[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 13/28] ARM: vGICv3: handle virtual LPI pending and property tables



Hi Andre,

On 30/01/17 18:31, Andre Przywara wrote:
Allow a guest to provide the address and size for the memory regions
it has reserved for the GICv3 pending and property tables.
We sanitise the various fields of the respective redistributor
registers and map those pages into Xen's address space to have easy
access.

Signed-off-by: Andre Przywara <andre.przywara@xxxxxxx>
---
 xen/arch/arm/vgic-v3.c       | 220 +++++++++++++++++++++++++++++++++++++++----
 xen/arch/arm/vgic.c          |   4 +
 xen/include/asm-arm/domain.h |   8 +-
 xen/include/asm-arm/vgic.h   |  24 ++++-
 4 files changed, 233 insertions(+), 23 deletions(-)

diff --git a/xen/arch/arm/vgic-v3.c b/xen/arch/arm/vgic-v3.c
index b0653c2..c6db2d7 100644
--- a/xen/arch/arm/vgic-v3.c
+++ b/xen/arch/arm/vgic-v3.c
@@ -20,12 +20,14 @@

 #include <xen/bitops.h>
 #include <xen/config.h>
+#include <xen/domain_page.h>
 #include <xen/lib.h>
 #include <xen/init.h>
 #include <xen/softirq.h>
 #include <xen/irq.h>
 #include <xen/sched.h>
 #include <xen/sizes.h>
+#include <xen/vmap.h>
 #include <asm/current.h>
 #include <asm/mmio.h>
 #include <asm/gic_v3_defs.h>
@@ -229,12 +231,15 @@ static int __vgic_v3_rdistr_rd_mmio_read(struct vcpu *v, 
mmio_info_t *info,
         goto read_reserved;

     case VREG64(GICR_PROPBASER):
-        /* LPI's not implemented */
-        goto read_as_zero_64;
+        if ( !vgic_reg64_check_access(dabt) ) goto bad_width;
+        *r = vgic_reg64_extract(v->domain->arch.vgic.rdist_propbase, info);
+        return 1;

     case VREG64(GICR_PENDBASER):
-        /* LPI's not implemented */
-        goto read_as_zero_64;
+        if ( !vgic_reg64_check_access(dabt) ) goto bad_width;
+        *r = vgic_reg64_extract(v->arch.vgic.rdist_pendbase, info);
+        *r &= ~GICR_PENDBASER_PTZ;       /* WO, reads as 0 */
+        return 1;

     case 0x0080:
         goto read_reserved;
@@ -302,11 +307,6 @@ bad_width:
     domain_crash_synchronous();
     return 0;

-read_as_zero_64:
-    if ( !vgic_reg64_check_access(dabt) ) goto bad_width;
-    *r = 0;
-    return 1;
-
 read_as_zero_32:
     if ( dabt.size != DABT_WORD ) goto bad_width;
     *r = 0;
@@ -331,11 +331,179 @@ read_unknown:
     return 1;
 }

+static uint64_t vgic_sanitise_field(uint64_t reg, uint64_t field_mask,
+                                    int field_shift,
+                                    uint64_t (*sanitise_fn)(uint64_t))
+{
+    uint64_t field = (reg & field_mask) >> field_shift;
+
+    field = sanitise_fn(field) << field_shift;
+
+    return (reg & ~field_mask) | field;
+}
+
+/* We want to avoid outer shareable. */
+static uint64_t vgic_sanitise_shareability(uint64_t field)
+{
+    switch (field) {

Coding style:

switch ( field )
{

+    case GIC_BASER_OuterShareable:
+        return GIC_BASER_InnerShareable;
+    default:
+        return field;
+    }
+}
+
+/* Avoid any inner non-cacheable mapping. */
+static uint64_t vgic_sanitise_inner_cacheability(uint64_t field)
+{
+    switch (field) {

Ditto

+    case GIC_BASER_CACHE_nCnB:
+    case GIC_BASER_CACHE_nC:
+        return GIC_BASER_CACHE_RaWb;
+    default:
+        return field;
+    }
+}
+
+/* Non-cacheable or same-as-inner are OK. */
+static uint64_t vgic_sanitise_outer_cacheability(uint64_t field)
+{
+    switch (field) {

