Re: checkpointing processes with >2GB on x86_64

From: Paul H. Hargrove (PHHargrove_at_lbl_dot_gov)
Date: Thu Apr 19 2007 - 17:49:59 PDT

  • Next message: Paul H. Hargrove: "BLCR 0.5.4 released"
    Paul H. Hargove wrote:
    > I've manually verified that I can checkpoint as large as a 2936MB
    > working set on a 32-bit x86 machine.
    > 
    > With a 6GB working set on a 64-bit machine I see a file truncated at
    > (2^31-1) bytes (also on reiserfs).
    > 
    > I will investigate the cause.
    > 
    > -Paul
    [snip]
    
    Actually, the 32-bit checkpoints were no good either when > 2GB.
    I've traced the source of the problem to overflow in some 32bit counters
    in the page-writing code.
    The problem is now fixed in CVS, and will appear as 0.5.4, hopefully soon.
    
    -Paul
    
    
    -- 
    Paul H. Hargrove                          PHHargrove_at_lbl_dot_gov
    Future Technologies Group
    HPC Research Department                   Tel: +1-510-495-2352
    Lawrence Berkeley National Laboratory     Fax: +1-510-486-6900
    

  • Next message: Paul H. Hargrove: "BLCR 0.5.4 released"