IJ43790: GSKIT ISSUE WITH SPECIFIC AMD EPYC PROCESSORS

APAR status

• Closed as program error.

Error description

• Customer run mmcrcluster command hang on tsgskkm due to
  an known problem in GSKit (used by tsgskkm) with specific
  AMD EPYC processors.
  This is CPU family 25.
  Earlier fix covers only CPU family 23 and only for EPYC
  7F72 and EPYC 7302 model.
  

Local fix

• The known fix is to use the ICC_SHIFT=3 env variable,
  either in the process environment or in the ICCSIG.txt
  file for the GSKit library used (FIPS-certifed or not.)
  
  
  
  By setting the ICC_SHIFT=3 env variable in the root's
  profile; or by adding ICC_SHIFT=3 to the following files,
  as follows:
  
  
  
  /usr/lpp/mmfs/lib/gsk8/C/icc/icclib/ICCSIG.txt
  
  /usr/lpp/mmfs/lib/gsk8/N/icc/icclib/ICCSIG.txt
  
  
  
  # IBM Crypto for C.
  
  # ICC Version 8.6.0.0
  
  ...
  
  #
  
  #Do not edit before this line
  
  #
  
  # Global Settings
  
  ICC_ALLOW_2KEY3DES=1
  
  ICC_SHIFT=3
  
  #
  

Problem summary

• Commands like mmcrcluster or mmaddnode may hang in GSKIT
  layer on AMD EPYC family 25 processors.  A particular model
  from family 25 that is known to hang in GSKIT layer is
  AMD EPYC 7343.
  

Problem conclusion

• This problem is fixed in 5.1.6.1
  To see all Spectrum Scale APARs and their respective
  Fix solutions refer to page:
  https://public.dhe.ibm.com/storage/spectrumscale/spectrum_scale_
  apars.html
  
  Benefits of the solution:
  Apply the GSKIT hang workaround automatically on AMD EPYC
  family 25 processors
  
  Work Around:
  Add "ICC_SHIFT=3" line in
  /usr/lpp/mmfs/lib/gsk8/Cicc/icclib/ICCSIG.txt
  file on problem nodes.
  
  Problem trigger:
  This problem affects AMD EPYC family 25 processors
  
  Symptom:
  Admin commands hangs
  
  Platforms affected:
  Linux OS environments
  
  Functional Area affected:
  Admin Commands, gskit
  
  Customer Impact:
  High Importance
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  SPEC SCALE ADV

• Fixed component ID

  5737F35AP

Applicable component levels