From: Con Kolivas <kernel@kolivas.org>

A non-critical corner case has come up for interactivity that I believe needs 
to be addressed. It is only extensive testing and examination that revealed
this, as this interactivity work is in maintenance mode.

Description:

  It is possible for a highly interactive task (like X) to cause large
  latencies in tasks that are less 'niced' (eg negative nice number
  compared to X which should normally run at nice 0) if they are fully
  cpu bound.  This occurs due to expiration of the cpu bound tasks.

  This patch addresses this by not reinserting interactive tasks into
  the active array if there is a better static priority task running but
  has been placed on the expired array.  This causes a substantial
  reduction in the maximum scheduling latency a task with a less nice
  value can have. 

  This also has the positive side effect of maintaining better cpu%
  proportions for tasks of different nice levels.

Testers will only be able to discern a difference with highly cpu bound tasks 
of normal scheduling policy at different nice levels. Test cases are doing 
something cpu intensive relatively -niced in the presence of an interactive
load (eg capturing and encoding video at nice -10 while using X nice 0, or 
something nice 0 vs nice +10) and so on. Because of the crossover of 10 
'nice' levels of dynamic priorities between interactive and cpu bound tasks
this patch will have a more noticable effect as the nice difference is
greater, especially 11 or more.

From: Manfred Spraul <manfred@colorfullife.com>

fasync_helper and kill_fasync are helpers for managing F_SETFL fcntl 
calls that change FASYNC and sending the necessary signals. The locking 
uses one global rwlock that's acquired for read in all kill_fasync
calls, and that causes cache line trashing. This is not necessary: if
the fasync list is empty, then there is no need to acquire the rwlock.
Tests with reaim on a 4-way pIII on STP showed an 80% reduction of the
time within kill_fasync.

From: Davide Libenzi <davidel@xmailserver.org>

CONFIG_EPOLL=n space reduction in struct file.

From: Bram Stolk <bram@sara.nl>

All modes that exceed the native resolution of a flatpanel are discarded.
However, a CRT has native resolution set to 0, and therefore, tridentfb.c
cannot be used with a CRT.

From: Andres Salomon <dilinger@voxel.net>

Remove calls to CardServices(); final.

This removes the CardServices() calls the rest of the tree, as well as
CardServices itself from cs.c and cs.h.  My previous patches, along w/
Arjan's patch for ide-cs.c and Russell's patch for serial_cs.c should
remove all traces of it.  I've attached a gzipped tarball to hopefully
make things easier.

Files touched:
                                                                           =
 201-cs_remove.patch:+++ mod/sound/pcmcia/vx/vx_entry.c
202-cs_remove.patch:+++ mod/drivers/bluetooth/bluecard_cs.c
202-cs_remove.patch:+++ mod/drivers/bluetooth/bt3c_cs.c
202-cs_remove.patch:+++ mod/drivers/bluetooth/btuart_cs.c
202-cs_remove.patch:+++ mod/drivers/bluetooth/dtl1_cs.c
203-cs_remove.patch:+++ mod/drivers/isdn/hardware/avm/avm_cs.c
203-cs_remove.patch:+++ mod/drivers/isdn/hisax/avma1_cs.c
203-cs_remove.patch:+++ mod/drivers/isdn/hisax/elsa_cs.c
203-cs_remove.patch:+++ mod/drivers/isdn/hisax/sedlbauer_cs.c
204-cs_remove.patch:+++ mod/drivers/parport/parport_cs.c
205-cs_remove.patch:+++ mod/drivers/telephony/ixj_pcmcia.c
206-cs_remove.patch:+++ mod/drivers/mtd/maps/pcmciamtd.c
207-cs_remove.patch:+++ mod/drivers/scsi/pcmcia/aha152x_stub.c
207-cs_remove.patch:+++ mod/drivers/scsi/pcmcia/fdomain_stub.c
207-cs_remove.patch:+++ mod/drivers/scsi/pcmcia/nsp_cs.c
207-cs_remove.patch:+++ mod/drivers/scsi/pcmcia/nsp_cs.h
207-cs_remove.patch:+++ mod/drivers/scsi/pcmcia/qlogic_stub.c
208-cs_remove.patch:+++ mod/drivers/char/pcmcia/synclink_cs.c
209-cs_remove.patch:+++ mod/drivers/pcmcia/cs.c
209-cs_remove.patch:+++ mod/include/pcmcia/cs.h

From: Andres Salomon <dilinger@voxel.net>

Anyways, this removes the last of the CS calls.