Ditto

+    case GIC_BASER_CACHE_SameAsInner:
+    case GIC_BASER_CACHE_nC:
+        return field;
+    default:
+        return GIC_BASER_CACHE_nC;
+    }
+}
+
+static uint64_t sanitize_propbaser(uint64_t reg)
+{
+    reg = vgic_sanitise_field(reg, GICR_PROPBASER_SHAREABILITY_MASK,
+                              GICR_PROPBASER_SHAREABILITY_SHIFT,
+                              vgic_sanitise_shareability);
+    reg = vgic_sanitise_field(reg, GICR_PROPBASER_INNER_CACHEABILITY_MASK,
+                              GICR_PROPBASER_INNER_CACHEABILITY_SHIFT,
+                              vgic_sanitise_inner_cacheability);
+    reg = vgic_sanitise_field(reg, GICR_PROPBASER_OUTER_CACHEABILITY_MASK,
+                              GICR_PROPBASER_OUTER_CACHEABILITY_SHIFT,
+                              vgic_sanitise_outer_cacheability);
+
+    reg &= ~GICR_PROPBASER_RES0_MASK;

Newline here please.

+    return reg;
+}
+
+static uint64_t sanitize_pendbaser(uint64_t reg)
+{
+    reg = vgic_sanitise_field(reg, GICR_PENDBASER_SHAREABILITY_MASK,
+                              GICR_PENDBASER_SHAREABILITY_SHIFT,
+                              vgic_sanitise_shareability);
+    reg = vgic_sanitise_field(reg, GICR_PENDBASER_INNER_CACHEABILITY_MASK,
+                              GICR_PENDBASER_INNER_CACHEABILITY_SHIFT,
+                              vgic_sanitise_inner_cacheability);
+    reg = vgic_sanitise_field(reg, GICR_PENDBASER_OUTER_CACHEABILITY_MASK,
+                              GICR_PENDBASER_OUTER_CACHEABILITY_SHIFT,
+                              vgic_sanitise_outer_cacheability);
+
+    reg &= ~GICR_PENDBASER_RES0_MASK;

Newline here please.

+    return reg;
+}
+
+/*
+ * Mark a given number of guest pages as used (by increasing their refcount),
+ * starting with the given guest address. This needs to be called once before
+ * calling (possibly repeatedly) map_guest_pages().
+ * Before the domain gets destroyed, call put_guest_pages() to drop the
+ * reference.
+ */
+int get_guest_pages(struct domain *d, paddr_t gpa, int nr_pages)

Many comments here:
   * please use the type gfn_t rather paddr_t,
   * nr_pages should be unsigned int
   * this code does not belong to vgic-v3.c. It is a generic function.

+{
+    int i;

unsigned int.

+    struct page_info *page;
+
+    for ( i = 0; i < nr_pages; i++ )
+    {
+        page = get_page_from_gfn(d, (gpa >> PAGE_SHIFT) + i, NULL, P2M_ALLOC);

Get page may return a foreign page (e.g belonging to another domain) and we don't want to use this type of page for ITS memory.

+        if ( ! page )
+            return -EINVAL;
+    }

Should not you revert the reference on all the previous pages when you fail to get one page?

+
+    return 0;
+}
+
+void put_guest_pages(struct domain *d, paddr_t gpa, int nr_pages)

Same comments as above.

+{
+    mfn_t mfn;
+    int i;

unsigned int.

+
+    p2m_read_lock(&d->arch.p2m);
+    for ( i = 0; i < nr_pages; i++ )
+    {
+        mfn = p2m_get_entry(&d->arch.p2m, _gfn((gpa >> PAGE_SHIFT) + i),
+                            NULL, NULL, NULL);
+        if ( mfn_eq(mfn, INVALID_MFN) )
+            continue;
+        put_page(mfn_to_page(mfn_x(mfn)));

This function is completely wrong in the actual state. You assume that the stage-2 page table has not been modified by the guest between get_guest_pages and put_guest_pages. If it has been modified, you may remove a reference on the wrong page.

Furthermore, it is likely an error to have the mfn not valid in this case.

As we discussed earlier, the way forward is to protect the pages. It is not mandatory for DOM0, but a comment in the code is necessary to explain what is missing.

While I am here, for the next version I would like a todo list of all the items missing to support properly a guest. So we can track what is required before enabling ITS for guest.

+    }
+    p2m_read_unlock(&d->arch.p2m);
+}
+
+/*
+ * Provides easy access to guest memory by "mapping" some parts of it into
+ * Xen's VA space. In fact it relies on the memory being already mapped
+ * and just provides a pointer to it.
+ * This allows the ITS configuration data to be held in guest memory and
+ * avoids using Xen's memory for that.
+ */
+void *map_guest_pages(struct domain *d, paddr_t guest_addr, int nr_pages)
+{
+    int i;
+    void *ptr, *follow;
+
+    ptr = map_domain_page(_mfn(guest_addr >> PAGE_SHIFT));

This might be correct for DOM0 but will not work for guest. Even if you don't support guest right now, we should really avoid such assumption in the code. It will likely mean quite a lot of rework which I'd like to see now.

