Permalink
Commits on Dec 4, 2009
  1. The value to binary line in do_put was accidentally duplicated making…

    … extra work on a put.
    Jon Meredith committed with toland Jun 29, 2009
  2. Moved check_all_databases_closed inside check for G_DB_ENV != NULL, o…

    …therwise
    
    fails on LOCK_DATABASE when G_DATABASES_MUTEX is uninitialized.
    Jon Meredith committed with toland Jun 29, 2009
Commits on Aug 10, 2009
Commits on Jun 23, 2009
  1. Updated Jon's email address.

    Phillip Toland committed Jun 23, 2009
  2. Added a simple readme file.

    Phillip Toland committed Jun 23, 2009
  3. Added license information.

    Phillip Toland committed Jun 23, 2009
  4. Made test closer to what was happening in production. Sometimes fails…

    … with BDB Error: assert failure: ../dist/../mp/mp_fget.c/257: "bhp != NULL", also seen it run with CRC errors.
    Phillip Toland committed Jun 23, 2009
  5. Version bump for release 12.

    Phillip Toland committed Jun 19, 2009
  6. Cleaned up the const != var tests to var != const.

    Jon Meredith committed with Phillip Toland Jun 19, 2009
  7. Cleaned up driver

     - removed the lock checks now we are happy with the mutex
     - replaced 'const == var' tests with 'var == const' to make code consistent.
    Jon Meredith committed with Phillip Toland Jun 19, 2009
  8. Relaced G_DATABASES rwlock with a mutex.

    Jon Meredith committed with Phillip Toland Jun 19, 2009
  9. Added extra paranoia around G_DATABASES, checking it is valid at lock…

    … and unlock.
    Jon Meredith committed with Phillip Toland Jun 19, 2009
  10. The read lock checks for the G_DATABASES rwlock were incorrect. Multi…

    …ple readers
    
    are possible.  Now the check for exclusive port/thread is only done on writes.
    
    Added an integration test that reproduces conditions seen during the local-storage-ops-
    in-coord branch.
    Jon Meredith committed with Phillip Toland Jun 19, 2009
  11. Heavily instrumented rwlock code - rwlock seems to be misbehaving.

    About to replace with a mutex.wq
    Jon Meredith committed with Phillip Toland Jun 18, 2009
  12. Improved debug output on driver stop.

    Disabled active_dbs counter as coredump went away.
    Jon Meredith committed with Phillip Toland Jun 19, 2009
  13. Added active counts to dbref and portref to help see what is being co…

    …rrupted
    
    in the coredump.
    Jon Meredith committed with Phillip Toland Jun 18, 2009
  14. Added extra debug/paranoia on bdberl_drv_stop/bdberl_drv_finish while

    looking for the coredump on port close discovered on the local storage ops
    inside coordinator branch.
    Jon Meredith committed with Phillip Toland Jun 18, 2009
  15. Added versions of transaction() and update() that take a transaction

    timeout.
    Jon Meredith committed with Phillip Toland Jun 15, 2009
  16. Refactored environment variable checks. All go through common functions

    that validate conversion to unsigned int and prints an error message
    on stderr if not used (would prefer to use SASL but it's as good as it
    gets during driver setup).
    Jon Meredith committed with Phillip Toland Jun 11, 2009
  17. Refactored common receive loops from many functions into recv_val() a…

    …nd recv_ok().
    Jon Meredith committed with Phillip Toland Jun 10, 2009
  18. Added bdberl:driver_info() function to get driver parameters.

    Jon Meredith committed with Phillip Toland Jun 10, 2009
  19. Forgot to rename DBGCMDX for non-debug build.

    Jon Meredith committed with Phillip Toland Jun 10, 2009
  20. Cleaned up debugging code.

    Jon Meredith committed with Phillip Toland Jun 10, 2009
  21. Broke stats code out into bdberl_stats.h module.

    Had to rename some functions that became non-static.
    Added some accessors to module static variables as needed.
    Jon Meredith committed with Phillip Toland Jun 9, 2009
  22. Added more comprehensive debug output if built with debug=1.

    Made deadlock checker interval configurable.
    Jon Meredith committed with Phillip Toland Jun 9, 2009
  23. Made size of thread pools configurable from environment variables

    BDBERL_NUM_GENERAL_THREADS and BDBERL_NUM_TXN_THREADS.
    Jon Meredith committed with Phillip Toland Jun 8, 2009
  24. Forgot to remove a couple of unnecessary work_buffer copies.

    Jon Meredith committed with Phillip Toland Jun 8, 2009
  25. Converted truncate to use async_dbref.

    Jon Meredith committed with Phillip Toland Jun 8, 2009
  26. Changed stats functions to use async_dbref/async_flags to pass parame…

    …ters
    
    to async functions rather than work buffer.
    Jon Meredith committed with Phillip Toland Jun 8, 2009
  27. In cases where there was more than one pending request and no active …

    …requests,
    
    the linked list would be corrupted linking the pending request on the list of
    active requests.
    Jon Meredith committed with Phillip Toland Jun 5, 2009
  28. Incorporated Diz's crc generation/checking patch to bdberl.erl.

    Added crc checking before/after BDB accesses in C driver.
    Changed get to use malloc DBTs rather than the user supplied buffer
    previously.
    Jon Meredith committed with Phillip Toland Jun 4, 2009
  29. Removed unused line.

    Jon Meredith committed with Phillip Toland Jun 2, 2009
Commits on Jun 2, 2009
  1. Added stop(). Closes #318.

    Jon Meredith committed Jun 2, 2009
  2. Replaced calls to getenv with erl_drv_getenv/G_DB_ENV->get_home. The …

    …erlang
    
    docs say not to use getenv in erlang drivers as they are not thread safe.
    The places we were using them were very unlikley to have issues, but you never
    know.
    Jon Meredith committed Jun 2, 2009
  3. Merge branch 'master' of git.rascal:/var/git/commons/bdberl

    Conflicts:
    	c_src/bdberl_drv.c
    	c_src/buildlib.sh
    	src/bdberl.erl
    Dave Smith committed Jun 2, 2009