Conflicts:
drivers/acpi/apei/apei-base.c
This was a conflict between
15afae6046
(CPI, APEI: Fix incorrect APEI register bit width check and usage)
and
653f4b538f
(ACPICA: Expand OSL memory read/write interfaces to 64 bits)
The former changed a parameter in the call to acpi_os_read_memory64()
and the later replaced all calls to acpi_os_read_memory64()
with calls to acpi_os_read_memory().
Signed-off-by: Len Brown <len.brown@intel.com>
The function apei_estatus_print() and apei_estatus_check() forget to move ahead
the gdata pointer when dealing with multiple generic error data sections.
Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The current code incorrectly assumes that
(1) the APEI register bit width is always 8, 16, 32, or 64 and
(2) the APEI register bit width is always equal to the APEI
register access width.
ERST serialization instructions entries such as:
[030h 0048 1] Action : 00 [Begin Write Operation]
[031h 0049 1] Instruction : 03 [Write Register Value]
[032h 0050 1] Flags (decoded below) : 01
Preserve Register Bits : 1
[033h 0051 1] Reserved : 00
[034h 0052 12] Register Region : [Generic Address Structure]
[034h 0052 1] Space ID : 00 [SystemMemory]
[035h 0053 1] Bit Width : 03
[036h 0054 1] Bit Offset : 00
[037h 0055 1] Encoded Access Width : 03 [DWord Access:32]
[038h 0056 8] Address : 000000007F2D7038
[040h 0064 8] Value : 0000000000000001
[048h 0072 8] Mask : 0000000000000007
break this assumption by yielding:
[Firmware Bug]: APEI: Invalid bit width in GAR [0x7f2d7038/3/0]
I have found no ACPI specification requirements corresponding
with the above assumptions. There is even a good example in
the Serialization Instruction Entries section (ACPI 4.0 section
17.4,1.2, ACPI 4.0a section 2.5.1.2, ACPI 5.0 section 18.5.1.2)
that mentions a serialization instruction with a bit range of
[6:2] which is 5 bits wide, _not_ 8, 16, 32, or 64 bits wide.
Compile and boot tested with 3.3.0-rc7 on a IBM HX5.
Signed-off-by: Gary Hade <garyhade@us.ibm.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some APEI firmware implementation will access injected address
specified in param1 to trigger the error when injecting memory
error, which means if one SRAR error is injected, the crash
always happens because it is executed in kernel context. This
new parameter can disable trigger action and control is taken
over by the user. In this way, an SRAR error can happen in user
context instead of crashing the system. This function is highly
depended on BIOS implementation so please ensure you know the
BIOS trigger procedure before you enable this switch.
v2:
notrigger should be created together with param1/param2
Tested-by: Tony Luck <tony.luck@lintel.com>
Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
On the platforms with ACPI4.x support, parameter extension
is not always doable, which means only parameter extension
is enabled, einj_param can take effect.
v2->v1: stopping early in einj_get_parameter_address for einj_param
Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This fixes a trivial copy & paste error in ERST header length check.
It's just for future safety because sizeof(struct acpi_table_einj)
equals to sizeof(struct acpi_table_erst) with current ACPI5.0
specification. It applies to v3.3-rc6.
Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
Acked-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This change expands acpi_os_read_memory and acpi_os_write_memory to a
full 64 bits. This allows 64 bit transfers via the acpi_read and
acpi_write interfaces. Note: The internal acpi_hw_read and acpi_hw_write
interfaces remain at 32 bits, because 64 bits is not needed to
access the standard ACPI registers.
Signed-off-by: Bob Moore <robert.moore@intel.com>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
ioremap() has become more picky and is now spitting out console messages like:
ioremap error for 0xbddbd000-0xbddbe000, requested 0x10, got 0x0
when loading the einj driver. What we are trying to so here is map
a couple of data structures that the EINJ table points to. Perhaps
acpi_os_map_memory() is a better tool for this?
