and is probably redundant now
 - other minor improvements

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: acceleration constants change
Krzysztof Helt [Thu, 24 Jul 2008 04:31:04 +0000 (21:31 -0700)]
tridentfb: acceleration constants change

This patch replaces deprecated constant FB_ACCELF_TEXT with
FBINFO_HWACCEL_DISABLED and adds constants for Trident families of
accelerators.

The FBINFO_HWACCEL_DISABLED is correctly used so noaccel parameter works
now.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: various pixclock and timing improvements
Krzysztof Helt [Thu, 24 Jul 2008 04:31:04 +0000 (21:31 -0700)]
tridentfb: various pixclock and timing improvements

This patch fixes few issues related to timings and pixclock generation:

 - disallow the pixclocks with numerator lower than
   double denominator. This fixes display instability
   for some modes.
 - choose the pixelclock with the highest
   numerator and denominator values. This improve
   image quality and fixes display instability
   for some modes.
 - make interlaced modes work.
 - set synchronization pulses polarization
   correctly.
 - horizontal synchronization timing are now
   the same as generated by X.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: acceleration bug fixes
Krzysztof Helt [Thu, 24 Jul 2008 04:31:03 +0000 (21:31 -0700)]
tridentfb: acceleration bug fixes

This patch fixes two problems when acceleration is enabled:

 - bit for bitblt direction is corrected
   so scrolling down works as expected on 3DImage chips

 - initialization of acceleration is done later
   this helps with initial console malfuntion (on Blade3D
   chips) well documented here:
   http://marc.info/?l=linux-fbdev-users&m=111386953124478&w=2

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: acceleration code improvements
Krzysztof Helt [Thu, 24 Jul 2008 04:31:02 +0000 (21:31 -0700)]
tridentfb: acceleration code improvements

This patch brings various acceleration improvements:
- set  copyarea/fillrect for non-accelerated framebuffer (fix)
- remove 15 bpp depth handling to simplify code as it hardly
  works (15 bpp handling was obviously missing in some switches)
- add fb_sync call and move waiting before accelerated function
  to make acceleration more asynchronous to cpu (few % of speed
  improvement)
- add cpu_relax() call in waiting loops
- make longer register names and name more registers
- move registers' definition to header
- general code improvements (shortening, simplifying)

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: add acceleration for TGUI families
Krzysztof Helt [Thu, 24 Jul 2008 04:31:01 +0000 (21:31 -0700)]
tridentfb: add acceleration for TGUI families

This patch adds acceleration for TGUI 9440 and 96xx chips.  These chips
requires line length to be power of 2, so this is also changed.

It also moves the troubling enable_mmio() function to its final
destination.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: fix hi-color modes for TGUI 9440
Krzysztof Helt [Thu, 24 Jul 2008 04:31:01 +0000 (21:31 -0700)]
tridentfb: fix hi-color modes for TGUI 9440

The TGUI 9440 requires doubling clock for 16bpp (hi-color) modes.

The patch also moves back enable_mmio() call to the right position.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: preserve memory type settings
Krzysztof Helt [Thu, 24 Jul 2008 04:31:00 +0000 (21:31 -0700)]
tridentfb: preserve memory type settings

Do not overwrite bits which contain memory type settings.  It removes
noise pixels ("snow") on Blade3D and 3DImage chips.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: improve check_var function
Krzysztof Helt [Thu, 24 Jul 2008 04:31:00 +0000 (21:31 -0700)]
tridentfb: improve check_var function

Do some additional checks (like pixelclock versus ramdac speed) to
eliminate modes which do not work.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: fix unitialized pseudo_palette
Krzysztof Helt [Thu, 24 Jul 2008 04:30:59 +0000 (21:30 -0700)]
tridentfb: fix unitialized pseudo_palette

Initialize the pseudo_palette pointer properly.  This fixes crash when
16bpp or 32bpp mode is selected.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: add TGUI 9440 support
Krzysztof Helt [Thu, 24 Jul 2008 04:30:58 +0000 (21:30 -0700)]
tridentfb: add TGUI 9440 support

Add support for TGUI 9440 chip.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: improved register values on TGUI 9680
Krzysztof Helt [Thu, 24 Jul 2008 04:30:58 +0000 (21:30 -0700)]
tridentfb: improved register values on TGUI 9680

Improved values for some registers after Xorg Trident driver.  The main
problem was that values set by BIOS have been ignored.

This patch completely remove random pixels ("snow") on the TGUI 9680 and
9440 (not supported yet by the driver).  It does not help with the "snow"
on 3DImage and Blade3D cards.

There is also small improvement in timing calculations (hblank start and
vblank start)

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: improve probe function
Krzysztof Helt [Thu, 24 Jul 2008 04:30:57 +0000 (21:30 -0700)]
tridentfb: improve probe function

Add missing release of allocated fb_info structure and move enable_mmio() to
fix error path.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: fix clock settings for older Trident 96XX chips
Krzysztof Helt [Thu, 24 Jul 2008 04:30:56 +0000 (21:30 -0700)]
tridentfb: fix clock settings for older Trident 96XX chips

The Xorg code shows that Trident models 9660, 9680 and 9682 require a
different clock setting method.  Add the second clock setting method for older
models.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: use mmio access for clock setting
Krzysztof Helt [Thu, 24 Jul 2008 04:30:56 +0000 (21:30 -0700)]
tridentfb: use mmio access for clock setting

Use the mmio outb function instead of direct one.  The mmio registers are
already mapped (in the probe function).

