diff mbox series

[v7,2/5] xen/arm64: Rework the memory layout

Message ID 20230416143211.72227-3-julien@xen.org (mailing list archive)
State New, archived
Headers show
Series xen/arm: Don't switch TTBR while the MMU is on | expand

Commit Message

Julien Grall April 16, 2023, 2:32 p.m. UTC
From: Julien Grall <jgrall@amazon.com>

Xen is currently not fully compliant with the Arm Arm because it will
switch the TTBR with the MMU on.

In order to be compliant, we need to disable the MMU before
switching the TTBR. The implication is the page-tables should
contain an identity mapping of the code switching the TTBR.

In most of the case we expect Xen to be loaded in low memory. I am aware
of one platform (i.e AMD Seattle) where the memory start above 512GB.
To give us some slack, consider that Xen may be loaded in the first 2TB
of the physical address space.

The memory layout is reshuffled to keep the first four slots of the zeroeth
level free. All the regions currently in L0 slot 0 will not be part of
slot 4 (2TB). This requires a slight tweak of the boot code because
XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.

This reshuffle will make trivial to create a 1:1 mapping when Xen is
loaded below 2TB.

Lastly, take the opportunity to check a compile time if any of the
regions may overlap with the reserved area for identity mapping.

Signed-off-by: Julien Grall <jgrall@amazon.com>

----
    Changes in v7:
        - Remove all tags
        - Add BUILD_BUG_ON()s
        - Don't forget to update FRAMETABLE_VIRT_START and
          VMAP_VIRT_START

    Changes in v6:
        - Correct the BUILD_BUG_ON(), Xen virtual address should be
          above 2TB (i.e. slot0 > 4).
        - Add Bertrand's reviewed-by

    Changes in v5:
        - We are reserving 4 slots rather than 2.
        - Fix the addresses in the layout comment.
        - Fix the size of the region in the layout comment
        - Add Luca's tested-by (the reviewed-by was not added
          because of the changes requested by Michal)
        - Add Michal's reviewed-by

    Changes in v4:
        - Correct the documentation
        - The start address is 2TB, so slot0 is 4 not 2.

    Changes in v2:
        - Reword the commit message
        - Load Xen at 2TB + 2MB
        - Update the documentation to reflect the new layout
---
 xen/arch/arm/arm64/head.S         |  3 ++-
 xen/arch/arm/include/asm/config.h | 38 +++++++++++++++++++++----------
 xen/arch/arm/mm.c                 | 23 +++++++++++++++----
 3 files changed, 46 insertions(+), 18 deletions(-)

Comments

Henry Wang April 17, 2023, 4:58 a.m. UTC | #1
Hi Julien,

> -----Original Message-----
> Subject: [PATCH v7 2/5] xen/arm64: Rework the memory layout
> 
> From: Julien Grall <jgrall@amazon.com>
> 
> Xen is currently not fully compliant with the Arm Arm because it will
> switch the TTBR with the MMU on.
> 
> In order to be compliant, we need to disable the MMU before
> switching the TTBR. The implication is the page-tables should
> contain an identity mapping of the code switching the TTBR.
> 
> In most of the case we expect Xen to be loaded in low memory. I am aware
> of one platform (i.e AMD Seattle) where the memory start above 512GB.
> To give us some slack, consider that Xen may be loaded in the first 2TB
> of the physical address space.
> 
> The memory layout is reshuffled to keep the first four slots of the zeroeth
> level free. All the regions currently in L0 slot 0 will not be part of
> slot 4 (2TB). This requires a slight tweak of the boot code because
> XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.
> 
> This reshuffle will make trivial to create a 1:1 mapping when Xen is
> loaded below 2TB.
> 
> Lastly, take the opportunity to check a compile time if any of the
> regions may overlap with the reserved area for identity mapping.
> 
> Signed-off-by: Julien Grall <jgrall@amazon.com>

This time I used our CI to test this series patch by patch on top of staging
today (Apr 17), so that we can see if the qemu issue reported by Bertrand
in v6 still persists.

I can confirm all boards including the qemu-arm64 passed this time, so:

Tested-by: Henry Wang <Henry.Wang@arm.com>

Kind regards,
Henry
Luca Fancellu April 17, 2023, 7:21 a.m. UTC | #2
> On 16 Apr 2023, at 15:32, Julien Grall <julien@xen.org> wrote:
> 
> From: Julien Grall <jgrall@amazon.com>
> 
> Xen is currently not fully compliant with the Arm Arm because it will
> switch the TTBR with the MMU on.
> 
> In order to be compliant, we need to disable the MMU before
> switching the TTBR. The implication is the page-tables should
> contain an identity mapping of the code switching the TTBR.
> 
> In most of the case we expect Xen to be loaded in low memory. I am aware
> of one platform (i.e AMD Seattle) where the memory start above 512GB.
> To give us some slack, consider that Xen may be loaded in the first 2TB
> of the physical address space.
> 
> The memory layout is reshuffled to keep the first four slots of the zeroeth
> level free. All the regions currently in L0 slot 0 will not be part of
> slot 4 (2TB). This requires a slight tweak of the boot code because
> XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.
> 
> This reshuffle will make trivial to create a 1:1 mapping when Xen is
> loaded below 2TB.
> 
> Lastly, take the opportunity to check a compile time if any of the
> regions may overlap with the reserved area for identity mapping.
> 
> Signed-off-by: Julien Grall <jgrall@amazon.com>
> 

Hi Julien,

It looks fine to me.

