When the spinlock routines were moved out of line into
kernel/spinlock.c this made it so that the debugging
spinlocks record lock acquisition program counts in the
kernel/spinlock.c functions not in their callers.
This makes the debugging info kind of useless.

So record the correct caller's program counter and
now this feature is useful once more.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Removed sparc64 architecture specific users of asm/segment.h and
asm-sparc64/segment.h itself
Signed-off-by: NKumar Gala <kumar.gala@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Current uncorrectable error handling was poor enough
that the processor could just loop taking the same
trap over and over again.  Fix things up so that we
at least get a log message and perhaps even some register
state.

In the process, much consolidation became possible,
particularly with the correctable error handler.

Prefix assembler and C function names with "spitfire"
to indicate that these are for Ultra-I/II/IIi/IIe only.

More work is needed to make these routines robust and
featureful to the level of the Ultra-III error handlers.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Verify we really are taking a data access exception trap, at TL1, from
one of the window spill/fill handlers.

Else call a new function, data_access_exception_tl1, to log the error.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

1) Read ASI_IMMU SFSR not ASI_DMMU.
2) IMMU has no SFAR, read TPC instead
3) Delete old and incorrect comment about the DTLB protection
   trap having a dependency on the SFSR contents in order to
   function correctly
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It has been reported that the way Linux handles NODEFER for signals is
not consistent with the way other Unix boxes handle it.  I've written a
program to test the behavior of how this flag affects signals and had
several reports from people who ran this on various Unix boxes,
confirming that Linux seems to be unique on the way this is handled.

The way NODEFER affects signals on other Unix boxes is as follows:

1) If NODEFER is set, other signals in sa_mask are still blocked.

2) If NODEFER is set and the signal is in sa_mask, then the signal is
still blocked. (Note: this is the behavior of all tested but Linux _and_
NetBSD 2.0 *).

The way NODEFER affects signals on Linux:

1) If NODEFER is set, other signals are _not_ blocked regardless of
sa_mask (Even NetBSD doesn't do this).

2) If NODEFER is set and the signal is in sa_mask, then the signal being
handled is not blocked.

The patch converts signal handling in all current Linux architectures to
the way most Unix boxes work.

Unix boxes that were tested:  DU4, AIX 5.2, Irix 6.5, NetBSD 2.0, SFU
3.5 on WinXP, AIX 5.3, Mac OSX, and of course Linux 2.6.13-rcX.

* NetBSD was the only other Unix to behave like Linux on point #2. The
main concern was brought up by point #1 which even NetBSD isn't like
Linux.  So with this patch, we leave NetBSD as the lonely one that
behaves differently here with #2.
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

pcibios_bus_to_resource is exported on all architectures except ia64
and sparc.  Add exports for the two missing architectures.  Needed when
Yenta socket support is compiled as a module.
Signed-off-by: NKeith Owens <kaos@sgi.com>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

GCC 4.x really dislikes the games we are playing in
unaligned.c, and the cleanest way to fix this is to
move things into assembler.

Noted by Al Viro.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

1) cpufreq wants frequenceis in KHZ not MHZ
2) provide ->get() method so curfreq node is created
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Revert commit fec59a71, which is
breaking sparc64 that doesn't have a working pci_update_resource.

We'll re-do this after 2.6.13 when we'll do it all properly.

Some PCI devices (e.g. 3c905B, 3c556B) lose all configuration
(including BARs) when transitioning from D3hot->D0.  This leaves such
a device in an inaccessible state.  The patch below causes the BARs
to be restored when enabling such a device, so that its driver will
be able to access it.

The patch also adds pci_restore_bars as a new global symbol, and adds a
correpsonding EXPORT_SYMBOL_GPL for that.

Some firmware (e.g. Thinkpad T21) leaves devices in D3hot after a
(re)boot.  Most drivers call pci_enable_device very early, so devices
left in D3hot that lose configuration during the D3hot->D0 transition
will be inaccessible to their drivers.

Drivers could be modified to account for this, but it would
be difficult to know which drivers need modification.  This is
especially true since often many devices are covered by the same
driver.  It likely would be necessary to replicate code across dozens
of drivers.

The patch below should trigger only when transitioning from D3hot->D0
(or at boot), and only for devices that have the "no soft reset" bit
cleared in the PM control register.  I believe it is safe to include
this patch as part of the PCI infrastructure.

The cleanest implementation of pci_restore_bars was to call
pci_update_resource.  Unfortunately, that does not currently exist
for the sparc64 architecture.  The patch below includes a null
implemenation of pci_update_resource for sparc64.

