1.. include:: <isonum.txt>
2
3============================================
4Reliability, Availability and Serviceability
5============================================
6
7RAS concepts
8************
9
10Reliability, Availability and Serviceability (RAS) is a concept used on
11servers meant to measure their robustness.
12
13Reliability
14  is the probability that a system will produce correct outputs.
15
16  * Generally measured as Mean Time Between Failures (MTBF)
17  * Enhanced by features that help to avoid, detect and repair hardware faults
18
19Availability
20  is the probability that a system is operational at a given time
21
22  * Generally measured as a percentage of downtime per a period of time
23  * Often uses mechanisms to detect and correct hardware faults in
24    runtime;
25
26Serviceability (or maintainability)
27  is the simplicity and speed with which a system can be repaired or
28  maintained
29
30  * Generally measured on Mean Time Between Repair (MTBR)
31
32Improving RAS
33-------------
34
35In order to reduce systems downtime, a system should be capable of detecting
36hardware errors, and, when possible correcting them in runtime. It should
37also provide mechanisms to detect hardware degradation, in order to warn
38the system administrator to take the action of replacing a component before
39it causes data loss or system downtime.
40
41Among the monitoring measures, the most usual ones include:
42
43* CPU – detect errors at instruction execution and at L1/L2/L3 caches;
44* Memory – add error correction logic (ECC) to detect and correct errors;
45* I/O – add CRC checksums for transferred data;
46* Storage – RAID, journal file systems, checksums,
47  Self-Monitoring, Analysis and Reporting Technology (SMART).
48
49By monitoring the number of occurrences of error detections, it is possible
50to identify if the probability of hardware errors is increasing, and, on such
51case, do a preventive maintenance to replace a degraded component while
52those errors are correctable.
53
54Types of errors
55---------------
56
57Most mechanisms used on modern systems use use technologies like Hamming
58Codes that allow error correction when the number of errors on a bit packet
59is below a threshold. If the number of errors is above, those mechanisms
60can indicate with a high degree of confidence that an error happened, but
61they can't correct.
62
63Also, sometimes an error occur on a component that it is not used. For
64example, a part of the memory that it is not currently allocated.
65
66That defines some categories of errors:
67
68* **Correctable Error (CE)** - the error detection mechanism detected and
69  corrected the error. Such errors are usually not fatal, although some
70  Kernel mechanisms allow the system administrator to consider them as fatal.
71
72* **Uncorrected Error (UE)** - the amount of errors happened above the error
73  correction threshold, and the system was unable to auto-correct.
74
75* **Fatal Error** - when an UE error happens on a critical component of the
76  system (for example, a piece of the Kernel got corrupted by an UE), the
77  only reliable way to avoid data corruption is to hang or reboot the machine.
78
79* **Non-fatal Error** - when an UE error happens on an unused component,
80  like a CPU in power down state or an unused memory bank, the system may
81  still run, eventually replacing the affected hardware by a hot spare,
82  if available.
83
84  Also, when an error happens on a userspace process, it is also possible to
85  kill such process and let userspace restart it.
86
87The mechanism for handling non-fatal errors is usually complex and may
88require the help of some userspace application, in order to apply the
89policy desired by the system administrator.
90
91Identifying a bad hardware component
92------------------------------------
93
94Just detecting a hardware flaw is usually not enough, as the system needs
95to pinpoint to the minimal replaceable unit (MRU) that should be exchanged
96to make the hardware reliable again.
97
98So, it requires not only error logging facilities, but also mechanisms that
99will translate the error message to the silkscreen or component label for
100the MRU.
101
102Typically, it is very complex for memory, as modern CPUs interlace memory
103from different memory modules, in order to provide a better performance. The
104DMI BIOS usually have a list of memory module labels, with can be obtained
105using the ``dmidecode`` tool. For example, on a desktop machine, it shows::
106
107	Memory Device
108		Total Width: 64 bits
109		Data Width: 64 bits
110		Size: 16384 MB
111		Form Factor: SODIMM
112		Set: None
113		Locator: ChannelA-DIMM0
114		Bank Locator: BANK 0
115		Type: DDR4
116		Type Detail: Synchronous
117		Speed: 2133 MHz
118		Rank: 2
119		Configured Clock Speed: 2133 MHz
120
121On the above example, a DDR4 SO-DIMM memory module is located at the
122system's memory labeled as "BANK 0", as given by the *bank locator* field.
123Please notice that, on such system, the *total width* is equal to the
124*data width*. It means that such memory module doesn't have error
125detection/correction mechanisms.
126
127Unfortunately, not all systems use the same field to specify the memory
128bank. On this example, from an older server, ``dmidecode`` shows::
129
130	Memory Device
131		Array Handle: 0x1000
132		Error Information Handle: Not Provided
133		Total Width: 72 bits
134		Data Width: 64 bits
135		Size: 8192 MB
136		Form Factor: DIMM
137		Set: 1
138		Locator: DIMM_A1
139		Bank Locator: Not Specified
140		Type: DDR3
141		Type Detail: Synchronous Registered (Buffered)
142		Speed: 1600 MHz
143		Rank: 2
144		Configured Clock Speed: 1600 MHz
145
146There, the DDR3 RDIMM memory module is located at the system's memory labeled
147as "DIMM_A1", as given by the *locator* field. Please notice that this
148memory module has 64 bits of *data width* and 72 bits of *total width*. So,
149it has 8 extra bits to be used by error detection and correction mechanisms.
150Such kind of memory is called Error-correcting code memory (ECC memory).
