As pointed out by Linus CONFIG_X86 in drivers/acpi/numa.c is
ugly.
Builds and boots on ia64 (both normally and with maxcpus=8 to limit
the number of cpus).
Signed-off-by: Tony Luck <tony.luck@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Len Brown <len.brown@intel.com>
LKML-Reference: <4D2D6B5D.4080208@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86: Fix APIC ID sizing bug on larger systems, clean up MAX_APICS confusion
x86, acpi: Parse all SRAT cpu entries even above the cpu number limitation
x86, acpi: Add MAX_LOCAL_APIC for 32bit
x86: io_apic: Split setup_ioapic_ids_from_mpc()
x86: io_apic: Fix CONFIG_X86_IO_APIC=n breakage
x86: apic: Move probe_nr_irqs_gsi() into ioapic_init_mappings()
x86: Allow platforms to force enable apic
GPEs with corresponding _Lxx/_Exx control methods need to be disabled
during initialization in case they have been enabled by the BIOS, so
that they don't fire up until they are enabled by acpi_update_gpes().
References: https://bugzilla.kernel.org/show_bug.cgi?id=25412
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
Recent Intel new system have different order in MADT, aka will list all thread0
at first, then all thread1.
But SRAT table still old order, it will list cpus in one socket all together.
If the user have compiled limited NR_CPUS or boot with nr_cpus=, could have missed
to put some cpus apic id to node mapping into apicid_to_node[].
for example for 4 sockets system with 64 cpus with nr_cpus=32 will get crash...
[ 9.106288] Total of 32 processors activated (136190.88 BogoMIPS).
[ 9.235021] divide error: 0000 [#1] SMP
[ 9.235315] last sysfs file:
[ 9.235481] CPU 1
[ 9.235592] Modules linked in:
[ 9.245398]
[ 9.245478] Pid: 2, comm: kthreadd Not tainted 2.6.37-rc1-tip-yh-01782-ge92ef79-dirty #274 /Sun Fire x4800
[ 9.265415] RIP: 0010:[<ffffffff81075a8f>] [<ffffffff81075a8f>] select_task_rq_fair+0x4f0/0x623
...
[ 9.645938] RIP [<ffffffff81075a8f>] select_task_rq_fair+0x4f0/0x623
[ 9.665356] RSP <ffff88103f8d1c40>
[ 9.665568] ---[ end trace 2296156d35fdfc87 ]---
So let just parse all cpu entries in SRAT.
Also add apicid checking with MAX_LOCAL_APIC, in case We could out of boundaries of
apicid_to_node[].
it fixes following bug too.
https://bugzilla.kernel.org/show_bug.cgi?id=22662
-v2: expand to 32bit according to hpa
need to add MAX_LOCAL_APIC for 32bit
Reported-and-Tested-by: Wu Fengguang <fengguang.wu@intel.com>
Reported-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Tested-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4D0AD486.9020704@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
If a device is reported as inactive or not present by its _STA
control method, acpi_bus_check_add() skips it without evaluating its
_PRW method. This leads to a problem when the device's _PRW method
points to a GPE, because in that case the GPE may be enabled by
ACPICA during the subsequent acpi_update_gpes() call which, in
turn, may cause a GPE storm to appear.
To avoid this issue, make acpi_bus_check_add() evaluate _PRW for
inactive or not present devices and register the wakeup GPE
information returned by them, so that acpi_update_gpes() does not
enable their GPEs unnecessarily.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The saving of the NVS memory area during suspend and restoring it
during resume causes problems to appear on Sony Vaio VGN-NW130D, so
blacklist that machine to avoid those problems.
Addresses https://bugzilla.kernel.org/show_bug.cgi?id=23002
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Adriano <adriano.vilela@yahoo.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Properly const-, __init-, and __read_mostly-annotate this code.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Len Brown <len.brown@intel.com>
When CONFIG_CPU_FREQ=n ...
drivers/acpi/processor_thermal.c:159:12: warning: ‘acpi_thermal_cpufreq_increase’ defined but not used
drivers/acpi/processor_thermal.c:163:12: warning: ‘acpi_thermal_cpufreq_decrease’ defined but not used
Remove unused declaration of ‘acpi_thermal_cpufreq_increase’ and
‘acpi_thermal_cpufreq_decrease’
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
WARNING: drivers/acpi/acpi.o(.text+0xeda): Section mismatch in reference from the function acpi_os_initialize1() to the function .init.text:set_osi_linux()
The function acpi_os_initialize1() references
the function __init set_osi_linux().
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
ERST writing may be used in NMI or Machine Check Exception handler. So
it need to use raw spinlock instead of normal spinlock. This patch
fixes it.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
commit b0ed7a91(ACPICA/ACPI: Add new host interfaces for _OSI suppor)
introduced another regression that only one _OSI string can be added or
removed.
Now multiple _OSI strings can be added or removed, for example
acpi_osi=Linux acpi_osi=FreeBSD acpi_osi="!Windows 2006"
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
commit b0ed7a91(ACPICA/ACPI: Add new host interfaces for _OSI suppor)
introduced a regression that _OSI string setup fails.
