RLIMIT_RTPRIO is supposed to grant non privileged users the right to use
SCHED_FIFO/SCHED_RR scheduling policies with priorites bounded by the
RLIMIT_RTPRIO value via sched_setscheduler(). This is usually used by
audio users.

Unfortunately this is broken in 2.6.13rc3 as you can see in the excerpt
from sched_setscheduler below:

        /*
         * Allow unprivileged RT tasks to decrease priority:
         */
        if (!capable(CAP_SYS_NICE)) {
                /* can't change policy */
                if (policy != p->policy)
                        return -EPERM;

After the above unconditional test which causes sched_setscheduler to
fail with no regard to the RLIMIT_RTPRIO value the following check is made:

               /* can't increase priority */
                if (policy != SCHED_NORMAL &&
                    param->sched_priority > p->rt_priority &&
                    param->sched_priority >
                                p->signal->rlim[RLIMIT_RTPRIO].rlim_cur)
                        return -EPERM;

Thus I do believe that the RLIMIT_RTPRIO value must be taken into
account for the policy check, especially as the RLIMIT_RTPRIO limit is
of no use without this change.

The attached patch fixes this problem.
Signed-off-by: NAndreas Steinmetz <ast@domdv.de>
Acked-by: NIngo Molnar <mingo@elte.hu>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Remove bogus initialization that was re-done (correctly) later.

Patch from Ben Dooks

Split the s3c2440 specific clocks from the arch clock support, to
make the code clearer.
Signed-off-by: NBen Dooks <ben-linux@fluff.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Now that all of the code paths that call acpi_power_off
have been modified to call either call kernel_power_off
(which calls apci_sleep_prepare by way of acpi_shutdown)
or to call acpi_sleep_prepare directly it is redundant to call
acpi_sleep_prepare from acpi_power_off.

So simplify the code and simply don't call acpi_sleep_prepare.

In addition there is a little error handling done so if we
can't register the acpi class we don't hook pm_power_off.

I think I have done the right thing with the CONFIG_PM define
but I'm not certain.  Can this code even be compiled if
CONFIG_PM is false?
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

machine_power_off on i386 and x86_64 now switch to the
boot cpu out of paranoia and because the MP Specification indicates it
is a good idea on reboot, so for those architectures it is a noop.
I can't see anything in the acpi spec that requires you to be on
the boot cpu to power off the system, so this should not be an issue
for ia64.  In addition ia64 has the altix a massive multi-node
system where switching to the boot cpu sounds insane as we may
hot removed the boot cpu.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

i386 machine_power_off was disabling the local apic
and all of it's users wanted to be on the boot cpu.
So call machine_shutdown which places us on the boot
cpu and disables the apics.  This keeps us in sync
and reduces the number of cases we need to worry about in
the power management code.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

machine_power_off now always switches to the boot cpu so there
is no reason for APM to also do that.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Call machine_shutdown() to move to the boot cpu
and disable apics.  Both acpi_power_off and
apm_power_off want to move to the boot cpu.
and we are already disabling the local apics
so calling machine_shutdown simply reuses
code.

ia64 doesn't have a special path in power_off
for efi so there is no reason i386 should.  If
we really need to call the efi power off path
the efi driver can set pm_power_off like everyone
else.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This appears to be a typo I introduced when cleaning
this code up earlier. Ooops.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The call appears to come from process context so kernel_power_off
should be safe.  And acpi_power_off won't necessarily work if you just
call machine_power_off.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The suspend to disk code was a poor copy of the code in
sys_reboot now that we have kernel_power_off, kernel_restart
and kernel_halt use them instead of poorly duplicating them inline.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The 68328serial.c driver has a weird local reimplementation of
magic sysrq.  The code is architecture specific enough that calling
machine_restart() is probably ok.  But there is no reason not to call
emergency_restart() so do so.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

If we've hung a clean reboot does not sound like a real
option.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

If a watchdog driver has decided it is time to reboot the system
we know something is wrong and we are in interrupt context
so emergency_reboot() is what we want.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

sysrq calls into the reboot path from an interrupt handler
we can either push the code do into process context and
call kernel_restart and get a clean reboot or we can simply
reboot the machine, and increase our chances of actually
rebooting.  emergency_reboot() seems like the closest match
to what we have previously done, and what we want.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

We know the system is in trouble so there is no question if this
is an emergecy :)
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

We already do all of the gymnastics to run from process context
to call the power off code so call into the power off code cleanly.

