linux-2.6
18 years ago[PATCH] block/stat.txt
Andy Isaacson [Sun, 8 Jan 2006 09:04:00 +0000 (01:04 -0800)] 
[PATCH] block/stat.txt

I couldn't find any docs explaining the contents of /sys/block/<dev>/stat,
so I wrote up the following.  I'm not completely sure it's accurate - Jens,
could you give a yea or nay on this?

In particular, the counts of read/write IOs and read/write sectors are
incremented in different places - it looks like they both increment as the
request is being finished, but I'm not completely sure of that.

Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] setpgid: should not accept ptraced childs
Oleg Nesterov [Sun, 8 Jan 2006 09:03:59 +0000 (01:03 -0800)] 
[PATCH] setpgid: should not accept ptraced childs

sys_setpgid() allows to change ->pgrp of ptraced childs.

'man setpgid' does not tell anything about that, so I consider
this behaviour is a bug.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Oren Laadan <orenl@cs.columbia.edu>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] setpgid: should work for sub-threads
Oren Laadan [Sun, 8 Jan 2006 09:03:58 +0000 (01:03 -0800)] 
[PATCH] setpgid: should work for sub-threads

setsid() does not work unless the calling process is a
thread_group_leader().

'man setpgid' does not tell anything about that, so I consider this
behaviour is a bug.

Signed-off-by: Oren Laadan <orenl@cs.columbia.edu>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] setpgid: should work for sub-threads
Oleg Nesterov [Sun, 8 Jan 2006 09:03:53 +0000 (01:03 -0800)] 
[PATCH] setpgid: should work for sub-threads

setpgid(0, pgid) or setpgid(forked_child_pid, pgid) does not work unless
the calling process is a thread_group_leader().

'man setpgid' does not tell anything about that, so I consider this
behaviour is a bug.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Oren Laadan <orenl@cs.columbia.edu>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fork: fix race in setting child's pgrp and tty
Oren Laadan [Sun, 8 Jan 2006 09:03:51 +0000 (01:03 -0800)] 
[PATCH] fork: fix race in setting child's pgrp and tty

In fork, child should recopy parent's pgrp/tty after it has tasklist_lock.
Otherwise following a setpgid() on the parent, *after* copy_signal(), the
child will own a stale pgrp (which may be reused); (eg.  if copy_mm()
sleeps a long while due to memory pressure).  Similar issue for the tty.

Signed-off-by: Oren Laadan <orenl@cs.columbia.edu>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cciss: adds MSI and MSI-X support
Mike Miller [Sun, 8 Jan 2006 09:03:50 +0000 (01:03 -0800)] 
[PATCH] cciss: adds MSI and MSI-X support

This creates a new function, cciss_interrupt_mode called from
cciss_pci_init.  This function determines what type of interrupt vector to
use, i.e., MSI, MSI-X, or IO-APIC.

One noticeable difference is changing the interrupt field of the controller
struct to an array of 4 unsigned ints.  The Smart Array HW is capable of
generating 4 distinct interrupts depending on the transport method in use
during operation.  These are:

#define DOORBELL_INT 0
Used to notify the contoller of configuration updates. We only use
this feature when in polling mode.

#define PERF_MODE_INT 0
Used when the controller is in Performant Mode.

#define SIMPLE_MODE_INT 2
Used when the controller is in Simple Mode (current Linux implementation).

#define MEMQ_INT_MODE 3
Not used.

When using IO-APIC interrupts these 4 lines are OR'ed together so when any
one fires an interrupt an is generated.  In MSI or MSI-X mode this hardware
OR'ing is ignored.  We must register for our interrupt depending on what
mode the controller is running.  For Linux we use SIMPLE_MODE_INT
exclusively at this time.  Please consider this for inclusion.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] tpmdd: remove global event log
Kylene Jo Hall [Sun, 8 Jan 2006 09:03:48 +0000 (01:03 -0800)] 
[PATCH] tpmdd: remove global event log

Remove global event log in the tpm bios event measurement log code that
would have caused problems when the code was run concurrently.  A log is
now allocated and attached to the seq file upon open and destroyed
appropriately.

Signed-off-by: Kylene Jo Hall <kjhall@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Don't attempt to power off if power off is not implemented
Eric W. Biederman [Sun, 8 Jan 2006 09:03:46 +0000 (01:03 -0800)] 
[PATCH] Don't attempt to power off if power off is not implemented

The problem.  It is expected that /sbin/halt -p works exactly like
/sbin/halt, when the kernel does not implement power off functionality.

The kernel can do a lot of work in the reboot notifiers and in
device_shutdown before we even get to machine_power_off.  Some of that
shutdown is not safe if you are leaving the power on, and it definitely
gets in the way of using sysrq or pressing ctrl-alt-del.  Since the
shutdown happens in generic code there is no way to fix this in
architecture specific code :(

Some machines are kernel oopsing today because of this.

The simple solution is to turn LINUX_REBOOT_CMD_POWER_OFF into
LINUX_REBOOT_CMD_HALT if power_off functionality is not implemented.

This has the unfortunate side effect of disabling the power off
functionality on architectures that leave pm_power_off to null and still
implement something in machine_power_off.  And it will break the build on
some architectures that don't have a pm_power_off variable at all.

On both counts I say tough.

For architectures like alpha that don't implement the pm_power_off variable
pm_power_off is declared in linux/pm.h and it is a generic part of our
power management code, and all architectures should implement it.

For architectures like parisc that have a default power off method in
machine_power_off if pm_power_off is not implemented or fails.  It is easy
enough to set the pm_power_off variable.  And nothing bad happens there,
the machines just stop powering off.

The current semantics are impossible without a flag at the top level so we
can avoid the problem code if a power off is not implemented.  pm_power_off
is as good a flag as any with the bonus that it works without modification
on at least x86, x86_64, powerpc, and ppc today.

Andrew can you pick this up and put this in the mm tree.  Kernels that
don't compile or don't power off seem saner than kernels that oops or
panic.  Until we get the arch specific patches for the problem
architectures this probably isn't smart to push into the stable kernel.
Unfortunately I don't have the time at the moment to walk through every
architecture and make them work.  And even if I did I couldn't test it :(

From: Hirokazu Takata <takata@linux-m32r.org>

    Add pm_power_off() for build fix of arch/m32r/kernel/process.c.

From: Miklos Szeredi <miklos@szeredi.hu>

    UML build fix

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Hayato Fujiwara <fujiwara@linux-m32r.org>
Signed-off-by: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fadvise: return ESPIPE on FIFO/pipe
Valentine Barshak [Sun, 8 Jan 2006 09:03:44 +0000 (01:03 -0800)] 
[PATCH] fadvise: return ESPIPE on FIFO/pipe

The patch makes posix_fadvise return ESPIPE on FIFO/pipe in order to be
fully POSIX-compliant.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] update to the initramfs docs
Rob Landley [Sun, 8 Jan 2006 09:03:43 +0000 (01:03 -0800)] 
[PATCH] update to the initramfs docs

Based on questions people have asked me.  Repeatedly.

Signed-off-by: Rob Landley <rob@landley.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Extend RCU torture module to test tickless idle CPU
Srivatsa Vaddagiri [Sun, 8 Jan 2006 09:03:42 +0000 (01:03 -0800)] 
[PATCH] Extend RCU torture module to test tickless idle CPU

This patch forces RCU torture threads off various CPUs in the system
allowing them to become idle and go tickless.  Meant to test support for
such tickless idle CPU in RCU.

Signed-off-by: Srivatsa Vaddagiri <vatsa@in.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Add tainting for proprietary helper modules
Dave Jones [Sun, 8 Jan 2006 09:03:41 +0000 (01:03 -0800)] 
[PATCH] Add tainting for proprietary helper modules

Kernels that have had Windows drivers loaded into them are undebuggable.
I've wasted a number of hours chasing bugs filed in Fedora bugzilla only to
find out much later that the user had used such 'helpers', and their
problems were unreproducable without them loaded.