151
152To make things even worse, it is not uncommon that systems with different
153labels on their system's board to use exactly the same BIOS, meaning that
154the labels provided by the BIOS won't match the real ones.
155
156ECC memory
157----------
158
159As mentioned on the previous section, ECC memory has extra bits to be
160used for error correction. So, on 64 bit systems, a memory module
161has 64 bits of *data width*, and 74 bits of *total width*. So, there are
1628 bits extra bits to be used for the error detection and correction
163mechanisms. Those extra bits are called *syndrome*\ [#f1]_\ [#f2]_.
164
165So, when the cpu requests the memory controller to write a word with
166*data width*, the memory controller calculates the *syndrome* in real time,
167using Hamming code, or some other error correction code, like SECDED+,
168producing a code with *total width* size. Such code is then written
169on the memory modules.
170
171At read, the *total width* bits code is converted back, using the same
172ECC code used on write, producing a word with *data width* and a *syndrome*.
173The word with *data width* is sent to the CPU, even when errors happen.
174
175The memory controller also looks at the *syndrome* in order to check if
176there was an error, and if the ECC code was able to fix such error.
177If the error was corrected, a Corrected Error (CE) happened. If not, an
178Uncorrected Error (UE) happened.
179
180The information about the CE/UE errors is stored on some special registers
181at the memory controller and can be accessed by reading such registers,
182either by BIOS, by some special CPUs or by Linux EDAC driver. On x86 64
183bit CPUs, such errors can also be retrieved via the Machine Check
184Architecture (MCA)\ [#f3]_.
185
186.. [#f1] Please notice that several memory controllers allow operation on a
187  mode called "Lock-Step", where it groups two memory modules together,
188  doing 128-bit reads/writes. That gives 16 bits for error correction, with
189  significantly improves the error correction mechanism, at the expense
190  that, when an error happens, there's no way to know what memory module is
191  to blame. So, it has to blame both memory modules.
192
193.. [#f2] Some memory controllers also allow using memory in mirror mode.
194  On such mode, the same data is written to two memory modules. At read,
195  the system checks both memory modules, in order to check if both provide
196  identical data. On such configuration, when an error happens, there's no
197  way to know what memory module is to blame. So, it has to blame both
198  memory modules (or 4 memory modules, if the system is also on Lock-step
199  mode).
200
201.. [#f3] For more details about the Machine Check Architecture (MCA),
202  please read Documentation/x86/x86_64/machinecheck at the Kernel tree.
203
204EDAC - Error Detection And Correction
205*************************************
206
207.. note::
208
209   "bluesmoke" was the name for this device driver subsystem when it
210   was "out-of-tree" and maintained at http://bluesmoke.sourceforge.net.
211   That site is mostly archaic now and can be used only for historical
212   purposes.
213
214   When the subsystem was pushed upstream for the first time, on
215   Kernel 2.6.16, for the first time, it was renamed to ``EDAC``.
216
217Purpose
218-------
219
220The ``edac`` kernel module's goal is to detect and report hardware errors
221that occur within the computer system running under linux.
222
223Memory
224------
225
226Memory Correctable Errors (CE) and Uncorrectable Errors (UE) are the
227primary errors being harvested. These types of errors are harvested by
228the ``edac_mc`` device.
229
230Detecting CE events, then harvesting those events and reporting them,
231**can** but must not necessarily be a predictor of future UE events. With
232CE events only, the system can and will continue to operate as no data
233has been damaged yet.
234
235However, preventive maintenance and proactive part replacement of memory
236modules exhibiting CEs can reduce the likelihood of the dreaded UE events
237and system panics.
238
239Other hardware elements
240-----------------------
241
242A new feature for EDAC, the ``edac_device`` class of device, was added in
243the 2.6.23 version of the kernel.
244
245This new device type allows for non-memory type of ECC hardware detectors
246to have their states harvested and presented to userspace via the sysfs
247interface.
248
249Some architectures have ECC detectors for L1, L2 and L3 caches,
250along with DMA engines, fabric switches, main data path switches,
251interconnections, and various other hardware data paths. If the hardware
252reports it, then a edac_device device probably can be constructed to
253harvest and present that to userspace.
254
255
256PCI bus scanning
257----------------
258
259In addition, PCI devices are scanned for PCI Bus Parity and SERR Errors
260in order to determine if errors are occurring during data transfers.
261
262The presence of PCI Parity errors must be examined with a grain of salt.
263There are several add-in adapters that do **not** follow the PCI specification
264with regards to Parity generation and reporting. The specification says
265the vendor should tie the parity status bits to 0 if they do not intend
266to generate parity.  Some vendors do not do this, and thus the parity bit
267can "float" giving false positives.
268
269There is a PCI device attribute located in sysfs that is checked by
270the EDAC PCI scanning code. If that attribute is set, PCI parity/error
271scanning is skipped for that device. The attribute is::
272
273	broken_parity_status
274
275and is located in ``/sys/devices/pci<XXX>/0000:XX:YY.Z`` directories for
276PCI devices.
277
278
279Versioning
280----------
281
282EDAC is composed of a "core" module (``edac_core.ko``) and several Memory
283Controller (MC) driver modules. On a given system, the CORE is loaded
284and one MC driver will be loaded. Both the CORE and the MC driver (or
285``edac_device`` driver) have individual versions that reflect current
286release level of their respective modules.
287
288Thus, to "report" on what version a system is running, one must report
289both the CORE's and the MC driver's versions.