Reviewed-by: Luca Fancellu <luca.fancellu@arm.com>
Michal Orzel April 17, 2023, 7:28 a.m. UTC | #3
Hi,

On 16/04/2023 16:32, Julien Grall wrote:
> 
> 
> From: Julien Grall <jgrall@amazon.com>
> 
> Xen is currently not fully compliant with the Arm Arm because it will
> switch the TTBR with the MMU on.
> 
> In order to be compliant, we need to disable the MMU before
> switching the TTBR. The implication is the page-tables should
> contain an identity mapping of the code switching the TTBR.
> 
> In most of the case we expect Xen to be loaded in low memory. I am aware
> of one platform (i.e AMD Seattle) where the memory start above 512GB.
> To give us some slack, consider that Xen may be loaded in the first 2TB
> of the physical address space.
> 
> The memory layout is reshuffled to keep the first four slots of the zeroeth
> level free. All the regions currently in L0 slot 0 will not be part of
> slot 4 (2TB). This requires a slight tweak of the boot code because
> XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.
> 
> This reshuffle will make trivial to create a 1:1 mapping when Xen is
> loaded below 2TB.
> 
> Lastly, take the opportunity to check a compile time if any of the
s/a/at/ compile time

> regions may overlap with the reserved area for identity mapping.
> 
> Signed-off-by: Julien Grall <jgrall@amazon.com>
> 
> ----
>     Changes in v7:
>         - Remove all tags
>         - Add BUILD_BUG_ON()s
>         - Don't forget to update FRAMETABLE_VIRT_START and
>           VMAP_VIRT_START
> 
>     Changes in v6:
>         - Correct the BUILD_BUG_ON(), Xen virtual address should be
>           above 2TB (i.e. slot0 > 4).
>         - Add Bertrand's reviewed-by
> 
>     Changes in v5:
>         - We are reserving 4 slots rather than 2.
>         - Fix the addresses in the layout comment.
>         - Fix the size of the region in the layout comment
>         - Add Luca's tested-by (the reviewed-by was not added
>           because of the changes requested by Michal)
>         - Add Michal's reviewed-by
> 
>     Changes in v4:
>         - Correct the documentation
>         - The start address is 2TB, so slot0 is 4 not 2.
> 
>     Changes in v2:
>         - Reword the commit message
>         - Load Xen at 2TB + 2MB
>         - Update the documentation to reflect the new layout
> ---
>  xen/arch/arm/arm64/head.S         |  3 ++-
>  xen/arch/arm/include/asm/config.h | 38 +++++++++++++++++++++----------
>  xen/arch/arm/mm.c                 | 23 +++++++++++++++----
>  3 files changed, 46 insertions(+), 18 deletions(-)
> 
> diff --git a/xen/arch/arm/arm64/head.S b/xen/arch/arm/arm64/head.S
> index 4a3f87117c83..663f5813b12e 100644
> --- a/xen/arch/arm/arm64/head.S
> +++ b/xen/arch/arm/arm64/head.S
> @@ -607,7 +607,8 @@ create_page_tables:
>           * need an additional 1:1 mapping, the virtual mapping will
>           * suffice.
>           */
> -        cmp   x19, #XEN_VIRT_START
> +        ldr   x0, =XEN_VIRT_START
> +        cmp   x19, x0
>          bne   1f
>          ret
>  1:
> diff --git a/xen/arch/arm/include/asm/config.h b/xen/arch/arm/include/asm/config.h
> index 5df0e4c4959b..2cfe5e480256 100644
> --- a/xen/arch/arm/include/asm/config.h
> +++ b/xen/arch/arm/include/asm/config.h
> @@ -72,16 +72,13 @@
>  #include <xen/page-size.h>
> 
>  /*
> - * Common ARM32 and ARM64 layout:
> + * ARM32 layout:
>   *   0  -   2M   Unmapped
>   *   2M -   4M   Xen text, data, bss
>   *   4M -   6M   Fixmap: special-purpose 4K mapping slots
>   *   6M -  10M   Early boot mapping of FDT
>   *   10M - 12M   Livepatch vmap (if compiled in)
>   *
> - * ARM32 layout:
> - *   0  -  12M   <COMMON>
> - *
>   *  32M - 128M   Frametable: 32 bytes per page for 12GB of RAM
>   * 256M -   1G   VMAP: ioremap and early_ioremap use this virtual address
>   *                    space
> @@ -90,14 +87,23 @@
>   *   2G -   4G   Domheap: on-demand-mapped
>   *
>   * ARM64 layout:
> - * 0x0000000000000000 - 0x0000007fffffffff (512GB, L0 slot [0])
> - *   0  -  12M   <COMMON>
> + * 0x0000000000000000 - 0x000001ffffffffff (2TB, L0 slots [0..3])
> + *
> + *  Reserved to identity map Xen
> + *
> + * 0x0000020000000000 - 0x0000027fffffffff (512GB, L0 slot [4]
missing closing parenthesis at the end of line