Acked-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] remove unused blkp field in percpu_data
Eric Dumazet [Sun, 8 Jan 2006 09:03:40 +0000 (01:03 -0800)] 
[PATCH] remove unused blkp field in percpu_data

I found that blkp field was not used in kernel tree.

As most of the times NR_CPUS is a power of two and kmalloc() memory blocks
too, this extra field basically doubles the memory space allocated in
__alloc_percpu() to store the 'struct percpu_data'

(for example, if NR_CPUS=8 on i386, kmalloc(4*8+4) returns a 64 bytes block
instead of a 32 bytes block after this patch)

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fs: remove s_old_blocksize from struct super_block
Pekka Enberg [Sun, 8 Jan 2006 09:03:39 +0000 (01:03 -0800)] 
[PATCH] fs: remove s_old_blocksize from struct super_block

This patch inlines the single user of struct super_block field
s_old_blocksize and removes the field.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Documentation: Small applying-patches.txt update
Jesper Juhl [Sun, 8 Jan 2006 09:03:38 +0000 (01:03 -0800)] 
[PATCH] Documentation: Small applying-patches.txt update

Minor update to Documentation/applying-patches.txt

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] drivers/mfd: header included twice
Nicolas Kaiser [Sun, 8 Jan 2006 09:03:37 +0000 (01:03 -0800)] 
[PATCH] drivers/mfd: header included twice

linux/delay.h included twice

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix handling of ELF segments with zero filesize
David Gibson [Sun, 8 Jan 2006 09:03:35 +0000 (01:03 -0800)] 
[PATCH] Fix handling of ELF segments with zero filesize

mmap() returns -EINVAL if given a zero length, and thus elf_map() in
binfmt_elf.c does likewise if it attempts to map a (page-aligned) ELF
segment with zero filesize.  Such a situation never arises with the default
linker scripts, but there's nothing inherently wrong with zero-filesize
(but non-zero memsize) ELF segments.  Custom linker scripts can generate
them, and the kernel should be able to map them; this patch makes it so.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] drivers/connector/cn_proc.c typos
David S. Miller [Sun, 8 Jan 2006 09:03:34 +0000 (01:03 -0800)] 
[PATCH] drivers/connector/cn_proc.c typos

The parameter to put_cpu_var() is unreferenced by the implementation, and
the compiler doesn't try to comprehend comments, so this wouldn't cause any
problem, but if bugged me enough to post a fix :-)

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] shrink dentry struct
Eric Dumazet [Sun, 8 Jan 2006 09:03:32 +0000 (01:03 -0800)] 
[PATCH] shrink dentry struct

Some long time ago, dentry struct was carefully tuned so that on 32 bits
UP, sizeof(struct dentry) was exactly 128, ie a power of 2, and a multiple
of memory cache lines.

Then RCU was added and dentry struct enlarged by two pointers, with nice
results for SMP, but not so good on UP, because breaking the above tuning
(128 + 8 = 136 bytes)

This patch reverts this unwanted side effect, by using an union (d_u),
where d_rcu and d_child are placed so that these two fields can share their
memory needs.

At the time d_free() is called (and d_rcu is really used), d_child is known
to be empty and not touched by the dentry freeing.

Lockless lookups only access d_name, d_parent, d_lock, d_op, d_flags (so
the previous content of d_child is not needed if said dentry was unhashed
but still accessed by a CPU because of RCU constraints)

As dentry cache easily contains millions of entries, a size reduction is
worth the extra complexity of the ugly C union.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: Maneesh Soni <maneesh@in.ibm.com>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: Ian Kent <raven@themaw.net>
Cc: Paul Jackson <pj@sgi.com>
Cc: Al Viro <viro@ftp.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Cc: James Morris <jmorris@namei.org>
Cc: Stephen Smalley <sds@epoch.ncsc.mil>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] nfsroot: do not silently stop parsing on an unknown option
Jorn Dreyer [Sun, 8 Jan 2006 09:03:30 +0000 (01:03 -0800)] 
[PATCH] nfsroot: do not silently stop parsing on an unknown option

It would be helpful if the kernel did not silently stop parsing
nfs options, but instead warned about any he does not recognize. The
attached patch adds one printk to do just that.

It took me a couple of hours to find my configuration mistake.

Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] sigio: cleanup, don't take tasklist twice
Oleg Nesterov [Sun, 8 Jan 2006 09:03:29 +0000 (01:03 -0800)] 
[PATCH] sigio: cleanup, don't take tasklist twice

The only user of send_sigio_to_task() already holds tasklist_lock, so it is
better not to send the signal via send_group_sig_info() (which takes
tasklist recursively) but use group_send_sig_info().

The same change in send_sigurg()->send_sigurg_to_task().

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] remove unneeded sig->curr_target recalculation
Oleg Nesterov [Sun, 8 Jan 2006 09:03:28 +0000 (01:03 -0800)] 
[PATCH] remove unneeded sig->curr_target recalculation

This patch removes unneeded sig->curr_target recalculation under 'if
(atomic_dec_and_test(&sig->count))' in __exit_signal().

When sig->count == 0 the signal can't be sent to this task and
next_thread(tsk) == tsk anyway.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] MAINTAINERS: line duplication
Nicolas Kaiser [Sun, 8 Jan 2006 09:03:26 +0000 (01:03 -0800)] 
[PATCH] MAINTAINERS: line duplication

uniq -d MAINTAINERS

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ext3: use sbi instead of EXT3_SB() in resize code.
Glauber de Oliveira Costa [Sun, 8 Jan 2006 09:03:23 +0000 (01:03 -0800)] 
[PATCH] ext3: use sbi instead of EXT3_SB() in resize code.

There are places in the resize code in which EXT3_SB() macro is used after
an statement like sbi = EXT3_SB(sb) is done.  Inside the same function,
both sbi and EXT3_SB() are used to reference the super block Altough it is
not wrong, keeping it coherent increases legibility, IMHO.

Signed-off-by: Glauber de Oliveira Costa <glommer@br.ibm.com>
Cc: "Stephen C. Tweedie" <sct@redhat.com>
Cc: Andreas Dilger <adilger@clusterfs.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ext3: remove trailing newlines from ext3_warning() calls
Glauber de Oliveira Costa [Sun, 8 Jan 2006 09:03:22 +0000 (01:03 -0800)] 
[PATCH] ext3: remove trailing newlines from ext3_warning() calls

Remove the trailing newlines in calls to ext3_warning().  This function
already adds a trailing newline to the end of messages.

Signed-off-by: Glauber de Oliveira Costa <glommer@br.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] oprofile: Use vmalloc_node() in alloc_cpu_buffers()
Eric Dumazet [Sun, 8 Jan 2006 09:03:21 +0000 (01:03 -0800)] 
[PATCH] oprofile: Use vmalloc_node() in alloc_cpu_buffers()

Make oprofile alloc_cpu_buffers() function NUMA aware, allocating each CPU
local buffer in its memory node if possible.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Cc: Philippe Elie <phil.el@wanadoo.fr>
Cc: John Levon <levon@movementarian.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ext3: external journal device as a mount option
Johann Lombardi [Sun, 8 Jan 2006 09:03:20 +0000 (01:03 -0800)] 
[PATCH] ext3: external journal device as a mount option

The patch below adds a new mount option to allow the external journal
device to be specified.

The syntax is as follows:
# mount -t ext3 -o journal_dev=0x0820 ...
where 0x0820 means major=8 and minor=32.

Signed-off-by: Johann Lombardi <johann.lombardi@bull.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] shared mounts: cleanup
Miklos Szeredi [Sun, 8 Jan 2006 09:03:19 +0000 (01:03 -0800)] 
[PATCH] shared mounts: cleanup

Small cleanups in shared mounts code.