Most importantly it works, but as a side benefit it maps the structures
into kernel virtual space so we can access them with normal C memory
dereferences, so instead of using:
writel(param1, &v5param->apicid);
we can use the more natural:
v5param->apicid = param1;
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This function is returning pointers. Sparse complains here:
drivers/acpi/apei/einj.c:262:32: warning:
Using plain integer as NULL pointer
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
According to the ACPI spec [1] section 18.6.4 the TRIGGER_ERROR action
table can consists of zero elements.
[1] Advanced Configuration and Power Interface Specification
Revision 5.0, December 6, 2011
http://www.acpi.info/DOWNLOADS/ACPIspec50.pdf
Signed-off-by: Niklas Söderlund <niklas.soderlund@ericsson.com>
Acked-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Base ACPI (CA) currently does not support atomic 64-bit reads and writes
(acpi_read() and acpi_write() split 64-bit loads/stores into two
32-bit transfers) yet APEI expects 64-bit transfer capability, even
when running on 32-bit systems.
This patch implements 64-bit read and write routines for APEI usage.
This patch re-factors similar functionality introduced in commit
04c25997c9, bringing it into the ACPI subsystem in preparation for
removing ./drivers/acpi/atomicio.[ch]. In the implementation I have
replicated acpi_os_read_memory() and acpi_os_write_memory(), creating
64-bit versions for APEI to utilize, as opposed to something more
elegant. My thinking is that we should attempt to see if we can get
ACPI's CA/OSL changed so that the existing acpi_read() and acpi_write()
interfaces are natively 64-bit capable and then subsequently remove the
replication.
Signed-off-by: Myron Stowe <myron.stowe@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This includes initial support for the recently published ACPI 5.0 spec.
In particular, support for the "hardware-reduced" bit that eliminates
the dependency on legacy hardware.
APEI has patches resulting from testing on real hardware.
Plus other random fixes.
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux: (52 commits)
acpi/apei/einj: Add extensions to EINJ from rev 5.0 of acpi spec
intel_idle: Split up and provide per CPU initialization func
ACPI processor: Remove unneeded variable passed by acpi_processor_hotadd_init V2
ACPI processor: Remove unneeded cpuidle_unregister_driver call
intel idle: Make idle driver more robust
intel_idle: Fix a cast to pointer from integer of different size warning in intel_idle
ACPI: kernel-parameters.txt : Add intel_idle.max_cstate
intel_idle: remove redundant local_irq_disable() call
ACPI processor: Fix error path, also remove sysdev link
ACPI: processor: fix acpi_get_cpuid for UP processor
intel_idle: fix API misuse
ACPI APEI: Convert atomicio routines
ACPI: Export interfaces for ioremapping/iounmapping ACPI registers
ACPI: Fix possible alignment issues with GAS 'address' references
ACPI, ia64: Use SRAT table rev to use 8bit or 16/32bit PXM fields (ia64)
ACPI, x86: Use SRAT table rev to use 8bit or 32bit PXM fields (x86/x86-64)
ACPI: Store SRAT table revision
ACPI, APEI, Resolve false conflict between ACPI NVS and APEI
ACPI, Record ACPI NVS regions
ACPI, APEI, EINJ, Refine the fix of resource conflict
...
ACPI 5.0 provides extensions to the EINJ mechanism to specify the
target for the error injection - by APICID for cpu related errors,
by address for memory related errors, and by segment/bus/device/function
for PCIe related errors. Also extensions for vendor specific error
injections.
Tested-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
APEI needs memory access in interrupt context. The obvious choice is
acpi_read(), but originally it couldn't be used in interrupt context
because it makes temporary mappings with ioremap(). Therefore, we added
drivers/acpi/atomicio.c, which provides:
acpi_pre_map_gar() -- ioremap in process context
acpi_atomic_read() -- memory access in interrupt context
acpi_post_unmap_gar() -- iounmap
Later we added acpi_os_map_generic_address() (2971852) and enhanced
acpi_read() so it works in interrupt context as long as the address has
been previously mapped (620242a). Now this sequence:
acpi_os_map_generic_address() -- ioremap in process context
acpi_read()/apei_read() -- now OK in interrupt context
acpi_os_unmap_generic_address()
is equivalent to what atomicio.c provides.