This can be done on commit, so:
Reviewed-by: Michal Orzel <michal.orzel@amd.com>

~Michal
Bertrand Marquis April 18, 2023, 1:53 p.m. UTC | #4
Hi Julien,

> On 16 Apr 2023, at 16:32, Julien Grall <julien@xen.org> wrote:
> 
> From: Julien Grall <jgrall@amazon.com>
> 
> Xen is currently not fully compliant with the Arm Arm because it will
> switch the TTBR with the MMU on.
> 
> In order to be compliant, we need to disable the MMU before
> switching the TTBR. The implication is the page-tables should
> contain an identity mapping of the code switching the TTBR.
> 
> In most of the case we expect Xen to be loaded in low memory. I am aware
> of one platform (i.e AMD Seattle) where the memory start above 512GB.
> To give us some slack, consider that Xen may be loaded in the first 2TB
> of the physical address space.
> 
> The memory layout is reshuffled to keep the first four slots of the zeroeth
> level free. All the regions currently in L0 slot 0 will not be part of
> slot 4 (2TB). This requires a slight tweak of the boot code because
> XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.
> 
> This reshuffle will make trivial to create a 1:1 mapping when Xen is
> loaded below 2TB.
> 
> Lastly, take the opportunity to check a compile time if any of the
> regions may overlap with the reserved area for identity mapping.
> 
> Signed-off-by: Julien Grall <jgrall@amazon.com>

With the 2 typos found by Michal fixed:
Reviewed-by: Bertrand Marquis <bertrand.marquis@arm.com>