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: Ram Pai <linuxram@us.ibm.com>
Cc: <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pivot_root: add comment
Neil Brown [Sun, 8 Jan 2006 09:03:18 +0000 (01:03 -0800)] 
[PATCH] pivot_root: add comment

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Updated CPU hotplug documentation
Ashok Raj [Sun, 8 Jan 2006 09:03:17 +0000 (01:03 -0800)] 
[PATCH] Updated CPU hotplug documentation

Thanks to Nathan Lynch for the review and comments.  Thanks to Joel Schopp
for the pointer to add user space scipts.

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Nathan Lynch <nathanl@austin.ibm.com>
Signed-off-by: Joel Schopp <jschopp@austin.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] tpm: add bios measurement log
Kylene Jo Hall [Sun, 8 Jan 2006 09:03:15 +0000 (01:03 -0800)] 
[PATCH] tpm: add bios measurement log

According to the TCG specifications measurements or hashes of the BIOS code
and data are extended into TPM PCRS and a log is kept in an ACPI table of
these extensions for later validation if desired.  This patch exports the
values in the ACPI table through a security-fs seq_file.

Signed-off-by: Seiji Munetoh <munetoh@jp.ibm.com>
Signed-off-by: Stefan Berger <stefanb@us.ibm.com>
Signed-off-by: Reiner Sailer <sailer@us.ibm.com>
Signed-off-by: Kylene Hall <kjhall@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] little do_group_exit() cleanup
Oleg Nesterov [Sun, 8 Jan 2006 09:03:14 +0000 (01:03 -0800)] 
[PATCH] little do_group_exit() cleanup

zap_other_threads() sets SIGNAL_GROUP_EXIT at the very start,
do_group_exit() doesn't need to do it.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] do_coredump() should reset group_stop_count earlier
Oleg Nesterov [Sun, 8 Jan 2006 09:03:13 +0000 (01:03 -0800)] 
[PATCH] do_coredump() should reset group_stop_count earlier

__group_complete_signal() sets ->group_stop_count in sig_kernel_coredump()
path and marks the target thread as ->group_exit_task.  So any thread
except group_exit_task will go to handle_group_stop()->finish_stop().

However, when group_exit_task actually starts do_coredump(), it sets
SIGNAL_GROUP_EXIT, but does not reset ->group_stop_count while killing
other threads.  If we have not yet stopped threads in the same thread
group, they all will spin in kernel mode until group_exit_task sends them
SIGKILL, because ->group_stop_count > 0 means:

recalc_sigpending_tsk() never clears TIF_SIGPENDING

get_signal_to_deliver() goes to handle_group_stop()

handle_group_stop() returns when SIGNAL_GROUP_EXIT set

syscall_exit/resume_userspace notice TIF_SIGPENDING,
call get_signal_to_deliver() again.

So we are wasting cpu cycles, and if one of these threads is rt_task() this
may be a serious problem.

NOTE: do_coredump() holds ->mmap_sem, so not stopped threads can't escape
coredumping after clearing ->group_stop_count.

See also this thread: http://marc.theaimsgroup.com/?t=112739139900002

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] kill_proc_info_as_uid: don't use hardcoded constants
Oleg Nesterov [Sun, 8 Jan 2006 09:03:09 +0000 (01:03 -0800)] 
[PATCH] kill_proc_info_as_uid: don't use hardcoded constants

Use symbolic names instead of hardcoded constants.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Harald Welte <laforge@gnumonks.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fix possible PAGE_CACHE_SHIFT overflows
Andrew Morton [Sun, 8 Jan 2006 09:03:05 +0000 (01:03 -0800)] 
[PATCH] fix possible PAGE_CACHE_SHIFT overflows

We've had two instances recently of overflows when doing

64_bit_value = (32_bit_value << PAGE_CACHE_SHIFT)

I did a tree-wide grep of `<<.*PAGE_CACHE_SHIFT' and this is the result.

- afs_rxfs_fetch_descriptor.offset is of type off_t, which seems broken.

- jfs and jffs are limited to 4GB anyway.

- reiserfs map_block_for_writepage() takes an unsigned long for the block -
  it should take sector_t.  (It'll fail for huge filesystems with
  blocksize<PAGE_CACHE_SIZE)

- cramfs_read() needs to use sector_t (I think cramsfs is busted on large
  filesystems anyway)

- affs is limited in file size anyway.

- I generally didn't fix 32-bit overflows in directory operations.

- arm's __flush_dcache_page() is peculiar.  What if the page lies beyond 4G?

- gss_wrap_req_priv() needs checking (snd_buf->page_base)

Cc: Oleg Drokin <green@linuxhacker.ru>
Cc: David Howells <dhowells@redhat.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: <reiserfs-dev@namesys.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Anton Altaparmakov <aia21@cantab.net>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: <linux-fsdevel@vger.kernel.org>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Unchecked alloc_percpu() return in __create_workqueue()
Ben Collins [Sun, 8 Jan 2006 09:03:04 +0000 (01:03 -0800)] 
[PATCH] Unchecked alloc_percpu() return in __create_workqueue()

__create_workqueue() not checking return of alloc_percpu()

NULL dereference was possible.

Signed-off-by: Ben Collins <bcollins@ubuntu.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] nbd: remove duplicate assignment
taneli.vahakangas@netsonic.fi [Sun, 8 Jan 2006 09:03:02 +0000 (01:03 -0800)] 
[PATCH] nbd: remove duplicate assignment

      <stuartm@connecttech.com>

Sent by Paul Clements <paul.clements@steeleye.com>, who needs to read
Documentation/SubmittingPatches..

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Add block_device_operations.getgeo block device method
Christoph Hellwig [Sun, 8 Jan 2006 09:02:50 +0000 (01:02 -0800)] 
[PATCH] Add block_device_operations.getgeo block device method

HDIO_GETGEO is implemented in most block drivers, and all of them have to
duplicate the code to copy the structure to userspace, as well as getting
the start sector.  This patch moves that to common code [1] and adds a
->getgeo method to fill out the raw kernel hd_geometry structure.  For many
drivers this means ->ioctl can go away now.

[1] the s390 block drivers are odd in this respect.  xpram sets ->start
    to 4 always which seems more than odd, and the dasd driver shifts
    the start offset around, probably because of it's non-standard
    sector size.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Cc: Jens Axboe <axboe@suse.de>
Cc: <mike.miller@hp.com>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Paolo Giarrusso <blaisorblade@yahoo.it>
Cc: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Cc: Markus Lidel <Markus.Lidel@shadowconnect.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] docs: updated some code docs
Xose Vazquez Perez [Sun, 8 Jan 2006 09:02:49 +0000 (01:02 -0800)] 
[PATCH] docs: updated some code docs

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] sigaction should clear all signals on SIG_IGN, not just < 32
George Anzinger [Sun, 8 Jan 2006 09:02:48 +0000 (01:02 -0800)] 
[PATCH] sigaction should clear all signals on SIG_IGN, not just < 32

While rooting aroung in the signal code trying to understand how to fix the
SIG_IGN ploy (set sig handler to SIG_IGN and flood system with high speed
repeating timers) I came across what, I think, is a problem in sigaction()
in that when processing a SIG_IGN request it flushes signals from 1 to
SIGRTMIN and leaves the rest.  Attempt to fix this.

Signed-off-by: George Anzinger <george@mvista.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] keys: Permit running process to instantiate keys
David Howells [Sun, 8 Jan 2006 09:02:47 +0000 (01:02 -0800)] 
[PATCH] keys: Permit running process to instantiate keys

Make it possible for a running process (such as gssapid) to be able to
instantiate a key, as was requested by Trond Myklebust for NFS4.

The patch makes the following changes:

 (1) A new, optional key type method has been added. This permits a key type
     to intercept requests at the point /sbin/request-key is about to be
     spawned and do something else with them - passing them over the
     rpc_pipefs files or netlink sockets for instance.

     The uninstantiated key, the authorisation key and the intended operation
     name are passed to the method.