290
291
292Loading
293-------
294
295If ``edac`` was statically linked with the kernel then no loading
296is necessary. If ``edac`` was built as modules then simply modprobe
297the ``edac`` pieces that you need. You should be able to modprobe
298hardware-specific modules and have the dependencies load the necessary
299core modules.
300
301Example::
302
303	$ modprobe amd76x_edac
304
305loads both the ``amd76x_edac.ko`` memory controller module and the
306``edac_mc.ko`` core module.
307
308
309Sysfs interface
310---------------
311
312EDAC presents a ``sysfs`` interface for control and reporting purposes. It
313lives in the /sys/devices/system/edac directory.
314
315Within this directory there currently reside 2 components:
316
317	======= ==============================
318	mc	memory controller(s) system
319	pci	PCI control and status system
320	======= ==============================
321
322
323
324Memory Controller (mc) Model
325----------------------------
326
327Each ``mc`` device controls a set of memory modules [#f4]_. These modules
328are laid out in a Chip-Select Row (``csrowX``) and Channel table (``chX``).
329There can be multiple csrows and multiple channels.
330
331.. [#f4] Nowadays, the term DIMM (Dual In-line Memory Module) is widely
332  used to refer to a memory module, although there are other memory
333  packaging alternatives, like SO-DIMM, SIMM, etc. Along this document,
334  and inside the EDAC system, the term "dimm" is used for all memory
335  modules, even when they use a different kind of packaging.
336
337Memory controllers allow for several csrows, with 8 csrows being a
338typical value. Yet, the actual number of csrows depends on the layout of
339a given motherboard, memory controller and memory module characteristics.
340
341Dual channels allow for dual data length (e. g. 128 bits, on 64 bit systems)
342data transfers to/from the CPU from/to memory. Some newer chipsets allow
343for more than 2 channels, like Fully Buffered DIMMs (FB-DIMMs) memory
344controllers. The following example will assume 2 channels:
345
346	+------------+-----------------------+
347	| CS Rows    |       Channels        |
348	+------------+-----------+-----------+
349	|            |  ``ch0``  |  ``ch1``  |
350	+============+===========+===========+
351	| ``csrow0`` |  DIMM_A0  |  DIMM_B0  |
352	+------------+           |           |
353	| ``csrow1`` |           |           |
354	+------------+-----------+-----------+
355	| ``csrow2`` |  DIMM_A1  | DIMM_B1   |
356	+------------+           |           |
357	| ``csrow3`` |           |           |
358	+------------+-----------+-----------+
359
360In the above example, there are 4 physical slots on the motherboard
361for memory DIMMs:
362
363	+---------+---------+
364	| DIMM_A0 | DIMM_B0 |
365	+---------+---------+
366	| DIMM_A1 | DIMM_B1 |
367	+---------+---------+
368
369Labels for these slots are usually silk-screened on the motherboard.
370Slots labeled ``A`` are channel 0 in this example. Slots labeled ``B`` are
371channel 1. Notice that there are two csrows possible on a physical DIMM.
372These csrows are allocated their csrow assignment based on the slot into
373which the memory DIMM is placed. Thus, when 1 DIMM is placed in each
374Channel, the csrows cross both DIMMs.
375
376Memory DIMMs come single or dual "ranked". A rank is a populated csrow.
377Thus, 2 single ranked DIMMs, placed in slots DIMM_A0 and DIMM_B0 above
378will have just one csrow (csrow0). csrow1 will be empty. On the other
379hand, when 2 dual ranked DIMMs are similarly placed, then both csrow0
380and csrow1 will be populated. The pattern repeats itself for csrow2 and
381csrow3.
382
383The representation of the above is reflected in the directory
384tree in EDAC's sysfs interface. Starting in directory
385``/sys/devices/system/edac/mc``, each memory controller will be
386represented by its own ``mcX`` directory, where ``X`` is the
387index of the MC::
388
389	..../edac/mc/
390		   |
391		   |->mc0
392		   |->mc1
393		   |->mc2
394		   ....
395
396Under each ``mcX`` directory each ``csrowX`` is again represented by a
397``csrowX``, where ``X`` is the csrow index::
398
399	.../mc/mc0/
400		|
401		|->csrow0
402		|->csrow2
403		|->csrow3
404		....
405
406Notice that there is no csrow1, which indicates that csrow0 is composed
407of a single ranked DIMMs. This should also apply in both Channels, in
408order to have dual-channel mode be operational. Since both csrow2 and
409csrow3 are populated, this indicates a dual ranked set of DIMMs for
410channels 0 and 1.
411
412Within each of the ``mcX`` and ``csrowX`` directories are several EDAC
413control and attribute files.
414
415``mcX`` directories
416-------------------
417
418In ``mcX`` directories are EDAC control and attribute files for
419this ``X`` instance of the memory controllers.
420
421For a description of the sysfs API, please see:
422
423	Documentation/ABI/testing/sysfs-devices-edac
424
425
426``dimmX`` or ``rankX`` directories
427----------------------------------
428
429The recommended way to use the EDAC subsystem is to look at the information
430provided by the ``dimmX`` or ``rankX`` directories [#f5]_.
