diff mbox series

[v2,14/40] xen/arm64: head: Jump to the runtime mapping in enable_mm()

Message ID 20230113052914.3845596-15-Penny.Zheng@arm.com (mailing list archive)
State New, archived
Headers show
Series xen/arm: Add Armv8-R64 MPU support to Xen - Part#1 | expand

Commit Message

Penny Zheng Jan. 13, 2023, 5:28 a.m. UTC 
At the moment, on MMU system, enable_mm() will return to an address in
the 1:1 mapping, then each path is responsible to switch to virtual runtime
mapping. Then remove_identity_mapping() is called to remove all 1:1 mapping.

Since remove_identity_mapping() is not necessary on MPU system, and we also
avoid creating empty function for MPU system, trying to keep only one codeflow
in arm64/head.S, we move path switch and remove_identity_mapping() in
enable_mm() on MMU system.

Signed-off-by: Penny Zheng <penny.zheng@arm.com>
Signed-off-by: Wei Chen <wei.chen@arm.com>
---
 xen/arch/arm/arm64/head.S     | 28 +++++++++++++---------------
 xen/arch/arm/arm64/head_mmu.S | 33 ++++++++++++++++++++++++++++++---
 2 files changed, 43 insertions(+), 18 deletions(-)

Comments

Julien Grall Feb. 5, 2023, 9:13 p.m. UTC | #1 
Hi Penny,

On 13/01/2023 05:28, Penny Zheng wrote:
> At the moment, on MMU system, enable_mm() will return to an address in
> the 1:1 mapping, then each path is responsible to switch to virtual runtime
> mapping. Then remove_identity_mapping() is called to remove all 1:1 mapping.
> 
> Since remove_identity_mapping() is not necessary on MPU system, and we also
> avoid creating empty function for MPU system, trying to keep only one codeflow
> in arm64/head.S, we move path switch and remove_identity_mapping() in
> enable_mm() on MMU system.

AFAICT, remove_identity_mapping() is still using ENTRY(). But you could 
avoid to introduce ENTRY() if you re-order your series so this patch 
happens before the MMU specific code is moved in a separate helper.

> 
> Signed-off-by: Penny Zheng <penny.zheng@arm.com>
> Signed-off-by: Wei Chen <wei.chen@arm.com>
> ---
>   xen/arch/arm/arm64/head.S     | 28 +++++++++++++---------------
>   xen/arch/arm/arm64/head_mmu.S | 33 ++++++++++++++++++++++++++++++---
>   2 files changed, 43 insertions(+), 18 deletions(-)

This will need to be rebased on top of [1] (which will be merged pretty 
soon). There are two main differences:

  1) enable_mmu has an extra parameter to take the root page-tables
  2) the remove_identity_mapping should only be called for the boot CPU.

So I think we want to introduce two functions:
  1) enable_boot_mmu
  2) enable_runtime_mmu

You might need the same for the MPU as I would expect it would be per-CPU.

Cheers,

[1] 20230127195508.2786-1-julien@xen.org
diff mbox series

Patch

diff --git a/xen/arch/arm/arm64/head.S b/xen/arch/arm/arm64/head.S
index a92883319d..6358305f03 100644
--- a/xen/arch/arm/arm64/head.S
+++ b/xen/arch/arm/arm64/head.S
@@ -258,20 +258,15 @@  real_start_efi:
          * and memory regions for MPU systems.
          */
         bl    prepare_early_mappings
+        /*
+         * Address in the runtime mapping to jump to after the
+         * MMU/MPU is enabled
+         */
+        ldr   lr, =primary_switched
         /* Turn on MMU or MPU */
-        bl    enable_mm
+        b    enable_mm
 
-        /* We are still in the 1:1 mapping. Jump to the runtime Virtual Address. */
-        ldr   x0, =primary_switched
-        br    x0
 primary_switched:
-        /*
-         * The 1:1 map may clash with other parts of the Xen virtual memory
-         * layout. As it is not used anymore, remove it completely to
-         * avoid having to worry about replacing existing mapping
-         * afterwards.
-         */
-        bl    remove_identity_mapping
         bl    setup_early_uart
 #ifdef CONFIG_EARLY_PRINTK
         /* Use a virtual address to access the UART. */
@@ -317,11 +312,14 @@  GLOBAL(init_secondary)
         bl    check_cpu_mode
         bl    cpu_init
         bl    prepare_early_mappings
-        bl    enable_mm
 
-        /* We are still in the 1:1 mapping. Jump to the runtime Virtual Address. */
-        ldr   x0, =secondary_switched
-        br    x0
+        /*
+         * Address in the runtime mapping to jump to after the
+         * MMU/MPU is enabled
+         */
+        ldr   lr, =secondary_switched
+        b    enable_mm
+
 secondary_switched:
         /*
          * Non-boot CPUs need to move on to the proper pagetables, which were
diff --git a/xen/arch/arm/arm64/head_mmu.S b/xen/arch/arm/arm64/head_mmu.S
index a19b7c873d..c9e83bbe2d 100644
--- a/xen/arch/arm/arm64/head_mmu.S
+++ b/xen/arch/arm/arm64/head_mmu.S
@@ -211,9 +211,11 @@  virtphys_clash:
 ENDPROC(prepare_early_mappings)
 
 /*
- * Turn on the Data Cache and the MMU. The function will return on the 1:1
- * mapping. In other word, the caller is responsible to switch to the runtime
- * mapping.
+ * Turn on the Data Cache and the MMU. The function will return
+ * to the virtual address provided in LR (e.g. the runtime mapping).
+ *
+ * Inputs:
+ * lr(x30): Virtual address to return to
  *
  * Clobbers x0 - x3
  */
@@ -238,6 +240,31 @@  ENTRY(enable_mm)
         dsb   sy                     /* Flush PTE writes and finish reads */
         msr   SCTLR_EL2, x0          /* now paging is enabled */
         isb                          /* Now, flush the icache */
+
+        /*
+         * The MMU is turned on and we are in the 1:1 mapping. Switch
+         * to the runtime mapping.
+         */
+        ldr   x0, =1f
+        br    x0
+1:
+        /*
+         * The 1:1 map may clash with other parts of the Xen virtual memory
+         * layout. As it is not used anymore, remove it completely to
+         * avoid having to worry about replacing existing mapping
+         * afterwards.
+         *
+         * On return this will jump to the virtual address requested by
+         * the caller
+         */
+        b     remove_identity_mapping
+
+        /*
+         * Here might not be reached, as "ret" in remove_identity_mapping
+         * will use the return address in LR in advance. But keep ret here
+         * might be more safe if "ret" in remove_identity_mapping is removed
+         * in future.
+         */
         ret
 ENDPROC(enable_mm)