diff mbox series

[3/9] xen/arm32: head: Don't map too much in boot_third

Message ID 20230625204907.57291-4-julien@xen.org (mailing list archive)
State New, archived
Headers show
Series xen/arm: Enable UBSAN support | expand

Commit Message

Julien Grall June 25, 2023, 8:49 p.m. UTC 
From: Julien Grall <jgrall@amazon.com>

At the moment, we are mapping the size of the reserved area for Xen
(i.e. 2MB) even if the binary is smaller. We don't exactly know what's
after Xen, so it is not a good idea to map more than necessary for a
couple of reasons:
    * We would need to use break-before-make if the extra PTE needs to
      be updated to point to another region
    * The extra area mapped may be mapped again by Xen with different
      memory attribute. This would result to attribue mismatch.

Therefore, rework the logic in create_page_tables() to map only what's
necessary. To simplify the logic, we also want to make sure _end
is page-aligned. So align the symbol in the linker and add an assert
to catch any change.

Signed-off-by: Julien Grall <jgrall@amazon.com>
---
 xen/arch/arm/arm32/head.S | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

Comments

Michal Orzel June 26, 2023, 11:30 a.m. UTC | #1 
On 25/06/2023 22:49, Julien Grall wrote:
> 
> 
> From: Julien Grall <jgrall@amazon.com>
> 
> At the moment, we are mapping the size of the reserved area for Xen
> (i.e. 2MB) even if the binary is smaller. We don't exactly know what's
> after Xen, so it is not a good idea to map more than necessary for a
> couple of reasons:
>     * We would need to use break-before-make if the extra PTE needs to
>       be updated to point to another region
>     * The extra area mapped may be mapped again by Xen with different
>       memory attribute. This would result to attribue mismatch.
> 
> Therefore, rework the logic in create_page_tables() to map only what's
> necessary. To simplify the logic, we also want to make sure _end
> is page-aligned. So align the symbol in the linker and add an assert
> to catch any change.
> 
> Signed-off-by: Julien Grall <jgrall@amazon.com>
The same comments apply as for the previous patch. But all in all:
Reviewed-by: Michal Orzel <michal.orzel@amd.com>

~Michal
diff mbox series

Patch

diff --git a/xen/arch/arm/arm32/head.S b/xen/arch/arm/arm32/head.S
index f9f7be9588b1..997c8a4fbbc1 100644
--- a/xen/arch/arm/arm32/head.S
+++ b/xen/arch/arm/arm32/head.S
@@ -462,6 +462,19 @@  create_page_tables:
         create_table_entry boot_pgtable, boot_second, r0, 1
         create_table_entry boot_second, boot_third, r0, 2
 
+        /*
+         * Find the size of Xen in pages and multiply by the size of a
+         * PTE. This will then be compared in the mapping loop below.
+         *
+         * Note the multiplication is just to avoid using an extra
+         * register/instruction per iteration.
+         */
+        mov_w r0, _start            /* r0 := vaddr(_start) */
+        mov_w r1, _end              /* r1 := vaddr(_end) */
+        sub   r0, r1, r0            /* r0 := effective size of Xen */
+        lsr   r0, r0, #PAGE_SHIFT   /* r0 := Number of pages for Xen */
+        lsl   r0, r0, #3            /* r0 := Number of pages * PTE size */
+
         /* Setup boot_third: */
         adr_l r4, boot_third
 
@@ -476,7 +489,7 @@  create_page_tables:
 1:      strd  r2, r3, [r4, r1]       /* Map vaddr(start) */
         add   r2, r2, #PAGE_SIZE     /* Next page */
         add   r1, r1, #8             /* Next slot */
-        cmp   r1, #(XEN_PT_LPAE_ENTRIES<<3) /* 512*8-byte entries per page */
+        cmp   r1, r0                 /* Loop until we map all of Xen */
         blo   1b
 
         /*