Cheers
Bertrand

> 
> ----
>    Changes in v7:
>        - Remove all tags
>        - Add BUILD_BUG_ON()s
>        - Don't forget to update FRAMETABLE_VIRT_START and
>          VMAP_VIRT_START
> 
>    Changes in v6:
>        - Correct the BUILD_BUG_ON(), Xen virtual address should be
>          above 2TB (i.e. slot0 > 4).
>        - Add Bertrand's reviewed-by
> 
>    Changes in v5:
>        - We are reserving 4 slots rather than 2.
>        - Fix the addresses in the layout comment.
>        - Fix the size of the region in the layout comment
>        - Add Luca's tested-by (the reviewed-by was not added
>          because of the changes requested by Michal)
>        - Add Michal's reviewed-by
> 
>    Changes in v4:
>        - Correct the documentation
>        - The start address is 2TB, so slot0 is 4 not 2.
> 
>    Changes in v2:
>        - Reword the commit message
>        - Load Xen at 2TB + 2MB
>        - Update the documentation to reflect the new layout
> ---
> xen/arch/arm/arm64/head.S         |  3 ++-
> xen/arch/arm/include/asm/config.h | 38 +++++++++++++++++++++----------
> xen/arch/arm/mm.c                 | 23 +++++++++++++++----
> 3 files changed, 46 insertions(+), 18 deletions(-)
> 
> diff --git a/xen/arch/arm/arm64/head.S b/xen/arch/arm/arm64/head.S
> index 4a3f87117c83..663f5813b12e 100644
> --- a/xen/arch/arm/arm64/head.S
> +++ b/xen/arch/arm/arm64/head.S
> @@ -607,7 +607,8 @@ create_page_tables:
>          * need an additional 1:1 mapping, the virtual mapping will
>          * suffice.
>          */
> -        cmp   x19, #XEN_VIRT_START
> +        ldr   x0, =XEN_VIRT_START
> +        cmp   x19, x0
>         bne   1f
>         ret
> 1:
> diff --git a/xen/arch/arm/include/asm/config.h b/xen/arch/arm/include/asm/config.h
> index 5df0e4c4959b..2cfe5e480256 100644
> --- a/xen/arch/arm/include/asm/config.h
> +++ b/xen/arch/arm/include/asm/config.h
> @@ -72,16 +72,13 @@
> #include <xen/page-size.h>
> 
> /*
> - * Common ARM32 and ARM64 layout:
> + * ARM32 layout:
>  *   0  -   2M   Unmapped
>  *   2M -   4M   Xen text, data, bss
>  *   4M -   6M   Fixmap: special-purpose 4K mapping slots
>  *   6M -  10M   Early boot mapping of FDT
>  *   10M - 12M   Livepatch vmap (if compiled in)
>  *
> - * ARM32 layout:
> - *   0  -  12M   <COMMON>
> - *
>  *  32M - 128M   Frametable: 32 bytes per page for 12GB of RAM
>  * 256M -   1G   VMAP: ioremap and early_ioremap use this virtual address
>  *                    space
> @@ -90,14 +87,23 @@
>  *   2G -   4G   Domheap: on-demand-mapped
>  *
>  * ARM64 layout:
> - * 0x0000000000000000 - 0x0000007fffffffff (512GB, L0 slot [0])
> - *   0  -  12M   <COMMON>
> + * 0x0000000000000000 - 0x000001ffffffffff (2TB, L0 slots [0..3])
> + *
> + *  Reserved to identity map Xen
> + *
> + * 0x0000020000000000 - 0x0000027fffffffff (512GB, L0 slot [4]
> + *  (Relative offsets)
> + *   0  -   2M   Unmapped
> + *   2M -   4M   Xen text, data, bss
> + *   4M -   6M   Fixmap: special-purpose 4K mapping slots
> + *   6M -  10M   Early boot mapping of FDT
> + *  10M -  12M   Livepatch vmap (if compiled in)
>  *
>  *   1G -   2G   VMAP: ioremap and early_ioremap
>  *
>  *  32G -  64G   Frametable: 56 bytes per page for 2TB of RAM
>  *
> - * 0x0000008000000000 - 0x00007fffffffffff (127.5TB, L0 slots [1..255])
> + * 0x0000028000000000 - 0x00007fffffffffff (125TB, L0 slots [5..255])
>  *  Unused
>  *
>  * 0x0000800000000000 - 0x000084ffffffffff (5TB, L0 slots [256..265])
> @@ -107,7 +113,17 @@
>  *  Unused
>  */
> 
> +#ifdef CONFIG_ARM_32
> #define XEN_VIRT_START          _AT(vaddr_t, MB(2))
> +#else
> +
> +#define SLOT0_ENTRY_BITS  39
> +#define SLOT0(slot) (_AT(vaddr_t,slot) << SLOT0_ENTRY_BITS)
> +#define SLOT0_ENTRY_SIZE  SLOT0(1)
> +
> +#define XEN_VIRT_START          (SLOT0(4) + _AT(vaddr_t, MB(2)))
