On Thu, Sep 29, 2016 at 01:54:04AM -0400, David Miller wrote: >From: Gavin Shan <gws...@linux.vnet.ibm.com> >Date: Thu, 29 Sep 2016 15:03:08 +1000 > >> This replaces the atomic access to NCSI channel's state with READ_ONCE() >> and WRITE_ONCE() to avoid the above build warning. We needn't hold the >> channel's lock when updating its state as well. No logical changes >> introduced. > >I don't understand this. > >If it's important to take the lock for the list add/del, then it must >be important to make the state change appear atomic wrt. that lock as >well. > >Can parallel threads of control enter these functions which change the >state? If so, then you need to make the state changes under the lock. >In fact, you probably have to make the state tests under the locks as >well. > >If not, please explain what prevents it from happening. >
Dave, thanks for your comments. I think it's occasionally working on AST2400 and AST2500 platforms. It's reasonable to grab the lock before fetching or updating the NCSI channel's state. Adding and removing the channel from the list also need taking the lock as well. I will modify the code accordingly in next revision. AST2400/AST2500 has single CPU. The channel's state (and the linked list) are changed in softirq context (packet Rx handler or timer), meaning they are not accessed in parallel mode. However, NCSI stack cannot make assumption to be run on single CPU platforms only. So yes, we need the lock to protect them. Thanks, Gavin