This patch introduces apei_read() and apei_write(), which currently are
functional equivalents of acpi_read() and acpi_write(). This is mainly
proactive, to prevent APEI breakages if acpi_read() and acpi_write()
are ever augmented to support the 'bit_offset' field of GAS, as APEI's
__apei_exec_write_register() precludes splitting up functionality
related to 'bit_offset' and APEI's 'mask' (see its
APEI_EXEC_PRESERVE_REGISTER block).
With apei_read() and apei_write() in place, usages of atomicio routines
are converted to apei_read()/apei_write() and existing calls within
osl.c and the CA, based on the re-factoring that was done in an earlier
patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2:
acpi_pre_map_gar() --> acpi_os_map_generic_address()
acpi_post_unmap_gar() --> acpi_os_unmap_generic_address()
acpi_atomic_read() --> apei_read()
acpi_atomic_write() --> apei_write()
Note that acpi_read() and acpi_write() currently use 'bit_width'
for accessing GARs which seems incorrect. 'bit_width' is the size of
the register, while 'access_width' is the size of the access the
processor must generate on the bus. The 'access_width' may be larger,
for example, if the hardware only supports 32-bit or 64-bit reads. I
wanted to minimize any possible impacts with this patch series so I
did *not* change this behavior.
Signed-off-by: Myron Stowe <myron.stowe@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some firmware will access memory in ACPI NVS region via APEI. That
is, instructions in APEI ERST/EINJ table will read/write ACPI NVS
region. The original resource conflict checking in APEI code will
check memory/ioport accessed by APEI via general resource management
mech. But ACPI NVS region is marked as busy already, so that the
false resource conflict will prevent APEI ERST/EINJ to work.
To fix this, this patch excludes ACPI NVS regions when APEI components
request resources. So that they will not conflict with ACPI NVS
regions.
Reported-and-tested-by: Pavel Ivanov <paivanof@gmail.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Current fix for resource conflict is to remove the address region <param1 &
param2, ~param2+1> from trigger resource, which is highly relies on valid user
input. This patch is trying to avoid such potential issues by fetching the
exact address region from trigger action table entry.
Signed-off-by: Xiao, Hui <hui.xiao@linux.intel.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some APEI firmware implementation will access injected address
specified in param1 to trigger the error when injecting memory error.
This will cause resource conflict with RAM.
On one of our testing machine, if injecting at memory address
0x10000000, the following error will be reported in dmesg:
APEI: Can not request iomem region <0000000010000000-0000000010000008> for GARs.
This patch removes the injecting memory address range from trigger
table resources to avoid conflict.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Tested-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Because printk is not safe inside NMI handler, the recoverable error
records received in NMI handler will be queued to be printked in a
delayed IRQ context via irq_work. If a fatal error occurs after the
recoverable error and before the irq_work processed, we lost a error
report.
To solve the issue, the queued error records are printked in NMI
handler if system will go panic.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
In most cases, printk only guarantees messages from different printk
calling will not be interleaved between each other. But, one APEI
GHES hardware error report will involve multiple printk calling,
normally each for one line. So it is possible that the hardware error
report comes from different generic hardware error source will be
interleaved.
In this patch, a sequence number is prefixed to each line of error
report. So that, even if they are interleaved, they still can be
distinguished by the prefixed sequence number.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Because APEI tables are optional, these message may confuse users, for
example,
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/599715
Reported-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Use the normal %pR-like format for MMIO and I/O port ranges.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
aer_recover_queue() is called when recoverable PCIe AER errors are
notified by firmware to do the recovery work.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
module_param(bool) used to counter-intuitively take an int. In
fddd5201 (mid-2009) we allowed bool or int/unsigned int using a messy
trick.
It's time to remove the int/unsigned int option. For this version
it'll simply give a warning, but it'll break next kernel version.
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This allows a backend to filter on the dmesg reason as well as the pstore
reason. When ramoops is switched to pstore, this is needed since it has
no interest in storing non-crash dmesg details.
