Things really _have_ calmed down, and hopefully we've also resolved a lot

of the regressions in -rc3.
That said, the diffstat shows two things standing out:
 - continued rename of arch include files (if you don't enable rename

   detection and filter it out). ARM, AVR32 and m68knommu this time.
 - the inclusion of the ath9k wireless driver (and to a smaller degree,

   watchdog driver updates). The dirstat says it all:
	   2.0% arch/avr32/configs/

	   2.1% arch/avr32/

	   2.2% arch/

	   2.4% drivers/char/

	   3.4% drivers/net/wan/

	  58.8% drivers/net/wireless/ath9k/

	  58.9% drivers/net/wireless/

	  65.3% drivers/net/

	   7.9% drivers/watchdog/

	  84.1% drivers/
   ie that ath9k driver is the bulk of the whole patch.
But all the small changes are where the regression fixes are, and other

random improvements. And they're all over. The ShortLog (appended)

probably gives a taste of it.
		Linus
---

Adam Langley (1):

      tcp: (whitespace only) fix confusing indentation
Adrian Bunk (18):

      mm: unexport ksize

      V4L/DVB (8562): DVB_DRX397XD: remove FW_LOADER select

      V4L/DVB (8563): fix drivers/media/video/arv.c compilation

      xen-blkfront.c: make blkif_ioctl() static

      remove bogus CONFIG_GFAR_NAPI's

      ath9k: work around gcc ICEs

      [ARM] 5177/1: arm/mach-sa1100/Makefile: remove CONFIG_SA1100_USB

      [NFSD] uninline nfsd4_op_name()

      uninline atl1e_setup_mac_ctrl()

      [WATCHDOG] MAINTAINERS: remove ZF MACHZ WATCHDOG entry

      [WATCHDOG] fix watchdog/txx9wdt.c compilation

      [WATCHDOG] fix watchdog/shwdt.c compilation

      [WATCHDOG] fix watchdog/at91rm9200_wdt.c compilation

      [WATCHDOG] fix watchdog/wdt285.c compilation

      [WATCHDOG] fix watchdog/ixp4xx_wdt.c compilation

      make struct scsi_dh_devlist's static

      [IA64] use bcd2bin/bin2bcd

      feature-removal-schedule.txt: remove the NCR53C9x entry
Alan Cox (59):

      [WATCHDOG 01/57] Clean acquirewdt and check for BKL dependancies

      [W...
[ message continues ]

On Wed, Aug 13, 2008 at 4:33 AM, Linus Torvalds
I found a new one in -rc3 vs. -rc2.

Only one of my cpus seems to powered down by cpufreq, because the

topology gets filled out wrong:
correct with -rc2:

#grep . /sys/devices/system/cpu/cpu*/cpufreq/affected_cpus

/sys/devices/system/cpu/cpu0/cpufreq/affected_cpus:0 1

/sys/devices/system/cpu/cpu1/cpufreq/affected_cpus:0 1

/sys/devices/system/cpu/cpu2/cpufreq/affected_cpus:2 3

/sys/devices/system/cpu/cpu3/cpufreq/affected_cpus:2 3

(The system has 2 dual core Opterons)
wrong with -rc3:

#grep . /sys/devices/system/cpu/cpu*/cpufreq/affected_cpus

/sys/devices/system/cpu/cpu0/cpufreq/affected_cpus:0 1

/sys/devices/system/cpu/cpu1/cpufreq/affected_cpus:0 1

/sys/devices/system/cpu/cpu2/cpufreq/affected_cpus:0 1

/sys/devices/system/cpu/cpu3/cpufreq/affected_cpus:0 1
HTH
Torsten

--

Well, that patch is likely going to get reverted because

it's causing a crash on Rafael's test box.
But before that happens, could you send me the disassembled

dump of your ACPI objects?  Iti's possible the code is wrong,

but it's also possible that your BIOS is passing a broken

_PSD.
-Mark Langsdorf

Operating System Research Center

AMD
--

On Wed, Aug 13, 2008 at 10:23 PM, Langsdorf, Mark
On the first look the ACPI seems correct.

The first two P00x objects have this PSD:

Package (0x05)

            {

                0x05,

                0x00,

                0x00000000,

                0x000000FD,

                0x00000002

            }

The other two have 0x00000001 at the third position.
But as I don't really have a clue about ACPI, I'm attaching the

complete SSDT so you can take look.

Just ask, if you need the other tables too.
Torsten

lockdep_count_forward_deps is only defined if CONFIG_PROVE_LOCKING but

used unconditionally in lockdep_proc.c.
kernel/built-in.o: In function `.lockdep_stats_show':

lockdep_proc.c:(.text+0x3fea8): undefined reference to `.lockdep_count_forward_deps'

kernel/built-in.o: In function `.l_show':

lockdep_proc.c:(.text+0x401f0): undefined reference to `.lockdep_count_forward_deps'

lockdep_proc.c:(.text+0x40210): undefined reference to `.lockdep_count_backward_deps'
Andreas.
--

Andreas Schwab, SuSE Labs, schwab@suse.de

SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany

PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5

"And now for something completely different."

--

I have so far been unable to build -rc3 because of this same error.

Just tried again with current git HEAD. Failing config attached.
Cheers,

FJP

Ok, I think this one is fixed by df60a8441866153d691ae69b77934904c2de5e0d,

just pushed out.
		Linus

--

cu

Adrian
-- 
       "Is there not promise of rain?" Ling Tan asked suddenly out

        of the darkness. There had been need of rain for many days.