+
+    /* Make sure subsequent pages are mapped in a virtually contigious way. */

s/contigious/contiguous/

+    for ( i = 1; i < nr_pages; i++ )
+    {
+        follow = map_domain_page(_mfn((guest_addr >> PAGE_SHIFT) + i));

map_domain_page should only be used for temporary mapping and this not the case here.

As mentioned in the first version of this series, the best solution is to map/unmap the page every time we need it.

+        if ( follow != ptr + ((long)i << PAGE_SHIFT) )

In the case of the guest, the region may not be contiguous in the physical address space.

+            return NULL;
+    }
+
+    return ptr + (guest_addr & ~PAGE_MASK);
+}
+
+/* "Unmap" previously mapped guest pages. Should be optimized away on arm64. */
+void unmap_guest_pages(void *va, int nr_pages)
+{
+    long i;
+
+    for ( i = nr_pages - 1; i >= 0; i-- )
+        unmap_domain_page(((uintptr_t)va & PAGE_MASK) + (i << PAGE_SHIFT));
+}

It does not make sense to implement the 4 functions above in this patch as they are never

+
+int vgic_lpi_get_priority(struct domain *d, uint32_t vlpi)
+{
+    if ( vlpi >= d->arch.vgic.nr_lpis )
+        return GIC_PRI_IRQ;

What if we get the priority before the ITS has been enabled?

+
+    return d->arch.vgic.proptable[vlpi - LPI_OFFSET] & LPI_PROP_PRIO_MASK;

This is memory shared with the guest, you need to make sure the compiler will not read twice.

+}
+
+bool vgic_lpi_is_enabled(struct domain *d, uint32_t vlpi)
+{
+    if ( vlpi >= d->arch.vgic.nr_lpis )
+        return false;
+
+    return d->arch.vgic.proptable[vlpi - LPI_OFFSET] & LPI_PROP_ENABLED;

Ditto.

+}
+
 static int __vgic_v3_rdistr_rd_mmio_write(struct vcpu *v, mmio_info_t *info,
                                           uint32_t gicr_reg,
                                           register_t r)
 {
     struct hsr_dabt dabt = info->dabt;
+    uint64_t reg;

     switch ( gicr_reg )
     {
@@ -366,36 +534,54 @@ static int __vgic_v3_rdistr_rd_mmio_write(struct vcpu *v, 
mmio_info_t *info,
         goto write_impl_defined;

     case VREG64(GICR_SETLPIR):
-        /* LPI is not implemented */
+        /* LPIs without an ITS are not implemented */
         goto write_ignore_64;

     case VREG64(GICR_CLRLPIR):
-        /* LPI is not implemented */
+        /* LPIs without an ITS are not implemented */
         goto write_ignore_64;

     case 0x0050:
         goto write_reserved;

     case VREG64(GICR_PROPBASER):
-        /* LPI is not implemented */
-        goto write_ignore_64;
+        if ( !vgic_reg64_check_access(dabt) ) goto bad_width;
+
+        /* Writing PROPBASER with LPIs enabled is UNPREDICTABLE. */
+        if ( v->arch.vgic.flags & VGIC_V3_LPIS_ENABLED )
+            return 1;
+
+        reg = v->domain->arch.vgic.rdist_propbase;
+        vgic_reg64_update(&reg, r, info);
+        reg = sanitize_propbaser(reg);
+        v->domain->arch.vgic.rdist_propbase = reg;
+        return 1;

     case VREG64(GICR_PENDBASER):
-        /* LPI is not implemented */
-        goto write_ignore_64;
+        if ( !vgic_reg64_check_access(dabt) ) goto bad_width;
+
+        /* Writing PENDBASER with LPIs enabled is UNPREDICTABLE. */
+        if ( v->arch.vgic.flags & VGIC_V3_LPIS_ENABLED )
+            return 1;
+
+       reg = v->arch.vgic.rdist_pendbase;
+       vgic_reg64_update(&reg, r, info);
+       reg = sanitize_pendbaser(reg);
+       v->arch.vgic.rdist_pendbase = reg;
+       return 1;

The indentation is wrong.