From: Russell King <rmk@arm.linux.org.uk>

From: Russell King <rmk+lkml@arm.linux.org.uk>

Ok, this is the last patch for removal of CardServices()

From: Andres Salomon <dilinger@voxel.net>

This wraps up drivers/net/wireless.  Still remaining:

- sound/pcmcia/vx
- drivers/{bluetooth,isdn,parport,telephony}
- drivers/mtd/maps
- drivers/scsi/pcmcia
- drivers/char/pcmcia/synclink_cs.c

From: Arjan van de Ven <arjanv@redhat.com>

From: Andres Salomon <dilinger@voxel.net>

Replace the various CardServices() calls w/ their pcmcia_* function.  The
pcmcia functions fit better with kernel conventions, and don't have the
nasty var args stuff.  These patches also fix a few places where the args
supplied to CardServices were either not supplied, or dealt with in
non-obvious ways.

From: Rik van Riel <riel@surriel.com>

In 2.6.0 both __alloc_pages() and the corresponding wakeup_kswapd()s walk
all zones in the zone list, possibly spanning multiple nodes in a low numa
factor system like AMD64.

Also, if lower_zone_protection is set in /proc, then it may be possible
that kswapd never cleans out data in zones further down the zonelist and
try_to_free_pages needs to do that.

However, in 2.6.0 try_to_free_pages() only frees pages in the pgdat the
first zone in the zonelist belongs to.

This is probably the wrong behaviour, since both the page allocator and the
kswapd wakeup free things from all zones on the zonelist.  The following
patch makes try_to_free_pages() consistent with the allocator, by passing
the zonelist as an argument and freeing pages from all zones in the list.

I do not have any numa systems myself, so I have only tested it on my own
little smp box.  Testing on NUMA systems may be useful, though the patch
really only should have an impact in those rare cases where kswapd can't
keep up with allocations...

As a side effect, the patch shrinks the kernel by 2 lines and replaces some
subtle magic by a simpler array walk.

From: Paul Mackerras <paulus@samba.org>

Since there are serial drivers (particularly the USB serial driver) that
call the line discipline receive_buf and write_wakeup routines at hard
interrupt level, I have changed the ppp_async code to cope with that.  It
now uses a tasklet so that it calls the generic PPP code at soft interrupt
level even if its receive_buf and write_wakeup entries are called at hard
interrupt level.

This patch has been lightly tested here with a keyspan USB serial adaptor
and also with the built-in modem on my tibook, which uses the pmac_zilog
driver (which hooks into the drivers/serial infrastructure).

From: Jens Axboe <axboe@suse.de>

Update the ATAPI MO support code to reflect the reorganisations and cleanups
which the Mt Ranier support patch added.

DESC
cdrom_open fix
EDESC
From: Jens Axboe <axboe@suse.de>

From: Jens Axboe <axboe@suse.de>

Following patch adds mt rainier support to the cdrom uniform layer (it
works with atapi and scsi/usb).

From: Pascal Schmidt <der.eremit@email.de>

The below patch is needed to support ATAPI MO drives in 2.6. ide-scsi
doesn't work any more for this, so ide-cd has to take over the job of
running the MO drive. Without this, there is no way to write to an
ATAPI MO drive.

This patch has been discussed with Linus and Jens already around test9
time and it was agreed this is the right way to go about it. I have
rediffed it against 2.6.0. Compiles, runs, works just fine for me.

