hi, > Emmanuel Dreyfus <[email protected]> wrote: > >> > i found the following patch in my tree. >> > unfortunately i forgot details and if there were more cases which needs >> > similar barriers. >> This fixes the problem with no detectable performance hit. Is there >> any reason why it should not be committed?
i forgot. :-) > > Extensive testing shows that the same kind of bug exists if puffs_init() > is called without PUFFS_KFLAG_WTCACHE. This happens on netbsd-5 with and > without your patch. > > The difference is that the chunk of zero spans multiple pages (but its > length is not page-aligned) and is always located at offset 0. Example > of output from my test case ( size of zero-filled chunk @ offset ) > 21071 @ 0 > 9145 @ 0 > 26493 @ 0 > 20772 @ 0 have you tried to analyze the problem with PUFFS_FLAG_OPDUMP? does your server process requests from kernel simultaneously? eg. multithread YAMAMOTO Takashi > -- > Emmanuel Dreyfus > http://hcpnet.free.fr/pubz > [email protected]