There are 2 paths to setup _OSI string.
DMI:
acpi_dmi_osi_linux -> set_osi_linux -> acpi_osi_setup -> copy _OSI
string to osi_setup_string
Boot command line:
acpi_osi_setup -> copy _OSI string to osi_setup_string
Later, acpi_osi_setup_late will be called to handle osi_setup_string.
If _OSI string is "Linux" or "!Linux", then the call path is,
acpi_osi_setup_late -> acpi_cmdline_osi_linux -> set_osi_linux ->
acpi_osi_setup -> copy _OSI string to osi_setup_string
This actually never installs _OSI string(acpi_install_interface not
called), but just copy the _OSI string to osi_setup_string.
This patch fixes the regression.
Reported-and-tested-by: Lukas Hejtmanek <xhejtman@ics.muni.cz>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Sometimes the Battery driver doesn't get notifications when it's
plugged/unplugged. And this results in the incorrect Battery
status reported by the power supply sysfs I/F.
Update Battery status first when querying from sysfs.
http://marc.info/?l=linux-acpi&m=128855015826728&w=2
Tested_by: Seblu <seblu@seblu.net>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
If turning on a power resource fails, do not reference count it,
since it cannot be in use in that case.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
Commit 3e384ee6c6 (ACPI / PM: Fix
reference counting of power resources) introduced a regression by
causing fan power resources to be turned on and reference counted
unnecessarily during resume, so on some boxes fans are always on
after resume.
Fix the problem by checking if the current device state is different
from the new state before reference counting and turning on power
resources in acpi_power_transition().
Addresses https://bugzilla.kernel.org/show_bug.cgi?id=22932 .
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Maciej Rutecki <maciej.rutecki@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Now that the bulk of the old nmi_watchdog is gone, remove all
the stub variables and hooks associated with it.
This touches lots of files mainly because of how the io_apic
nmi_watchdog was implemented. Now that the io_apic nmi_watchdog
is forever gone, remove all its fingers.
Most of this code was not being exercised by virtue of
nmi_watchdog != NMI_IO_APIC, so there shouldn't be anything to
risky here.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Cc: fweisbec@gmail.com
Cc: gorcunov@openvz.org
LKML-Reference: <1289578944-28564-3-git-send-email-dzickus@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently we have:
--w--w--w-. 1 root root 0 2010-11-11 14:56 /sys/kernel/debug/acpi/custom_method
which is just crazy. Change this to --w-------.
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Cc: stable@kernel.org (for 2.6.36)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (53 commits)
ACPI: install ACPI table handler before any dynamic tables being loaded
ACPI / PM: Blacklist another machine that needs acpi_sleep=nonvs
ACPI: Page based coalescing of I/O remappings optimization
ACPI: Convert simple locking to RCU based locking
ACPI: Pre-map 'system event' related register blocks
ACPI: Add interfaces for ioremapping/iounmapping ACPI registers
ACPI: Maintain a list of ACPI memory mapped I/O remappings
ACPI: Fix ioremap size for MMIO reads and writes
ACPI / Battery: Return -ENODEV for unknown values in get_property()
ACPI / PM: Fix reference counting of power resources
Subject: [PATCH] ACPICA: Fix Scope() op in module level code
ACPI battery: support percentage battery remaining capacity
ACPI: Make Embedded Controller command timeout delay configurable
ACPI dock: move some functions to .init.text
ACPI: thermal: remove unused limit code
ACPI: static sleep_states[] and acpi_gts_bfs_check
ACPI: remove dead code
ACPI: delete dedicated MAINTAINERS entries for ACPI EC and BATTERY drivers
ACPI: Only processor needs CPU_IDLE
ACPICA: Update version to 20101013
...
* 'idle-release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-idle-2.6:
intel_idle: do not use the LAPIC timer for ATOM C2
intel_idle: add initial Sandy Bridge support
acpi_idle: delete bogus data from cpuidle_state.power_usage
intel_idle: delete bogus data from cpuidle_state.power_usage
intel_idle: simplify test for leave_mm()
ACPI table sysfs I/F is broken by commit
78f1699659
Author: Alex Chiang <achiang@hp.com>
Date: Sun Dec 20 12:19:09 2009 -0700
ACPI: processor: call _PDC early
because dynamic SSDT tables may be loaded in _PDC,
before installing the ACPI table handler.
As a result, the sysfs I/F of these dynamic tables are
located at /sys/firmware/acpi/tables instead of
/sys/firmware/acpi/tables/dynamic, which is not true.
Invoke acpi_sysfs_init() before acpi_early_processor_set_pdc(),
so that the table handler is installed before any dynamic tables loaded.
https://bugzilla.kernel.org/show_bug.cgi?id=21142
CC: Dennis Jansen <dennis.jansen@web.de>
CC: Alex Chiang <achiang@hp.com>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Sony Vaio VPCEB1Z1E is reported to require acpi_sleep=nonvs for
suspend/resume to work on it correctly, so blacklist it.