Drop pstore_write() as it has no users, and handling the "reason" here
has no obviously correct value.
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
The buf_lock cannot be held while populating the inodes, so make the backend
pass forward an allocated and filled buffer instead. This solves the following
backtrace. The effect is that "buf" is only ever used to notify the backends
that something was written to it, and shouldn't be used in the read path.
To replace the buf_lock during the read path, isolate the open/read/close
loop with a separate mutex to maintain serialized access to the backend.
Note that is is up to the pstore backend to cope if the (*write)() path is
called in the middle of the read path.
[ 59.691019] BUG: sleeping function called from invalid context at .../mm/slub.c:847
[ 59.691019] in_atomic(): 0, irqs_disabled(): 1, pid: 1819, name: mount
[ 59.691019] Pid: 1819, comm: mount Not tainted 3.0.8 #1
[ 59.691019] Call Trace:
[ 59.691019] [<810252d5>] __might_sleep+0xc3/0xca
[ 59.691019] [<810a26e6>] kmem_cache_alloc+0x32/0xf3
[ 59.691019] [<810b53ac>] ? __d_lookup_rcu+0x6f/0xf4
[ 59.691019] [<810b68b1>] alloc_inode+0x2a/0x64
[ 59.691019] [<810b6903>] new_inode+0x18/0x43
[ 59.691019] [<81142447>] pstore_get_inode.isra.1+0x11/0x98
[ 59.691019] [<81142623>] pstore_mkfile+0xae/0x26f
[ 59.691019] [<810a2a66>] ? kmem_cache_free+0x19/0xb1
[ 59.691019] [<8116c821>] ? ida_get_new_above+0x140/0x158
[ 59.691019] [<811708ea>] ? __init_rwsem+0x1e/0x2c
[ 59.691019] [<810b67e8>] ? inode_init_always+0x111/0x1b0
[ 59.691019] [<8102127e>] ? should_resched+0xd/0x27
[ 59.691019] [<8137977f>] ? _cond_resched+0xd/0x21
[ 59.691019] [<81142abf>] pstore_get_records+0x52/0xa7
[ 59.691019] [<8114254b>] pstore_fill_super+0x7d/0x91
[ 59.691019] [<810a7ff5>] mount_single+0x46/0x82
[ 59.691019] [<8114231a>] pstore_mount+0x15/0x17
[ 59.691019] [<811424ce>] ? pstore_get_inode.isra.1+0x98/0x98
[ 59.691019] [<810a8199>] mount_fs+0x5a/0x12d
[ 59.691019] [<810b9174>] ? alloc_vfsmnt+0xa4/0x14a
[ 59.691019] [<810b9474>] vfs_kern_mount+0x4f/0x7d
[ 59.691019] [<810b9d7e>] do_kern_mount+0x34/0xb2
[ 59.691019] [<810bb15f>] do_mount+0x5fc/0x64a
[ 59.691019] [<810912fb>] ? strndup_user+0x2e/0x3f
[ 59.691019] [<810bb3cb>] sys_mount+0x66/0x99
[ 59.691019] [<8137b537>] sysenter_do_call+0x12/0x26
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
* 'pstore' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux:
pstore: make pstore write function return normal success/fail value
pstore: change mutex locking to spin_locks
pstore: defer inserting OOPS entries into pstore
* 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (46 commits)
llist: Add back llist_add_batch() and llist_del_first() prototypes
sched: Don't use tasklist_lock for debug prints
sched: Warn on rt throttling
sched: Unify the ->cpus_allowed mask copy
sched: Wrap scheduler p->cpus_allowed access
sched: Request for idle balance during nohz idle load balance
sched: Use resched IPI to kick off the nohz idle balance
sched: Fix idle_cpu()
llist: Remove cpu_relax() usage in cmpxchg loops
sched: Convert to struct llist
llist: Add llist_next()
irq_work: Use llist in the struct irq_work logic
llist: Return whether list is empty before adding in llist_add()
llist: Move cpu_relax() to after the cmpxchg()
llist: Remove the platform-dependent NMI checks
llist: Make some llist functions inline
sched, tracing: Show PREEMPT_ACTIVE state in trace_sched_switch
sched: Remove redundant test in check_preempt_tick()
sched: Add documentation for bandwidth control
sched: Return unused runtime on group dequeue
...