This especially helps acpi as part of it's shutdown logic should
run acpi_shutdown called from device_shutdown which was not
being called from here.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

It is not safe to call set_cpus_allowed() in interrupt
context and disabling the apics is complicated code.
So unconditionally skip machine_shutdown in machine_emergency_reboot
on x86_64.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

We only want to shutdown the apics if reboot_force
is not specified.  Be we are doing this both
in machine_shutdown which is called unconditionally
and if (!reboot_force).  So simply call machine_shutdown
if (!reboot_force).  It looks like something
went weird with merging some of the kexec patches for
x86_64, and caused this.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

set_cpus_allowed is not safe in interrupt context
and disabling apics is complicated code so don't
call machine_shutdown on i386 from emergency_restart().
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

machine_restart, machine_halt and machine_power_off are machine
specific hooks deep into the reboot logic, that modules
have no business messing with.  Usually code should be calling
kernel_restart, kernel_halt, kernel_power_off, or
emergency_restart. So don't export machine_restart,
machine_halt, and machine_power_off so we can catch buggy users.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

It appears machine_restart has been working cris just
by luck.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

When the kernel is working well and we want to restart cleanly
kernel_restart is the function to use.   But in many instances
the kernel wants to reboot when thing are expected to be working
very badly such as from panic or a software watchdog handler.

This patch adds the function emergency_restart() so that
callers can be clear what semantics they expect when calling
restart.  emergency_restart() is expected to be callable
from interrupt context and possibly reliable in even more
trying circumstances.

This is an initial generic implementation for all architectures.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

It is obvious we wanted to call kernel_restart here
but since we don't have it the code was expanded inline and hasn't
been correct since sometime in 2.4.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Because the factors of sys_reboot don't exist people calling
into the reboot path duplicate the code badly, leading to
inconsistent expectations of code in the reboot path.

This patch should is just code motion.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

In the recent addition of device_suspend calls into
sys_reboot two code paths were missed.
Signed-off-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

*** Warning: "fc_remote_port_block" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "scsi_is_fc_rport" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "fc_remote_port_unblock" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "fc_remote_port_rolechg" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "fc_release_transport" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "fc_remove_host" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "fc_remote_port_add" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!
*** Warning: "fc_attach_transport" [drivers/scsi/qla2xxx/qla2xxx.ko] undefined!

Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Older gcc's dont support anonymous unions, so this driver gets hundreds of
error.

Fortunately the fix is easy...

Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Cc: Robert Love <rml@novell.com>
Cc: John McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Add inotify syscall entries to x86-64.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Add missing fsnotify_open() hook to sys32_open().

Add fsnotify_open() hook to sys32_open() on x86-64.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Check for (unlikely) errors in the filesystem initialization stuff in
our module_init() function.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Change default inotify limits: Maximum instances per user to 128 and
maximum events per queue to 16k.  The max instances used to be 128; the
change to 8 was a mistake.  Memory consumption is fine.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Handle error out paths better.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Bug fix: Ensure that the fd passed to inotify_add_watch() and
inotify_rm_watch() belongs to inotify.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

As an optimization, use fget_light() and fput_light() where possible.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Miscellaneous invariant clean up, comment fixes, and so on.  Trivial
stuff.
Signed-off-by: NRobert Love <rml@novell.com>
Signed-off-by: NJohn McCutchan <ttb@tentacle.dhs.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>