IBM Support

VM66254: USER HANG DUE TO LOST PAGING READ/WRITE

A fix is available

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as program error.

Error description

  • A z/VM guest can appear to be hung either during normal guest
    execution or at logoff.
    
    A request to page in or out a guest page can be lost due to
    incorrect paging request queue management.
    

Local fix

  • N/A
    

Problem summary

  • ****************************************************************
    * USERS AFFECTED: All users of z/VM systems with more than 512 *
    *                 GB of real storage/memory.                   *
    ****************************************************************
    * PROBLEM DESCRIPTION:                                         *
    ****************************************************************
    * RECOMMENDATION: APPLY PTF                                    *
    ****************************************************************
    Guest hangs can occur due to lost paging requests to read or
    write a guest page.  This can cause a guest to hang during
    normal execution or logoff.
    
    In the paging code, there is incorrect queue management
    logic.  The code loads a 64-bit value, but then uses a
    32-bit test for zero.  As such, if the low-order 32 bits of
    the address are zero, but high-order bits are non-zero, the
    test will take the wrong branch.  This will result in
    overwriting the queue anchors, instead of appending to the
    queue, and losing the FRMTEs that were already in the queue.
    In the specific area of code, the 64-bit address is a FRMTE
    address.  Owing to the way the frame table is organized, FRMTEs
    with a zero low half, and non-zero high half represent the
    frames at 512 GB, 1 TB, and 1.5 TB.  So if one of these
    specific FRMTEs happens to be at the front of a read or write
    queue, the read/write request for this frame and any that
    happen to be after it will be lost.
    
    Only systems with greater than 512 GB of real memory are
    affected by this problem.
    

Problem conclusion

  • Code in module HCPPAI is corrected to do a 64-bit check
    for zero on the queue anchors.
    

Temporary fix

Comments

APAR Information

  • APAR number

    VM66254

  • Reported component name

    VM CP

  • Reported component ID

    568411202

  • Reported release

    710

  • Status

    CLOSED PER

  • PE

    NoPE

  • HIPER

    YesHIPER

  • Special Attention

    NoSpecatt / Xsystem

  • Submitted date

    2019-02-12

  • Closed date

    2019-02-27

  • Last modified date

    2020-12-16

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

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

    UM35434 UM35435

Modules/Macros

  • HCPPAI
    

Fix information

  • Fixed component name

    VM CP

  • Fixed component ID

    568411202

Applicable component levels

  • R640 PSY UM35434

       UP19/03/06 P 2001 ¢

  • R710 PSY UM35435

       UP19/09/26 P 1902 ¢

Fix is available

  • Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.

[{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SG27M","label":"APARs - z\/VM environment"},"Platform":[{"code":"PF054","label":"z\/OS"}],"Version":"710","Line of Business":{"code":"LOB16","label":"Mainframe HW"}}]

Document Information

Modified date:
12 January 2021