linux-toradex.git/arch/powerpc/sysdev/xics, branch v3.4.73

powerpc: Make sure IPI handlers see data written by IPI senders

2012-09-14T17:00:22+00:00

commit 9fb1b36ca1234e64a5d1cc573175303395e3354d upstream.

We have been observing hangs, both of KVM guest vcpu tasks and more
generally, where a process that is woken doesn't properly wake up and
continue to run, but instead sticks in TASK_WAKING state.  This
happens because the update of rq->wake_list in ttwu_queue_remote()
is not ordered with the update of ipi_message in
smp_muxed_ipi_message_pass(), and the reading of rq->wake_list in
scheduler_ipi() is not ordered with the reading of ipi_message in
smp_ipi_demux().  Thus it is possible for the IPI receiver not to see
the updated rq->wake_list and therefore conclude that there is nothing
for it to do.

In order to make sure that anything done before smp_send_reschedule()
is ordered before anything done in the resulting call to scheduler_ipi(),
this adds barriers in smp_muxed_message_pass() and smp_ipi_demux().
The barrier in smp_muxed_message_pass() is a full barrier to ensure that
there is a full ordering between the smp_send_reschedule() caller and
scheduler_ipi().  In smp_ipi_demux(), we use xchg() rather than
xchg_local() because xchg() includes release and acquire barriers.
Using xchg() rather than xchg_local() makes sense given that
ipi_message is not just accessed locally.

This moves the barrier between setting the message and calling the
cause_ipi() function into the individual cause_ipi implementations.
Most of them -- those that used outb, out_8 or similar -- already had
a full barrier because out_8 etc. include a sync before the MMIO
store.  This adds an explicit barrier in the two remaining cases.

These changes made no measurable difference to the speed of IPIs as
measured using a simple ping-pong latency test across two CPUs on
different cores of a POWER7 machine.

The analysis of the reason why processes were not waking up properly
is due to Milton Miller.

Reported-by: Milton Miller 
Signed-off-by: Paul Mackerras 
Signed-off-by: Benjamin Herrenschmidt 
Signed-off-by: Greg Kroah-Hartman

powerpc/irqdomain: Fix broken NR_IRQ references

2012-04-30T00:45:26+00:00

The switch from using irq_map to irq_alloc_desc*() for managing irq
number allocations introduced new bugs in some of the powerpc
interrupt code.  Several functions rely on the value of NR_IRQS to
determine the maximum irq number that could get allocated.  However,
with sparse_irq and using irq_alloc_desc*() the maximum possible irq
number is now specified with 'nr_irqs' which may be a number larger
than NR_IRQS.  This has caused breakage on powermac when
CONFIG_NR_IRQS is set to 32.

This patch removes most of the direct references to NR_IRQS in the
powerpc code and replaces them with either a nr_irqs reference or by
using the common for_each_irq_desc() macro.  The powerpc-specific
for_each_irq() macro is removed at the same time.

Also, the Cell axon_msi driver is refactored to remove the global
build assumption on the size of NR_IRQS and instead add a limit to the
maximum irq number when calling irq_domain_add_nomap().

Signed-off-by: Grant Likely 
Signed-off-by: Benjamin Herrenschmidt

irq_domain: Replace irq_alloc_host() with revmap-specific initializers

2012-02-16T13:11:22+00:00

Each revmap type has different arguments for setting up the revmap.
This patch splits up the generator functions so that each revmap type
can do its own setup and the user doesn't need to keep track of how
each revmap type handles the arguments.

This patch also adds a host_data argument to the generators.  There are
cases where the host_data pointer will be needed before the function returns.
ie. the legacy map calls the .map callback for each irq before returning.

v2: - Add void *host_data argument to irq_domain_add_*() functions
    - fixed failure to compile
    - Moved IRQ_DOMAIN_MAP_* defines into irqdomain.c

Signed-off-by: Grant Likely 
Cc: Rob Herring 
Cc: Benjamin Herrenschmidt 
Cc: Thomas Gleixner 
Cc: Milton Miller 
Tested-by: Olof Johansson

irq_domain/powerpc: Eliminate virq_is_host()

2012-02-16T08:36:46+00:00

There is only one user, and it is trivial to open-code.

Signed-off-by: Grant Likely 
Cc: Benjamin Herrenschmidt 
Cc: Thomas Gleixner 
Cc: Milton Miller 
Tested-by: Olof Johansson

irq_domain/powerpc: Use common irq_domain structure instead of irq_host

2012-02-14T21:06:50+00:00

This patch drops the powerpc-specific irq_host structures and uses the common
irq_domain strucutres defined in linux/irqdomain.h.  It also fixes all
the users to use the new structure names.

Renaming irq_host to irq_domain has been discussed for a long time, and this
patch is a step in the process of generalizing the powerpc virq code to be
usable by all architecture.

An astute reader will notice that this patch actually removes the irq_host
structure instead of renaming it.  This is because the irq_domain structure
already exists in include/linux/irqdomain.h and has the needed data members.

Signed-off-by: Grant Likely 
Cc: Benjamin Herrenschmidt 
Cc: Thomas Gleixner 
Cc: Milton Miller 
Tested-by: Olof Johansson

powerpc/xics: Reset the CPPR if H_EOI fails

2011-12-08T03:02:23+00:00

I have an intermittent kdump fail where the hypervisor fails an H_EOI.
As a result our CPPR is never reset to 0xff and we no longer accept
interrupts.

This patch calls icp_hv_set_cppr to reset the CPPR if H_EOI fails,
fixing the kdump fail.

Signed-off-by: Anton Blanchard 
Signed-off-by: Benjamin Herrenschmidt

powerpc/xics: Harden xics hypervisor backend

2011-11-28T00:42:09+00:00

During kdump stress testing I sometimes see the kdump kernel panic
with:

  Interrupt 0x306 (real) is invalid, disabling it.
  Kernel panic - not syncing: bad return code EOI - rc = -4, value=ff000306

Instead of panicing print the error message, dump the stack the first
time it happens and continue on. Add some more information to the
debug messages as well.

Signed-off-by: Anton Blanchard 
Signed-off-by: Benjamin Herrenschmidt

powerpc: Mark IPI interrupts IRQF_NO_THREAD

2011-11-25T03:14:38+00:00

IPI handlers cannot be threaded. Remove the obsolete IRQF_DISABLED
flag (see commit e58aa3d2) while at it.

Signed-off-by: Thomas Gleixner 
Signed-off-by: Benjamin Herrenschmidt

powerpc/irq: Remove IRQF_DISABLED

2011-11-08T03:51:46+00:00

Since commit [e58aa3d2: genirq: Run irq handlers with interrupts disabled],
We run all interrupt handlers with interrupts disabled
and we even check and yell when an interrupt handler
returns with interrupts enabled (see commit [b738a50a:
genirq: Warn when handler enables interrupts]).

So now this flag is a NOOP and can be removed.

Signed-off-by: Yong Zhang 
Acked-by: Arnd Bergmann 
Acked-by: Geoff Levand 
Signed-off-by: Benjamin Herrenschmidt

powerpc/powernv: Add OPAL ICS backend

2011-09-20T06:09:59+00:00

OPAL handles HW access to the various ICS or equivalent chips
for us (with the exception of p5ioc2 based HEA which uses a

different backend) similarily to what RTAS does on pSeries.

Signed-off-by: Benjamin Herrenschmidt