commit 351fc4a99d49fde63fe5ab7412beb35c40d27269 upstream.
Intel IA32 SDM Table 15-14 defines channel 0xf as 'not specified', but
EDAC doesn't know about this and returns and INTERNAL ERROR when the
channel is greater than NUM_CHANNELS:
kernel: [ 1538.886456] CPU 0: Machine Check Exception: 0 Bank 1: 940000000000009f
kernel: [ 1538.886669] TSC 2bc68b22e7e812 ADDR 46dae7000 MISC 0 PROCESSOR 0:306e4 TIME 1390414572 SOCKET 0 APIC 0
kernel: [ 1538.971948] EDAC MC1: INTERNAL ERROR: channel value is out of range (15 >= 4)
kernel: [ 1538.972203] EDAC MC1: 0 CE memory read error on unknown memory (slot:0 page:0x46dae7 offset:0x0 grain:0 syndrome:0x0 - area:DRAM err_code:0000:009f socket:1 channel_mask:1 rank:0)
This commit changes sb_edac to forward a channel of -1 to EDAC if the
channel is not specified. edac_mc_handle_error() sets the channel to -1
internally after the error message anyway, so this commit should have no
effect other than avoiding the INTERNAL ERROR message when the channel
is not specified.
Signed-off-by: Seth Jennings <sjenning@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Cc: Vinson Lee <vlee@twopensource.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 75135da0d68419ef8a925f4c1d5f63d8046e314d upstream.
pci_get_device() decrements the reference count of "from" (last
argument) so when we break off the loop successfully we have only one
device reference - and we don't know which device we have. If we want
a reference to each device, we must take them explicitly and let
the pci_get_device() walk complete to avoid duplicate references.
This is serious, as over-putting device references will cause
the device to eventually disappear. Without this fix, the kernel
crashes after a few insmod/rmmod cycles.
Tested on an Intel S7000FC4UR system with a 7300 chipset.
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Link: http://lkml.kernel.org/r/20140224111656.09bbb7ed@endymion.delvare
Cc: Mauro Carvalho Chehab <m.chehab@samsung.com>
Cc: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c0f5eeed0f4cef4f05b74883a7160e7edde58b6a upstream.
The reference count changes done by pci_get_device can be a little
misleading when the usage diverges from the most common scheme. The
reference count of the device passed as the last parameter is always
decreased, even if the function returns no new device. So if we are
going to try alternative device IDs, we must manually increment the
device reference count before each retry. If we don't, we end up
decreasing the reference count, and after a few modprobe/rmmod cycles
the PCI devices will vanish.
In other words and as Alan put it: without this fix the EDAC code
corrupts the PCI device list.
This fixes kernel bug #50491:
https://bugzilla.kernel.org/show_bug.cgi?id=50491
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Link: http://lkml.kernel.org/r/20140224093927.7659dd9d@endymion.delvare
Reviewed-by: Alan Cox <alan@linux.intel.com>
Cc: Mauro Carvalho Chehab <m.chehab@samsung.com>
Cc: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit cb6ef42e516cb8948f15e4b70dc03af8020050a2 upstream.
We're using edac_mc_workq_setup() both on the init path, when
we load an edac driver and when we change the polling period
(edac_mc_reset_delay_period) through /sys/.../edac_mc_poll_msec.
On that second path we don't need to init the workqueue which has been
initialized already.
Thanks to Tejun for workqueue insights.
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: http://lkml.kernel.org/r/1391457913-881-1-git-send-email-prarit@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c542b53da9ffa4fe9de61149818a06aacae531f8 upstream.
The usage of strict_strtol() is not preferred, because strict_strtol()
is obsolete. Thus, kstrtol() should be used.
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 90ed4988b8c030d65b41b7d13140e9376dc6ec5a upstream.
In case the device 0, function 1 is not found using pci_get_device(),
pci_scan_single_device() will be used but, differently than
pci_get_device(), it allocates a pci_dev but doesn't does bump the usage
count on the pci_dev and after few module removals and loads the pci_dev
will be freed.
Signed-off-by: Aristeu Rozanski <aris@redhat.com>
Reviewed-by: mark gross <mark.gross@intel.com>
Link: http://lkml.kernel.org/r/20131205153755.GL4545@redhat.com
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a72b8859fd3941cc1d2940d5c43026d2c6fb959e upstream.