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: fix timing calculations
Krzysztof Helt [Thu, 24 Jul 2008 04:30:55 +0000 (21:30 -0700)]
tridentfb: fix timing calculations

Fix broken timings calculations. This patch helps with following
problems:
 - no left part of screen visible (up to half of the screen)
 - monitor's frequencies are not the ones intended for selected modes
 - if mode with resoultion y > 1024 is selected at least once then
   all modes with y < 1024 are "out of sync" (no display)

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: make use of functions and constants from the vga.h
Krzysztof Helt [Thu, 24 Jul 2008 04:30:54 +0000 (21:30 -0700)]
tridentfb: make use of functions and constants from the vga.h

Make use of functions and constants from the vga.h header to compact the code
and make it more readable.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: move global acceleration hooks into structure
Krzysztof Helt [Thu, 24 Jul 2008 04:30:54 +0000 (21:30 -0700)]
tridentfb: move global acceleration hooks into structure

This patch moves acceleration hooks into the tridentfb_par structure and
removes global hooks.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: convert is_blade and is_xp macros into functions
Krzysztof Helt [Thu, 24 Jul 2008 04:30:53 +0000 (21:30 -0700)]
tridentfb: convert is_blade and is_xp macros into functions

This patch converts the is_blade() and is_xp() macros into local functions.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: move global flat panel variable into structure
Krzysztof Helt [Thu, 24 Jul 2008 04:30:53 +0000 (21:30 -0700)]
tridentfb: move global flat panel variable into structure

This patch moves flat panel indicator into tridentfb_par structure and removes
related global variables and macros.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: move global chip_id into structure
Krzysztof Helt [Thu, 24 Jul 2008 04:30:52 +0000 (21:30 -0700)]
tridentfb: move global chip_id into structure

This patch moves the chip_id into tridentfb_par structure and removes global
chip_id related constants.

It also bumps version of the driver to 0.7.9

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: move global pseudo palette into structure
Krzysztof Helt [Thu, 24 Jul 2008 04:30:51 +0000 (21:30 -0700)]
tridentfb: move global pseudo palette into structure

This patch moves pseudo palette int tridentfb_par structure and removes global
default_var.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: convert fb_info into allocated one
Krzysztof Helt [Thu, 24 Jul 2008 04:30:51 +0000 (21:30 -0700)]
tridentfb: convert fb_info into allocated one

This patch converts fb_info structure from global variable to allocatable one.

The global default_par is moved into function variable.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotridentfb: replace macros with functions
Krzysztof Helt [Thu, 24 Jul 2008 04:30:50 +0000 (21:30 -0700)]
tridentfb: replace macros with functions

This patch replaces macros with static functions and puts tridentfb_par
pointer as the first argument of these functions.  These is a step toward
multihead support.

Additionally, bogus TRIDENT_MMIO define is removed as the driver supports
graphics cards only through the mmio mode.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdev: add the carmine FB driver
Sebastian Siewior [Thu, 24 Jul 2008 04:30:49 +0000 (21:30 -0700)]
fbdev: add the carmine FB driver

Basic FB driver for the carmine chip.  The driver registers two FB devices for
the two possible screens.  The DRAM settings can be be switched via Kconfig
(between eval board and custom).

Signed-off-by: Sebastian Siewior <bigeasy@linutronix.de>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc-vr41xx: add irq_set_freq() and irq_set_state()
Yoichi Yuasa [Thu, 24 Jul 2008 04:30:48 +0000 (21:30 -0700)]
rtc-vr41xx: add irq_set_freq() and irq_set_state()

Implement the ioctls RTC_PIE_ON, RTC_PIE_OFF, RTC_IRQP_SET and
RTC_IRQP_READ in the standard RTC way.

Thanks Dave for noticing it.

Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp>
Cc: David Brownell <david-b@pacbell.net>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc-cmos: avoid spurious irqs
David Brownell [Thu, 24 Jul 2008 04:30:47 +0000 (21:30 -0700)]
rtc-cmos: avoid spurious irqs

This fixes kernel http://bugzilla.kernel.org/show_bug.cgi?id=11112 (bogus
RTC update IRQs reported) for rtc-cmos, in two ways:

  - When HPET is stealing the IRQs, use the first IRQ to grab
    the seconds counter which will be monitored (instead of
    using whatever was previously in that memory);

  - In sane IRQ handling modes, scrub out old IRQ status before
    enabling IRQs.

That latter is done by tightening up IRQ handling for rtc-cmos everywhere,
also ensuring that when HPET is used it's the only thing triggering IRQ
reports to userspace; net object shrink.

Also fix a bogus HPET message related to its RTC emulation.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Report-by: W Unruh <unruh@physics.ubc.ca>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc-at91rm9200: avoid spurious irqs
David Brownell [Thu, 24 Jul 2008 04:30:46 +0000 (21:30 -0700)]
rtc-at91rm9200: avoid spurious irqs

This fixes kernel http://bugzilla.kernel.org/show_bug.cgi?id=11112 (bogus
RTC update IRQs reported) for rtc-at91rm9200 by scrubbing old IRQ status
before enabling IRQs.