 (2) The callout_info is no longer passed as an argument to /sbin/request-key
     to prevent unauthorised viewing of this data using ps or by looking in
     /proc/pid/cmdline.

     This means that the old /sbin/request-key program will not work with the
     patched kernel as it will expect to see an extra argument that is no
     longer there.

     A revised keyutils package will be made available tomorrow.

 (3) The callout_info is now attached to the authorisation key. Reading this
     key will retrieve the information.

 (4) A new field has been added to the task_struct. This holds the
     authorisation key currently active for a thread. Searches now look here
     for the caller's set of keys rather than looking for an auth key in the
     lowest level of the session keyring.

     This permits a thread to be servicing multiple requests at once and to
     switch between them. Note that this is per-thread, not per-process, and
     so is usable in multithreaded programs.

     The setting of this field is inherited across fork and exec.

 (5) A new keyctl function (KEYCTL_ASSUME_AUTHORITY) has been added that
     permits a thread to assume the authority to deal with an uninstantiated
     key. Assumption is only permitted if the authorisation key associated
     with the uninstantiated key is somewhere in the thread's keyrings.

     This function can also clear the assumption.

 (6) A new magic key specifier has been added to refer to the currently
     assumed authorisation key (KEY_SPEC_REQKEY_AUTH_KEY).

 (7) Instantiation will only proceed if the appropriate authorisation key is
     assumed first. The assumed authorisation key is discarded if
     instantiation is successful.

 (8) key_validate() is moved from the file of request_key functions to the
     file of permissions functions.

 (9) The documentation is updated.

From: <Valdis.Kletnieks@vt.edu>

    Build fix.

Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Alexander Zangerl <az@bond.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] keys: Discard duplicate keys from a keyring on link
David Howells [Sun, 8 Jan 2006 09:02:45 +0000 (01:02 -0800)] 
[PATCH] keys: Discard duplicate keys from a keyring on link

Cause any links within a keyring to keys that match a key to be linked into
that keyring to be discarded as a link to the new key is added.  The match is
contingent on the type and description strings being the same.

This permits requests, adds and searches to displace negative, expired,
revoked and dead keys easily.  After some discussion it was concluded that
duplicate valid keys should probably be discarded also as they would otherwise
hide the new key.

Since request_key() is intended to be the primary method by which keys are
added to a keyring, duplicate valid keys wouldn't be an issue there as that
function would return an existing match in preference to creating a new key.

Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Alexander Zangerl <az@bond.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] keys: Permit key expiry time to be set
David Howells [Sun, 8 Jan 2006 09:02:43 +0000 (01:02 -0800)] 
[PATCH] keys: Permit key expiry time to be set

Add a new keyctl function that allows the expiry time to be set on a key or
removed from a key, provided the caller has attribute modification access.

Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Alexander Zangerl <az@bond.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] kmsg_write: don't return printk return value
Guillaume Chazarain [Sun, 8 Jan 2006 09:02:43 +0000 (01:02 -0800)] 
[PATCH] kmsg_write: don't return printk return value

kmsg_write returns with printk, so some programs may be confused by a
successful write() with a return value different than the buffer length.

# /bin/echo something > /dev/kmsg
/bin/echo: write error: Inappropriate ioctl for device

The drawbacks is that the printk return value can no more be quickly
checked from userspace.

Signed-off-by: Guillaume Chazarain <guichaz@yahoo.fr>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] printk return value: fix it
Guillaume Chazarain [Sun, 8 Jan 2006 09:02:41 +0000 (01:02 -0800)] 
[PATCH] printk return value: fix it

What's the true meaning of the printk return value?  Should it include the
priority prefix length of 3?  and what about the timing information?  In
both cases it was broken:

strace -e write echo 1 > /dev/kmsg
=> write(1, "1\n", 2)                      = 5
strace -e write echo "<1>1" > /dev/kmsg
=> write(1, "<1>1\n", 5)                   = 8

The returned length was "length of input string + 3", I made it "length
of string output to the log buffer".

Note that I couldn't find any printk caller in the kernel interested by its
return value besides kmsg_write.

Signed-off-by: Guillaume Chazarain <guichaz@yahoo.fr>
Acked-By: Tim Bird <tim.bird@am.sony.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix overflow tests for compat_sys_fcntl64 locking
NeilBrown [Sun, 8 Jan 2006 09:02:40 +0000 (01:02 -0800)] 
[PATCH] Fix overflow tests for compat_sys_fcntl64 locking

When making an fctl locking call through compat_sys_fcntl64 (i.e.  a 32bit
app on a 64bit kernel), the syscall can return a locking range that is in
conflict with the queried lock.

If some aspect of this range does not fit in the 32bit structure, something
needs to be done.

The current code is wrong in several respects:

- It returns data to userspace even if no conflict was found
   i.e. it should check l_type for F_UNLCK
- It returns -EOVERFLOW too agressively.   A lock range covering
  the last possible byte of the file (start = COMPAT_OFF_T_MAX,
  len = 1) should be possible, but is rejected with the current test.
- A extra-long 'len' should not be a problem.  If only that part
  of the conflicting lock that would be visible to the 32bit
  app needs to be reported to the 32bit app anyway.

This patch addresses those three issues and adds a comment to (hopefully)
record it for posterity.

Note: this patch mainly affects test-cases.  Real applications rarely is
ever see the problems.

This patch has been tested (LSB test suite), and works.

Signed-off-by: Neil Brown <neilb@suse.de>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Matthew Wilcox <willy@debian.org>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix some problems with truncate and mtime semantics.
NeilBrown [Sun, 8 Jan 2006 09:02:39 +0000 (01:02 -0800)] 
[PATCH] Fix some problems with truncate and mtime semantics.

SUS requires that when truncating a file to the size that it currently
is:
  truncate and ftruncate should NOT modify ctime or mtime
  O_TRUNC SHOULD modify ctime and mtime.

Currently mtime and ctime are always modified on most local
filesystems (side effect of ->truncate) or never modified (on NFS).

With this patch:
  ATTR_CTIME|ATTR_MTIME are sent with ATTR_SIZE precisely when
    an update of these times is required whether size changes or not
    (via a new argument to do_truncate).  This allows NFS to do
    the right thing for O_TRUNC.
  inode_setattr nolonger forces ATTR_MTIME|ATTR_CTIME when the ATTR_SIZE
    sets the size to it's current value.  This allows local filesystems
    to do the right thing for f?truncate.

Also, the logic in inode_setattr is changed a bit so there are two return
points.  One returns the error from vmtruncate if it failed, the other
returns 0 (there can be no other failure).

Finally, if vmtruncate succeeds, and ATTR_SIZE is the only change
requested, we now fall-through and mark_inode_dirty.  If a filesystem did
not have a ->truncate function, then vmtruncate will have changed i_size,
without marking the inode as 'dirty', and I think this is wrong.

Signed-off-by: Neil Brown <neilb@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Permit multiple inclusion of linux/pagevec.h
David Howells [Sun, 8 Jan 2006 09:02:37 +0000 (01:02 -0800)] 
[PATCH] Permit multiple inclusion of linux/pagevec.h

Make it possible to include linux/pagevec.h multiple times without
incurring errors due to duplicate definitions.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] vgacon: Workaround for resize bug in some chipsets
Antonino A. Daplas [Sun, 8 Jan 2006 09:02:36 +0000 (01:02 -0800)] 
[PATCH] vgacon: Workaround for resize bug in some chipsets

Reported from Redhat Bugzilla Bug 170450

"I updated to the development kernel and now during boot only the top of the
text is visable. For example the monitor screen the is the lines and I can
only see text in the asterisk area.

18 years ago[PATCH] vgacon: fix doublescan mode
Samuel Thibault [Sun, 8 Jan 2006 09:02:34 +0000 (01:02 -0800)] 
[PATCH] vgacon: fix doublescan mode

When doublescan mode is in use, scanlines must be doubled.

Thanks to Jason Dravet <dravet@hotmail.com> for reporting and testing.

Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] udf: remove bogus inode == NULL check in inode_bmap
Christoph Hellwig [Sun, 8 Jan 2006 09:02:33 +0000 (01:02 -0800)] 
[PATCH] udf: remove bogus inode == NULL check in inode_bmap

inode can never be NULL when calling this function.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] use ptrace_get_task_struct in various places
Christoph Hellwig [Sun, 8 Jan 2006 09:02:33 +0000 (01:02 -0800)] 
[PATCH] use ptrace_get_task_struct in various places

The ptrace_get_task_struct() helper that I added as part of the ptrace
consolidation is useful in variety of places that currently opencode it.
Switch them to the common helpers.

Add a ptrace_traceme() helper that needs to be explicitly called, and simplify
the ptrace_get_task_struct() interface.  We don't need the request argument
now, and we return the task_struct directly, using ERR_PTR() for error
returns.  It's a bit more code in the callers, but we have two sane routines
that do one thing well now.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: Documentation cleanup, remove obsolete info
Tom Zanussi [Sun, 8 Jan 2006 09:02:32 +0000 (01:02 -0800)] 
[PATCH] relayfs: Documentation cleanup, remove obsolete info

librelay and relay-app.h have been retired - update Documentation to reflect
that.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: cleanup, change relayfs_file_* to relay_file_*
Tom Zanussi [Sun, 8 Jan 2006 09:02:31 +0000 (01:02 -0800)] 
[PATCH] relayfs: cleanup, change relayfs_file_* to relay_file_*

This patch renames relayfs_file_operations to relay_file_operations, and the
file operations themselves from relayfs_XXX to relay_file_XXX, to make it more
clear that they refer to relay files.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: add Documentation on global relay buffers
Tom Zanussi [Sun, 8 Jan 2006 09:02:30 +0000 (01:02 -0800)] 
[PATCH] relayfs: add Documentation on global relay buffers

Documentation update for creating global buffers.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: add support for global relay buffers
Tom Zanussi [Sun, 8 Jan 2006 09:02:29 +0000 (01:02 -0800)] 
[PATCH] relayfs: add support for global relay buffers

This patch adds the optional is_global outparam to the create_buf_file()
callback.  This can be used by clients to create a single global relayfs
buffer instead of the default per-cpu buffers.  This was suggested as being
useful for certain debugging applications where it's more convenient to be
able to get all the data from a single channel without having to go to the
bother of dealing with per-cpu files.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: add Documentation on relay files in other filesystems
Tom Zanussi [Sun, 8 Jan 2006 09:02:29 +0000 (01:02 -0800)] 
[PATCH] relayfs: add Documentation on relay files in other filesystems

Documentation update for creating relay files in other filesystems.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: add support for relay files in other filesystems
Tom Zanussi [Sun, 8 Jan 2006 09:02:28 +0000 (01:02 -0800)] 
[PATCH] relayfs: add support for relay files in other filesystems

This patch adds a couple of callback functions that allow a client to hook
into relay_open()/close() and supply the files that will be used to represent
the channel buffers; the default implementation if no callbacks are defined is
to create the files in relayfs.  This is to support the creation and use of
relay files in other filesystems such as debugfs, as implied by the fact that
relayfs_file_operations are exported.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: add Documention for non-relay files
Tom Zanussi [Sun, 8 Jan 2006 09:02:27 +0000 (01:02 -0800)] 
[PATCH] relayfs: add Documention for non-relay files

Documentation update for non-relay files.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: remove unused alloc/destroy_inode()
Tom Zanussi [Sun, 8 Jan 2006 09:02:26 +0000 (01:02 -0800)] 
[PATCH] relayfs: remove unused alloc/destroy_inode()

Since we're no longer using relayfs_inode_info, remove relayfs_alloc_inode()
and relayfs_destroy_inode() along with the relayfs inode cache.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: use generic_ip for private data
Tom Zanussi [Sun, 8 Jan 2006 09:02:26 +0000 (01:02 -0800)] 
[PATCH] relayfs: use generic_ip for private data

Use inode->u.generic_ip instead of relayfs_inode_info to store pointer to user
data.  Clients using relayfs_file_create() to create their own files would
probably more expect their data to be stored in generic_ip; we also intend in
the next set of patches to get rid of relayfs-specific stuff in the file
operations, so we might as well do it here.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: add relayfs_remove_file()
Tom Zanussi [Sun, 8 Jan 2006 09:02:25 +0000 (01:02 -0800)] 
[PATCH] relayfs: add relayfs_remove_file()

This patch adds and exports relayfs_remove_file(), for API symmetry (with
relayfs_create_file()).

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: export relayfs_create_file() with fileops param
Tom Zanussi [Sun, 8 Jan 2006 09:02:24 +0000 (01:02 -0800)] 
[PATCH] relayfs: export relayfs_create_file() with fileops param

This patch adds a mandatory fileops param to relayfs_create_file() and exports
that function so that clients can use it to create files defined by their own
set of file operations, in relayfs.  The purpose is to allow relayfs
applications to create their own set of 'control' files alongside their relay
files in relayfs rather than having to create them in /proc or debugfs for
instance.  relayfs_create_file() is also used by relay_open_buf() to create
the relay files for a channel.  In this case, a pointer to
relayfs_file_operations is passed in, along with a pointer to the buffer
associated with the file.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] relayfs: decouple buffer creation from inode creation
Tom Zanussi [Sun, 8 Jan 2006 09:02:23 +0000 (01:02 -0800)] 
[PATCH] relayfs: decouple buffer creation from inode creation

The patch series implementa or fixes 3 things that were specifically requested
or suggested by relayfs users:

- support for non-relay files (patches 1-6)

Currently, the relayfs API only supports the creation of directories
(relayfs_create_dir()) and relay files (relay_open()).  These patches adds
support for non-relay files (relayfs_create_file()).  This is so relayfs
applications can create 'control files' in relayfs itself rather than in /proc
or via a netlink channel, as is currently done in the relay-app examples.
Basically what this amounts to is exporting relayfs_create_file() with an
additional file_ops param that clients can use to supply file operations for
their own special-purpose files in relayfs.

- make exported relay file ops useful (patches 7-8)

The relayfs relay_file_operations have always been exported, the intent being
to make it possible to create relay files in other filesystems such as
debugfs.  The problem, though, is that currently the file operations are too
tightly coupled to relayfs to actually be used for this purpose.  This patch
fixes that by adding a couple of callback functions that allow a client to
hook into relay_open()/close() and supply the files that will be used to
represent the channel buffers; the default implementation if no callbacks are
defined is to create the files in relayfs.

- add an option to create global relay buffer (patches 9-10) The file creation
callback also supplies an optional param, is_global, that can be used by
clients to create a single global relayfs buffer instead of the default
per-cpu buffers.  This was suggested as being useful for certain debugging
applications where it's more convenient to be able to get all the data from a
single channel without having to go to the bother of dealing with per-cpu
files.

- cleanup, some renaming and Documentation updates (patches 11-12)

There were several comments that the use of netlink in the example code was
non-intuitive and in fact the whole relay-app business was needlessly
confusing.  Based on that feedback, the example code has been completely
converted over to relayfs control files as supported by this patch, and have
also been made completely self-contained.

The converted examples along with a couple of new examples that demonstrate
using exported relay files can be found in relay-apps tarball:
http://prdownloads.sourceforge.net/relayfs/relay-apps-0.9.tar.gz?download

This patch:

Separate buffer create/destroy from inode create/destroy.  We want to be able
to associate other data and not just relay buffers with inodes.  Buffer
create/destroy is moved out of inode.c and into relayfs core code.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ipc: expand shm_flags
Andrew Morton [Sun, 8 Jan 2006 09:02:21 +0000 (01:02 -0800)] 
[PATCH] ipc: expand shm_flags

Unobfsucate this struct member

Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] ELF: symbol table type additions
Jan Beulich [Sun, 8 Jan 2006 09:02:20 +0000 (01:02 -0800)] 
[PATCH] ELF: symbol table type additions

Needed for the Novell kernel debugger and perhaps some per-cpu data on x86_64
in the future.

Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] rcu file: use atomic primitives
Nick Piggin [Sun, 8 Jan 2006 09:02:19 +0000 (01:02 -0800)] 
[PATCH] rcu file: use atomic primitives

Use atomic_inc_not_zero for rcu files instead of special case rcuref.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] atomic: dec_and_lock use atomic primitives
Nick Piggin [Sun, 8 Jan 2006 09:02:19 +0000 (01:02 -0800)] 
[PATCH] atomic: dec_and_lock use atomic primitives

Convert atomic_dec_and_lock to use new atomic primitives.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] pktcdvd: Use bd_claim to get exclusive access
Peter Osterlund [Sun, 8 Jan 2006 09:02:17 +0000 (01:02 -0800)] 
[PATCH] pktcdvd: Use bd_claim to get exclusive access

Use bd_claim() when opening the cdrom device to prevent user space programs
such as cdrecord, hald and kded from interfering with the burning process.

Signed-off-by: Peter Osterlund <petero2@telia.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] kernel/: small cleanups
Adrian Bunk [Sun, 8 Jan 2006 09:02:17 +0000 (01:02 -0800)] 
[PATCH] kernel/: small cleanups

This patch contains the following cleanups:
- make needlessly global functions static
- every file should include the headers containing the prototypes for
  it's global functions

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] drivers/isdn/: "extern inline" -> "static inline"
Adrian Bunk [Sun, 8 Jan 2006 09:02:16 +0000 (01:02 -0800)] 
[PATCH] drivers/isdn/: "extern inline" -> "static inline"

"extern inline" -> "static inline"

Since there's no pullphone() function this patch removes the dead
prototype.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Karsten Keil <kkeil@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] move rtc_interrupt() prototype to rtc.h
Adrian Bunk [Sun, 8 Jan 2006 09:02:15 +0000 (01:02 -0800)] 
[PATCH] move rtc_interrupt() prototype to rtc.h

This patch moves the rtc_interrupt() prototype to rtc.h and removes the
prototypes from C files.

It also renames static rtc_interrupt() functions in
arch/arm/mach-integrator/time.c and arch/sh64/kernel/time.c to avoid compile
problems.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Paul Gortmaker <p_gortmaker@yahoo.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix and add EXPORT_SYMBOL(filemap_write_and_wait)
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:14 +0000 (01:02 -0800)] 
[PATCH] Fix and add EXPORT_SYMBOL(filemap_write_and_wait)

This patch add EXPORT_SYMBOL(filemap_write_and_wait) and use it.

See mm/filemap.c:

And changes the filemap_write_and_wait() and filemap_write_and_wait_range().

Current filemap_write_and_wait() doesn't wait if filemap_fdatawrite()
returns error.  However, even if filemap_fdatawrite() returned an
error, it may have submitted the partially data pages to the device.
(e.g. in the case of -ENOSPC)

<quotation>
Andrew Morton writes,

If filemap_fdatawrite() returns an error, this might be due to some
I/O problem: dead disk, unplugged cable, etc.  Given the generally
crappy quality of the kernel's handling of such exceptions, there's a
good chance that the filemap_fdatawait() will get stuck in D state
forever.
</quotation>

So, this patch doesn't wait if filemap_fdatawrite() returns the -EIO.

Trond, could you please review the nfs part?  Especially I'm not sure,
nfs must use the "filemap_fdatawrite(inode->i_mapping) == 0", or not.

Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fat: support a truncate() for expanding size (generic_cont_expand)
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:13 +0000 (01:02 -0800)] 
[PATCH] fat: support a truncate() for expanding size (generic_cont_expand)

This patch changes generic_cont_expand(), in order to share the code
with fatfs.

  - Use vmtruncate() if ->prepare_write() returns a error.

Even if ->prepare_write() returns an error, it may already have added some
blocks.  So, this truncates blocks outside of ->i_size by vmtruncate().

  - Add generic_cont_expand_simple().

The generic_cont_expand_simple() assumes that ->prepare_write() can handle
the block boundary.  With this, we don't need to care the extra byte.

And for expanding a file size by truncate(), fatfs uses the
added generic_cont_expand_simple().

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] export/change sync_page_range/_nolock()
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:12 +0000 (01:02 -0800)] 
[PATCH] export/change sync_page_range/_nolock()

This exports/changes the sync_page_range/_nolock().  The fatfs needs
sync_page_range/_nolock() for expanding truncate, and changes "size_t count"
to "loff_t count".

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fat: support ->direct_IO()
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:11 +0000 (01:02 -0800)] 
[PATCH] fat: support ->direct_IO()

This patch add to support of ->direct_IO() for mostly read.

The user of this seems to want to use for streaming read.  So, current direct
I/O has limitation, it can only overwrite.  (For write operation, mainly we
need to handle the hole etc..)

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fat: s/EXPORT_SYMBOL/EXPORT_SYMBOL_GPL/
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:10 +0000 (01:02 -0800)] 
[PATCH] fat: s/EXPORT_SYMBOL/EXPORT_SYMBOL_GPL/

All EXPORT_SYMBOL of fatfs is only for vfat/msdos. _GPL would be proper.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fat: add the read/writepages()
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:10 +0000 (01:02 -0800)] 
[PATCH] fat: add the read/writepages()

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fat: use sb_find_get_block() instead of sb_getblk()
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:09 +0000 (01:02 -0800)] 
[PATCH] fat: use sb_find_get_block() instead of sb_getblk()

We don't need to allocate buffer for checking the buffer is uptodate.  This
use sb_find_get_block() instead, and if it returns NULL it's not uptodate.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fat: move fat_clusters_flush() to write_super()
OGAWA Hirofumi [Sun, 8 Jan 2006 09:02:08 +0000 (01:02 -0800)] 
[PATCH] fat: move fat_clusters_flush() to write_super()

It is overkill to update the FS_INFO whenever modifying
prev_free/free_clusters, because those are just a hint.

So, this patch uses ->write_super() for updating FS_INFO instead.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] IRQ type flags
Russell King [Sun, 8 Jan 2006 09:02:07 +0000 (01:02 -0800)] 
[PATCH] IRQ type flags

Some ARM platforms have the ability to program the interrupt controller to
detect various interrupt edges and/or levels.  For some platforms, this is
critical to setup correctly, particularly those which the setting is dependent
on the device.

Currently, ARM drivers do (eg) the following:

err = request_irq(irq, ...);

set_irq_type(irq, IRQT_RISING);

However, if the interrupt has previously been programmed to be level sensitive
(for whatever reason) then this will cause an interrupt storm.

Hence, if we combine set_irq_type() with request_irq(), we can then safely set
the type prior to unmasking the interrupt.  The unfortunate problem is that in
order to support this, these flags need to be visible outside of the ARM
architecture - drivers such as smc91x need these flags and they're
cross-architecture.

Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the
property that the device would like.  The IRQ controller code should do its
best to select the most appropriate supported mode.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] new driver synclink_gt
Paul Fulghum [Sun, 8 Jan 2006 09:02:06 +0000 (01:02 -0800)] 
[PATCH] new driver synclink_gt

New character device driver for the SyncLink GT and SyncLink AC families of
synchronous and asynchronous serial adapters

Signed-off-by: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] fix more missing includes
Tim Schmielau [Sun, 8 Jan 2006 09:02:05 +0000 (01:02 -0800)] 
[PATCH] fix more missing includes

Include fixes for 2.6.14-git11.  Should allow to remove sched.h from
module.h on i386, x86_64, arm, ia64, ppc, ppc64, and s390.  Probably more
to come since I haven't yet checked the other archs.

