On the Xeon 55XX series cpus the pci deives are not exposed via acpi so
we much explicitly probe them to make the usable as a Linux PCI device.
This moves the detection of this state to before pci_register_driver is
called. Its present position was not working on my systems, the driver
would complain about not finding a specific device.
This patch allows the driver to load on my systems.
Signed-off-by: Keith Mannthey <kmannth@us.ibm.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Instead of assuming that the entire machine has either registered or
unregistered memories, do it at CPU socket based.
While here, fix a bug at i7core_mce_output_error(), where the we're
using m->cpu directly as if it would represent a socket. Instead, the
proper socket_id is given by cpu_data[m->cpu].phys_proc_id.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
---
Nehalem and upper chipsets provide an special device that has corrected memory
error counters detected with registered dimms. This device is only seen if
there are registered memories plugged.
After this patch, on a machine fully equiped with RDIMM's, it will use the
Device 3 function 2 to count corrected errors instead on relying at mcelog.
For unregistered DIMMs, it will keep the old behavior, counting errors
via mcelog.
This patch were developed together with Keith Mannthey <kmannth@us.ibm.com>
Signed-off-by: Keith Mannthey <kmannth@us.ibm.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
From: Keith Mannthey <kmannth@us.ibm.com>
Simple correction to a shift value.
ECC_ENABLED is bit 4 of MC_STATUS, Dev 3 Fun 0 Offset 0x4c
This correctly identifies the state of the ECC at the machine.
Signed-off-by: Keith Mannthey <kmannth@us.ibm.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There were two stupid error injection bugs introduced by wrong
cut-and-paste: one at socket store, and another at the error inject
register. The last one were causing the code to not work at all.
While here, adds debug messages to allow seeing what registers are being
set while sending error injection.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
i7core_get_devices() were preparet to get just the first found device of each type.
Due to that, on Xeon 55xx, only socket 1 were retrived.
Rework i7core_get_devices() to clean it and to properly support Xeon 55xx.
While here, fix a small typo.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Xeon55xx fails to probe with this error message:
EDAC DEBUG: in drivers/edac/i7core_edac.c, line at 1660: MC: drivers/edac/i7core_edac.c: i7core_init()
EDAC i7core: Device not found: dev 00:00.0 PCI ID 8086:2c41
i7core_edac: probe of 0000:00:14.0 failed with error -22
This is due to the fact that, on Xeon35xx (and i7core), device 00.0 has
PCI ID 8086:2c40.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
m->bank is not related to the memory bank but, instead, to the MCA Error
register bank. Fix it accordingly. While here, improves the comments for
Nehalem bank.
A later fix is needed, in order to get bank/rank information from MCA
error log.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Enriches mcelog error by using the encoded information at MCE status and
misc registers (IA32_MCx_STATUS, IA32_MCx_MISC).
Some fixes are still needed here, in order to properly fill the EDAC
fields.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Some Nehalem architectures have more than one MC socket. Socket 0 is
located at bus 255.
Currently, it is using up to 2 sockets, but increasing it to a larger
number is just a matter of increasing MAX_SOCKETS definition.
This seems to be required for properly support of Xeon 55xx.
Still needs testing with Xeon 55xx.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This code changes the detection procedure of i7core_edac. Instead of
directly probing for MC registers, it probes for another register found
on Nehalem. If found, it tries to pick the first MC PCI BUS. This should
work fine with Xeon 35xx, but, on Xeon 55xx, this is at bus 254 and 255
that are not properly detected by the non-legacy PCI methods.
The new detection code scans specifically at buses 254 and 255 for the
Xeon 55xx devices.
This code has not tested yet. After working, a change at the code will
be needed, since the i7core is not yet ready for working with 2 sets of
MC.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The public Intel Xeon 5500 volume 2 datasheet describes, on page 53,
session 2.6.7 a register that can lock/unlock Memory Controller the
configuration register, called MC_CFG_CONTROL.
Adds support for it in the hope that software error injection would
work. With my tests with Xeon 35xx, there's still something missing.
With a program that does sequencial bit writes at dev 0.0, sometimes, it
produces error injection, after unblocking the MC_CFG_CONTROL (and,
sometimes, it just locks my testing machine).
I'll try later to discover by trial and error what's the register that
solves this issue on Xeon 35xx.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Adds a glue code to allow i7core to work with mcelog. With the glue,
i7core registers itself on edac_mce. At mce, when an error is detected,
it calls all registered drivers (in this case, i7core), for EDAC error
handling.