It also removes nonfunctional periodic IRQ support from this driver;
only update IRQs are reported, or provided by the hardware.

I suspect some other RTCs probably have versions of #11112; it's easy to
overlook, since most non-RTC drivers don't care about spurious IRQs:
they're not reported to userspace.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Report-by: W Unruh <unruh@physics.ubc.ca>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: rtc-s3c: update IRQ handling
Ben Dooks [Thu, 24 Jul 2008 04:30:45 +0000 (21:30 -0700)]
rtc: rtc-s3c: update IRQ handling

The rtc-s3c.c driver has been using its own ioctl() handling to deal with
alarm and periodic interrupts to handle what should now be done with the
rtc core code.

Change to using the .irq_set_freq and .irq_set_state driver entries and
remove the .ioctl handling.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: rtc-s3c: add __devexit and __devinit markers
Ben Dooks [Thu, 24 Jul 2008 04:30:44 +0000 (21:30 -0700)]
rtc: rtc-s3c: add __devexit and __devinit markers

Add the relevant __devinit and __devexit attributes to the rtc-s3c driver.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc-cmos: improve HPET IRQ glue
David Brownell [Thu, 24 Jul 2008 04:30:43 +0000 (21:30 -0700)]
rtc-cmos: improve HPET IRQ glue

Resolve http://bugzilla.kernel.org/show_bug.cgi?id=11051 and other bugs
related to the way the HPET glue code in rtc-cmos was incomplete and
inconsistent:

 * Switch the approach so that the basic driver code flow isn't
   changed by having HPET ... instead, just have HPET shadow the
   RTC_CONTROL irq enables and RTC_FREQ_SELECT data.  It's only
   coping with IRQ thievery, after all.

 * Do that consistently (!!) to avoid problems when the HPET code
   is out of sync with the real RTC intent.  Examples include:

   - cmos_procfs(), which now reports correct data

   - cmos_irq_set_state() ... also removing the previous PIE_{ON,OFF}
     ioctl support so only one code path manages "periodic" IRQs

   - cmos_do_shutdown() ... currently a "just in case" change.

   - cmos_suspend() and cmos_resume() ... also handling a bug that
     was specific to HPET's IRQ thievery, where the alarm wasn't
     disabled after waking the system

 * Always call that HPET code under the RTC spinlock (it doesn't do
   its own locking)

Also clean up the HPET glue:

 * Add some comments explaining what's going on.

 * Switch to having just one #ifdef for the HPET glue, and inline
   functions (not #defines) to avoid some compiler warnings.

 * Have the probe message also report when HPET IRQs are involved

This still leaves various holes in the HPET glue, like the emulated update
IRQs being out of sync with the RTC, alarms never using day or month
matches, and many extra IRQs (at 64 Hz).

[akpm@linux-foundation.org: fix build]
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Tomas Janousek <tomi@nomi.cz>
Cc: Bernhard Walle <bwalle@suse.de>
Cc: Carlos R. Mafra <crmafra@ift.unesp.br>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: remove and clarify unneeded externs
Carlos R. Mafra [Thu, 24 Jul 2008 04:30:40 +0000 (21:30 -0700)]
rtc: remove and clarify unneeded externs

When CONFIG_HPET_EMULATE_RTC is defined the external declaration of
hpet_rtc_interrupt is redundant due to the inclusion of hpet.h.

When !CONFIG_HPET_EMULATE_RTC we make it clear that hpet_rtc_interrupt is
not used by defining it to return zero.

Signed-off-by: Carlos R. Mafra <crmafra@ift.unesp.br>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: convert the PCF8583 driver to the new I2C style framework with device_ids
Wolfram Sang [Thu, 24 Jul 2008 04:30:39 +0000 (21:30 -0700)]
rtc: convert the PCF8583 driver to the new I2C style framework with device_ids

Convert the PCF8583 driver to the new I2C style framework with device_ids

Signed-off-by: Juergen Beisert <j.beisert@pengutronix.de>
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: David Brownell <david-b@pacbell.net>
Acked-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: rtc-omap footprint shrinkage
David Brownell [Thu, 24 Jul 2008 04:30:38 +0000 (21:30 -0700)]
rtc: rtc-omap footprint shrinkage

Shrink the runtime footprint of the OMAP1 RTC driver a bunch by removing
some old hacks and switching to platform_driver_probe().

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: BCD codeshrink
David Brownell [Thu, 24 Jul 2008 04:30:37 +0000 (21:30 -0700)]
rtc: BCD codeshrink

This updates <linux/bcd.h> to define the key routines as constant
functions, which the macros will then call.  Newer code can now call
bcd2bin() instead of SCREAMING BCD2BIN() TO THE FOUR WINDS.

This lets each driver shrink their codespace by using N function calls to
a single (global) copy of those routines, instead of N inlined copies of
these functions per driver.