Signed-off-by: Tim Schmielau <tim@physik3.uni-rostock.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: skip rcu check if task is in root cpuset
Paul Jackson [Sun, 8 Jan 2006 09:02:04 +0000 (01:02 -0800)] 
[PATCH] cpuset: skip rcu check if task is in root cpuset

For systems that aren't using cpusets, but have them CONFIG_CPUSET enabled in
their kernel (eventually this may be most distribution kernels), this patch
removes even the minimal rcu_read_lock() from the memory page allocation path.

Actually, it removes that rcu call for any task that is in the root cpuset
(top_cpuset), which on systems not actively using cpusets, is all tasks.

We don't need the rcu check for tasks in the top_cpuset, because the
top_cpuset is statically allocated, so at no risk of being freed out from
underneath us.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: mark number_of_cpusets read_mostly
Paul Jackson [Sun, 8 Jan 2006 09:02:03 +0000 (01:02 -0800)] 
[PATCH] cpuset: mark number_of_cpusets read_mostly

Mark cpuset global 'number_of_cpusets' as __read_mostly.

This global is accessed everytime a zone is considered in the zonelist loops
beneath __alloc_pages, looking for a free memory page.  If number_of_cpusets
is just one, then we can short circuit the mems_allowed check.

Since this global is read alot on a hot path, and written rarely, it is an
excellent candidate for __read_mostly.

Thanks to Christoph Lameter for the suggestion.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: use rcu directly optimization
Paul Jackson [Sun, 8 Jan 2006 09:02:02 +0000 (01:02 -0800)] 
[PATCH] cpuset: use rcu directly optimization

Optimize the cpuset impact on page allocation, the most performance critical
cpuset hook in the kernel.

On each page allocation, the cpuset hook needs to check for a possible change
in the current tasks cpuset.  It can now handle the common case, of no change,
without taking any spinlock or semaphore, thanks to RCU.

Convert a spinlock on the current task to an rcu_read_lock(), saving
approximately a memory barrier and an atomic op, depending on architecture.

This is done by adding rcu_assign_pointer() and synchronize_rcu() calls to the
write side of the task->cpuset pointer, in cpuset.c:attach_task(), to delay
freeing up a detached cpuset until after any critical sections referencing
that pointer.

Thanks to Andi Kleen, Nick Piggin and Eric Dumazet for ideas.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: remove test for null cpuset from alloc code path
Paul Jackson [Sun, 8 Jan 2006 09:02:01 +0000 (01:02 -0800)] 
[PATCH] cpuset: remove test for null cpuset from alloc code path

Remove a couple of more lines of code from the cpuset hooks in the page
allocation code path.

There was a check for a NULL cpuset pointer in the routine
cpuset_update_task_memory_state() that was only needed during system boot,
after the memory subsystem was initialized, before the cpuset subsystem was
initialized, to catch a NULL task->cpuset pointer.

Add a cpuset_init_early() routine, just before the mem_init() call in
init/main.c, that sets up just enough of the init tasks cpuset structure to
render cpuset_update_task_memory_state() calls harmless.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: migrate all tasks in cpuset at once
Paul Jackson [Sun, 8 Jan 2006 09:02:00 +0000 (01:02 -0800)] 
[PATCH] cpuset: migrate all tasks in cpuset at once

Given the mechanism in the previous patch to handle rebinding the per-vma
mempolicies of all tasks in a cpuset that changes its memory placement, it is
now easier to handle the page migration requirements of such tasks at the same
time.

The previous code didn't actually attempt to migrate the pages of the tasks in
a cpuset whose memory placement changed until the next time each such task
tried to allocate memory.  This was undesirable, as users invoking memory page
migration exected to happen when the placement changed, not some unspecified
time later when the task needed more memory.

It is now trivial to handle the page migration at the same time as the per-vma
rebinding is done.

The routine cpuset.c:update_nodemask(), which handles changing a cpusets
memory placement ('mems') now checks for the special case of being asked to
write a placement that is the same as before.  It was harmless enough before
to just recompute everything again, even though nothing had changed.  But page
migration is a heavy weight operation - moving pages about.  So now it is
worth avoiding that if asked to move a cpuset to its current location.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: rebind vma mempolicies fix
Paul Jackson [Sun, 8 Jan 2006 09:01:59 +0000 (01:01 -0800)] 
[PATCH] cpuset: rebind vma mempolicies fix

Fix more of longstanding bug in cpuset/mempolicy interaction.

NUMA mempolicies (mm/mempolicy.c) are constrained by the current tasks cpuset
to just the Memory Nodes allowed by that cpuset.  The kernel maintains
internal state for each mempolicy, tracking what nodes are used for the
MPOL_INTERLEAVE, MPOL_BIND or MPOL_PREFERRED policies.

When a tasks cpuset memory placement changes, whether because the cpuset
changed, or because the task was attached to a different cpuset, then the
tasks mempolicies have to be rebound to the new cpuset placement, so as to
preserve the cpuset-relative numbering of the nodes in that policy.

An earlier fix handled such mempolicy rebinding for mempolicies attached to a
task.

This fix rebinds mempolicies attached to vma's (address ranges in a tasks
address space.) Due to the need to hold the task->mm->mmap_sem semaphore while
updating vma's, the rebinding of vma mempolicies has to be done when the
cpuset memory placement is changed, at which time mmap_sem can be safely
acquired.  The tasks mempolicy is rebound later, when the task next attempts
to allocate memory and notices that its task->cpuset_mems_generation is
out-of-date with its cpusets mems_generation.

Because walking the tasklist to find all tasks attached to a changing cpuset
requires holding tasklist_lock, a spinlock, one cannot update the vma's of the
affected tasks while doing the tasklist scan.  In general, one cannot acquire
a semaphore (which can sleep) while already holding a spinlock (such as
tasklist_lock).  So a list of mm references has to be built up during the
tasklist scan, then the tasklist lock dropped, then for each mm, its mmap_sem
acquired, and the vma's in that mm rebound.

Once the tasklist lock is dropped, affected tasks may fork new tasks, before
their mm's are rebound.  A kernel global 'cpuset_being_rebound' is set to
point to the cpuset being rebound (there can only be one; cpuset modifications
are done under a global 'manage_sem' semaphore), and the mpol_copy code that
is used to copy a tasks mempolicies during fork catches such forking tasks,
and ensures their children are also rebound.

When a task is moved to a different cpuset, it is easier, as there is only one
task involved.  It's mm->vma's are scanned, using the same
mpol_rebind_policy() as used above.

It may happen that both the mpol_copy hook and the update done via the
tasklist scan update the same mm twice.  This is ok, as the mempolicies of
each vma in an mm keep track of what mems_allowed they are relative to, and
safely no-op a second request to rebind to the same nodes.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: number_of_cpusets optimization
Paul Jackson [Sun, 8 Jan 2006 09:01:57 +0000 (01:01 -0800)] 
[PATCH] cpuset: number_of_cpusets optimization

Easy little optimization hack to avoid actually having to call
cpuset_zone_allowed() and check mems_allowed, in the main page allocation
routine, __alloc_pages().  This saves several CPU cycles per page allocation
on systems not using cpusets.

A counter is updated each time a cpuset is created or removed, and whenever
there is only one cpuset in the system, it must be the root cpuset, which
contains all CPUs and all Memory Nodes.  In that case, when the counter is
one, all allocations are allowed.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: numa_policy_rebind cleanup
Paul Jackson [Sun, 8 Jan 2006 09:01:56 +0000 (01:01 -0800)] 
[PATCH] cpuset: numa_policy_rebind cleanup

Cleanup, reorganize and make more robust the mempolicy.c code to rebind
mempolicies relative to the containing cpuset after a tasks memory placement
changes.

The real motivator for this cleanup patch is to lay more groundwork for the
upcoming patch to correctly rebind NUMA mempolicies that are attached to vma's
after the containing cpuset memory placement changes.