Some have expressed interest in making general use of the the
pci_restore_bars function, so that has been exported to GPL licensed
modules.
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

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>

These two bits were accesses non-atomically from assembler
code.  So, in order to eliminate any potential races resulting
from that, move these pieces of state into two bytes elsewhere
in struct thread_info.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It is only used by some localized code in irq.c, and also
delete enable_prom_timer() as that is totally unused.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

arch/sparc64/kernel/smp.c:48: error: parse error before "__attribute__"
arch/sparc64/kernel/smp.c:49: error: parse error before "__attribute__"
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Also fix a bug in 32-bit syscall tracing.  We forgot to update
this code when we moved over to the convention that all 32-bit
syscall arguments are zero extended by default.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Noticed by Eddie C. Dost
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Do not cat bucket->irq_info to struct irqaction * directly,
but go through struct irq_desc *.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The following renames arch_init, a kprobes function for performing any
architecture specific initialization, to arch_init_kprobes in order to
cleanup the namespace.

Also, this patch adds arch_init_kprobes to sparc64 to fix the sparc64 kprobes
build from the last return probe patch.
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Use '5' instead of 'infinity'.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use macro instead of magic value for Tomatillo discard-
timeout interrupt enable register bit.

Leave OBP programming PTO value unless Tomatillo and
version >= 0x2.

If no-bus-parking property is present, explicitly clear
PCICTRL_PARK bit.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This was the main impetus behind adding the PCI IRQ shim.

In order to properly order DMA writes wrt. interrupts, you have to
write to a PCI controller register, then poll for that bit clearing.
There is one bit for each interrupt source, and setting this register
bit tells Tomatillo to drain all pending DMA from that device.

Furthermore, Tomatillo's with revision less than 4 require us to do a
block store due to some memory transaction ordering issues it has on
JBUS.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This allows a PCI controller to shim into IRQ delivery
so that DMA queues can be drained, if necessary.

If some bus specific code needs to run before an IRQ
handler is invoked, the bus driver simply needs to setup
the function pointer in bucket->irq_info->pre_handler and
the two args bucket->irq_info->pre_handler_arg[12].

The Schizo PCI driver is converted over to use a pre-handler
for the DMA write-sync processing it needs when a device
is behind a PCI->PCI bus deeper than the top-level APB
bridges.

While we're here, clean up all of the action allocation
and handling.  Now, we allocate the irqaction as part of
the bucket->irq_info area.  There is an array of 4 irqaction
(for PCI irq sharing) and a bitmask saying which entries
are active.

The bucket->irq_info is allocated at build_irq() time, not
at request_irq() time.  This simplifies request_irq() and
free_irq() tremendously.

The SMP dynamic IRQ retargetting code got removed in this
change too.  It was disabled for a few months now, and we
can resurrect it in the future if we want.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The only real user was the assembler floppy interrupt
handler, which does not need to be in assembly.

This makes it so that there are less pieces of code which
know about the internal layout of ivector_table[] and
friends.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In particular, avoid membar instructions in the delay
slot of a jmpl instruction.

UltraSPARC-I, II, IIi, and IIe have a bug, documented in
the UltraSPARC-IIi User's Manual, Appendix K, Erratum 51

The long and short of it is that if the IMU unit misses
on a branch or jmpl, and there is a store buffer synchronizing
membar in the delay slot, the chip can stop fetching instructions.

If interrupts are enabled or some other trap is enabled, the
chip will unwedge itself, but performance will suffer.

We already had a workaround for this bug in a few spots, but
it's better to have the entire tree sanitized for this rule.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch is based on work by Carlos O'Donell and Matthew Wilcox.  It
introduces/updates the compat_time_t type and uses it for compat siginfo
structures.  I have built this on ppc64 and x86_64.
Signed-off-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This patch includes sparc64 architecture specific changes to support temporary
disarming on reentrancy of probes.
Signed-of-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The architecture independent code of the current kprobes implementation is
arming and disarming kprobes at registration time.  The problem is that the
code is assuming that arming and disarming is a just done by a simple write
of some magic value to an address.  This is problematic for ia64 where our
instructions look more like structures, and we can not insert break points
by just doing something like:

*p->addr = BREAKPOINT_INSTRUCTION;

The following patch to 2.6.12-rc4-mm2 adds two new architecture dependent
functions:

     * void arch_arm_kprobe(struct kprobe *p)
     * void arch_disarm_kprobe(struct kprobe *p)

and then adds the new functions for each of the architectures that already
implement kprobes (spar64/ppc64/i386/x86_64).