These routines aren't used in speed-critical code.  Almost all callers are
in the RTC framework.  Typical per-driver savings is near 300 bytes.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Acked-by: Adrian Bunk <bunk@kernel.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: ds1305/ds1306 driver
David Brownell [Thu, 24 Jul 2008 04:30:36 +0000 (21:30 -0700)]
rtc: ds1305/ds1306 driver

Support the Dallas/Maxim DS1305 and DS1306 RTC chips.  These use SPI, and
support alarms, NVRAM, and a trickle charger for use when their backup
power supply is a supercap or rechargeable cell.

This basic driver doesn't yet support suspend/resume or wakealarms.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: add support for ST M41T94 SPI RTC
Kim B. Heino [Thu, 24 Jul 2008 04:30:34 +0000 (21:30 -0700)]
rtc: add support for ST M41T94 SPI RTC

This patch adds kernel driver for M41T94 RTC chip connected via SPI.
I've tested it on two different AT91-based hardwares.

This is third revision of the patch: some comments made by
Alessandro Zummo fixed.

Revision two added support for century bit and fixes.

Signed-off-by: Kim B. Heino <Kim.Heino@bluegiga.com>
Signed-off-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: remove BKL for ioctl()
David Brownell [Thu, 24 Jul 2008 04:30:33 +0000 (21:30 -0700)]
rtc: remove BKL for ioctl()

Remove implicit use of BKL in ioctl() from the RTC framework.

Instead, the rtc->ops_lock is used.  That's the same lock that already
protects the RTC operations when they're issued through the exported
rtc_*() calls in drivers/rtc/interface.c ...  making this a bugfix, not
just a cleanup, since both ioctl calls and set_alarm() need to update IRQ
enable flags and that implies a common lock (which RTC drivers as a rule
do not provide on their own).

A new comment at the declaration of "struct rtc_class_ops" summarizes
current locking rules.  It's not clear to me that the exceptions listed
there should exist ...  if not, those are pre-existing problems which can
be fixed in a patch that doesn't relate to BKL removal.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Jonathan Corbet <corbet@lwn.net>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: push the BKL down into the driver ioctl method
Alan Cox [Thu, 24 Jul 2008 04:30:32 +0000 (21:30 -0700)]
rtc: push the BKL down into the driver ioctl method

For now just wrap the main logic, but this driver is a prime candidate for
someone wanting to eliminate the lock entirely

[lizf@cn.fujitsu.com: fix build failure]
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: m41t80: use pr_info() as appropriate
Maciej W. Rozycki [Thu, 24 Jul 2008 04:30:32 +0000 (21:30 -0700)]
rtc: m41t80: use pr_info() as appropriate

Replace printk(KERN_INFO ...) calls with appropriate pr_info(...)
equivalents.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Alexander Bigga <ab@mycable.de>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: m41t80: sort header inclusions for readability
Maciej W. Rozycki [Thu, 24 Jul 2008 04:30:29 +0000 (21:30 -0700)]
rtc: m41t80: sort header inclusions for readability

Sort the header inclusions for readability.  No functional changes.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Alexander Bigga <ab@mycable.de>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: remove unused ioctls
Ian Kent [Thu, 24 Jul 2008 04:30:29 +0000 (21:30 -0700)]
autofs4: remove unused ioctls

The ioctls AUTOFS_IOC_TOGGLEREGHOST and AUTOFS_IOC_ASKREGHOST were added
several years ago but what they were intended for has never been
implemented (as far as I'm aware noone uses them) so remove them.

Signed-off-by: Ian Kent <raven@themaw.net>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: reorganize expire pending wait function calls
Ian Kent [Thu, 24 Jul 2008 04:30:28 +0000 (21:30 -0700)]
autofs4: reorganize expire pending wait function calls

This patch re-orgnirzes the checking for and waiting on active expires and
elininates redundant checks.

Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix direct mount pending expire race - correction
Ian Kent [Thu, 24 Jul 2008 04:30:28 +0000 (21:30 -0700)]
autofs4: fix direct mount pending expire race - correction

Appologies, somehow I seem to have sent an out dated version of this
patch. Here is an additional patch that brings the patch up to date.

Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix direct mount pending expire race
Ian Kent [Thu, 24 Jul 2008 04:30:27 +0000 (21:30 -0700)]
autofs4: fix direct mount pending expire race

For direct and offset type mounts that are covered by another mount we
cannot check the AUTOFS_INF_EXPIRING flag during a path walk which leads
to lookups walking into an expiring mount while it is being expired.

For example, for the direct multi-mount map entry with a couple of
offsets:

/race/mm1  /      <server1>:/<path1>
           /om1   <server2>:/<path2>
           /om2   <server1>:/<path3>

an autofs trigger mount is mounted on /race/mm1 and when accessed it is
over mounted and trigger mounts made for /race/mm1/om1 and /race/mm1/om2.
So it isn't possible for path walks to see the expiring flag at all and
they happily walk into the file system while it is expiring.

When expiring these mounts follow_down() must stop at the autofs mount and
all processes must block in the ->follow_link() method (except the daemon)
until the expire is complete.  This is done by decrementing the d_mounted
field of the autofs trigger mount root dentry until the expire is
completed.  In ->follow_link() all processes wait on the expire and the
mount following is completed for the daemon until the expire is complete.

Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix indirect mount pending expire race
Ian Kent [Thu, 24 Jul 2008 04:30:26 +0000 (21:30 -0700)]
autofs4: fix indirect mount pending expire race

The selection of a dentry for expiration and the setting of the
AUTOFS_INF_EXPIRING flag isn't done atomically which can lead to lookups
walking into an expiring mount.

What happens is that an expire is initiated by the daemon and a dentry is
selected for expire but, since there is no lock held between the selection
and setting of the expiring flag, a process may find the flag clear and
continue walking into the mount tree at the same time the daemon attempts
the expire it.

Signed-off-by: Ian Kent <raven@themaw.net>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix pending checks
Ian Kent [Thu, 24 Jul 2008 04:30:25 +0000 (21:30 -0700)]
autofs4: fix pending checks

There are two cases for which a dentry that has a pending mount request
does not wait for completion.  One is via autofs4_revalidate() and the
other via autofs4_follow_link().

In revalidate, after the mount point directory is created, but before the
mount is done, the check in try_to_fill_dentry() can can fail to send the
dentry to the wait queue since the dentry is positive and the lookup flags
may contain only LOOKUP_FOLLOW.  Although we don't trigger a mount for the
LOOKUP_FOLLOW flag, if ther's one pending we might as well wait and use
the mounted dentry for the lookup.

In autofs4_follow_link() the dentry is not checked to see if it is pending
so it may fail to call try_to_fill_dentry() and not wait for mount
completion.

A dentry that is pending must always be sent to the wait queue.

Signed-off-by: Ian Kent <raven@themaw.net>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: cleanup redundant readir code
Ian Kent [Thu, 24 Jul 2008 04:30:24 +0000 (21:30 -0700)]
autofs4: cleanup redundant readir code

The mount triggering functionality of readdir and related functions is no
longer used (and is quite broken as well).  The unused portions have been
removed.

Signed-off-by: Ian Kent <raven@themaw.net>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: indirect dentry must almost always be positive
Ian Kent [Thu, 24 Jul 2008 04:30:23 +0000 (21:30 -0700)]
autofs4: indirect dentry must almost always be positive

We have been seeing mount requests comming to the automount daemon for
keys of the form "<map key>/<non key directory>" which are lookups for
invalid map keys.  But we can check for this in the kernel module and
return a fail immediately, without having to send a request to the daemon.

It is possible to recognise these requests are invalid based on whether
the request dentry is negative and its relation to the autofs file system
root.

For example, given the indirect multi-mount map entry:

idm1  \
    /mm1  <server>:/<path1>
    /mm2  <server>:/<path2>

For a request to mount idm1, IS_ROOT((idm1)->d_parent) will be always be
true and the dentry may be negative.  But directories idm1/mm1 and
idm1/mm2 will always be created as part of the mount request for idm1.  So
any mount request within idm1 itself must have a positive dentry otherwise
the map key is invalid.

In version 4 these multi-mount entries are all mounted and umounted as a
single request and in version 5 the directories idm1/mm1 and idm1/mm2 are
created and an autofs fs mounted on them to act as a mount trigger so the
above is also true.

This also holds true for the autofs version 4 pseudo direct mount feature.
 When this feature is used without the "--ghost" option automount(8) will
create internal submounts as we go down the map key paths which are
essentially normal indirect mounts for which the above holds.  If the
"--ghost" option is given the directories for map keys are created at
daemon startup so valid map entries correspond to postive dentries in the
autofs fs.

autofs version 5 direct mount maps are similar except that the IS_ROOT
check is not needed.  This has been addressed in a previous patch tittled
"autofs4 - detect invalid direct mount requests".

For example, given the direct multi-mount map entry:

/test/dm1  \
    /mm1  <server>:/<path1>
    /mm2  <server>:/<path2>

An autofs fs is mounted on /test/dm1 as a trigger mount and when a mount
is triggered for /test/dm1, the multi-mount offset directories
/test/dm1/mm1 and /test/dm1/mm2 are created and an autofs fs is mounted on
them to act as mount triggers.  So valid direct mount requests must always
have a positive dentry if they correspond to a valid map entry.

Signed-off-by: Ian Kent <raven@themaw.net>
Acked-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: detect invalid direct mount requests
Ian Kent [Thu, 24 Jul 2008 04:30:22 +0000 (21:30 -0700)]
autofs4: detect invalid direct mount requests

autofs v5 direct and offset mounts within an autofs filesystem are
triggered by existing autofs triger mounts so the mount point dentry must
be positive.  If the mount point dentry is negative then the trigger
doesn't exist so we can return fail immediately.

Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix waitq memory leak
Ian Kent [Thu, 24 Jul 2008 04:30:21 +0000 (21:30 -0700)]
autofs4: fix waitq memory leak

If an autofs mount becomes catatonic before autofs4_wait_release() is
called the wait queue counter will not be decremented down to zero and the
entry will never be freed.  There are also races decrementing the wait
counter in the wait release function.  To deal with this the counter needs
to be updated while holding the wait queue mutex and waiters need to be
woken up unconditionally when the wait is removed from the queue to ensure
we eventually free the wait.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: check kernel communication pipe is valid for write
Ian Kent [Thu, 24 Jul 2008 04:30:20 +0000 (21:30 -0700)]
autofs4: check kernel communication pipe is valid for write

It is possible for an autofs mount to become catatonic (and for the daemon
communication pipe to become NULL) after a wait has been initiallized but
before the request has been sent to the daemon.  We need to check for this
before sending the request packet.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: add missing kfree
Ian Kent [Thu, 24 Jul 2008 04:30:19 +0000 (21:30 -0700)]
autofs4: add missing kfree

It see that the patch tittled "autofs4 - fix pending mount race" is
missing a change that I had recently made.