NUMA mempolicies are constrained by the cpuset their task is a member of.
When either (1) a task is moved to a different cpuset, or (2) the 'mems'
mems_allowed of a cpuset is changed, then the NUMA mempolicies have embedded
node numbers (for MPOL_BIND, MPOL_INTERLEAVE and MPOL_PREFERRED) that need to
be recalculated, relative to their new cpuset placement.

The old code used an unreliable method of determining what was the old
mems_allowed constraining the mempolicy.  It just looked at the tasks
mems_allowed value.  This sort of worked with the present code, that just
rebinds the -task- mempolicy, and leaves any -vma- mempolicies broken,
referring to the old nodes.  But in an upcoming patch, the vma mempolicies
will be rebound as well.  Then the order in which the various task and vma
mempolicies are updated will no longer be deterministic, and one can no longer
count on the task->mems_allowed holding the old value for as long as needed.
It's not even clear if the current code was guaranteed to work reliably for
task mempolicies.

So I added a mems_allowed field to each mempolicy, stating exactly what
mems_allowed the policy is relative to, and updated synchronously and reliably
anytime that the mempolicy is rebound.

Also removed a useless wrapper routine, numa_policy_rebind(), and had its
caller, cpuset_update_task_memory_state(), call directly to the rewritten
policy_rebind() routine, and made that rebind routine extern instead of
static, and added a "mpol_" prefix to its name, making it
mpol_rebind_policy().

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: implement cpuset_mems_allowed
Paul Jackson [Sun, 8 Jan 2006 09:01:55 +0000 (01:01 -0800)] 
[PATCH] cpuset: implement cpuset_mems_allowed

Provide a cpuset_mems_allowed() method, which the sys_migrate_pages() code
needed, to obtain the mems_allowed vector of a cpuset, and replaced the
workaround in sys_migrate_pages() to call this new method.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: combine refresh_mems and update_mems
Paul Jackson [Sun, 8 Jan 2006 09:01:54 +0000 (01:01 -0800)] 
[PATCH] cpuset: combine refresh_mems and update_mems

The important code paths through alloc_pages_current() and alloc_page_vma(),
by which most kernel page allocations go, both called
cpuset_update_current_mems_allowed(), which in turn called refresh_mems().
-Both- of these latter two routines did a tasklock, got the tasks cpuset
pointer, and checked for out of date cpuset->mems_generation.

That was a silly duplication of code and waste of CPU cycles on an important
code path.

Consolidated those two routines into a single routine, called
cpuset_update_task_memory_state(), since it updates more than just
mems_allowed.

Changed all callers of either routine to call the new consolidated routine.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: fork hook fix
Paul Jackson [Sun, 8 Jan 2006 09:01:53 +0000 (01:01 -0800)] 
[PATCH] cpuset: fork hook fix

Fix obscure, never seen in real life, cpuset fork race.  The cpuset_fork()
call in fork.c was setting up the correct task->cpuset pointer after the
tasklist_lock was dropped, which briefly exposed the newly forked process with
an unsafe (copied from parent without locks or usage counter increment) cpuset
pointer.

In theory, that exposed cpuset pointer could have been pointing at a cpuset
that was already freed and removed, and in theory another task that had been
sitting on the tasklist_lock waiting to scan the task list could have raced
down the entire tasklist, found our new child at the far end, and dereferenced
that bogus cpuset pointer.

To fix, setup up the correct cpuset pointer in the new child by calling
cpuset_fork() before the new task is linked into the tasklist, and with that,
add a fork failure case, to dereference that cpuset, if the fork fails along
the way, after cpuset_fork() was called.

Had to remove a BUG_ON() from cpuset_exit(), because it was no longer valid -
the call to cpuset_exit() from a failed fork would not have PF_EXITING set.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: update_nodemask code reformat
Paul Jackson [Sun, 8 Jan 2006 09:01:52 +0000 (01:01 -0800)] 
[PATCH] cpuset: update_nodemask code reformat

Restructure code layout of the kernel/cpuset.c update_nodemask() routine,
removing embedded returns and nested if's in favor of goto completion labels.
This is being done in anticipation of adding more logic to this routine, which
will favor the goto style structure.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: minor spacing initializer fixes
Paul Jackson [Sun, 8 Jan 2006 09:01:51 +0000 (01:01 -0800)] 
[PATCH] cpuset: minor spacing initializer fixes

Four trivial cpuset fixes: remove extra spaces, remove useless initializers,
mark one __read_mostly.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: remove marker_pid documentation
Paul Jackson [Sun, 8 Jan 2006 09:01:51 +0000 (01:01 -0800)] 
[PATCH] cpuset: remove marker_pid documentation

Remove documentation for the cpuset 'marker_pid' feature, that was in the
patch "cpuset: change marker for relative numbering" That patch was previously
pulled from *-mm at my (pj) request.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: document additional features
Paul Jackson [Sun, 8 Jan 2006 09:01:50 +0000 (01:01 -0800)] 
[PATCH] cpuset: document additional features

Document the additional cpuset features:
notify_on_release
marker_pid
memory_pressure
memory_pressure_enabled

Rearrange and improve formatting of existing documentation for
cpu_exclusive and mem_exclusive features.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: memory pressure meter
Paul Jackson [Sun, 8 Jan 2006 09:01:49 +0000 (01:01 -0800)] 
[PATCH] cpuset: memory pressure meter

Provide a simple per-cpuset metric of memory pressure, tracking the -rate-
that the tasks in a cpuset call try_to_free_pages(), the synchronous
(direct) memory reclaim code.

This enables batch managers monitoring jobs running in dedicated cpusets to
efficiently detect what level of memory pressure that job is causing.

This is useful both on tightly managed systems running a wide mix of
submitted jobs, which may choose to terminate or reprioritize jobs that are
trying to use more memory than allowed on the nodes assigned them, and with
tightly coupled, long running, massively parallel scientific computing jobs
that will dramatically fail to meet required performance goals if they
start to use more memory than allowed to them.

This patch just provides a very economical way for the batch manager to
monitor a cpuset for signs of memory pressure.  It's up to the batch
manager or other user code to decide what to do about it and take action.

==> Unless this feature is enabled by writing "1" to the special file
    /dev/cpuset/memory_pressure_enabled, the hook in the rebalance
    code of __alloc_pages() for this metric reduces to simply noticing
    that the cpuset_memory_pressure_enabled flag is zero.  So only
    systems that enable this feature will compute the metric.

Why a per-cpuset, running average:

    Because this meter is per-cpuset, rather than per-task or mm, the
    system load imposed by a batch scheduler monitoring this metric is
    sharply reduced on large systems, because a scan of the tasklist can be
    avoided on each set of queries.

    Because this meter is a running average, instead of an accumulating
    counter, a batch scheduler can detect memory pressure with a single
    read, instead of having to read and accumulate results for a period of
    time.

    Because this meter is per-cpuset rather than per-task or mm, the
    batch scheduler can obtain the key information, memory pressure in a
    cpuset, with a single read, rather than having to query and accumulate
    results over all the (dynamically changing) set of tasks in the cpuset.

A per-cpuset simple digital filter (requires a spinlock and 3 words of data
per-cpuset) is kept, and updated by any task attached to that cpuset, if it
enters the synchronous (direct) page reclaim code.

A per-cpuset file provides an integer number representing the recent
(half-life of 10 seconds) rate of direct page reclaims caused by the tasks
in the cpuset, in units of reclaims attempted per second, times 1000.

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] cpuset: mempolicy one more nodemask conversion
Paul Jackson [Sun, 8 Jan 2006 09:01:47 +0000 (01:01 -0800)] 
[PATCH] cpuset: mempolicy one more nodemask conversion

Finish converting mm/mempolicy.c from bitmaps to nodemasks.  The previous
conversion had left one routine using bitmaps, since it involved a
corresponding change to kernel/cpuset.c

Fix that interface by replacing with a simple macro that calls nodes_subset(),
or if !CONFIG_CPUSET, returns (1).

Signed-off-by: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <christoph@lameter.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>