On venerdì 20 gennaio 2023 06:56:01 CET Al Viro wrote:
> On Fri, Jan 20, 2023 at 05:07:48AM +0000, Al Viro wrote:
> > On Fri, Jan 20, 2023 at 04:54:51AM +0000, Matthew Wilcox wrote:
> > > > Sure, but... there's also this:
> > > >
> > > > static inline void __kunmap_local(const void *addr)
> > > > {
> > > > #ifdef ARCH_HAS_FLUSH_ON_KUNMAP
> > > >
> > > > kunmap_flush_on_unmap(addr);
> > > >
> > > > #endif
> > > > }
> > > >
> > > > Are you sure that the guts of that thing will be happy with address
that
> > > > is not page-aligned? I've looked there at some point, got scared of
> > > > parisc (IIRC) MMU details and decided not to rely upon that...
> > >
> > > Ugh, PA-RISC (the only implementer) definitely will flush the wrong
> > > addresses. I think we should do this, as having bugs that only manifest
> > > on one not-well-tested architecture seems Bad.
> > >
> > > static inline void __kunmap_local(const void *addr)
> > > {
> > > #ifdef ARCH_HAS_FLUSH_ON_KUNMAP
> > >
> > > - kunmap_flush_on_unmap(addr);
> > > + kunmap_flush_on_unmap(PAGE_ALIGN_DOWN(addr));
> > >
> > > #endif
> > > }
> >
> > PTR_ALIGN_DOWN(addr, PAGE_SIZE), perhaps?
>
> Anyway, that's a question to parisc folks; I _think_ pdtlb
> quietly ignores the lower bits of address, so that part seems
> to be safe, but I wouldn't bet upon that. And when I got to
> flush_kernel_dcache_page_asm I gave up - it's been a long time
> since I've dealt with parisc assembler.
There seems to be consensus that __kunmap_local() needs to be fixed for the
parisc case (ARCH_HAS_FLUSH_ON_KUNMAP).
Is anyone doing this task?
If you agree, I could make this change and give proper credits for the tip.
Thank you,
Fabio
