在 2012年5月3日星期四，MORITA Kazutaka 写道： > At Thu, 03 May 2012 18:34:54 +0800, > Liu Yuan wrote: > > > > On 05/03/2012 06:22 PM, MORITA Kazutaka wrote: > > > > > The old code also has the completely same problem; worker threads call > > > get_ordered_sd_vnode_list()/free_ordered_sd_vnode_list(). The correct > > > approach is to modify the code so that all > > > get_vnode_info()/put_vnode_info() are called only in the main thread. > > > I'll do the work if no one is trying to fix it now. I'm going to fix it. > > > > > how about introducing lockless mechanism such as Read Copy Update (RCU)? > > It seems that we can make use of shared variables between main and > > worker threads without locks and also get the benefit to avoid a rework? > > get/put_vnode_info() might be a good starting point to use it. > > > > There is also a user level implementation URCU (http://lttng.org/urcu). > > I'm not against introducing it if the library is mature, but I'm > afraid that it would make it harder to ensure the correctness of the > program. If possible, I don't want to access shared variables from > other than the main thread. > Thanks, > > Kazutaka > -- > sheepdog mailing list > sheepdog at lists.wpkg.org > http://lists.wpkg.org/mailman/listinfo/sheepdog > -- Yunkai Zhang Work at Taobao -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.wpkg.org/pipermail/sheepdog/attachments/20120503/f8ecb0e2/attachment.html> |