[PATCH for 4.21 v5 1/3] xen/riscv: implement software page table walking

[Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>]

RISC-V doesn't have hardware feature to ask MMU to translate
virtual address to physical address ( like Arm has, for example ),
so software page table walking is implemented.

Signed-off-by: Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>
Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
---
Changes in v5:
 - Update the comment above _pt_walk() about returning of optional
   level of the found pte.
 - Rename local variable `pte_p *entry` to `ptep` in pt_walk() function.
 - Add Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>.
---
Changes in v4:
 - Update the comment message above _pt_walk(): add information that
   `pte_level` is optional and add a note that `table` should be
   unmapped by a caller.
 - Unmap `table` in pt_walk().
---
Changes in v3:
 - Remove circular dependency.
 - Move declaration of pt_walk() to asm/page.h.
 - Revert other not connected to pt_walk() changes.
 - Update the commit message.
 - Drop unnessary anymore local variables of pt_walk().
 - Refactor pt_walk() to use for() loop instead of while() loop
   as it was suggested by Jan B.
 - Introduce _pt_walk() which returns pte_t * and update prototype
   of pt_walk() to return pte_t by value.
---
Changes in v2:
 - Update the code of pt_walk() to return pte_t instead of paddr_t.
 - Fix typos and drop blankets inside parentheses in the comment.
 - Update the `ret` handling; there is no need for `mfn` calculation
   anymore as pt_walk() returns or pte_t of a leaf node or non-present
   pte_t.
 - Drop the comment before unmap_table().
 - Drop local variable `pa` as pt_walk() is going to return pte_t
   instead of paddr_t.
 - Add the comment above pt_walk() to explain what it does and returns.
 - Add additional explanation about possible return values of pt_next_level()
   used inside while loop in pt_walk().
 - Change `pa` to `table` in the comment before while loop in pt_walk()
   as actually this loop finds a page table where paga table entry for `va`
   is located.
 - After including <asm/page.h> in <asm/mm.h>, the following compilation
   error occurs:
    ./arch/riscv/include/asm/cmpxchg.h:56:9: error: implicit declaration of
    function 'GENMASK'
   To resolve this, <xen/bitops.h> needs to be included at the top of
   <asm/cmpxchg.h>.
 - To avoid an issue with the implicit declaration of map_domain_page() and
   unmap_domain_page() after including <asm/page.h> in <asm/mm.h>, the
   implementation of flush_page_to_ram() has been moved to mm.c. (look for
   more detailed explanation in the commit message) As a result drop
   inclusion of <xen/domain.h> in <asm/page.h>.
 - Update the commit message.
---
 xen/arch/riscv/include/asm/page.h |  2 ++
 xen/arch/riscv/pt.c               | 60 +++++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+)

diff --git a/xen/arch/riscv/include/asm/page.h 
b/xen/arch/riscv/include/asm/page.h
index 7a6174a109..0439a1a9ee 100644
--- a/xen/arch/riscv/include/asm/page.h
+++ b/xen/arch/riscv/include/asm/page.h
@@ -208,6 +208,8 @@ static inline pte_t pte_from_mfn(mfn_t mfn, unsigned int 
flags)
     return (pte_t){ .pte = pte };
 }
 
+pte_t pt_walk(vaddr_t va, unsigned int *pte_level);
+
 #endif /* __ASSEMBLY__ */
 
 #endif /* ASM__RISCV__PAGE_H */
diff --git a/xen/arch/riscv/pt.c b/xen/arch/riscv/pt.c
index a703e0f1bd..9c1f8f6b55 100644
--- a/xen/arch/riscv/pt.c
+++ b/xen/arch/riscv/pt.c
@@ -185,6 +185,66 @@ static int pt_next_level(bool alloc_tbl, pte_t **table, 
unsigned int offset)
     return XEN_TABLE_NORMAL;
 }
 
+/*
+ * _pt_walk() performs software page table walking and returns the pte_t of
+ * a leaf node or the leaf-most not-present pte_t if no leaf node is found
+ * for further analysis.
+ *
+ * _pt_walk() can optionally return the level of the found pte. Pass NULL
+ * for `pte_level` if this information isn't needed.
+ *
+ * Note: unmapping of final `table` should be done by a caller.
+ */
+static pte_t *_pt_walk(vaddr_t va, unsigned int *pte_level)
+{
+    const mfn_t root = get_root_page();
+    unsigned int level;
+    pte_t *table;
+
+    DECLARE_OFFSETS(offsets, va);
+
+    table = map_table(root);
+
+    /*
+     * Find `table` of an entry which corresponds to `va` by iterating for each
+     * page level and checking if the entry points to a next page table or
+     * to a page.
+     *
+     * Two cases are possible:
+     * - ret == XEN_TABLE_SUPER_PAGE means that the entry was found;
+     *   (Despite the name) XEN_TABLE_SUPER_PAGE also covers 4K mappings. If
+     *   pt_next_level() is called for page table level 0, it results in the
+     *   entry being a pointer to a leaf node, thereby returning
+     *   XEN_TABLE_SUPER_PAGE, despite of the fact this leaf covers 4k mapping.
+     * - ret == XEN_TABLE_MAP_NONE means that requested `va` wasn't actually
+     *   mapped.
+     */
+    for ( level = HYP_PT_ROOT_LEVEL; ; --level )
+    {
+        int ret = pt_next_level(false, &table, offsets[level]);
+
+        if ( ret == XEN_TABLE_MAP_NONE || ret == XEN_TABLE_SUPER_PAGE )
+            break;
+
+        ASSERT(level);
+    }
+
+    if ( pte_level )
+        *pte_level = level;
+
+    return table + offsets[level];
+}
+
+pte_t pt_walk(vaddr_t va, unsigned int *pte_level)
+{
+    pte_t *ptep = _pt_walk(va, pte_level);
+    pte_t pte = *ptep;
+
+    unmap_table(ptep);
+
+    return pte;
+}
+
 /* Update an entry at the level @target. */
 static int pt_update_entry(mfn_t root, vaddr_t virt,
                            mfn_t mfn, unsigned int target,
-- 
2.48.1