Register and enable interrupts after the edac registration. Otherwise
incomming ecc error interrupts lead to crashes during device setup.
Fixing this in drivers for mc and l2.
Signed-off-by: Robert Richter <robert.richter@linaro.org>
Acked-by: Rob Herring <rob.herring@calxeda.com>
Signed-off-by: Robert Richter <rric@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit f0a56c480196a98479760862468cc95879df3de0 upstream.
It can happen that configurations are running in a single-channel mode
even with a dual-channel memory controller, by, say, putting the DIMMs
only on the one channel and leaving the other empty. This causes a
problem in init_csrows which implicitly assumes that when the second
channel is enabled, i.e. channel 1, the struct dimm hierarchy will be
present. Which is not.
So always allocate two channels unconditionally.
This provides for the nice side effect that the data structures are
initialized so some day, when memory hotplug is supported, it should
just work out of the box when all of a sudden a second channel appears.
Reported-and-tested-by: Roger Leigh <rleigh@debian.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix yet another issue caught by 8f46baaa7e ("base: core: WARN() about
bogus permissions on device attributes").
Signed-off-by: Borislav Petkov <bp@suse.de>
Pull two small EDAC fixes from Borislav Petkov.
* tag 'edac_fixes_for_3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
EDAC: Don't give write permission to read-only files
EDAC, mc_sysfs.c: Fix string array pointer types
I get the following warning on boot:
------------[ cut here ]------------
WARNING: at drivers/base/core.c:575 device_create_file+0x9a/0xa0()
Hardware name: -[8737R2A]-
Write permission without 'store'
...
</snip>
Drilling down, this is related to dynamic channel ce_count attribute
files sporting a S_IWUSR mode without a ->store() function. Looking
around, it appears that they aren't supposed to have a ->store()
function. So remove the bogus write permission to get rid of the
warning.
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>
Cc: <stable@vger.kernel.org> # 3.[89]
[ shorten commit message ]
Signed-off-by: Borislav Petkov <bp@suse.de>
Pull edac fixes from Mauro Carvalho Chehab:
"Two edac fixes:
- i7300_edac currently reports a wrong number of DIMMs when the
memory controller is in single channel mode
- on some Sandy Bridge machines, the EDAC driver bails out as one of
the PCI IDs used by the driver is hidden by BIOS. As the driver
uses it only to detect the type of memory, make it optional at the
driver"
* 'linux_next' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edac:
edac: sb_edac.c should not require prescence of IMC_DDRIO device
i7300_edac: Fix memory detection in single mode
The Sandy Bridge EDAC driver uses a register in the IMC_DDRIO CSR
space to determine the type of DIMMs (registered or unregistered).
But this device does not exist on some single socket Sandy Bridge
servers. While the type of DIMMs is nice to know, it is not essential
for this driver's other functions. So it seems harsh to have it
refuse to load at all when it cannot find this device.
Make the check for this device be optional. If it isn't present
just report the memory type as "MEM_UNKNOWN".
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Add code to handle DRAM ECC errors decoding for Fam16h.
Tested on Fam16h with ECC turned on using the mce_amd_inj facility and
works fine.
Signed-off-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@amd.com>
[ Boris: cleanups and clarifications ]
Signed-off-by: Borislav Petkov <bp@suse.de>
Both mci.mem_is_per_rank and mci.csbased denote the same thing: the
memory controller is csrows based. Merge both fields into one.
There's no need for the driver to actually fill it, as the core detects
it by checking if one of the layers has the csrows type as part of the
memory hierarchy:
if (layers[i].type == EDAC_MC_LAYER_CHIP_SELECT)
per_rank = true;
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
We were filling the csrow size with a wrong value. 16a528ee39 ("EDAC:
Fix csrow size reported in sysfs") tried to address the issue. It fixed
the report with the old API but not with the new one. Correct it for the
new API too.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
[ make it a per-csrow accounting regardless of ->channel_count ]
Signed-off-by: Borislav Petkov <bp@suse.de>
Pull EDAC fixes and ghes-edac from Mauro Carvalho Chehab:
"For:
- Some fixes at edac drivers (i7core_edac, sb_edac, i3200_edac);
- error injection support for i5100, when EDAC debug is enabled;
- fix edac when it is loaded builtin (early init for the subsystem);
- a "Firmware First" EDAC driver, allowing ghes to report errors via
EDAC (ghes-edac).
