Skip to content

L
linux
Switch branch/tag
  1. 27 Jan, 2012 1 commit
  3. 03 Jan, 2012 6 commits
    • Tony Luck's avatar
      x86/mce: Recognise machine check bank signature for data path error · 5f7b88d5
      Tony Luck authored 
      Action required data path signature is defined in table 15-19 of SDM:

+-----------------------------------------------------------------------------+
| SRAR Error | Valid | OVER | UC | EN | MISCV | ADDRV | PCC | S | AR | MCACOD |
| Data Load  |     1 |    0 |  1 |  1 |     1 |     1 |   0 | 1 |  1 |  0x134 |
+-----------------------------------------------------------------------------+

Recognise this, and pass MCE_AR_SEVERITY code back to do_machine_check() if
we have the action handler configured (CONFIG_MEMORY_FAILURE=y)
Acked-by: default avatarBorislav Petkov <bp@amd64.org>
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
      5f7b88d5
    • Tony Luck's avatar
      x86/mce: Handle "action required" errors · a8c321fb
      Tony Luck authored 
      All non-urgent actions (reporting low severity errors and handling
"action-optional" errors) are now handled by a work queue. This
means that TIF_MCE_NOTIFY can be used to block execution for a
thread experiencing an "action-required" fault until we get all
cpus out of the machine check handler (and the thread that hit
the fault into mce_notify_process().

We use the new mce_{save,find,clear}_info() API to get information
from do_machine_check() to mce_notify_process(), and then use the
newly improved memory_failure(..., MF_ACTION_REQUIRED) to handle
the error (possibly signalling the process).

Update some comments to make the new code flows clearer.
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
      a8c321fb
    • Tony Luck's avatar
      x86/mce: Add mechanism to safely save information in MCE handler · af104e39
      Tony Luck authored 
      Machine checks on Intel cpus interrupt execution on all cpus, regardless
of interrupt masking.  We have a need to save some data about the cause
of the machine check (physical address) in the machine check handler that
can be retrieved later to attempt recovery in a more flexible execution
state.
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
      af104e39
    • Tony Luck's avatar
      x86/mce: Create helper function to save addr/misc when needed · 85f92694
      Tony Luck authored 
      The MCI_STATUS_MISCV and MCI_STATUS_ADDRV bits in the bank status
registers define whether the MISC and ADDR registers respectively
contain valid data - provide a helper function to check these bits
and read the registers when needed.

In addition, processors that support software error recovery (as
indicated by the MCG_SER_P bit in the MCG_CAP register) may include
some undefined bits in the ADDR register - mask these out.
Acked-by: default avatarBorislav Petkov <bp@amd64.org>
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
      85f92694
    • Tony Luck's avatar
      HWPOISON: Add code to handle "action required" errors. · 7329bbeb
      Tony Luck authored 
      Add new flag bit "MF_ACTION_REQUIRED" to be used by machine check
code to force a signal with si_code = BUS_MCEERR_AR in the case
where the error occurs in processor execution context. Pass the
flags argument along call chain:
	memory_failure()
	  hwpoison_user_mappings()
	    kill_procs()
	      kill_proc()

Drop the "_ao" suffix from kill_procs_ao() and kill_proc_ao() since
they can now handle "action required" as well as "action optional" errors.
Acked-by: default avatarBorislav Petkov <bp@amd64.org>
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
      7329bbeb
    • Tony Luck's avatar
      HWPOISON: Clean up memory_failure() vs. __memory_failure() · cd42f4a3
      Tony Luck authored 
      There is only one caller of memory_failure(), all other users call
__memory_failure() and pass in the flags argument explicitly. The
lone user of memory_failure() will soon need to pass flags too.

Add flags argument to the callsite in mce.c. Delete the old memory_failure()
function, and then rename __memory_failure() without the leading "__".

Provide clearer message when action optional memory errors are ignored.
Acked-by: default avatarBorislav Petkov <bp@amd64.org>
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
      cd42f4a3
  5. 09 Dec, 2011 33 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7