Currently pstore write interface employs record id as return
value, but it is not enough because it can't tell caller if
the write operation is successful. Pass the record id back via
an argument pointer and return zero for success, non-zero for
failure.
Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Just convert all the files that have an nmi handler to the new routines.
Most of it is straight forward conversion. A couple of places needed some
tweaking like kgdb which separates the debug notifier from the nmi handler
and mce removes a call to notify_die.
[Thanks to Ying for finding out the history behind that mce call
https://lkml.org/lkml/2010/5/27/114
And Boris responding that he would like to remove that call because of it
https://lkml.org/lkml/2011/9/21/163]
The things that get converted are the registeration/unregistration routines
and the nmi handler itself has its args changed along with code removal
to check which list it is on (most are on one NMI list except for kgdb
which has both an NMI routine and an NMI Unknown routine).
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Corey Minyard <minyard@acm.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Robert Richter <robert.richter@amd.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Corey Minyard <minyard@acm.org>
Cc: Jack Steiner <steiner@sgi.com>
Link: http://lkml.kernel.org/r/1317409584-23662-4-git-send-email-dzickus@redhat.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Because llist code will be used in performance critical scheduler
code path, make llist_add() and llist_del_all() inline to avoid
function calling overhead and related 'glue' overhead.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1315461646-1379-2-git-send-email-ying.huang@intel.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
pstore was using mutex locking to protect read/write access to the
backend plug-ins. This causes problems when pstore is executed in
an NMI context through panic() -> kmsg_dump().
This patch changes the mutex to a spin_lock_irqsave then also checks to
see if we are in an NMI context. If we are in an NMI and can't get the
lock, just print a message stating that and blow by the locking.
All this is probably a hack around the bigger locking problem but it
solves my current situation of trying to sleep in an NMI context.
Tested by loading the lkdtm module and executing a HARDLOCKUP which
will cause the machine to panic inside the nmi handler.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Acked-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
IRQ_WORK is used by GHES, but it is selected by PERF_EVENT.
For now PERF_EVENT is selected by x86 by default, but
in concept, IRQ_WORK should be selected by GHES, not by others.
Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Bit 0 of the support parameter to the OSC call should be set in order to
indicate that the OS supports the WHEA mechanism. Stuart Hayes tracked
an APEI issue on some Dell platforms down to this.
Reported-by: Stuart Hayes <Stuart_Hayes@Dell.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some trivial conflicts due to other various merges
adding to the end of common lists sooner than this one.
arch/ia64/Kconfig
arch/powerpc/Kconfig
arch/x86/Kconfig
lib/Kconfig
lib/Makefile
Signed-off-by: Len Brown <len.brown@intel.com>
EINJ parameter support is only usable for some specific BIOS.
Originally, it is expected to have no harm for BIOS does not support
it. But now, we found it will cause issue (memory overwriting) for
some BIOS. So param support is disabled by default and only enabled
when newly added module parameter named "param_extension" is
explicitly specified.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Cc: Matthew Garrett <mjg@redhat.com>
Acked-by: Don Zickus <dzickus@redhat.com>
Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
drivers/acpi/apei/ghes.c:542: warning: integer overflow in expression
drivers/acpi/apei/ghes.c:619: warning: integer overflow in expression
ghes.c:(.text+0x46289): undefined reference to `__udivdi3'
in function ghes_estatus_cache_add().
Reported-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Len Brown <len.brown@intel.com>
memory_failure_queue() is called when recoverable memory errors are
notified by firmware to do the recovery work.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
printk is used by GHES to report hardware errors. Ratelimit is
enforced on the printk to avoid too many hardware error reports in
kernel log. Because there may be thousands or even millions of
corrected hardware errors during system running.
Currently, a simple scheme is used. That is, the total number of
hardware error reporting is ratelimited. This may cause some issues
in practice.