> +#endif
> +
> #define XEN_VIRT_SIZE           _AT(vaddr_t, MB(2))
> 
> #define FIXMAP_VIRT_START       (XEN_VIRT_START + XEN_VIRT_SIZE)
> @@ -163,14 +179,12 @@
> 
> #else /* ARM_64 */
> 
> -#define SLOT0_ENTRY_BITS  39
> -#define SLOT0(slot) (_AT(vaddr_t,slot) << SLOT0_ENTRY_BITS)
> -#define SLOT0_ENTRY_SIZE  SLOT0(1)
> +#define IDENTITY_MAPPING_AREA_NR_L0  4
> 
> -#define VMAP_VIRT_START  GB(1)
> +#define VMAP_VIRT_START  (SLOT0(4) + GB(1))
> #define VMAP_VIRT_SIZE   GB(1)
> 
> -#define FRAMETABLE_VIRT_START  GB(32)
> +#define FRAMETABLE_VIRT_START  (SLOT0(4) + GB(32))
> #define FRAMETABLE_SIZE        GB(32)
> #define FRAMETABLE_NR          (FRAMETABLE_SIZE / sizeof(*frame_table))
> 
> diff --git a/xen/arch/arm/mm.c b/xen/arch/arm/mm.c
> index b99806af996c..1d09d61dd922 100644
> --- a/xen/arch/arm/mm.c
> +++ b/xen/arch/arm/mm.c
> @@ -153,7 +153,19 @@ static void __init __maybe_unused build_assertions(void)
> #endif
>     /* Page table structure constraints */
> #ifdef CONFIG_ARM_64
> -    BUILD_BUG_ON(zeroeth_table_offset(XEN_VIRT_START));
> +    /*
> +     * The first few slots of the L0 table is reserved for the identity
> +     * mapping. Check that none of the other regions are overlapping
> +     * with it.
> +     */
> +#define CHECK_OVERLAP_WITH_IDMAP(virt) \
> +    BUILD_BUG_ON(zeroeth_table_offset(virt) < IDENTITY_MAPPING_AREA_NR_L0)
> +
> +    CHECK_OVERLAP_WITH_IDMAP(XEN_VIRT_START);
> +    CHECK_OVERLAP_WITH_IDMAP(VMAP_VIRT_START);
> +    CHECK_OVERLAP_WITH_IDMAP(FRAMETABLE_VIRT_START);
> +    CHECK_OVERLAP_WITH_IDMAP(DIRECTMAP_VIRT_START);
> +#undef CHECK_OVERLAP_WITH_IDMAP
> #endif
>     BUILD_BUG_ON(first_table_offset(XEN_VIRT_START));
> #ifdef CONFIG_ARCH_MAP_DOMAIN_PAGE
> @@ -496,10 +508,11 @@ void __init setup_pagetables(unsigned long boot_phys_offset)
>     phys_offset = boot_phys_offset;
> 
> #ifdef CONFIG_ARM_64
> -    p = (void *) xen_pgtable;
> -    p[0] = pte_of_xenaddr((uintptr_t)xen_first);
> -    p[0].pt.table = 1;
> -    p[0].pt.xn = 0;
> +    pte = pte_of_xenaddr((uintptr_t)xen_first);
> +    pte.pt.table = 1;
> +    pte.pt.xn = 0;
> +    xen_pgtable[zeroeth_table_offset(XEN_VIRT_START)] = pte;
> +
>     p = (void *) xen_first;
> #else
>     p = (void *) cpu0_pgtable;
> -- 
> 2.39.2
>
Bertrand Marquis April 18, 2023, 1:58 p.m. UTC | #5
> On 18 Apr 2023, at 15:53, Bertrand Marquis <Bertrand.Marquis@arm.com> wrote:
> 
> Hi Julien,
> 
>> On 16 Apr 2023, at 16:32, Julien Grall <julien@xen.org> wrote:
>> 
>> From: Julien Grall <jgrall@amazon.com>
>> 
>> Xen is currently not fully compliant with the Arm Arm because it will
>> switch the TTBR with the MMU on.
>> 
>> In order to be compliant, we need to disable the MMU before
>> switching the TTBR. The implication is the page-tables should
>> contain an identity mapping of the code switching the TTBR.
>> 
>> In most of the case we expect Xen to be loaded in low memory. I am aware
>> of one platform (i.e AMD Seattle) where the memory start above 512GB.
>> To give us some slack, consider that Xen may be loaded in the first 2TB
>> of the physical address space.
>> 
>> The memory layout is reshuffled to keep the first four slots of the zeroeth
>> level free. All the regions currently in L0 slot 0 will not be part of
>> slot 4 (2TB). This requires a slight tweak of the boot code because
>> XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.
>> 
>> This reshuffle will make trivial to create a 1:1 mapping when Xen is
>> loaded below 2TB.
>> 
>> Lastly, take the opportunity to check a compile time if any of the
>> regions may overlap with the reserved area for identity mapping.
>> 
>> Signed-off-by: Julien Grall <jgrall@amazon.com>
> 
> With the 2 typos found by Michal fixed:
> Reviewed-by: Bertrand Marquis <bertrand.marquis@arm.com>