With regards to ghes-edac, this fixes a longstanding BZ at Red Hat
that happens with Nehalem and Sandy Bridge CPUs: when both GHES and
i7core_edac or sb_edac are running, the error reports are
unpredictable, as both BIOS and OS race to access the registers. With
ghes-edac, the EDAC core will refuse to register any other concurrent
memory error driver.
This patchset moves the ghes struct definitions to a separate header
file (include/acpi/ghes.h) and adds 3 hooks at apei/ghes.c to
register/unregister and to report errors via ghes-edac. Those changes
were acked by ghes driver maintainer (Huang)."
* 'linux_next' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edac: (30 commits)
i5100_edac: convert to use simple_open()
ghes_edac: fix to use list_for_each_entry_safe() when delete list items
ghes_edac: Fix RAS tracing
ghes_edac: Make it compliant with UEFI spec 2.3.1
ghes_edac: Improve driver's printk messages
ghes_edac: Don't credit the same memory dimm twice
ghes_edac: do a better job of filling EDAC DIMM info
ghes_edac: add support for reporting errors via EDAC
ghes_edac: Register at EDAC core the BIOS report
ghes: add the needed hooks for EDAC error report
ghes: move structures/enum to a header file
edac: add support for error type "Info"
edac: add support for raw error reports
edac: reduce stack pressure by using a pre-allocated buffer
edac: lock module owner to avoid error report conflicts
edac: remove proc_name from mci structure
edac: add a new memory layer type
edac: initialize the core earlier
edac: better report error conditions in debug mode
i5100_edac: Remove two checkpatch warnings
...
This removes an open coded simple_open() function and
replaces file operations references to the function
with simple_open() instead.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Since we will remove items off the list using list_del() we need
to use a safe version of the list_for_each_entry() macro aptly named
list_for_each_entry_safe().
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
With the current version of CPER, there's no way to associate an
error with the memory error. So, the error location in EDAC
layers is unused.
As CPER has its own idea about memory architectural layers, just
output whatever is there inside the driver's detail at the RAS
tracepoint.
The EDAC location keeps untouched, in the case that, in some future,
we could actually map the error into the dimm labels.
Now, the error message:
[ 72.396625] {1}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 0
[ 72.396627] {1}[Hardware Error]: APEI generic hardware error status
[ 72.396628] {1}[Hardware Error]: severity: 2, corrected
[ 72.396630] {1}[Hardware Error]: section: 0, severity: 2, corrected
[ 72.396632] {1}[Hardware Error]: flags: 0x01
[ 72.396634] {1}[Hardware Error]: primary
[ 72.396635] {1}[Hardware Error]: section_type: memory error
[ 72.396637] {1}[Hardware Error]: error_status: 0x0000000000000400
[ 72.396638] {1}[Hardware Error]: node: 3
[ 72.396639] {1}[Hardware Error]: card: 0
[ 72.396640] {1}[Hardware Error]: module: 0
[ 72.396641] {1}[Hardware Error]: device: 0
[ 72.396643] {1}[Hardware Error]: error_type: 18, unknown
[ 72.396666] EDAC MC0: 1 CE reserved error (18) on unknown label (node:3 card:0 module:0 page:0x0 offset:0x0 grain:0 syndrome:0x0 - status(0x0000000000000400): Storage error in DRAM memory)
Is properly represented on the trace event:
kworker/0:2-584 [000] .... 72.396657: mc_event: 1 Corrected error: reserved error (18) on unknown label (mc:0 location👎-1:-1 address:0x00000000 grain:1 syndrome:0x00000000 APEI location: node:3 card:0 module:0 status(0x0000000000000400): Storage error in DRAM memory)
Tested on a 4 sockets E5-4650 Sandy Bridge machine.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Provide a better infrastructure for printk's inside the driver:
- use edac_dbg() for debug messages;
- standardize the usage of pr_info();
- provide warning about the risk of relying on this
driver.
