On Tue, 1 Jun 2010, Tony Luck wrote: > On Sat, May 29, 2010 at 10:56 PM, Julia Lawall <julia@xxxxxxx> wrote: > > I have the impression that ioremap should not be called with locks held. > > Is this the case? If so, what should be done in the following code (from > > the function iosapic_init): > > This code is probably safe because on any sane system the I/O address > being mapped should not fall into any of the complex cases in ioremap(). > So we'll just end up in __ioremap() ... which just does some simple math > to get the region 6 address that provides uncacheable access to the > physical address - which can safely be done while holding a lock. > > Did you find this because you have a code scanner that is flagging > this as a problem? Yes, but the question was more because I was trying to understand whether the scanner should be looking for ioremap problems or not. Thanks for the explanation. julia
|