On Tue, May 03, 2016 at 05:48:47PM +0200, Alexander Gordeev wrote:
> On Tue, May 03, 2016 at 02:55:17PM +0200, Andrew Jones wrote:
> > On Tue, May 03, 2016 at 01:48:11PM +0200, Alexander Gordeev wrote:
> > > Cc: Andrew Jones <drjones@xxxxxxxxxx>
> > > Cc: Thomas Huth <thuth@xxxxxxxxxx>
> > > Cc: Radim Krčmář <rkrcmar@xxxxxxxxxx>
> > > Signed-off-by: Alexander Gordeev <agordeev@xxxxxxxxxx>
> > > ---
> > > lib/x86/asm/io.h | 2 ++
> > > lib/x86/io.c | 10 ++++++++++
> > > x86/vmexit.c | 10 ++--------
> > > 3 files changed, 14 insertions(+), 8 deletions(-)
> > >
> > > diff --git a/lib/x86/asm/io.h b/lib/x86/asm/io.h
> > > index 83387b5..2436822 100644
> > > --- a/lib/x86/asm/io.h
> > > +++ b/lib/x86/asm/io.h
> > > @@ -49,4 +49,6 @@ static inline void *phys_to_virt(unsigned long phys)
> > > return (void *)phys;
> > > }
> > >
> > > +void __iomem *ioremap(phys_addr_t phys_addr, size_t size);
> > > +
> > > #endif
> > > diff --git a/lib/x86/io.c b/lib/x86/io.c
> > > index d396d42..f106842 100644
> > > --- a/lib/x86/io.c
> > > +++ b/lib/x86/io.c
> > > @@ -1,6 +1,8 @@
> > > #include "libcflat.h"
> > > +#include "vm.h"
> > > #include "smp.h"
> > > #include "asm/io.h"
> > > +#include "asm/page.h"
> > > #ifndef USE_SERIAL
> > > #define USE_SERIAL
> > > #endif
> > > @@ -81,3 +83,11 @@ void exit(int code)
> > > asm volatile("out %0, %1" : : "a"(code), "d"((short)0xf4));
> > > #endif
> > > }
> > > +
> > > +void __iomem *ioremap(phys_addr_t phys_addr, size_t size)
> > > +{
> > > + phys_addr_t base = phys_addr & PAGE_MASK;
> > > + phys_addr_t offset = phys_addr - base;
> > > +
> > > + return vmap(phys_addr, size) + offset;
> >
> > yeah, vmap is the way to go! Now, one missing thing, but it wasn't
> > in x86/vmexit's implementation either. The kernel sets the PTEs
> > for an ioremap with PCD, page cache disabled. I guess we can live
> > without that in kvm-unit-tests, but it would make sense that I/O
> > mappings would be uncached - and may help us find bugs when we
> > properly map that way. (Note, ARM gets away with just using
> > asm-generic's ioremap, because it uses an identity map for the page
> > tables and all I/O is in a single range which is mapped as uncached)
>
> I would say, the above is better to put in the changelog for this patch,
> huh?
Sure, or a comment in the ioremap function.
thanks,
drew
>
> > > +}
> > > diff --git a/x86/vmexit.c b/x86/vmexit.c
> > > index db7dbd8..c2e1e49 100644
> > > --- a/x86/vmexit.c
> > > +++ b/x86/vmexit.c
> > > @@ -371,8 +371,7 @@ int main(int ac, char **av)
> > > {
> > > struct fadt_descriptor_rev1 *fadt;
> > > int i;
> > > - unsigned long membar = 0, base, offset;
> > > - void *m;
> > > + unsigned long membar = 0;
> > > pcidevaddr_t pcidev;
> > >
> > > smp_init();
> > > @@ -394,12 +393,7 @@ int main(int ac, char **av)
> > > }
> > > if (pci_bar_is_memory(pcidev, i)) {
> > > membar = pci_bar_addr(pcidev, i);
> > > - base = membar & ~4095;
> > > - offset = membar - base;
> > > - m = alloc_vpages(1);
> > > -
> > > - install_page((void *)read_cr3(), base, m);
> > > - pci_test.memaddr = m + offset;
> > > + pci_test.memaddr = ioremap(membar, PAGE_SIZE);
> > > } else {
> > > pci_test.iobar = pci_bar_addr(pcidev, i);
> > > }
> > > --
> > > 1.8.3.1
> > >
> >
> > Reviewed-by: Andrew Jones <drjones@xxxxxxxxxx>
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