     case 0x0080:
         goto write_reserved;

     case VREG64(GICR_INVLPIR):
-        /* LPI is not implemented */
+        /* LPIs without an ITS are not implemented */
         goto write_ignore_64;

     case 0x00A8:
         goto write_reserved;

     case VREG64(GICR_INVALLR):
-        /* LPI is not implemented */
+        /* LPIs without an ITS are not implemented */
         goto write_ignore_64;

     case 0x00B8:
diff --git a/xen/arch/arm/vgic.c b/xen/arch/arm/vgic.c
index 7e3440f..cf444f3 100644
--- a/xen/arch/arm/vgic.c
+++ b/xen/arch/arm/vgic.c
@@ -494,6 +494,10 @@ struct pending_irq *lpi_to_pending(struct vcpu *v, 
unsigned int lpi,
         empty->pirq.irq = lpi;
     }

+    /* Update the enabled status */
+    if ( vgic_lpi_is_enabled(v->domain, lpi) )
+        set_bit(GIC_IRQ_GUEST_ENABLED, &empty->pirq.status);
+
     spin_unlock(&v->arch.vgic.pending_lpi_list_lock);

     return &empty->pirq;
diff --git a/xen/include/asm-arm/domain.h b/xen/include/asm-arm/domain.h
index f44a84b..33c1851 100644
--- a/xen/include/asm-arm/domain.h
+++ b/xen/include/asm-arm/domain.h
@@ -110,6 +110,9 @@ struct arch_domain
         } *rdist_regions;
         int nr_regions;                     /* Number of rdist regions */
         uint32_t rdist_stride;              /* Re-Distributor stride */
+        int nr_lpis;

unsigned int.

+        uint64_t rdist_propbase;
+        uint8_t *proptable;
         struct rb_root its_devices;         /* devices mapped to an ITS */
         spinlock_t its_devices_lock;        /* protects the its_devices tree */
 #endif
@@ -255,7 +258,10 @@ struct arch_vcpu

         /* GICv3: redistributor base and flags for this vCPU */
         paddr_t rdist_base;
-#define VGIC_V3_RDIST_LAST  (1 << 0)        /* last vCPU of the rdist */
+#define VGIC_V3_RDIST_LAST      (1 << 0)        /* last vCPU of the rdist */
+#define VGIC_V3_LPIS_ENABLED    (1 << 1)
+        uint64_t rdist_pendbase;
+        unsigned long *pendtable;

unsigned long? Why?

         uint8_t flags;
         struct list_head pending_lpi_list;
         spinlock_t pending_lpi_list_lock;   /* protects the pending_lpi_list */
diff --git a/xen/include/asm-arm/vgic.h b/xen/include/asm-arm/vgic.h
index 03d4d2e..a882fe8 100644
--- a/xen/include/asm-arm/vgic.h
+++ b/xen/include/asm-arm/vgic.h
@@ -285,6 +285,11 @@ VGIC_REG_HELPERS(32, 0x3);

 #undef VGIC_REG_HELPERS

+int get_guest_pages(struct domain *d, paddr_t gpa, int nr_pages);
+void put_guest_pages(struct domain *d, paddr_t gpa, int nr_pages);
+void *map_guest_pages(struct domain *d, paddr_t guest_addr, int nr_pages);
+void unmap_guest_pages(void *va, int nr_pages);
+
 enum gic_sgi_mode;

 /*
@@ -312,14 +317,23 @@ extern struct vgic_irq_rank *vgic_rank_irq(struct vcpu 
*v, unsigned int irq);
 extern bool vgic_emulate(struct cpu_user_regs *regs, union hsr hsr);
 extern void vgic_disable_irqs(struct vcpu *v, uint32_t r, int n);
 extern void vgic_enable_irqs(struct vcpu *v, uint32_t r, int n);
-/* placeholder function until the property table gets introduced */
-static inline int vgic_lpi_get_priority(struct domain *d, uint32_t vlpi)
-{
-    return 0xa;
-}
 extern void register_vgic_ops(struct domain *d, const struct vgic_ops *ops);
 int vgic_v2_init(struct domain *d, int *mmio_count);
 int vgic_v3_init(struct domain *d, int *mmio_count);
+#ifdef CONFIG_HAS_GICV3
+extern int vgic_lpi_get_priority(struct domain *d, uint32_t vlpi);
+extern bool vgic_lpi_is_enabled(struct domain *d, uint32_t vlpi);
+#else
+static inline int vgic_lpi_get_priority(struct domain *d, uint32_t vlpi)
+{
+    return 0xa0;

Why 0xa0? This function should never be called when GICv3 ITS is not supported.

+}
+
+static inline bool vgic_lpi_is_enabled(struct domain *d, uint32_t vlpi)
+{
+    return false;
+}
+#endif

 extern int domain_vgic_register(struct domain *d, int *mmio_count);
 extern int vcpu_vgic_free(struct vcpu *v);


--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.