While here, changes the size of a fake memory to 1 page. This is
as good or as bad as 1000 pages, but it is easier for userspace to
detect, as I don't expect that any machine implementing GHES would
provide just 1 page available ;)
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Conflicts:
drivers/edac/ghes_edac.c
Instead of just faking a random value for the DIMM data, get
the information that it is available via DMI table.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Now that the EDAC core is capable of just forward the errors via
the userspace API, add a report mechanism for the GHES errors.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Register GHES at EDAC MC core, in order to avoid other
drivers to also handle errors and mangle with error data.
The edac core will warrant that just one driver will be used,
so the first one to register (BIOS first) will be the one that
will be reporting the hardware errors.
For now, the EDAC driver does nothing but to register at the
EDAC core, preventing the hardware-driven mechanism to
interfere with GHES.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Pull driver core patches from Greg Kroah-Hartman:
"Here is the big driver core merge for 3.9-rc1
There are two major series here, both of which touch lots of drivers
all over the kernel, and will cause you some merge conflicts:
- add a new function called devm_ioremap_resource() to properly be
able to check return values.
- remove CONFIG_EXPERIMENTAL
Other than those patches, there's not much here, some minor fixes and
updates"
Fix up trivial conflicts
* tag 'driver-core-3.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core: (221 commits)
base: memory: fix soft/hard_offline_page permissions
drivercore: Fix ordering between deferred_probe and exiting initcalls
backlight: fix class_find_device() arguments
TTY: mark tty_get_device call with the proper const values
driver-core: constify data for class_find_device()
firmware: Ignore abort check when no user-helper is used
firmware: Reduce ifdef CONFIG_FW_LOADER_USER_HELPER
firmware: Make user-mode helper optional
firmware: Refactoring for splitting user-mode helper code
Driver core: treat unregistered bus_types as having no devices
watchdog: Convert to devm_ioremap_resource()
thermal: Convert to devm_ioremap_resource()
spi: Convert to devm_ioremap_resource()
power: Convert to devm_ioremap_resource()
mtd: Convert to devm_ioremap_resource()
mmc: Convert to devm_ioremap_resource()
mfd: Convert to devm_ioremap_resource()
media: Convert to devm_ioremap_resource()
iommu: Convert to devm_ioremap_resource()
drm: Convert to devm_ioremap_resource()
...
The number of variables at the stack is too big.
Reduces the stack usage by using a pre-allocated error
buffer.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
APEI GHES and i7core_edac/sb_edac currently can be loaded at
the same time, but those are Highlander modules:
"There can be only one".
There are two reasons for that:
1) Each driver assumes that it is the only one registering at
the EDAC core, as it is driver's responsibility to number
the memory controllers, and all of them start from 0;
2) If BIOS is handling the memory errors, the OS can't also be
doing it, as one will mangle with the other.
So, we need to add an module owner's lock at the EDAC core,
in order to avoid having two different modules handling memory
errors at the same time. The best way for doing this lock seems
to use the driver's name, as this is unique, and won't require
changes on every driver.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
There are some cases where the memory controller layout is
completely hidden. This is the case of firmware-driven error
code, like the one provided by GHES. Add a new layer to be
used on such memory error report mechanisms.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
In order for it to work with it builtin, the EDAC core should
be initialized earlier, otherwise the ghes_edac driver initializes
before edac_mc_sysfs_init() being called:
...
[ 4.998373] EDAC MC0: Giving out device to 'ghes_edac.c' 'ghes_edac': DEV ghes
...
[ 4.998373] EDAC MC1: Giving out device to 'ghes_edac.c' 'ghes_edac': DEV ghes
[ 6.519495] EDAC MC: Ver: 3.0.0
[ 6.523749] EDAC DEBUG: edac_mc_sysfs_init: device mc created
The net result is that no EDAC sysfs nodes will appear.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The last changeset introduced a few checkpatch warnings:
WARNING: debugfs_remove_recursive(NULL) is safe this check is probably not required
261: FILE: drivers/edac/i5100_edac.c:1207:
+ if (priv->debugfs)
+ debugfs_remove_recursive(priv->debugfs);
WARNING: debugfs_remove(NULL) is safe this check is probably not required
290: FILE: drivers/edac/i5100_edac.c:1250:
+ if (i5100_debugfs)
+ debugfs_remove(i5100_debugfs);
Get rid of them.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Emmanouil Maroudas