431
432A typical EDAC system has the following structure under
433``/sys/devices/system/edac/``\ [#f6]_::
434
435	/sys/devices/system/edac/
436	├── mc
437	│   ├── mc0
438	│   │   ├── ce_count
439	│   │   ├── ce_noinfo_count
440	│   │   ├── dimm0
441	│   │   │   ├── dimm_ce_count
442	│   │   │   ├── dimm_dev_type
443	│   │   │   ├── dimm_edac_mode
444	│   │   │   ├── dimm_label
445	│   │   │   ├── dimm_location
446	│   │   │   ├── dimm_mem_type
447	│   │   │   ├── dimm_ue_count
448	│   │   │   ├── size
449	│   │   │   └── uevent
450	│   │   ├── max_location
451	│   │   ├── mc_name
452	│   │   ├── reset_counters
453	│   │   ├── seconds_since_reset
454	│   │   ├── size_mb
455	│   │   ├── ue_count
456	│   │   ├── ue_noinfo_count
457	│   │   └── uevent
458	│   ├── mc1
459	│   │   ├── ce_count
460	│   │   ├── ce_noinfo_count
461	│   │   ├── dimm0
462	│   │   │   ├── dimm_ce_count
463	│   │   │   ├── dimm_dev_type
464	│   │   │   ├── dimm_edac_mode
465	│   │   │   ├── dimm_label
466	│   │   │   ├── dimm_location
467	│   │   │   ├── dimm_mem_type
468	│   │   │   ├── dimm_ue_count
469	│   │   │   ├── size
470	│   │   │   └── uevent
471	│   │   ├── max_location
472	│   │   ├── mc_name
473	│   │   ├── reset_counters
474	│   │   ├── seconds_since_reset
475	│   │   ├── size_mb
476	│   │   ├── ue_count
477	│   │   ├── ue_noinfo_count
478	│   │   └── uevent
479	│   └── uevent
480	└── uevent
481
482In the ``dimmX`` directories are EDAC control and attribute files for
483this ``X`` memory module:
484
485- ``size`` - Total memory managed by this csrow attribute file
486
487	This attribute file displays, in count of megabytes, the memory
488	that this csrow contains.
489
490- ``dimm_ue_count`` - Uncorrectable Errors count attribute file
491
492	This attribute file displays the total count of uncorrectable
493	errors that have occurred on this DIMM. If panic_on_ue is set
494	this counter will not have a chance to increment, since EDAC
495	will panic the system.
496
497- ``dimm_ce_count`` - Correctable Errors count attribute file
498
499	This attribute file displays the total count of correctable
500	errors that have occurred on this DIMM. This count is very
501	important to examine. CEs provide early indications that a
502	DIMM is beginning to fail. This count field should be
503	monitored for non-zero values and report such information
504	to the system administrator.
505
506- ``dimm_dev_type``  - Device type attribute file
507
508	This attribute file will display what type of DRAM device is
509	being utilized on this DIMM.
510	Examples:
511
512		- x1
513		- x2
514		- x4
515		- x8
516
517- ``dimm_edac_mode`` - EDAC Mode of operation attribute file
518
519	This attribute file will display what type of Error detection
520	and correction is being utilized.
521
522- ``dimm_label`` - memory module label control file
523
524	This control file allows this DIMM to have a label assigned
525	to it. With this label in the module, when errors occur
526	the output can provide the DIMM label in the system log.
527	This becomes vital for panic events to isolate the
528	cause of the UE event.
529
530	DIMM Labels must be assigned after booting, with information
531	that correctly identifies the physical slot with its
532	silk screen label. This information is currently very
533	motherboard specific and determination of this information
534	must occur in userland at this time.
535
536- ``dimm_location`` - location of the memory module
537
538	The location can have up to 3 levels, and describe how the
539	memory controller identifies the location of a memory module.
540	Depending on the type of memory and memory controller, it
541	can be:
542
543		- *csrow* and *channel* - used when the memory controller
544		  doesn't identify a single DIMM - e. g. in ``rankX`` dir;
545		- *branch*, *channel*, *slot* - typically used on FB-DIMM memory
546		  controllers;
547		- *channel*, *slot* - used on Nehalem and newer Intel drivers.
548
549- ``dimm_mem_type`` - Memory Type attribute file
550
551	This attribute file will display what type of memory is currently
552	on this csrow. Normally, either buffered or unbuffered memory.
553	Examples:
554
555		- Registered-DDR
556		- Unbuffered-DDR
557
558.. [#f5] On some systems, the memory controller doesn't have any logic
559  to identify the memory module. On such systems, the directory is called ``rankX`` and works on a similar way as the ``csrowX`` directories.
560  On modern Intel memory controllers, the memory controller identifies the
561  memory modules directly. On such systems, the directory is called ``dimmX``.
562
563.. [#f6] There are also some ``power`` directories and ``subsystem``
564  symlinks inside the sysfs mapping that are automatically created by
565  the sysfs subsystem. Currently, they serve no purpose.
566
567``csrowX`` directories
568----------------------
569
570When CONFIG_EDAC_LEGACY_SYSFS is enabled, sysfs will contain the ``csrowX``
571directories. As this API doesn't work properly for Rambus, FB-DIMMs and
572modern Intel Memory Controllers, this is being deprecated in favor of
573``dimmX`` directories.
574
575In the ``csrowX`` directories are EDAC control and attribute files for
576this ``X`` instance of csrow:
577
578
579- ``ue_count`` - Total Uncorrectable Errors count attribute file
580
581	This attribute file displays the total count of uncorrectable
582	errors that have occurred on this csrow. If panic_on_ue is set
583	this counter will not have a chance to increment, since EDAC
584	will panic the system.
585
586
587- ``ce_count`` - Total Correctable Errors count attribute file
588
589	This attribute file displays the total count of correctable
590	errors that have occurred on this csrow. This count is very
591	important to examine. CEs provide early indications that a
592	DIMM is beginning to fail. This count field should be
593	monitored for non-zero values and report such information
594	to the system administrator.