From: Mike Christie <michaelc@cs.wisc.edu>

DM, MD, rd and loop use blk_alloc_queue and blk_queue_make_request to
initialize their queue, because they only use the make_request_fn.  The
attached patch prevents the queue from being registered if only
blk_alloc_queue was called.

From: Rusty Russell <rusty@rustcorp.com.au>

It's complaining about:

#define per_cpu(var, cpu)			((void)cpu, per_cpu__##var)

There are several ways of fixing this, but the simplest is:

#define per_cpu(var, cpu)			(*((void)cpu, &per_cpu__##var))

From: Tom Rini <trini@kernel.crashing.org>

PPC32: Minor cleanups to IBM4xx and MPC82xx headers.

- Make sure that if <asm/ibm4xx.h> is included on !40x && !440, there is no real effect.
- Delete arch/ppc/platforms/mpc82xx.h
- Make sure that if CONFIG_8260 isn't set, there is no effect in <asm/mpc8260.h>.
- Add a __ASSEMBLY__ test around the extern for __res in <asm/mpc8260.h>.

From: Tom Rini <trini@kernel.crashing.org>

PPC32: Select arch/ppc/kernel/head.S on CONFIG_PPC_STD_MMU.

- Don't pick a head*.S by default, instead select head.S on
  CONFIG_PPC_STD_MMU.  This is more consistent with how we
  case things in this file.

From: Tom Rini <trini@kernel.crashing.org>

PPC32: Change all EXPORT_SYMBOL_NOVERS to EXPORT_SYMBOL in ppc_ksyms.c

- Change all EXPORT_SYMBOL_NOVERS to EXPORT_SYMBOL in ppc_ksyms.c

From: Tom Rini <trini@kernel.crashing.org>

We must export the consistent_sync_page symbol.  It is used by inline
functions which implement the PCI DMA API.

From: Nick Piggin <piggin@cyberone.com.au>

The big regression from deadline is tiobench random reads with TCQ disks,
however it is present in -linus as well, and would have been since day 1 of
AS, but nobody has complained too loudly.

http://developer.osdl.org/judith/tiobench/4CPU/rr.html

This problem is probably a distilation of what causes lower database
throughput, because I have only ever seen it with TCQ drives, and pgbench
and OraSim are actually getting higher throughput here with a non TCQ
drive.

That is not to say that TCQ is useless, it obviously can provide a very
real and significant boost.  What I might do in the (near) future is get AS
to detect TCQ and turn itself off indefinitely unless/until the a sysfs
flag is set, and default that flag to off.

This patch changes the AS tunables a bit to be more on par with deadline.  It
lowers the threshold for random reading processes to be considered unsuitable
for anticipation, and it slightly rearranges and comments the "cooperative
seek distance" logic.

With this patch, AS is now very competitive with deadline on the single IDE
and SCSI (non TCQ) disks here.  In fact, I don't have any regressions
anywhere.  Even when TCQ is on, although throughput can be lower, AS still
has benefits because of its much better read vs write latency and general
tendancy to keep number of outstanding tags smaller.

From: Nick Piggin <piggin@cyberone.com.au>

Sometimes a processes thinktime shouldn't be measured on how soon it
submits its next request, but how soon any close request is submitted.

Some processes, such as those in make -j, find | xargs blah, etc. Should be
waited upon even if they never submit another request, because they work with
cooperating or child processes. This helps to take that into account.

From: Nick Piggin <piggin@cyberone.com.au>

This one gathers better statistics about the new process problem.  It
improves estimation for initial process IO.  That is, better calculations
for whether it will be worth waiting after a process submits its first
read.

This is done with a per queue average thinktime and seek time for a second
read submitted from a process.

When combined with 3/3, numbers are around the same as mm1 for most long
lived tasks, but much better for things like the top 4 benchmarks.

