Attention is currently required from: Patrick Georgi, Rizwan Qureshi, Stefan Reinauer, Sridhar Siricilla, Angel Pons, Alex Levin, YH Lin, Martin Roth, Subrata Banik, Caveh Jalali, Tim Wawrzynczak, Nick Vaccaro, Boris Mittelberg.

View Change

1 comment:

• Patchset:

  • Patch Set #6:

    A version of this change that takes the applicable context into account
    (e.g. other waiting loops in this driver, implications when multiple
    instances would reach this point in the code simultaneously) would be
    accepted immediately.

    I understand that would be a comprehensive fix across flashrom.

    Doesn't need to be. All I'm asking for is to keep the whole code sound
    and not just the single spot where a check would be added. This could
    for instance be a simple

      if (REGREAD8(ICH9_REG_HSFS) & HSFS_SCIP)
              return -1;

    That would adhere to the datasheet, AFAICT and wouldn't make it look
    like we don't know what we are doing (not knowing what we are doing,
    for instance, would be writing a program that tries to synchronize
    multiple processes but fails to do so properly).

    Another solution would be to check the bit once on programmer initiali-
    zation and after each transaction.

    However that will still not guarantee that another flashrom like program will wait on the same file/mutex to synchronize the access.

    This check on H_SCIP OTOH, which should be adhered by all the programs at the least prevents the program from cancelling out another's transaction. So let us make flashrom compliant with the recommendation.

    Then check the bit once in programmer init. That should cover another
    previously running program. Multiple, concurrently running programs
    are out of the scope of SCIP.

To view, visit change 61854. To unsubscribe, or for help writing mail filters, visit settings.