595
596
597- ``size_mb`` - Total memory managed by this csrow attribute file
598
599	This attribute file displays, in count of megabytes, the memory
600	that this csrow contains.
601
602
603- ``mem_type`` - Memory Type attribute file
604
605	This attribute file will display what type of memory is currently
606	on this csrow. Normally, either buffered or unbuffered memory.
607	Examples:
608
609		- Registered-DDR
610		- Unbuffered-DDR
611
612
613- ``edac_mode`` - EDAC Mode of operation attribute file
614
615	This attribute file will display what type of Error detection
616	and correction is being utilized.
617
618
619- ``dev_type`` - Device type attribute file
620
621	This attribute file will display what type of DRAM device is
622	being utilized on this DIMM.
623	Examples:
624
625		- x1
626		- x2
627		- x4
628		- x8
629
630
631- ``ch0_ce_count`` - Channel 0 CE Count attribute file
632
633	This attribute file will display the count of CEs on this
634	DIMM located in channel 0.
635
636
637- ``ch0_ue_count`` - Channel 0 UE Count attribute file
638
639	This attribute file will display the count of UEs on this
640	DIMM located in channel 0.
641
642
643- ``ch0_dimm_label`` - Channel 0 DIMM Label control file
644
645
646	This control file allows this DIMM to have a label assigned
647	to it. With this label in the module, when errors occur
648	the output can provide the DIMM label in the system log.
649	This becomes vital for panic events to isolate the
650	cause of the UE event.
651
652	DIMM Labels must be assigned after booting, with information
653	that correctly identifies the physical slot with its
654	silk screen label. This information is currently very
655	motherboard specific and determination of this information
656	must occur in userland at this time.
657
658
659- ``ch1_ce_count`` - Channel 1 CE Count attribute file
660
661
662	This attribute file will display the count of CEs on this
663	DIMM located in channel 1.
664
665
666- ``ch1_ue_count`` - Channel 1 UE Count attribute file
667
668
669	This attribute file will display the count of UEs on this
670	DIMM located in channel 0.
671
672
673- ``ch1_dimm_label`` - Channel 1 DIMM Label control file
674
675	This control file allows this DIMM to have a label assigned
676	to it. With this label in the module, when errors occur
677	the output can provide the DIMM label in the system log.
678	This becomes vital for panic events to isolate the
679	cause of the UE event.
680
681	DIMM Labels must be assigned after booting, with information
682	that correctly identifies the physical slot with its
683	silk screen label. This information is currently very
684	motherboard specific and determination of this information
685	must occur in userland at this time.
686
687
688System Logging
689--------------
690
691If logging for UEs and CEs is enabled, then system logs will contain
692information indicating that errors have been detected::
693
694  EDAC MC0: CE page 0x283, offset 0xce0, grain 8, syndrome 0x6ec3, row 0, channel 1 "DIMM_B1": amd76x_edac
695  EDAC MC0: CE page 0x1e5, offset 0xfb0, grain 8, syndrome 0xb741, row 0, channel 1 "DIMM_B1": amd76x_edac
696
697
698The structure of the message is:
699
700	+---------------------------------------+-------------+
701	| Content                               | Example     |
702	+=======================================+=============+
703	| The memory controller                 | MC0         |
704	+---------------------------------------+-------------+
705	| Error type                            | CE          |
706	+---------------------------------------+-------------+
707	| Memory page                           | 0x283       |
708	+---------------------------------------+-------------+
709	| Offset in the page                    | 0xce0       |
710	+---------------------------------------+-------------+
711	| The byte granularity                  | grain 8     |
712	| or resolution of the error            |             |
713	+---------------------------------------+-------------+
714	| The error syndrome                    | 0xb741      |
715	+---------------------------------------+-------------+
716	| Memory row                            | row 0       |
717	+---------------------------------------+-------------+
718	| Memory channel                        | channel 1   |
719	+---------------------------------------+-------------+
720	| DIMM label, if set prior              | DIMM B1     |
721	+---------------------------------------+-------------+
722	| And then an optional, driver-specific |             |
723	| message that may have additional      |             |
724	| information.                          |             |
725	+---------------------------------------+-------------+
726
727Both UEs and CEs with no info will lack all but memory controller, error
728type, a notice of "no info" and then an optional, driver-specific error
729message.
730
731
732PCI Bus Parity Detection
733------------------------
734
735On Header Type 00 devices, the primary status is looked at for any
736parity error regardless of whether parity is enabled on the device or
737not. (The spec indicates parity is generated in some cases). On Header
738Type 01 bridges, the secondary status register is also looked at to see
739if parity occurred on the bus on the other side of the bridge.
740
741
742Sysfs configuration
743-------------------
744
745Under ``/sys/devices/system/edac/pci`` are control and attribute files as
746follows:
747
748
749- ``check_pci_parity`` - Enable/Disable PCI Parity checking control file
750
751	This control file enables or disables the PCI Bus Parity scanning
752	operation. Writing a 1 to this file enables the scanning. Writing
753	a 0 to this file disables the scanning.
754
755	Enable::
756
757		echo "1" >/sys/devices/system/edac/pci/check_pci_parity
758
759	Disable::
760
761		echo "0" >/sys/devices/system/edac/pci/check_pci_parity
762
763
764- ``pci_parity_count`` - Parity Count
765
766	This attribute file will display the number of parity errors that
767	have been detected.
