On Fri, 19 Oct 2012, Greg Freemyer wrote:

I saw a thread about disk i/o that may relate.  At least with disk
buffers in the block layer, it appeared it was an all or nothing
situation for flushing the queues.  So if a upper level wanted to
ensure write A completed prior to write B, the only option was:

- write A (places data in the block queue)
- flush buffers and write all data in the block queue regardless of source
- write B (places data in the block queue)

I gathered that several of the people in the thread were surprised
at the lack of granularity, but no better solution was offered, nor
even proposed for the future.

Thus, I "assume" the calls you see in the block i/o stack are truly needed.

  i'm still not convinced of the logic here.  i can see that while one
can create one's own workqueue, there is a single kernel-wide work
queue that you can take advantage of.  but is that single kernel-wide
WQ meant for things that are really important?  or really trivial?

  and if some kernel subsystem truly needs a work queue, wouldn't it
make sense to create its own?  whoops, hang on, it appears that
flush_scheduled_work() has been deprecated for a while:

https://www.google.ca/webhp?sourceid=chrome-instant&ie=UTF-8#hl=en&output=search&sclient=psy-ab&q=flush_scheduled_work%20deprecated&oq=&gs_l=&pbx=1&fp=7ac18a866ba087&bpcl=35466521&bav=on.2,or.r_gc.r_pw.r_cp.r_qf.&biw=1018&bih=912

it just hasn't gone away yet.  i will read further ...

rday

-- 

========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================