You can fix them on commit definitely :-)

Cheers
Bertrand

> 
> Cheers
> Bertrand
> 
>> 
>> ----
>>   Changes in v7:
>>       - Remove all tags
>>       - Add BUILD_BUG_ON()s
>>       - Don't forget to update FRAMETABLE_VIRT_START and
>>         VMAP_VIRT_START
>> 
>>   Changes in v6:
>>       - Correct the BUILD_BUG_ON(), Xen virtual address should be
>>         above 2TB (i.e. slot0 > 4).
>>       - Add Bertrand's reviewed-by
>> 
>>   Changes in v5:
>>       - We are reserving 4 slots rather than 2.
>>       - Fix the addresses in the layout comment.
>>       - Fix the size of the region in the layout comment
>>       - Add Luca's tested-by (the reviewed-by was not added
>>         because of the changes requested by Michal)
>>       - Add Michal's reviewed-by
>> 
>>   Changes in v4:
>>       - Correct the documentation
>>       - The start address is 2TB, so slot0 is 4 not 2.
>> 
>>   Changes in v2:
>>       - Reword the commit message
>>       - Load Xen at 2TB + 2MB
>>       - Update the documentation to reflect the new layout
>> ---
>> xen/arch/arm/arm64/head.S         |  3 ++-
>> xen/arch/arm/include/asm/config.h | 38 +++++++++++++++++++++----------
>> xen/arch/arm/mm.c                 | 23 +++++++++++++++----
>> 3 files changed, 46 insertions(+), 18 deletions(-)
>> 
>> diff --git a/xen/arch/arm/arm64/head.S b/xen/arch/arm/arm64/head.S
>> index 4a3f87117c83..663f5813b12e 100644
>> --- a/xen/arch/arm/arm64/head.S
>> +++ b/xen/arch/arm/arm64/head.S
>> @@ -607,7 +607,8 @@ create_page_tables:
>>         * need an additional 1:1 mapping, the virtual mapping will
>>         * suffice.
>>         */
>> -        cmp   x19, #XEN_VIRT_START
>> +        ldr   x0, =XEN_VIRT_START
>> +        cmp   x19, x0
>>        bne   1f
>>        ret
>> 1:
>> diff --git a/xen/arch/arm/include/asm/config.h b/xen/arch/arm/include/asm/config.h
>> index 5df0e4c4959b..2cfe5e480256 100644
>> --- a/xen/arch/arm/include/asm/config.h
>> +++ b/xen/arch/arm/include/asm/config.h
>> @@ -72,16 +72,13 @@
>> #include <xen/page-size.h>
>> 
>> /*
>> - * Common ARM32 and ARM64 layout:
>> + * ARM32 layout:
>> *   0  -   2M   Unmapped
>> *   2M -   4M   Xen text, data, bss
>> *   4M -   6M   Fixmap: special-purpose 4K mapping slots
>> *   6M -  10M   Early boot mapping of FDT
>> *   10M - 12M   Livepatch vmap (if compiled in)
>> *
>> - * ARM32 layout:
>> - *   0  -  12M   <COMMON>
>> - *
>> *  32M - 128M   Frametable: 32 bytes per page for 12GB of RAM
>> * 256M -   1G   VMAP: ioremap and early_ioremap use this virtual address
>> *                    space
>> @@ -90,14 +87,23 @@
>> *   2G -   4G   Domheap: on-demand-mapped
>> *
>> * ARM64 layout:
>> - * 0x0000000000000000 - 0x0000007fffffffff (512GB, L0 slot [0])
>> - *   0  -  12M   <COMMON>
>> + * 0x0000000000000000 - 0x000001ffffffffff (2TB, L0 slots [0..3])
>> + *
>> + *  Reserved to identity map Xen
>> + *
>> + * 0x0000020000000000 - 0x0000027fffffffff (512GB, L0 slot [4]
>> + *  (Relative offsets)
>> + *   0  -   2M   Unmapped
>> + *   2M -   4M   Xen text, data, bss
>> + *   4M -   6M   Fixmap: special-purpose 4K mapping slots
>> + *   6M -  10M   Early boot mapping of FDT
>> + *  10M -  12M   Livepatch vmap (if compiled in)
>> *
>> *   1G -   2G   VMAP: ioremap and early_ioremap
>> *
>> *  32G -  64G   Frametable: 56 bytes per page for 2TB of RAM
>> *
>> - * 0x0000008000000000 - 0x00007fffffffffff (127.5TB, L0 slots [1..255])
>> + * 0x0000028000000000 - 0x00007fffffffffff (125TB, L0 slots [5..255])
>> *  Unused
>> *
>> * 0x0000800000000000 - 0x000084ffffffffff (5TB, L0 slots [256..265])
>> @@ -107,7 +113,17 @@
>> *  Unused
>> */
>> 
>> +#ifdef CONFIG_ARM_32
>> #define XEN_VIRT_START          _AT(vaddr_t, MB(2))
>> +#else
>> +
>> +#define SLOT0_ENTRY_BITS  39
>> +#define SLOT0(slot) (_AT(vaddr_t,slot) << SLOT0_ENTRY_BITS)
>> +#define SLOT0_ENTRY_SIZE  SLOT0(1)
>> +
>> +#define XEN_VIRT_START          (SLOT0(4) + _AT(vaddr_t, MB(2)))
>> +#endif
>> +
>> #define XEN_VIRT_SIZE           _AT(vaddr_t, MB(2))
>> 
>> #define FIXMAP_VIRT_START       (XEN_VIRT_START + XEN_VIRT_SIZE)
>> @@ -163,14 +179,12 @@
>> 
>> #else /* ARM_64 */
>> 
>> -#define SLOT0_ENTRY_BITS  39
>> -#define SLOT0(slot) (_AT(vaddr_t,slot) << SLOT0_ENTRY_BITS)
>> -#define SLOT0_ENTRY_SIZE  SLOT0(1)
>> +#define IDENTITY_MAPPING_AREA_NR_L0  4
>> 
>> -#define VMAP_VIRT_START  GB(1)
>> +#define VMAP_VIRT_START  (SLOT0(4) + GB(1))
>> #define VMAP_VIRT_SIZE   GB(1)
>> 
>> -#define FRAMETABLE_VIRT_START  GB(32)
>> +#define FRAMETABLE_VIRT_START  (SLOT0(4) + GB(32))
>> #define FRAMETABLE_SIZE        GB(32)
>> #define FRAMETABLE_NR          (FRAMETABLE_SIZE / sizeof(*frame_table))
>> 
>> diff --git a/xen/arch/arm/mm.c b/xen/arch/arm/mm.c
>> index b99806af996c..1d09d61dd922 100644
>> --- a/xen/arch/arm/mm.c
>> +++ b/xen/arch/arm/mm.c
>> @@ -153,7 +153,19 @@ static void __init __maybe_unused build_assertions(void)
>> #endif
>>    /* Page table structure constraints */
>> #ifdef CONFIG_ARM_64
>> -    BUILD_BUG_ON(zeroeth_table_offset(XEN_VIRT_START));
>> +    /*
>> +     * The first few slots of the L0 table is reserved for the identity
>> +     * mapping. Check that none of the other regions are overlapping
>> +     * with it.
>> +     */
>> +#define CHECK_OVERLAP_WITH_IDMAP(virt) \
>> +    BUILD_BUG_ON(zeroeth_table_offset(virt) < IDENTITY_MAPPING_AREA_NR_L0)
>> +
>> +    CHECK_OVERLAP_WITH_IDMAP(XEN_VIRT_START);
>> +    CHECK_OVERLAP_WITH_IDMAP(VMAP_VIRT_START);
>> +    CHECK_OVERLAP_WITH_IDMAP(FRAMETABLE_VIRT_START);
>> +    CHECK_OVERLAP_WITH_IDMAP(DIRECTMAP_VIRT_START);
>> +#undef CHECK_OVERLAP_WITH_IDMAP
>> #endif
>>    BUILD_BUG_ON(first_table_offset(XEN_VIRT_START));
>> #ifdef CONFIG_ARCH_MAP_DOMAIN_PAGE
>> @@ -496,10 +508,11 @@ void __init setup_pagetables(unsigned long boot_phys_offset)
>>    phys_offset = boot_phys_offset;
>> 
>> #ifdef CONFIG_ARM_64
>> -    p = (void *) xen_pgtable;
>> -    p[0] = pte_of_xenaddr((uintptr_t)xen_first);
>> -    p[0].pt.table = 1;
>> -    p[0].pt.xn = 0;
>> +    pte = pte_of_xenaddr((uintptr_t)xen_first);
>> +    pte.pt.table = 1;
>> +    pte.pt.xn = 0;
>> +    xen_pgtable[zeroeth_table_offset(XEN_VIRT_START)] = pte;
>> +
>>    p = (void *) xen_first;
>> #else
>>    p = (void *) cpu0_pgtable;
>> -- 
>> 2.39.2
Julien Grall April 19, 2023, 6:38 p.m. UTC | #6
Hi Michal,