768
769
770Module parameters
771-----------------
772
773- ``edac_mc_panic_on_ue`` - Panic on UE control file
774
775	An uncorrectable error will cause a machine panic.  This is usually
776	desirable.  It is a bad idea to continue when an uncorrectable error
777	occurs - it is indeterminate what was uncorrected and the operating
778	system context might be so mangled that continuing will lead to further
779	corruption. If the kernel has MCE configured, then EDAC will never
780	notice the UE.
781
782	LOAD TIME::
783
784		module/kernel parameter: edac_mc_panic_on_ue=[0|1]
785
786	RUN TIME::
787
788		echo "1" > /sys/module/edac_core/parameters/edac_mc_panic_on_ue
789
790
791- ``edac_mc_log_ue`` - Log UE control file
792
793
794	Generate kernel messages describing uncorrectable errors.  These errors
795	are reported through the system message log system.  UE statistics
796	will be accumulated even when UE logging is disabled.
797
798	LOAD TIME::
799
800		module/kernel parameter: edac_mc_log_ue=[0|1]
801
802	RUN TIME::
803
804		echo "1" > /sys/module/edac_core/parameters/edac_mc_log_ue
805
806
807- ``edac_mc_log_ce`` - Log CE control file
808
809
810	Generate kernel messages describing correctable errors.  These
811	errors are reported through the system message log system.
812	CE statistics will be accumulated even when CE logging is disabled.
813
814	LOAD TIME::
815
816		module/kernel parameter: edac_mc_log_ce=[0|1]
817
818	RUN TIME::
819
820		echo "1" > /sys/module/edac_core/parameters/edac_mc_log_ce
821
822
823- ``edac_mc_poll_msec`` - Polling period control file
824
825
826	The time period, in milliseconds, for polling for error information.
827	Too small a value wastes resources.  Too large a value might delay
828	necessary handling of errors and might loose valuable information for
829	locating the error.  1000 milliseconds (once each second) is the current
830	default. Systems which require all the bandwidth they can get, may
831	increase this.
832
833	LOAD TIME::
834
835		module/kernel parameter: edac_mc_poll_msec=[0|1]
836
837	RUN TIME::
838
839		echo "1000" > /sys/module/edac_core/parameters/edac_mc_poll_msec
840
841
842- ``panic_on_pci_parity`` - Panic on PCI PARITY Error
843
844
845	This control file enables or disables panicking when a parity
846	error has been detected.
847
848
849	module/kernel parameter::
850
851			edac_panic_on_pci_pe=[0|1]
852
853	Enable::
854
855		echo "1" > /sys/module/edac_core/parameters/edac_panic_on_pci_pe
856
857	Disable::
858
859		echo "0" > /sys/module/edac_core/parameters/edac_panic_on_pci_pe
860
861
862
863EDAC device type
864----------------
865
866In the header file, edac_pci.h, there is a series of edac_device structures
867and APIs for the EDAC_DEVICE.
868
869User space access to an edac_device is through the sysfs interface.
870
871At the location ``/sys/devices/system/edac`` (sysfs) new edac_device devices
872will appear.
873
874There is a three level tree beneath the above ``edac`` directory. For example,
875the ``test_device_edac`` device (found at the http://bluesmoke.sourceforget.net
876website) installs itself as::
877
878	/sys/devices/system/edac/test-instance
879
880in this directory are various controls, a symlink and one or more ``instance``
881directories.
882
883The standard default controls are:
884
885	==============	=======================================================
886	log_ce		boolean to log CE events
887	log_ue		boolean to log UE events
888	panic_on_ue	boolean to ``panic`` the system if an UE is encountered
889			(default off, can be set true via startup script)
890	poll_msec	time period between POLL cycles for events
891	==============	=======================================================
892
893The test_device_edac device adds at least one of its own custom control:
894
895	==============	==================================================
896	test_bits	which in the current test driver does nothing but
897			show how it is installed. A ported driver can
898			add one or more such controls and/or attributes
899			for specific uses.
900			One out-of-tree driver uses controls here to allow
901			for ERROR INJECTION operations to hardware
902			injection registers
903	==============	==================================================
904
905The symlink points to the 'struct dev' that is registered for this edac_device.
906
907Instances
908---------
909
910One or more instance directories are present. For the ``test_device_edac``
911case:
912
913	+----------------+
914	| test-instance0 |
915	+----------------+
916
917
918In this directory there are two default counter attributes, which are totals of
919counter in deeper subdirectories.
920
921	==============	====================================
922	ce_count	total of CE events of subdirectories
923	ue_count	total of UE events of subdirectories
924	==============	====================================
925
926Blocks
927------
928
929At the lowest directory level is the ``block`` directory. There can be 0, 1
930or more blocks specified in each instance:
931
932	+-------------+
933	| test-block0 |
934	+-------------+
935
936In this directory the default attributes are:
937
938	==============	================================================
939	ce_count	which is counter of CE events for this ``block``
940			of hardware being monitored
941	ue_count	which is counter of UE events for this ``block``
942			of hardware being monitored
943	==============	================================================
944
945
946The ``test_device_edac`` device adds 4 attributes and 1 control:
947
948	================== ====================================================
949	test-block-bits-0	for every POLL cycle this counter
950				is incremented
951	test-block-bits-1	every 10 cycles, this counter is bumped once,
952				and test-block-bits-0 is set to 0
953	test-block-bits-2	every 100 cycles, this counter is bumped once,
954				and test-block-bits-1 is set to 0
955	test-block-bits-3	every 1000 cycles, this counter is bumped once,
956				and test-block-bits-2 is set to 0
957	================== ====================================================
958
959
960	================== ====================================================
961	reset-counters		writing ANY thing to this control will
962				reset all the above counters.
