[lkml]   [2012]   [Jan]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    Subject[PATCH RFCv2 00/16] This is the version 2 of the HERM patches
    This patch series is there to address some troubles with the
    EDAC subsystem.

    There are two groups of change in this series:

    a) a trace-based class of events for hardware errors is
    added (Hardware Events Report Mecanism - HERM);

    The need of moving for a tracepoint-based approach were
    widely discussed already at the ML. Basically, it offers
    more flexibility than message dumps at the console, allowing
    events filtering and other sorts of improvements.

    The long-term target is that memory errors will generate
    events like:

    Corrected error: memory read error on DIMM_1A (row 1, channel 0, rank=5, cpu=0, Err=0001:0090, addr = 0x7a789f03e)
    Uncorrected error: memory write error on DIMM_2B (row 2, channel 3, rank=4, cpu=1, Err=0001:0091, addr = 0xdeadbeef)

    E. g. putting the user-relevant information first while
    keeping the technical details that could help the
    hardware manufacturers and the ones that might want to replace
    a DRAM chip in parenthesis.

    b) the edac core was changed to better support memory
    controllers that aren't able to see csrows.

    The EDAC subsystem were originally written to work with
    memory controllers directly connected to the DIMM chips.
    Not all memory architectures use this concept. For example,
    FBDIMM memories are connected via a buffer, called AMB [1].

    When an AMB is present, the memory controller only sees
    its communication bus, called "channel". This has nothing
    to do with the "csrow channel" concept, widely used at
    the subsystem, and mandatory. All drivers that work with
    such architectures currently need to fake data, lying to
    the edac core, in order for them to work.

    Lying to the subsystem in general is not a good idea ;)

    So, this series addresses it by splitting the DIMM information
    from the EDAC csrow_info struct, and creating a new set of
    DIMM-oriented sysfs nodes:

    ├── dimm0
    │   ├── dimm_dev_type
    │   ├── dimm_edac_mode
    │   ├── dimm_label
    │   ├── dimm_location
    │   ├── dimm_mem_type
    │   └── dimm_size
    └── dimm3
       ├── dimm_dev_type
       ├── dimm_edac_mode
       ├── dimm_label
       ├── dimm_location
       ├── dimm_mem_type
       └── dimm_size

    The DIMM description looks like:

    dimm_location:branch 1 channel 0 dimm 1

    Currently, the existing struct was not touched. The next step
    (as indicated at the last patch on this series) is to
    create the error counters.

    Currently, is still an RFC, as it is not complete, and some
    changes will require more test. Also, didn't try to compile
    it yet on non x86 archs.


    Please review.



    Mauro Carvalho Chehab (16):
    events/hw_event: Create a Hardware Events Report Mecanism (HERM)
    events/hw_event: use __string() trace macros for events
    hw_event: Consolidate uncorrected/corrected error msgs into one
    drivers/edac: rename channel_info to csrow_channel_info
    edac: Create a dimm struct and move the labels into it
    edac_mc_sysfs: Fix error handling
    edac: Add per dimm's sysfs nodes
    edac: Prepare to push down to drivers the filling of the dimm_info
    i5400_edac: Convert it to report memory with the new location
    i7300_edac: Convert it to report memory with the new location
    edac: move dimm properties to struct dimm_info
    edac: Don't initialize csrow's first_page & friends when not needed
    edac: move nr_pages to dimm struct
    edac: Add per-dimm sysfs show nodes
    edac: DIMM location cleanup
    edac: Add an error scope logic

    drivers/edac/amd64_edac.c | 72 +++-------
    drivers/edac/amd76x_edac.c | 14 +-
    drivers/edac/cell_edac.c | 18 ++-
    drivers/edac/cpc925_edac.c | 70 +++++-----
    drivers/edac/e752x_edac.c | 48 ++++---
    drivers/edac/e7xxx_edac.c | 49 ++++---
    drivers/edac/edac_mc.c | 168 ++++++++++++++++++-----
    drivers/edac/edac_mc_sysfs.c | 283 ++++++++++++++++++++++++++++++++++++---
    drivers/edac/i3000_edac.c | 24 ++--
    drivers/edac/i3200_edac.c | 24 ++--
    drivers/edac/i5000_edac.c | 31 ++---
    drivers/edac/i5100_edac.c | 67 +++++-----
    drivers/edac/i5400_edac.c | 46 +++----
    drivers/edac/i7300_edac.c | 47 ++++---
    drivers/edac/i7core_edac.c | 46 +++----
    drivers/edac/i82443bxgx_edac.c | 15 ++-
    drivers/edac/i82860_edac.c | 13 +-
    drivers/edac/i82875p_edac.c | 22 ++-
    drivers/edac/i82975x_edac.c | 28 +++--
    drivers/edac/mpc85xx_edac.c | 16 ++-
    drivers/edac/mv64x60_edac.c | 22 ++--
    drivers/edac/pasemi_edac.c | 24 ++--
    drivers/edac/ppc4xx_edac.c | 25 ++--
    drivers/edac/r82600_edac.c | 13 +-
    drivers/edac/sb_edac.c | 44 ++++---
    drivers/edac/tile_edac.c | 17 +--
    drivers/edac/x38_edac.c | 24 ++--
    include/linux/edac.h | 90 +++++++++++--
    include/trace/events/hw_event.h | 133 ++++++++++++++++++
    29 files changed, 1018 insertions(+), 475 deletions(-)
    create mode 100644 include/trace/events/hw_event.h


    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2012-01-28 16:37    [W:0.025 / U:80.888 seconds]
    ©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site