Re: Overlapping ioremap() calls, set_memory_*() semantics

[Toshi Kani <toshi.kani@hpe.com>]

On Wed, 2016-04-13 at 23:16 +0200, Luis R. Rodriguez wrote:
> On Thu, Mar 17, 2016 at 04:44:53PM -0600, Toshi Kani wrote:
> > On Wed, 2016-03-16 at 02:45 +0100, Luis R. Rodriguez wrote:
> > > On Fri, Mar 11, 2016 at 03:13:52PM -0700, Toshi Kani wrote:
:
> > > 
> > > If aliasing with different cache attributes is not allowed for x86
> > > and if its also rare for other architectures that just leaves the
> > > hunt for valid aliasing uses. That still may be hard to hunt for, but
> > > I also suspect it may be rare.
> >
> > Yes, I'd fail the different cache attribute case if we are to place
> > more strict check.
>
> OK it seems this is a good starting point. How can we get a general
> architecture consensus aliasing with different cache attributes is a
> terrible idea ? Perhaps a patch to WARN/error out and let architectures
> opt in to this piece of code?

I expect aliasing with different cache attributes is a bad idea on most
architectures.  Given the fact that track_pfn_remap(), track_pfn_insert(),
etc. are only implemented on x86, I suspect that other architectures would
not be able to implement such check easily, though.  

On x86, ioremap() and remap_pfn_range() already fail on a conflicting cache
type if it is not allowed by the rule defined in is_new_memtype_allowed().
 This exception handling is necessary for remap_pfn_range() called by
/dev/mem, but I do not think it's necessary for ioremap().  I think we can
start from adding a warning message to ioremap().


> > > > > Are there cases where we change the caching attribute of RAM for
> > > > > valid reasons, outside of legacy quirks?
> > > >
> > > > ati_create_page_map() is one example that it gets a RAM page
> > > > by __get_free_page(), and changes it to UC by calling set_memory_uc
> > > > ().
> > >
> > > Should we instead have an API that lets it ask for RAM and of UC
> > > type? That would seem a bit cleaner. BTW do you happen to know *why*
> > > it needs UC RAM types?
> >
> > This RAM page is then shared between graphic card and CPU.  I think
> > this is because graphic card cannot snoop the cache.
>
> Was this reason alone sufficient to open such APIs broadly for RAM?

According to commit 75cbade8ea3, such APIs were introduced because drivers
previously had to deal with low-level staff.  So, I think we need to keep
them as long as we have such drivers...

Thanks,
-Toshi