963	================== ====================================================
964
965
966Use of the ``test_device_edac`` driver should enable any others to create their own
967unique drivers for their hardware systems.
968
969The ``test_device_edac`` sample driver is located at the
970http://bluesmoke.sourceforge.net project site for EDAC.
971
972
973Usage of EDAC APIs on Nehalem and newer Intel CPUs
974--------------------------------------------------
975
976On older Intel architectures, the memory controller was part of the North
977Bridge chipset. Nehalem, Sandy Bridge, Ivy Bridge, Haswell, Sky Lake and
978newer Intel architectures integrated an enhanced version of the memory
979controller (MC) inside the CPUs.
980
981This chapter will cover the differences of the enhanced memory controllers
982found on newer Intel CPUs, such as ``i7core_edac``, ``sb_edac`` and
983``sbx_edac`` drivers.
984
985.. note::
986
987   The Xeon E7 processor families use a separate chip for the memory
988   controller, called Intel Scalable Memory Buffer. This section doesn't
989   apply for such families.
990
9911) There is one Memory Controller per Quick Patch Interconnect
992   (QPI). At the driver, the term "socket" means one QPI. This is
993   associated with a physical CPU socket.
994
995   Each MC have 3 physical read channels, 3 physical write channels and
996   3 logic channels. The driver currently sees it as just 3 channels.
997   Each channel can have up to 3 DIMMs.
998
999   The minimum known unity is DIMMs. There are no information about csrows.
1000   As EDAC API maps the minimum unity is csrows, the driver sequentially
1001   maps channel/DIMM into different csrows.
1002
1003   For example, supposing the following layout::
1004
1005	Ch0 phy rd0, wr0 (0x063f4031): 2 ranks, UDIMMs
1006	  dimm 0 1024 Mb offset: 0, bank: 8, rank: 1, row: 0x4000, col: 0x400
1007	  dimm 1 1024 Mb offset: 4, bank: 8, rank: 1, row: 0x4000, col: 0x400
1008        Ch1 phy rd1, wr1 (0x063f4031): 2 ranks, UDIMMs
1009	  dimm 0 1024 Mb offset: 0, bank: 8, rank: 1, row: 0x4000, col: 0x400
1010	Ch2 phy rd3, wr3 (0x063f4031): 2 ranks, UDIMMs
1011	  dimm 0 1024 Mb offset: 0, bank: 8, rank: 1, row: 0x4000, col: 0x400
1012
1013   The driver will map it as::
1014
1015	csrow0: channel 0, dimm0
1016	csrow1: channel 0, dimm1
1017	csrow2: channel 1, dimm0
1018	csrow3: channel 2, dimm0
1019
1020   exports one DIMM per csrow.
1021
1022   Each QPI is exported as a different memory controller.
1023
10242) The MC has the ability to inject errors to test drivers. The drivers
1025   implement this functionality via some error injection nodes:
1026
1027   For injecting a memory error, there are some sysfs nodes, under
1028   ``/sys/devices/system/edac/mc/mc?/``:
1029
1030   - ``inject_addrmatch/*``:
1031      Controls the error injection mask register. It is possible to specify
1032      several characteristics of the address to match an error code::
1033
1034         dimm = the affected dimm. Numbers are relative to a channel;
1035         rank = the memory rank;
1036         channel = the channel that will generate an error;
1037         bank = the affected bank;
1038         page = the page address;
1039         column (or col) = the address column.
1040
1041      each of the above values can be set to "any" to match any valid value.
1042
1043      At driver init, all values are set to any.
1044
1045      For example, to generate an error at rank 1 of dimm 2, for any channel,
1046      any bank, any page, any column::
1047
1048		echo 2 >/sys/devices/system/edac/mc/mc0/inject_addrmatch/dimm
1049		echo 1 >/sys/devices/system/edac/mc/mc0/inject_addrmatch/rank
1050
1051	To return to the default behaviour of matching any, you can do::
1052
1053		echo any >/sys/devices/system/edac/mc/mc0/inject_addrmatch/dimm
1054		echo any >/sys/devices/system/edac/mc/mc0/inject_addrmatch/rank
1055
1056   - ``inject_eccmask``:
1057          specifies what bits will have troubles,
1058
1059   - ``inject_section``:
1060       specifies what ECC cache section will get the error::
1061
1062		3 for both
1063		2 for the highest
1064		1 for the lowest
1065
1066   - ``inject_type``:
1067       specifies the type of error, being a combination of the following bits::
1068
1069		bit 0 - repeat
1070		bit 1 - ecc
1071		bit 2 - parity
1072
1073   - ``inject_enable``:
1074       starts the error generation when something different than 0 is written.
1075
1076   All inject vars can be read. root permission is needed for write.
1077
1078   Datasheet states that the error will only be generated after a write on an
1079   address that matches inject_addrmatch. It seems, however, that reading will
1080   also produce an error.
1081
1082   For example, the following code will generate an error for any write access
1083   at socket 0, on any DIMM/address on channel 2::
1084
1085	echo 2 >/sys/devices/system/edac/mc/mc0/inject_addrmatch/channel
1086	echo 2 >/sys/devices/system/edac/mc/mc0/inject_type
1087	echo 64 >/sys/devices/system/edac/mc/mc0/inject_eccmask
1088	echo 3 >/sys/devices/system/edac/mc/mc0/inject_section
1089	echo 1 >/sys/devices/system/edac/mc/mc0/inject_enable
1090	dd if=/dev/mem of=/dev/null seek=16k bs=4k count=1 >& /dev/null
1091
1092   For socket 1, it is needed to replace "mc0" by "mc1" at the above
1093   commands.
