[v3] x86/vmap: handle superpages in vmap_to_mfn()

Commit Message

Hongyan Xia Dec. 7, 2020, 12:49 p.m. UTC

From: Hongyan Xia <[e-mail]>

There is simply no guarantee that vmap won't return superpages to the
caller. It can happen if the list of MFNs are contiguous, or we simply
have a large granularity. Although rare, if such things do happen, we
will simply hit BUG_ON() and crash.

Introduce xen_map_to_mfn() to translate any mapped Xen address to mfn
regardless of page size, and wrap vmap_to_mfn() around it.

Signed-off-by: Hongyan Xia <[e-mail]>
Reviewed-by: Jan Beulich <[e-mail]>

---
Changed in v3:
- switch to do-while.
- move the declaration close to map_pages_to_xen().
- add missing parentheses to vmap_to_mfn().

Changed in v2:
- const pl*e
- introduce xen_map_to_mfn().
- goto to a single exit path.
- ASSERT_UNREACHABLE instead of ASSERT.
---
 xen/arch/x86/mm.c          | 56 ++++++++++++++++++++++++++++++++++++++
 xen/include/asm-x86/page.h |  2 +-
 xen/include/xen/mm.h       |  1 +
 3 files changed, 58 insertions(+), 1 deletion(-)

Comments

Jan Beulich Dec. 7, 2020, 1:14 p.m. UTC | #1

On 07.12.2020 13:49, Hongyan Xia wrote:
> --- a/xen/include/xen/mm.h
> +++ b/xen/include/xen/mm.h
> @@ -175,6 +175,7 @@ bool scrub_free_pages(void);
>  } while ( false )
>  #define FREE_XENHEAP_PAGE(p) FREE_XENHEAP_PAGES(p, 0)
>  
> +mfn_t xen_map_to_mfn(unsigned long va);
>  /* Map machine page range in Xen virtual address space. */
>  int map_pages_to_xen(
>      unsigned long virt,
> 

This really would have wanted to be below the sibling functions
and then, following suit, have a one line comment. Will adjust
while committing; sorry to be picky.

Jan

Patch

diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c
index 5a50339284c7..723cc1070f16 100644
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -5194,6 +5194,62 @@  l1_pgentry_t *virt_to_xen_l1e(unsigned long v)
         }                                          \
     } while ( false )
 
+/* Translate mapped Xen address to MFN. */
+mfn_t xen_map_to_mfn(unsigned long va)
+{
+#define CHECK_MAPPED(cond)          \
+    do {                            \
+        if ( !(cond) )              \
+        {                           \
+            ASSERT_UNREACHABLE();   \
+            ret = INVALID_MFN;      \
+            goto out;               \
+        }                           \
+    } while ( false )
+
+    bool locking = system_state > SYS_STATE_boot;
+    unsigned int l2_offset = l2_table_offset(va);
+    unsigned int l1_offset = l1_table_offset(va);
+    const l3_pgentry_t *pl3e = virt_to_xen_l3e(va);
+    const l2_pgentry_t *pl2e = NULL;
+    const l1_pgentry_t *pl1e = NULL;
+    struct page_info *l3page;
+    mfn_t ret;
+
+    L3T_INIT(l3page);
+    CHECK_MAPPED(pl3e);
+    l3page = virt_to_page(pl3e);
+    L3T_LOCK(l3page);
+
+    CHECK_MAPPED(l3e_get_flags(*pl3e) & _PAGE_PRESENT);
+    if ( l3e_get_flags(*pl3e) & _PAGE_PSE )
+    {
+        ret = mfn_add(l3e_get_mfn(*pl3e),
+                      (l2_offset << PAGETABLE_ORDER) + l1_offset);
+        goto out;
+    }
+
+    pl2e = map_l2t_from_l3e(*pl3e) + l2_offset;
+    CHECK_MAPPED(l2e_get_flags(*pl2e) & _PAGE_PRESENT);
+    if ( l2e_get_flags(*pl2e) & _PAGE_PSE )
+    {
+        ret = mfn_add(l2e_get_mfn(*pl2e), l1_offset);
+        goto out;
+    }
+
+    pl1e = map_l1t_from_l2e(*pl2e) + l1_offset;
+    CHECK_MAPPED(l1e_get_flags(*pl1e) & _PAGE_PRESENT);
+    ret = l1e_get_mfn(*pl1e);
+
+#undef CHECK_MAPPED
+ out:
+    L3T_UNLOCK(l3page);
+    unmap_domain_page(pl1e);
+    unmap_domain_page(pl2e);
+    unmap_domain_page(pl3e);
+    return ret;
+}
+
 int map_pages_to_xen(
     unsigned long virt,
     mfn_t mfn,
diff --git a/xen/include/asm-x86/page.h b/xen/include/asm-x86/page.h
index 7a771baf7cb3..082c14a66226 100644
--- a/xen/include/asm-x86/page.h
+++ b/xen/include/asm-x86/page.h
@@ -291,7 +291,7 @@  void copy_page_sse2(void *, const void *);
 #define pfn_to_paddr(pfn)   __pfn_to_paddr(pfn)
 #define paddr_to_pfn(pa)    __paddr_to_pfn(pa)
 #define paddr_to_pdx(pa)    pfn_to_pdx(paddr_to_pfn(pa))
-#define vmap_to_mfn(va)     l1e_get_mfn(*virt_to_xen_l1e((unsigned long)(va)))
+#define vmap_to_mfn(va)     xen_map_to_mfn((unsigned long)(va))
 #define vmap_to_page(va)    mfn_to_page(vmap_to_mfn(va))
 
 #endif /* !defined(__ASSEMBLY__) */
diff --git a/xen/include/xen/mm.h b/xen/include/xen/mm.h
index f7975b2df00b..1475f352e411 100644
--- a/xen/include/xen/mm.h
+++ b/xen/include/xen/mm.h
@@ -175,6 +175,7 @@  bool scrub_free_pages(void);
 } while ( false )
 #define FREE_XENHEAP_PAGE(p) FREE_XENHEAP_PAGES(p, 0)
 
+mfn_t xen_map_to_mfn(unsigned long va);
 /* Map machine page range in Xen virtual address space. */
 int map_pages_to_xen(
     unsigned long virt,