       "Only a promise," Lao Er said.

                                       Pearl S. Buck - Dragon Seed
--

On Tue, 12 Aug 2008 19:33:16 -0700 (PDT)
Must admit it still seemed pretty broken to me. The firmware changes

still seem to break if you cross build and then nfs make modules_install

and a readonly mount "make modules_install" now fails. UDF mount appears

to blow up on the spot.
The tty resize race turns out to be fun. The reason people are now seeing

it bisects down to a scheduler change showing up the lack of locking

going back forever.
This is a bit of a big patch for -rc3 time so I don't know what you feel

about it. Basically we turn it inside out - we make resize a tty method

so that the driver takes its own locks first as you need to in this

situation then can call back into the default helper.
(If it looks ok let me know .. it'll also apply for testing versus -rc but

 wants a final polish in that form)
--
tty: remove resize window special case
From: Alan Cox <alan@redhat.com>
This moves it to being a tty operation. That removes special cases and now

also means that resize can be picked up by um and other non vt consoles

which may have a resize operation.
Signed-off-by: Alan Cox <alan@redhat.com>

---
 drivers/char/tty_io.c      |   76 +++++++++++++++++++++--------------------

 drivers/char/vt.c          |   82 +++++++++++++++++++++++++++++++++-----------

 drivers/char/vt_ioctl.c    |    7 +++-

 include/linux/tty.h        |    2 +

 include/linux/tty_driver.h |   14 ++++++++

 include/linux/vt_kern.h    |    1 -

 6 files changed, 124 insertions(+), 58 deletions(-)
diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c

index 0e6866f..ddf5562 100644

--- a/drivers/char/tty_io.c

+++ b/drivers/char/tty_io.c

@@ -2496,45 +2496,25 @@ static int tiocgwinsz(struct tty_struct *tty, struct winsize __user *arg)

 }
 /**

- *	tiocswinsz		-	implement window size set ioctl

- *	@tty; tty

- *	@arg: user buffer for result

+ *	tty_do_resize		-	resize event

+ *	@tty: tty being resized

+ *	@real_tty: real tty (if using a pty/tty pair)

+ *	@rows: rows...
[ message continues ]

Looks ok. It's biggish, but looks pretty straightforward, and a fair chunk

of it seems to be the comment changes/additions. So send some final

version over when the polish is done..
		Linus

--

On Wed, 13 Aug 2008 15:24:15 -0700 (PDT)
tty: remove resize window special case
From: Alan Cox <alan@redhat.com>
This moves it to being a tty operation. That removes special cases and now

also means that resize can be picked up by um and other non vt consoles

which may have a resize operation.
Signed-off-by: Alan Cox <alan@redhat.com>

---
 drivers/char/tty_io.c      |   72 +++++++++++++++++++++------------------

 drivers/char/vt.c          |   82 +++++++++++++++++++++++++++++++++-----------

 drivers/char/vt_ioctl.c    |    4 ++

 include/linux/tty.h        |    2 +

 include/linux/tty_driver.h |   14 ++++++++

 include/linux/vt_kern.h    |    1 -

 6 files changed, 121 insertions(+), 54 deletions(-)
diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c

index b499a38..ddf5562 100644

--- a/drivers/char/tty_io.c

+++ b/drivers/char/tty_io.c

@@ -2496,45 +2496,25 @@ static int tiocgwinsz(struct tty_struct *tty, struct winsize __user *arg)

 }
 /**

- *	tiocswinsz		-	implement window size set ioctl

- *	@tty; tty

- *	@arg: user buffer for result

+ *	tty_do_resize		-	resize event

+ *	@tty: tty being resized

+ *	@real_tty: real tty (if using a pty/tty pair)

+ *	@rows: rows (character)

+ *	@cols: cols (character)

  *

- *	Copies the user idea of the window size to the kernel. Traditionally

- *	this is just advisory information but for the Linux console it

- *	actually has driver level meaning and triggers a VC resize.

- *

- *	Locking:

- *		Called function use the console_sem is used to ensure we do

- *	not try and resize the console twice at once.

- *		The tty->termios_mutex is used to ensure we don't double

- *	resize and get confused. Lock order - tty->termios_mutex before

- *	console sem

+ *	Update the termios variables and send the neccessary signals to

+ *	peform a terminal resize correctly

  */
-static int tiocswinsz(struct tty_struct *tty, struct tty_struct *real_tty,

-	struct winsize __user *arg)

+int tty_do_resize(struct tty_struct *t...
[ message continues ]

FWIW guys, I added this patch to my first build of -rc3, and so far, nothing

has raised its hand to object.  The logs are squeaky clean over the last few

hours.
--

Cheers, Gene

"There are four boxes to be used in defense of liberty:

 soap, ballot, jury, and ammo. Please use in that order."

-Ed Howdershelt (Author)

"I will make no bargains with terrorist hardware."

-- Peter da Silva

--

[mostly generic hand-waving]
LTP overnight run resulted in a box which

a) pings

b) reSysRq+Bbootable
And that's all.
SysRq+<the rest> doesn't work (doesn't print anything on console and on

netconsole) nor does ssh to to box. Pressing keys on keyboard doesn't wakeup

console.
Printout stopped at waitpid06 test. This is second time I see this.

First time it was more or less reproducible, there were waitpid06

processes eating all CPU.
--