Probably wants rwhron and the OSDL database guys to give it some testing.

test                                    2.6.0-test9-mm1  2.6.0-test9-mm1-np
Cat kernel source during seq read       0:26.89          0:24.75
Cat kernel source during seq write      9:17.80          0:23.48
ls -lr kernel source during seq read    0:11.03          0:14.68
ls -lr kernel source during seq write   0:49.95          0:08.06

contest no_load                         143s 0 loads     144s 0 loads
contest io_load                         193s 40.2 loads  193s 40.1 loads
contest read_load                       186s 11.6 loads  190s 10.5 loads
contest list_load                       201s 5.0 loads   200s 5.0 loads

pgbench 1 client                        31.3TPS          31.5TPS
pgbench 4 clients                       37.7TPS          37.5TPS
pgbench 16 clients                      42.1TPS          48.1TPS

From: Nick Piggin <piggin@cyberone.com.au>

This fixes all known bugs with as in mm2.  That constitutes small fixes for
2 WARNs getting triggered.  It looks like Prakash's lost interrupt problem
was due to as spewing bazillions of warnings, and he must have had his
kernel.printk configured not to show them or something.  I have to just get
final confirmation from him that this final cut of the patch fixes his
problem too.

From: Nick Piggin <piggin@cyberone.com.au>

From: Nick Piggin <piggin@cyberone.com.au>

I have tested this on my disks and cdroms, but they don't represent
what all drivers might do. I have asked Jarkko Lehti with his dvd writing
problem to try it...

From: Nick Piggin <piggin@cyberone.com.au>

Randy has just reported that this fixes up his regression. Its now as good
as test5 in his tests with this patch. He is also seeing quite large
increases (above test5) when IO scheduler barriers are disabled (this patch
doesn't do that). Perhaps something is using them too liberally.

Anyway, this reverts AS back to defaulting to not anticipate IO for a
program that submits its first request (this really hurts find | xargs grep
sort of things).  I am working on something to fix this up properly.  That
can go in after 2.6.0 anyway.

Use daemonize() rather than open-coding half of it.

(I don't know if this has been tested - it has been in -mm for three months).

From: Anton Blanchard <anton@samba.org>

vi fixup

From: Anton Blanchard <anton@samba.org>

Gone are the days of initialising stuff we touch in prom_init just
to keep it out of the BSS. There are a few things the hypervisor
writes to in the iseries case, hard code them into the data segment
and add a comment.

Remove the -fno-zero-initialized-in-bss hack, it was required when
gcc got smart and put zero initialised stuff into the BSS

From: Anton Blanchard <anton@samba.org>

Fixed rtas_extended_busy_delay_time() to calculate correct value, from John
Rose

From: Anton Blanchard <anton@samba.org>

Xmon changes to make breakpoints and single-stepping work on pSeries LPARs.
Also changed help text to reflect obsolete cmds.

From: Anton Blanchard <anton@samba.org>

Looks like glibc is using stat in some places, so we should modify
it to copy the nanosecond fields. Also speed up stat by only checking
the region once instead of each put_user call.

From: Anton Blanchard <anton@samba.org>

correct epoll syscall names

From: Anton Blanchard <anton@samba.org>

setup_cpu was being called for the boot cpu after all other cpus had been
spun up.  The init thread can sleep during secondary cpu spinup (eg migration
thread init) and sometimes the init thread would switch to another cpu at
this time.  The end result was setup_cpu was called twice on one cpu and
never on the boot cpu.

The original fix called setup_cpu on the boot cpu before all others but that
wont work for G5, so we now use cpu affinity calls to enforce it.

From: Anton Blanchard <anton@samba.org>

- Remove a bunch of unnecessary sections
- Always declare section labels inside the section (bug found on ppc32)
- Rearrange sections to waste less space

From: Anton Blanchard <anton@samba.org>

 - remove old checks target, no longer used
 - add -mtraceback=none, we dont use traceback tables any more
 - add check for -mcpu=power4, older toolchains dont support this option

From: Anton Blanchard <anton@samba.org>

Use an atomic_t instead of a volatile unsigned long, from Stephen Rothwell