It's missing a kfree for the case mutex_lock_interruptible() fails
to aquire the wait queue mutex.

Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix pending mount race
Ian Kent [Thu, 24 Jul 2008 04:30:19 +0000 (21:30 -0700)]
autofs4: fix pending mount race

Close a race between a pending mount that is about to finish and a new
lookup for the same directory.

Process P1 triggers a mount of directory foo.  It sets
DCACHE_AUTOFS_PENDING in the ->lookup routine, creates a waitq entry for
'foo', and calls out to the daemon to perform the mount.  The autofs
daemon will then create the directory 'foo', using a new dentry that will
be hashed in the dcache.

Before the mount completes, another process, P2, tries to walk into the
'foo' directory.  The vfs path walking code finds an entry for 'foo' and
calls the revalidate method.  Revalidate finds that the entry is not
PENDING (because PENDING was never set on the dentry created by the
mkdir), but it does find the directory is empty.  Revalidate calls
try_to_fill_dentry, which sets the PENDING flag and then calls into the
autofs4 wait code to trigger or wait for a mount of 'foo'.  The wait code
finds the entry for 'foo' and goes to sleep waiting for the completion of
the mount.

Yet another process, P3, tries to walk into the 'foo' directory.  This
process again finds a dentry in the dcache for 'foo', and calls into the
autofs revalidate code.

The revalidate code finds that the PENDING flag is set, and so calls
try_to_fill_dentry.

a) try_to_fill_dentry sets the PENDING flag redundantly for this
   dentry, then calls into the autofs4 wait code.

b) the autofs4 wait code takes the waitq mutex and searches for an
   entry for 'foo'

Between a and b, P1 is woken up because the mount completed.  P1 takes the
wait queue mutex, clears the PENDING flag from the dentry, and removes the
waitqueue entry for 'foo' from the list.

When it releases the waitq mutex, P3 (eventually) acquires it.  At this
time, it looks for an existing waitq for 'foo', finds none, and so creates
a new one and calls out to the daemon to mount the 'foo' directory.

Now, the reason that three processes are required to trigger this race is
that, because the PENDING flag is not set on the dentry created by mkdir,
the window for the race would be way to slim for it to ever occur.
Basically, between the testing of d_mountpoint(dentry) and the taking of
the waitq mutex, the mount would have to complete and the daemon would
have to be woken up, and that in turn would have to wake up P1.  This is
simply impossible.  Add the third process, though, and it becomes slightly
more likely.

Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix waitq locking
Ian Kent [Thu, 24 Jul 2008 04:30:17 +0000 (21:30 -0700)]
autofs4: fix waitq locking

The autofs4_catatonic_mode() function accesses the wait queue without any
locking but can be called at any time.  This could lead to a possible
double free of the name field of the wait and a double fput of the daemon
communication pipe or an fput of a NULL file pointer.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: use struct qstr in waitq.c
Jeff Moyer [Thu, 24 Jul 2008 04:30:16 +0000 (21:30 -0700)]
autofs4: use struct qstr in waitq.c

The autofs_wait_queue already contains all of the fields of the
struct qstr, so change it into a qstr.

This patch, from Jeff Moyer, has been modified a liitle by myself.

Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: use lookup intent flags to trigger mounts
Ian Kent [Thu, 24 Jul 2008 04:30:15 +0000 (21:30 -0700)]
autofs4: use lookup intent flags to trigger mounts

When an open(2) call is made on an autofs mount point directory that
already exists and the O_DIRECTORY flag is not used the needed mount
callback to the daemon is not done. This leads to the path walk
continuing resulting in a callback to the daemon with an incorrect
key. open(2) is called without O_DIRECTORY by the "find" utility but
this should be handled properly anyway.

This happens because autofs needs to use the lookup flags to decide
when to callback to the daemon to perform a mount to prevent mount
storms. For example, an autofs indirect mount map that has the "browse"
option will have the mount point directories are pre-created and the
stat(2) call made by a color ls against each directory will cause all
these directories to be mounted. It is unfortunate we need to resort
to this but mount maps can be quite large. Additionally, if a user
manually umounts an autofs indirect mount the directory isn't removed
which also leads to this situation.

To resolve this autofs needs to use the lookup intent flags to enable
it to make this decision. This patch adds this check and triggers a
call back if any of the lookup intent flags are set as all these calls
warrant a mount attempt be requested.

I know that external VFS code which uses the lookup flags is something
that the VFS would like to eliminate but I have no choice as I can't
see any other way to do this. A VFS dentry or inode operation callback
which returns the lookup "type" (requires a definition) would be
sufficient. But this change is needed now and I'm not aware of the form
that coming VFS changes will take so I'm not willing to propose anything
along these lines.