On 17/04/2023 08:28, Michal Orzel wrote:
> On 16/04/2023 16:32, Julien Grall wrote:
>>
>>
>> From: Julien Grall <jgrall@amazon.com>
>>
>> Xen is currently not fully compliant with the Arm Arm because it will
>> switch the TTBR with the MMU on.
>>
>> In order to be compliant, we need to disable the MMU before
>> switching the TTBR. The implication is the page-tables should
>> contain an identity mapping of the code switching the TTBR.
>>
>> In most of the case we expect Xen to be loaded in low memory. I am aware
>> of one platform (i.e AMD Seattle) where the memory start above 512GB.
>> To give us some slack, consider that Xen may be loaded in the first 2TB
>> of the physical address space.
>>
>> The memory layout is reshuffled to keep the first four slots of the zeroeth
>> level free. All the regions currently in L0 slot 0 will not be part of
>> slot 4 (2TB). This requires a slight tweak of the boot code because
>> XEN_VIRT_START (2TB + 2MB) cannot be used as an immediate.
>>
>> This reshuffle will make trivial to create a 1:1 mapping when Xen is
>> loaded below 2TB.
>>
>> Lastly, take the opportunity to check a compile time if any of the
> s/a/at/ compile time
> 
>> regions may overlap with the reserved area for identity mapping.
>>
>> Signed-off-by: Julien Grall <jgrall@amazon.com>
>>
>> ----
>>      Changes in v7:
>>          - Remove all tags
>>          - Add BUILD_BUG_ON()s
>>          - Don't forget to update FRAMETABLE_VIRT_START and
>>            VMAP_VIRT_START
>>
>>      Changes in v6:
>>          - Correct the BUILD_BUG_ON(), Xen virtual address should be
>>            above 2TB (i.e. slot0 > 4).
>>          - Add Bertrand's reviewed-by
>>
>>      Changes in v5:
>>          - We are reserving 4 slots rather than 2.
>>          - Fix the addresses in the layout comment.
>>          - Fix the size of the region in the layout comment
>>          - Add Luca's tested-by (the reviewed-by was not added
>>            because of the changes requested by Michal)
>>          - Add Michal's reviewed-by
>>
>>      Changes in v4:
>>          - Correct the documentation
>>          - The start address is 2TB, so slot0 is 4 not 2.
>>
>>      Changes in v2:
>>          - Reword the commit message
>>          - Load Xen at 2TB + 2MB
>>          - Update the documentation to reflect the new layout
>> ---
>>   xen/arch/arm/arm64/head.S         |  3 ++-
>>   xen/arch/arm/include/asm/config.h | 38 +++++++++++++++++++++----------
>>   xen/arch/arm/mm.c                 | 23 +++++++++++++++----
>>   3 files changed, 46 insertions(+), 18 deletions(-)
>>
>> diff --git a/xen/arch/arm/arm64/head.S b/xen/arch/arm/arm64/head.S
>> index 4a3f87117c83..663f5813b12e 100644
>> --- a/xen/arch/arm/arm64/head.S
>> +++ b/xen/arch/arm/arm64/head.S
>> @@ -607,7 +607,8 @@ create_page_tables:
>>            * need an additional 1:1 mapping, the virtual mapping will
>>            * suffice.
>>            */
>> -        cmp   x19, #XEN_VIRT_START
>> +        ldr   x0, =XEN_VIRT_START
>> +        cmp   x19, x0
>>           bne   1f
>>           ret
>>   1:
>> diff --git a/xen/arch/arm/include/asm/config.h b/xen/arch/arm/include/asm/config.h
>> index 5df0e4c4959b..2cfe5e480256 100644
>> --- a/xen/arch/arm/include/asm/config.h
>> +++ b/xen/arch/arm/include/asm/config.h
>> @@ -72,16 +72,13 @@
>>   #include <xen/page-size.h>
>>
>>   /*
>> - * Common ARM32 and ARM64 layout:
>> + * ARM32 layout:
>>    *   0  -   2M   Unmapped
>>    *   2M -   4M   Xen text, data, bss
>>    *   4M -   6M   Fixmap: special-purpose 4K mapping slots
>>    *   6M -  10M   Early boot mapping of FDT
>>    *   10M - 12M   Livepatch vmap (if compiled in)
>>    *
>> - * ARM32 layout:
>> - *   0  -  12M   <COMMON>
>> - *
>>    *  32M - 128M   Frametable: 32 bytes per page for 12GB of RAM
>>    * 256M -   1G   VMAP: ioremap and early_ioremap use this virtual address
>>    *                    space
>> @@ -90,14 +87,23 @@
>>    *   2G -   4G   Domheap: on-demand-mapped
>>    *
>>    * ARM64 layout:
>> - * 0x0000000000000000 - 0x0000007fffffffff (512GB, L0 slot [0])
>> - *   0  -  12M   <COMMON>
>> + * 0x0000000000000000 - 0x000001ffffffffff (2TB, L0 slots [0..3])
>> + *
>> + *  Reserved to identity map Xen
>> + *
>> + * 0x0000020000000000 - 0x0000027fffffffff (512GB, L0 slot [4]
> missing closing parenthesis at the end of line
> 
> This can be done on commit, so:

Thanks for spotting the typoes. I will update on commit.