1094
1095   The generated error message will look like::
1096
1097	EDAC MC0: UE row 0, channel-a= 0 channel-b= 0 labels "-": NON_FATAL (addr = 0x0075b980, socket=0, Dimm=0, Channel=2, syndrome=0x00000040, count=1, Err=8c0000400001009f:4000080482 (read error: read ECC error))
1098
10993) Corrected Error memory register counters
1100
1101   Those newer MCs have some registers to count memory errors. The driver
1102   uses those registers to report Corrected Errors on devices with Registered
1103   DIMMs.
1104
1105   However, those counters don't work with Unregistered DIMM. As the chipset
1106   offers some counters that also work with UDIMMs (but with a worse level of
1107   granularity than the default ones), the driver exposes those registers for
1108   UDIMM memories.
1109
1110   They can be read by looking at the contents of ``all_channel_counts/``::
1111
1112     $ for i in /sys/devices/system/edac/mc/mc0/all_channel_counts/*; do echo $i; cat $i; done
1113	/sys/devices/system/edac/mc/mc0/all_channel_counts/udimm0
1114	0
1115	/sys/devices/system/edac/mc/mc0/all_channel_counts/udimm1
1116	0
1117	/sys/devices/system/edac/mc/mc0/all_channel_counts/udimm2
1118	0
1119
1120   What happens here is that errors on different csrows, but at the same
1121   dimm number will increment the same counter.
1122   So, in this memory mapping::
1123
1124	csrow0: channel 0, dimm0
1125	csrow1: channel 0, dimm1
1126	csrow2: channel 1, dimm0
1127	csrow3: channel 2, dimm0
1128
1129   The hardware will increment udimm0 for an error at the first dimm at either
1130   csrow0, csrow2  or csrow3;
1131
1132   The hardware will increment udimm1 for an error at the second dimm at either
1133   csrow0, csrow2  or csrow3;
1134
1135   The hardware will increment udimm2 for an error at the third dimm at either
1136   csrow0, csrow2  or csrow3;
1137
11384) Standard error counters
1139
1140   The standard error counters are generated when an mcelog error is received
1141   by the driver. Since, with UDIMM, this is counted by software, it is
1142   possible that some errors could be lost. With RDIMM's, they display the
1143   contents of the registers
1144
1145Reference documents used on ``amd64_edac``
1146------------------------------------------
1147
1148``amd64_edac`` module is based on the following documents
1149(available from http://support.amd.com/en-us/search/tech-docs):
1150
11511. :Title:  BIOS and Kernel Developer's Guide for AMD Athlon 64 and AMD
1152	   Opteron Processors
1153   :AMD publication #: 26094
1154   :Revision: 3.26
1155   :Link: http://support.amd.com/TechDocs/26094.PDF
1156
11572. :Title:  BIOS and Kernel Developer's Guide for AMD NPT Family 0Fh
1158	   Processors
1159   :AMD publication #: 32559
1160   :Revision: 3.00
1161   :Issue Date: May 2006
1162   :Link: http://support.amd.com/TechDocs/32559.pdf
1163
11643. :Title:  BIOS and Kernel Developer's Guide (BKDG) For AMD Family 10h
1165	   Processors
1166   :AMD publication #: 31116
1167   :Revision: 3.00
1168   :Issue Date: September 07, 2007
1169   :Link: http://support.amd.com/TechDocs/31116.pdf
1170
11714. :Title: BIOS and Kernel Developer's Guide (BKDG) for AMD Family 15h
1172	  Models 30h-3Fh Processors
1173   :AMD publication #: 49125
1174   :Revision: 3.06
1175   :Issue Date: 2/12/2015 (latest release)
1176   :Link: http://support.amd.com/TechDocs/49125_15h_Models_30h-3Fh_BKDG.pdf
1177
11785. :Title: BIOS and Kernel Developer's Guide (BKDG) for AMD Family 15h
1179	  Models 60h-6Fh Processors
1180   :AMD publication #: 50742
1181   :Revision: 3.01
1182   :Issue Date: 7/23/2015 (latest release)
1183   :Link: http://support.amd.com/TechDocs/50742_15h_Models_60h-6Fh_BKDG.pdf
1184
11856. :Title: BIOS and Kernel Developer's Guide (BKDG) for AMD Family 16h
1186	  Models 00h-0Fh Processors
1187   :AMD publication #: 48751
1188   :Revision: 3.03
1189   :Issue Date: 2/23/2015 (latest release)
1190   :Link: http://support.amd.com/TechDocs/48751_16h_bkdg.pdf
1191
1192Credits
1193=======
1194
1195* Written by Doug Thompson <dougthompson@xmission.com>
1196
1197  - 7 Dec 2005
1198  - 17 Jul 2007	Updated
1199
1200* |copy| Mauro Carvalho Chehab
1201
1202  - 05 Aug 2009	Nehalem interface
1203  - 26 Oct 2016 Converted to ReST and cleanups at the Nehalem section
1204
1205* EDAC authors/maintainers:
1206
1207  - Doug Thompson, Dave Jiang, Dave Peterson et al,
1208  - Mauro Carvalho Chehab
1209  - Borislav Petkov
1210  - original author: Thayne Harbaugh
1211