If anyone can provide an alternate method I would be happy to use it.

[akpm@linux-foundation.org: fix build for concurrent VFS changes]
Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: don't release directory mutex if called in oz_mode
Ian Kent [Thu, 24 Jul 2008 04:30:14 +0000 (21:30 -0700)]
autofs4: don't release directory mutex if called in oz_mode

Since we now delay hashing of dentrys until the ->mkdir() call, droping
and re-taking the directory mutex within the ->lookup() function when we
are being called by user space is not needed.  This can lead to a race
when other processes are attempting to access the same directory during
mount point directory creation.

In this case we need to hang onto the mutex to ensure we don't get user
processes trying to create a mount request for a newly created dentry
after the mount point entry has already been created.  This ensures that
when we need to check a dentry passed to autofs4_wait(), if it is hashed,
it is always the mount point dentry and not a new dentry created by
another lookup during directory creation.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: fix symlink name allocation
Ian Kent [Thu, 24 Jul 2008 04:30:13 +0000 (21:30 -0700)]
autofs4: fix symlink name allocation

The length of the symlink name has been moved but it needs to be set
before allocating space for it in the dentry info struct.  This corrects a
mistake in a recent patch.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: use look aside list for lookups
Ian Kent [Thu, 24 Jul 2008 04:30:12 +0000 (21:30 -0700)]
autofs4: use look aside list for lookups

A while ago a patch to resolve a deadlock during directory creation was
merged.  This delayed the hashing of lookup dentrys until the ->mkdir()
(or ->symlink()) operation completed to ensure we always went through
->lookup() instead of also having processes go through ->revalidate() so
our VFS locking remained consistent.

Now we are seeing a couple of side affects of that change in situations
with heavy mount activity.

Two cases have been identified:

1) When a mount request is triggered, due to the delayed hashing, the
   directory created by user space for the mount point doesn't have the
   DCACHE_AUTOFS_PENDING flag set.  In the case of an autofs multi-mount
   where a tree of mount point directories are created this can lead to
   the path walk continuing rather than the dentry being sent to the wait
   queue to wait for request completion.  This is because, if the pending
   flag isn't set, the criteria for deciding this is a mount in progress
   fails to hold, namely that the dentry is not a mount point and has no
   subdirectories.

2) A mount request dentry is initially created negative and unhashed.
   It remains this way until the ->mkdir() callback completes.  Since it
   is unhashed a fresh dentry is used when the user space mount request
   creates the mount point directory.  This leaves the original dentry
   negative and unhashed.  But revalidate has no way to tell the VFS that
   the dentry has changed, other than to force another ->lookup() by
   returning false, which is at best wastefull and at worst not possible.
   This results in an -ENOENT return from the original path walk when in
   fact the mount succeeded.

To resolve this we need to ensure that the same dentry is used in all
calls to ->lookup() during the course of a mount request.  This patch
achieves that by adding the initial dentry to a look aside list and
removes it at ->mkdir() or ->symlink() completion (or when the dentry is
released), since these are the only create operations autofs4 supports.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: revert - redo lookup in ttfd
Ian Kent [Thu, 24 Jul 2008 04:30:11 +0000 (21:30 -0700)]
autofs4: revert - redo lookup in ttfd

This patch series enables the use of a single dentry for lookups prior to
the dentry being hashed and so we no longer need to redo the lookup.  This
patch reverts the patch of commit
033790449ba9c4dcf8478a87693d33df625c23b5.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: don't make expiring dentry negative
Ian Kent [Thu, 24 Jul 2008 04:30:09 +0000 (21:30 -0700)]
autofs4: don't make expiring dentry negative

Correct the error of making a positive dentry negative after it has been
instantiated.

The code that makes this error attempts to re-use the dentry from a
concurrent expire and mount to resolve a race and the dentry used for the
lookup must be negative for mounts to trigger in the required cases.  The
fact is that the dentry doesn't need to be re-used because all that is
needed is to preserve the flag that indicates an expire is still
incomplete at the time of the mount request.

This change uses the the dentry to check the flag and wait for the expire
to complete then discards it instead of attempting to re-use it.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoeCryptfs: Make all persistent file opens delayed
Michael Halcrow [Thu, 24 Jul 2008 04:30:08 +0000 (21:30 -0700)]
eCryptfs: Make all persistent file opens delayed

There is no good reason to immediately open the lower file, and that can
cause problems with files that the user does not intend to immediately
open, such as device nodes.

This patch removes the persistent file open from the interpose step and
pushes that to the locations where eCryptfs really does need the lower
persistent file, such as just before reading or writing the metadata
stored in the lower file header.

Two functions are jumping to out_dput when they should just be jumping to
out on error paths.  This patch also fixes these.

Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoeCryptfs: do not try to open device files on mknod
Michael Halcrow [Thu, 24 Jul 2008 04:30:07 +0000 (21:30 -0700)]
eCryptfs: do not try to open device files on mknod

When creating device nodes, eCryptfs needs to delay actually opening the lower
persistent file until an application tries to open.  Device handles may not be
backed by anything when they first come into existence.