For example, there are two kinds of hardware errors occurred in
system. One is corrected memory error, because the fault memory
address is accessed frequently, there may be hundreds error report
per-second. The other is corrected PCIe AER error, it will be
reported once per-second. Because they share one ratelimit control
structure, it is highly possible that only memory error is reported.
To avoid the above issue, an error record content based throttle
algorithm is implemented in the patch. Where after the first
successful reporting, all error records that are same are throttled for
some time, to let other kinds of error records have the opportunity to
be reported.
In above example, the memory errors will be throttled for some time,
after being printked. Then the PCIe AER error will be printked
successfully.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some APEI GHES recoverable errors are reported via NMI, but printk is
not safe in NMI context.
To solve the issue, a lock-less memory allocator is used to allocate
memory in NMI handler, save the error record into the allocated
memory, put the error record into a lock-less list. On the other
hand, an irq_work is used to delay the operation from NMI context to
IRQ context. The irq_work IRQ handler will remove nodes from
lock-less list, printk the error record and do some further processing
include recovery operation, then free the memory.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
We'll never have a negative part, so just make this an unsigned int.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
EFI only provides small amounts of individual storage, and conventionally
puts metadata in the storage variable name. Rather than add a metadata
header to the (already limited) variable storage, it's easier for us to
modify pstore to pass all the information we need to construct a unique
variable name to the appropriate functions.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Some pstore implementations may not have a static context, so extend the
API to pass the pstore_info struct to all calls and allow for a context
pointer.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
APEI firmware first mode must be turned on explicitly on some
machines, otherwise there may be no GHES hardware error record for
hardware error notification. APEI bit in generic _OSC call can be
used to do that, but on some machine, a special WHEA _OSC call must be
used. This patch adds the support to that WHEA _OSC call.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some machine may have broken firmware so that GHES and firmware first
mode should be disabled. This patch adds support to that.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
GHES (Generic Hardware Error Source) is used to process hardware error
notification in firmware first mode. But because firmware first mode
can be turned on but can not be turned off, it is unreasonable to
unload the GHES module with firmware first mode turned on. To avoid
confusion, this patch makes GHES can be enabled/disabled in
configuration time, but not built as module and unloaded at run time.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This patch changes APEI EINJ and ERST to use apei_exec_run for
mandatory actions, and apei_exec_run_optional for optional actions.
Cc: Thomas Renninger <trenn@novell.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some actions in APEI ERST and EINJ tables are optional, for example,
ACPI_EINJ_BEGIN_OPERATION action is used to do some preparation for
error injection, and firmware may choose to do nothing here. While
some other actions are mandatory, for example, firmware must provide
ACPI_EINJ_GET_ERROR_TYPE implementation.
Original implementation treats all actions as optional (that is, can
have no instructions), that may cause issue if firmware does not
provide some mandatory actions. To fix this, this patch adds
apei_exec_run_optional, which should be used for optional actions.
The original apei_exec_run should be used for mandatory actions.
Cc: Thomas Renninger <trenn@novell.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
printk is used by GHES to report hardware errors. Normally, the
printk will be ratelimited to avoid too many hardware error reports in
kernel log. Because there may be thousands or even millions of
corrected hardware errors during system running.
That is different for fatal hardware error, because system will go
panic as soon as possible, there will be no more than several error
records. And these error records are valuable for system fault
diagnosis, so they should not be ratelimited.
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
When we debug ERST table with erst-dbg, if the error record in ERST
table is too long(>4K), it can't be read out. So this patch increases
the buffer size to 16K to ensure such error records can be read from
ERST table.
Signed-off-by: Chen Gong <gong.chen@linux.intel.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Len Brown
Jiang Liu
Gary Hade
Chen Gong
Bob Moore
Luck, Tony
Dan Carpenter
Niklas Söderlund
Myron Stowe
Linus Torvalds
Huang Ying
Xiao, Hui
Bjorn Helgaas
Rusty Russell
Kees Cook
Don Zickus
Matthew Garrett