TODO: It currently just prints the MCE error log using about the same
format as mce panic messages. The error message should be enhanced
with mcelog userspace info and converted into the proper EDAC format,
to feed the EDAC error counts.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
edac_mce module is an interface module that gets mcelog data and
forwards to any registered edac module that expects to receive data via
mce.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This function appears only on Xeon 5500 datasheet. Yet, testing with a
Xeon 3503 showed that this is also implemented on other Nehalem
processors.
At the first read, MC_TEST_ERR_RCV1 and MC_TEST_ERR_RCV0 can contain any
value. Modify CE error logic to update the error count only after the
second read.
An alternative approach would be to do a write at rcv0 and rcv1
registers, but it seemed better to keep they untouched, since BIOS might
eventually assume that they are exclusive for their usage.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There are some locking troubles with edac_core: if you don't declare an
edac_check, module may suffer from soft lock.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Implements set_inject_error() with the low-level code needed to inject
memory errors at Nehalem, and adds some sysfs nodes to allow error injection
The next patch will add an API for error injection.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This driver is meant to support i7 core/i7core extreme desktop
processors and Xeon 35xx/55xx series with integrated memory controller.
It is likely that it can be expanded in the future to work with other
processor series based at the same Memory Controller design.
For now, it has just a few MCH status reads.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Print the CPU associated with the error only when the field is valid.
Cc: <stable@kernel.org> # .32.x .33.x
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Add support to scrub DRAM using the e752x integrated memory scrubbing
engine. The e7320/7520/e7525 chipsets support scrubbing at one rate while
the i3100 chipset supports a normal and fast rate.
A similar patch was originally sent back in 2008:
http://sourceforge.net/mailarchive/forum.php?thread_name=1204835866.25206.70.camel@localhost.localdomain&forum_name=bluesmoke-devel
This version has the following updates:
- Use 16-bit PCI config cycles to access MCHSCRB register
e7320/7520/e7525 docs say register is 16bits wide, i3100 says 8. I
tested 16bits on the i3100 to be safe.
- Recalcuate and round actual scrub rates
The changes have been tested on an i3100-based board.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
FSB parity is only supported on the Xeon processor. Previously it was
incorrectly enabled for the Celeron as well.
Signed-off-by: Konstantin Olifer <kolifer@gmail.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Use resource_size() instead of arithmetic.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Acked-by: Dave Jiang <djiang@mvista.com>
Cc: Peter Tyser <ptyser@xes-inc.com>
Cc: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add the ability to detect the specific data line or ECC line which failed
when printing out SDRAM single-bit errors. An example of a single-bit
SDRAM ECC error is below:
EDAC MPC85xx MC1: Err Detect Register: 0x80000004
EDAC MPC85xx MC1: Faulty data bit: 59
EDAC MPC85xx MC1: Expected Data / ECC: 0x7f80d000_409effa0 / 0x6d
EDAC MPC85xx MC1: Captured Data / ECC: 0x7780d000_409effa0 / 0x6d
EDAC MPC85xx MC1: Err addr: 0x00031ca0
EDAC MPC85xx MC1: PFN: 0x00000031
Knowning which specific data or ECC line caused an error can be useful in
tracking down hardware issues such as improperly terminated signals, loose
pins, etc.
Note that this feature is only currently enabled for 64-bit wide data
buses, 32-bit wide bus support should be added.
I don't have any 32-bit wide systems to test on. If someone has one and
is willing to give this patch a shot with the check for a 64-bit data bus
removed it would be much appreciated and I can re-submit with both 32 and
64 bit buses supported.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Kumar Gala <galak@gate.crashing.org>
Cc: Dave Jiang <djiang@mvista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With a 64-bit wide data bus only the lowest 8-bits of the ECC syndrome are
relevant. With a 32-bit wide data bus only the lowest 16-bits are
relevant on most architectures.
Without this change, the ECC syndrome displayed can be mildly confusing,
eg:
EDAC MPC85xx MC1: syndrome: 0x25252525
When in reality the ECC syndrome is 0x25.
A variety of Freescale manuals say a variety of different things about how
to decode the CAPTURE_ECC (syndrome) register. I don't have a system with
a 32-bit bus to test on, but I believe the change is correct. It'd be
good to get an ACK from someone at Freescale about this change though.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Doug Thompson <dougthompson@xmission.com>
Cc: Kumar Gala <galak@gate.crashing.org>
Cc: Dave Jiang <djiang@mvista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