[Valdis.Kletnieks@vt.edu: build fix]
Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
Cc: <Valdis.Kletnieks@vt.edu}
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoecryptfs: inode.c mmap.c use unaligned byteorder helpers
Harvey Harrison [Thu, 24 Jul 2008 04:30:07 +0000 (21:30 -0700)]
ecryptfs: inode.c mmap.c use unaligned byteorder helpers

Fixe sparse warnings:
fs/ecryptfs/inode.c:368:15: warning: cast to restricted __be64
fs/ecryptfs/mmap.c:385:12: warning: incorrect type in assignment (different base types)
fs/ecryptfs/mmap.c:385:12:    expected unsigned long long [unsigned] [assigned] [usertype] file_size
fs/ecryptfs/mmap.c:385:12:    got restricted __be64 [usertype] <noident>
fs/ecryptfs/mmap.c:428:12: warning: incorrect type in assignment (different base types)
fs/ecryptfs/mmap.c:428:12:    expected unsigned long long [unsigned] [assigned] [usertype] file_size
fs/ecryptfs/mmap.c:428:12:    got restricted __be64 [usertype] <noident>

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoecryptfs: crypto.c use unaligned byteorder helpers
Harvey Harrison [Thu, 24 Jul 2008 04:30:06 +0000 (21:30 -0700)]
ecryptfs: crypto.c use unaligned byteorder helpers

Fixes the following sparse warnings:
fs/ecryptfs/crypto.c:1036:8: warning: cast to restricted __be32
fs/ecryptfs/crypto.c:1038:8: warning: cast to restricted __be32
fs/ecryptfs/crypto.c:1077:10: warning: cast to restricted __be32
fs/ecryptfs/crypto.c:1103:6: warning: incorrect type in assignment (different base types)
fs/ecryptfs/crypto.c:1105:6: warning: incorrect type in assignment (different base types)
fs/ecryptfs/crypto.c:1124:8: warning: incorrect type in assignment (different base types)
fs/ecryptfs/crypto.c:1241:21: warning: incorrect type in assignment (different base types)
fs/ecryptfs/crypto.c:1244:30: warning: incorrect type in assignment (different base types)
fs/ecryptfs/crypto.c:1414:23: warning: cast to restricted __be32
fs/ecryptfs/crypto.c:1417:32: warning: cast to restricted __be16

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoecryptfs: string copy cleanup
Miklos Szeredi [Thu, 24 Jul 2008 04:30:05 +0000 (21:30 -0700)]
ecryptfs: string copy cleanup

Clean up overcomplicated string copy, which also gets rid of this
bogus warning:

fs/ecryptfs/main.c: In function 'ecryptfs_parse_options':
include/asm/arch/string_32.h:75: warning: array subscript is above array bounds

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoecryptfs: propagate key errors up at mount time
Eric Sandeen [Thu, 24 Jul 2008 04:30:04 +0000 (21:30 -0700)]
ecryptfs: propagate key errors up at mount time

Mounting with invalid key signatures should probably fail, if they were
specifically requested but not available.

Also fix case checks in process_request_key_err() for the right sign of
the errnos, as spotted by Jan Tluka.

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Jan Tluka <jtluka@redhat.com>
Acked-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoecryptfs: discard ecryptfsd registration messages in miscdev
Tyler Hicks [Thu, 24 Jul 2008 04:30:04 +0000 (21:30 -0700)]
ecryptfs: discard ecryptfsd registration messages in miscdev

The userspace eCryptfs daemon sends HELO and QUIT messages to the kernel
for per-user daemon (un)registration.  These messages are required when
netlink is used as the transport, but (un)registration is handled by
opening and closing the device file when miscdev is the transport.  These
messages should be discarded in the miscdev transport so that a daemon
isn't registered twice.

Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoeCryptfs: Privileged kthread for lower file opens
Michael Halcrow [Thu, 24 Jul 2008 04:30:02 +0000 (21:30 -0700)]
eCryptfs: Privileged kthread for lower file opens

eCryptfs would really like to have read-write access to all files in the
lower filesystem.  Right now, the persistent lower file may be opened
read-only if the attempt to open it read-write fails.  One way to keep
from having to do that is to have a privileged kthread that can open the
lower persistent file on behalf of the user opening the eCryptfs file;
this patch implements this functionality.

This patch will properly allow a less-privileged user to open the eCryptfs
file, followed by a more-privileged user opening the eCryptfs file, with
the first user only being able to read and the second user being able to
both read and write.  eCryptfs currently does this wrong; it will wind up
calling vfs_write() on a file that was opened read-only.  This is fixed in
this patch.

Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com>
Cc: Dave Kleikamp <shaggy@austin.ibm.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoI2O: handle sysfs_create_link() failures
Wang Chen [Thu, 24 Jul 2008 04:30:01 +0000 (21:30 -0700)]
I2O: handle sysfs_create_link() failures

Compile warning:
ignoring return value of `sysfs_create_link', declared with attribute warn_unused_result.

If sysfs_create_link failed, take care of the return value and do some
error handle after the failure.

Since sysfs_remove_link() will check whether a link exists, when removing the
link in error path, we don't need to care whether a link was created.

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
Cc: Markus Lidel <Markus.Lidel@shadowconnect.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>