I thought arch_[dis]arm_kprobe was the most descriptive of what was really
happening, but each of the architectures already had a disarm_kprobe()
function that was really a "disarm and do some other clean-up items as
needed when you stumble across a recursive kprobe." So...  I took the
liberty of changing the code that was calling disarm_kprobe() to call
arch_disarm_kprobe(), and then do the cleanup in the block of code dealing
with the recursive kprobe case.

So far this patch as been tested on i386, x86_64, and ppc64, but still
needs to be tested in sparc64.
Signed-off-by: NRusty Lynch <rusty.lynch@intel.com>
Signed-off-by: NAnil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Ingo recently introduced a great speedup for allocating new mmaps using the
free_area_cache pointer which boosts the specweb SSL benchmark by 4-5% and
causes huge performance increases in thread creation.

The downside of this patch is that it does lead to fragmentation in the
mmap-ed areas (visible via /proc/self/maps), such that some applications
that work fine under 2.4 kernels quickly run out of memory on any 2.6
kernel.

The problem is twofold:

  1) the free_area_cache is used to continue a search for memory where
     the last search ended.  Before the change new areas were always
     searched from the base address on.

     So now new small areas are cluttering holes of all sizes
     throughout the whole mmap-able region whereas before small holes
     tended to close holes near the base leaving holes far from the base
     large and available for larger requests.

  2) the free_area_cache also is set to the location of the last
     munmap-ed area so in scenarios where we allocate e.g.  five regions of
     1K each, then free regions 4 2 3 in this order the next request for 1K
     will be placed in the position of the old region 3, whereas before we
     appended it to the still active region 1, placing it at the location
     of the old region 2.  Before we had 1 free region of 2K, now we only
     get two free regions of 1K -> fragmentation.

The patch addresses thes issues by introducing yet another cache descriptor
cached_hole_size that contains the largest known hole size below the
current free_area_cache.  If a new request comes in the size is compared
against the cached_hole_size and if the request can be filled with a hole
below free_area_cache the search is started from the base instead.

The results look promising: Whereas 2.6.12-rc4 fragments quickly and my
(earlier posted) leakme.c test program terminates after 50000+ iterations
with 96 distinct and fragmented maps in /proc/self/maps it performs nicely
(as expected) with thread creation, Ingo's test_str02 with 20000 threads
requires 0.7s system time.

Taking out Ingo's patch (un-patch available per request) by basically
deleting all mentions of free_area_cache from the kernel and starting the
search for new memory always at the respective bases we observe: leakme
terminates successfully with 11 distinctive hardly fragmented areas in
/proc/self/maps but thread creating is gringdingly slow: 30+s(!) system
time for Ingo's test_str02 with 20000 threads.

Now - drumroll ;-) the appended patch works fine with leakme: it ends with
only 7 distinct areas in /proc/self/maps and also thread creation seems
sufficiently fast with 0.71s for 20000 threads.
Signed-off-by: NWolfgang Wander <wwc@rentec.com>
Credit-to: "Richard Purdie" <rpurdie@rpsys.net>
Signed-off-by: NKen Chen <kenneth.w.chen@intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu> (partly)
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The initial peek read PIO of the match register is just a waste.
Just do the flush writes first, as that is more efficient.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Firstly, if the direction is TODEVICE, then dirty data in the
streaming cache is impossible so we can elide the flush-flag
synchronization in that case.

Next, the context allocator is broken.  It is highly likely
that contexts get used multiple times for different dma
mappings, which confuses the strbuf flushing code and makes
it run inefficiently.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Older UltraSPARC-III chips have a P-Cache bug that makes us disable it
by default at boot time.

However, this does hurt performance substantially, particularly with
memcpy(), and the bug is _incredibly_ obscure.  I have never seen it
triggered in practice, ever.

So provide a "-P" boot option that forces the P-Cache on.  It taints
the kernel, so if it does trigger and cause some data corruption or
OOPS, we will find out in the logs that this option was on when it
happened.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The recent change to add a timeout to strbuf flushing had
a negative performance impact.  The udelay()'s are too long,
and they were done in the wrong order wrt. the register read
checks.  Fix both, and things are happy again.

There are more possible improvements in this area.  In fact,
PCI streaming buffer flushing seems to be part of the bottleneck
in network receive performance on my SunBlade1000 box.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If some hardware error occurs and the flush flag never updates,
we will hang forever in these routines.  Add a timeout, and
print out a diagnostic if it is reached.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>