Re: [PATCH 4/6] xen/pvh: bootup and setup related changes.

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Mon, Oct 22, 2012 at 02:34:44PM +0100, Stefano Stabellini wrote:
> On Sat, 20 Oct 2012, Konrad Rzeszutek Wilk wrote:
> > From: Mukesh Rathor <mukesh.rathor@oracle.com>
> > 
> > In enlighten.c for PVH we can trap cpuid via vmexit, so don't
> > need to use emulated prefix call. We also check for vector callback
> > early on, as it is a required feature. PVH also runs at default kernel
> > IOPL.
> > 
> > In setup.c, in xen_add_extra_mem() we can skip updating P2M as it's managed
> > by Xen. PVH maps the entire IO space, but only RAM pages need to be repopulated.
> > 
> > Finally, pure PV settings are moved to a separate function that is only called
> > for pure PV, ie, pv with pvmmu.
> > 
> > Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
> > Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> >
> > diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
> > index 8971a26..8cce47b 100644
> > --- a/arch/x86/xen/setup.c
> > +++ b/arch/x86/xen/setup.c
> > @@ -27,6 +27,7 @@
> >  #include <xen/interface/memory.h>
> >  #include <xen/interface/physdev.h>
> >  #include <xen/features.h>
> > +#include "mmu.h"
> >  #include "xen-ops.h"
> >  #include "vdso.h"
> >  
> > @@ -78,6 +79,9 @@ static void __init xen_add_extra_mem(u64 start, u64 size)
> >  
> >  	memblock_reserve(start, size);
> >  
> > +	if (xen_feature(XENFEAT_auto_translated_physmap))
> > +		return;
> > +
> >  	xen_max_p2m_pfn = PFN_DOWN(start + size);
> >  	for (pfn = PFN_DOWN(start); pfn < xen_max_p2m_pfn; pfn++) {
> >  		unsigned long mfn = pfn_to_mfn(pfn);
> > @@ -100,6 +104,7 @@ static unsigned long __init xen_do_chunk(unsigned long start,
> >  		.domid        = DOMID_SELF
> >  	};
> >  	unsigned long len = 0;
> > +	int xlated_phys = xen_feature(XENFEAT_auto_translated_physmap);
> >  	unsigned long pfn;
> >  	int ret;
> >  
> > @@ -113,7 +118,7 @@ static unsigned long __init xen_do_chunk(unsigned long start,
> >  				continue;
> >  			frame = mfn;
> >  		} else {
> > -			if (mfn != INVALID_P2M_ENTRY)
> > +			if (!xlated_phys && mfn != INVALID_P2M_ENTRY)
> >  				continue;
> >  			frame = pfn;
> >  		}
> 
> Shouldn't we add a "!xlated_phys &&" to the other case as well?

No. That is handled in xen_pvh_identity_map_chunk which
just does a xen_set_clr_mmio_pvh_pte call for the "released"
pages. But that is a bit of ... well, extra logic. I think
if we did this it would work and look much nicer:


diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c
index 8cce47b..b451a77 100644
--- a/arch/x86/xen/setup.c
+++ b/arch/x86/xen/setup.c
@@ -114,9 +114,15 @@ static unsigned long __init xen_do_chunk(unsigned long start,
 
 		if (release) {
 			/* Make sure pfn exists to start with */
-			if (mfn == INVALID_P2M_ENTRY || mfn_to_pfn(mfn) != pfn)
+			if (mfn == INVALID_P2M_ENTRY || (!xlated_phys && (mfn_to_pfn(mfn) != pfn)))
 				continue;
 			frame = mfn;
+			/* The hypercall PHYSDEVOP_map_iomem to release memory has already
+			 * happend, so we just do a nop here. */
+			if (xlated_phys) {
+				len++;
+				continue;
+			}
 		} else {
 			if (!xlated_phys && mfn != INVALID_P2M_ENTRY)
 				continue;
@@ -219,15 +225,24 @@ static void __init xen_set_identity_and_release_chunk(
 {
 	unsigned long pfn;
 
+	/* For PVH, the pfns [0..MAX] are mapped to mfn's in the EPT/NPT. The mfns
+	 * are released as part of this 1:1 mapping hypercall back to the dom heap.
+	 * Also, we map the entire IO space, ie, beyond max_pfn_mapped.
+	 */
+	int xlated_phys = xen_feature(XENFEAT_auto_translated_physmap);
+
 	/*
 	 * If the PFNs are currently mapped, the VA mapping also needs
 	 * to be updated to be 1:1.
 	 */
-	for (pfn = start_pfn; pfn <= max_pfn_mapped && pfn < end_pfn; pfn++)
-		(void)HYPERVISOR_update_va_mapping(
-			(unsigned long)__va(pfn << PAGE_SHIFT),
-			mfn_pte(pfn, PAGE_KERNEL_IO), 0);
-
+	for (pfn = start_pfn; pfn <= max_pfn_mapped && pfn < end_pfn; pfn++) {
+		if (xlated_phys)
+			xen_set_clr_mmio_pvh_pte(pfn, pfn, 1 /* one pfn */, 1 /* add mapping */);
+		else
+			(void)HYPERVISOR_update_va_mapping(
+				(unsigned long)__va(pfn << PAGE_SHIFT),
+				mfn_pte(pfn, PAGE_KERNEL_IO), 0);
+	}
 	if (start_pfn < nr_pages)
 		*released += xen_release_chunk(
 			start_pfn, min(end_pfn, nr_pages));
@@ -235,27 +250,6 @@ static void __init xen_set_identity_and_release_chunk(
 	*identity += set_phys_range_identity(start_pfn, end_pfn);
 }
 
-/* For PVH, the pfns [0..MAX] are mapped to mfn's in the EPT/NPT. The mfns
- * are released as part of this 1:1 mapping hypercall back to the dom heap.
- * Also, we map the entire IO space, ie, beyond max_pfn_mapped.
- */
-static void __init xen_pvh_identity_map_chunk(unsigned long start_pfn,
-		unsigned long end_pfn, unsigned long *released,
-		unsigned long *identity, unsigned long max_pfn)
-{
-	unsigned long pfn;
-	int numpfns = 1, add_mapping = 1;
-
-	for (pfn = start_pfn; pfn < end_pfn; pfn++)
-		xen_set_clr_mmio_pvh_pte(pfn, pfn, numpfns, add_mapping);
-
-	if (start_pfn <= max_pfn) {
-		unsigned long end = min(max_pfn_mapped, end_pfn);
-		*released += end - start_pfn;
-	}
-	*identity += end_pfn - start_pfn;
-}
-
 static unsigned long __init xen_set_identity_and_release(
 	const struct e820entry *list, size_t map_size, unsigned long nr_pages)
 {
@@ -286,17 +280,10 @@ static unsigned long __init xen_set_identity_and_release(
 			if (entry->type == E820_RAM)
 				end_pfn = PFN_UP(entry->addr);
 
-			if (start_pfn < end_pfn) {
-				if (xlated_phys) {
-					xen_pvh_identity_map_chunk(start_pfn,
-						end_pfn, &released, &identity,
-						nr_pages);
-				} else {
-					xen_set_identity_and_release_chunk(
+			if (start_pfn < end_pfn)
+				xen_set_identity_and_release_chunk(
 						start_pfn, end_pfn, nr_pages,
 						&released, &identity);
-				}
-			}
 			start = end;
 		}
 	}