Reported-by: Emanuele Bigiarini <pulmro@gmail.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
This patch optimizes ACPI MMIO remappings by keeping track of the
remappings on a PAGE_SIZE granularity.
When an ioremap() occurs, the underlying infrastructure works on a 'page'
based granularity. As such, an ioremap() request for 1 byte for example,
will end up mapping in an entire (PAGE_SIZE) page. Huang Ying took
advantage of this in commit 15651291a2 by
checking if subsequent ioremap() requests reside within any of the list's
existing remappings still in place, and if so, incrementing a reference
count on the existing mapping as opposed to performing another ioremap().
Signed-off-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Convert the simple locking introduced earlier for the ACPI MMIO
remappings list to an RCU based locking scheme.
Signed-off-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
During ACPI initialization, pre-map fixed hardware registers that are
accessed during ACPI's 'system event' related IRQ handing.
ACPI's 'system event' handing accesses specific fixed hardware
registers; namely PM1a event, PM1b event, GPE0, and GPE1 register
blocks which are declared within the FADT. If these registers are
backed by MMIO, as opposed to I/O port space, accessing them within
interrupt context will cause a panic as acpi_os_read_memory()
depends on ioremap() in such cases - BZ 18012.
By utilizing the functionality provided in the previous two patches -
ACPI: Maintain a list of ACPI memory mapped I/O remappings, and, ACPI:
Add interfaces for ioremapping/iounmapping ACPI registers - accesses
to ACPI MMIO areas will now be safe from within interrupt contexts (IRQ
and/or NMI) provided the area was pre-mapped. This solves BZ 18012.
ACPI "System Event" reference(s):
ACPI Specification, Revision 4.0, Section 3 "ACPI Overview",
3.8 "System Events", 5.6 "ACPI Event Programming Model".
Reference: https://bugzilla.kernel.org/show_bug.cgi?id=18012
Reported-by: <bjorn.helgaas@hp.com>
Signed-off-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Add remapping and unmapping interfaces for ACPI registers that are
backed by memory mapped I/O (MMIO). These interfaces, along with
the MMIO remapping list, enable accesses of such registers from within
interrupt context.
ACPI Generic Address Structure (GAS) reference (ACPI's fixed/generic
hardware registers use the GAS format):
ACPI Specification, Revision 4.0, Section 5.2.3.1, "Generic Address
Structure".
Signed-off-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
For memory mapped I/O (MMIO) remappings, add a list to maintain the
remappings and augment the corresponding mapping and unmapping interface
routines (acpi_os_map_memory() and acpi_os_unmap_memory()) to
dynamically add to, and delete from, the list.
The current ACPI I/O accessing methods - acpi_read() and acpi_write() -
end up calling ioremap() when accessing MMIO. This prevents use of these
methods within interrupt context (IRQ and/or NMI), since ioremap() may
block to allocate memory. Maintaining a list of MMIO remappings enables
accesses to such areas from within interrupt context provided they have
been pre-mapped.
Signed-off-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The size used for I/O remapping MMIO read and write accesses has not
accounted for the basis of ACPI's Generic Address Structure (GAS)
'Register Bit Width' field which is bits, not bytes. This patch
adjusts the ioremap() 'size' argument accordingly.
ACPI "Generic Register" reference:
ACPI Specification, Revision 4.0, Section 5.2.3.1, "Generic Address
Structure".
Signed-off-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The function acpi_battery_get_property() is called by the
power supply framework's function power_supply_show_property()
implementing the sysfs interface for power supply devices as the
ACPI battery driver's ->get_property() callback. Thus it is supposed
to return error code if the value of the given property is unknown.
Unfortunately, however, it returns 0 in those cases and puts a
wrong (negative) value into the intval field of the
union power_supply_propval object provided by
power_supply_show_property(). In consequence, wrong negative
values are read by user space from the battery's sysfs files.
Fix this by making acpi_battery_get_property() return -ENODEV
for properties with unknown values (-ENODEV is returned, because
power_supply_uevent() returns with error for any other error code
returned by power_supply_show_property()).
Reported-and-tested-by: Sitsofe Wheeler <sitsofe@yahoo.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
The reference counting of ACPI power resources is currently broken
for a few reasons. First, instead of using a simple reference
counter per power resource it uses a list of objects representing
refereces to the given power resource from devices. This leads to
the second breakage, because it prevents power resources from
being referenced more than once by one device, which is necessary
if the device is configured to signal wakeup. Namely, when putting
the device into a low power state we first call
acpi_enable_wakeup_device_power() that should reference count power
resources needed for signaling wakeup and then we call
acpi_power_transition() to power off the device. The latter call
drops references to the device's power resources, possibly including
the ones added by acpi_enable_wakeup_device_power(), so the device
can't signal wakeup as a result. Apart from this, the locking
in acpi_power_on() and acpi_power_off_device() doesn't prevent
all possible races from happening, which may be problematic for
runtime PM and asynchronous suspend and resume.
Fix the problem by using a counter for power resources reference
counting and putting the evaluation of ACPI _ON and _OFF methods
under the power resource mutex.
Reported-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