> Reviewed-by: Michal Orzel <michal.orzel@amd.com>

Cheers,
diff mbox series

Patch

diff --git a/xen/arch/arm/arm64/head.S b/xen/arch/arm/arm64/head.S
index 4a3f87117c83..663f5813b12e 100644
--- a/xen/arch/arm/arm64/head.S
+++ b/xen/arch/arm/arm64/head.S
@@ -607,7 +607,8 @@  create_page_tables:
          * need an additional 1:1 mapping, the virtual mapping will
          * suffice.
          */
-        cmp   x19, #XEN_VIRT_START
+        ldr   x0, =XEN_VIRT_START
+        cmp   x19, x0
         bne   1f
         ret
 1:
diff --git a/xen/arch/arm/include/asm/config.h b/xen/arch/arm/include/asm/config.h
index 5df0e4c4959b..2cfe5e480256 100644
--- a/xen/arch/arm/include/asm/config.h
+++ b/xen/arch/arm/include/asm/config.h
@@ -72,16 +72,13 @@ 
 #include <xen/page-size.h>
 
 /*
- * Common ARM32 and ARM64 layout:
+ * ARM32 layout:
  *   0  -   2M   Unmapped
  *   2M -   4M   Xen text, data, bss
  *   4M -   6M   Fixmap: special-purpose 4K mapping slots
  *   6M -  10M   Early boot mapping of FDT
  *   10M - 12M   Livepatch vmap (if compiled in)
  *
- * ARM32 layout:
- *   0  -  12M   <COMMON>
- *
  *  32M - 128M   Frametable: 32 bytes per page for 12GB of RAM
  * 256M -   1G   VMAP: ioremap and early_ioremap use this virtual address
  *                    space
@@ -90,14 +87,23 @@ 
  *   2G -   4G   Domheap: on-demand-mapped
  *
  * ARM64 layout:
- * 0x0000000000000000 - 0x0000007fffffffff (512GB, L0 slot [0])
- *   0  -  12M   <COMMON>
+ * 0x0000000000000000 - 0x000001ffffffffff (2TB, L0 slots [0..3])
+ *
+ *  Reserved to identity map Xen
+ *
+ * 0x0000020000000000 - 0x0000027fffffffff (512GB, L0 slot [4]
+ *  (Relative offsets)
+ *   0  -   2M   Unmapped
+ *   2M -   4M   Xen text, data, bss
+ *   4M -   6M   Fixmap: special-purpose 4K mapping slots
+ *   6M -  10M   Early boot mapping of FDT
+ *  10M -  12M   Livepatch vmap (if compiled in)
  *
  *   1G -   2G   VMAP: ioremap and early_ioremap
  *
  *  32G -  64G   Frametable: 56 bytes per page for 2TB of RAM
  *
- * 0x0000008000000000 - 0x00007fffffffffff (127.5TB, L0 slots [1..255])
+ * 0x0000028000000000 - 0x00007fffffffffff (125TB, L0 slots [5..255])
  *  Unused
  *
  * 0x0000800000000000 - 0x000084ffffffffff (5TB, L0 slots [256..265])
@@ -107,7 +113,17 @@ 
  *  Unused
  */
 
+#ifdef CONFIG_ARM_32
 #define XEN_VIRT_START          _AT(vaddr_t, MB(2))
+#else
+
+#define SLOT0_ENTRY_BITS  39
+#define SLOT0(slot) (_AT(vaddr_t,slot) << SLOT0_ENTRY_BITS)
+#define SLOT0_ENTRY_SIZE  SLOT0(1)
+
+#define XEN_VIRT_START          (SLOT0(4) + _AT(vaddr_t, MB(2)))
+#endif
+
 #define XEN_VIRT_SIZE           _AT(vaddr_t, MB(2))
 
 #define FIXMAP_VIRT_START       (XEN_VIRT_START + XEN_VIRT_SIZE)
@@ -163,14 +179,12 @@ 
 
 #else /* ARM_64 */
 
-#define SLOT0_ENTRY_BITS  39
-#define SLOT0(slot) (_AT(vaddr_t,slot) << SLOT0_ENTRY_BITS)
-#define SLOT0_ENTRY_SIZE  SLOT0(1)
+#define IDENTITY_MAPPING_AREA_NR_L0  4
 
-#define VMAP_VIRT_START  GB(1)
+#define VMAP_VIRT_START  (SLOT0(4) + GB(1))
 #define VMAP_VIRT_SIZE   GB(1)
 
-#define FRAMETABLE_VIRT_START  GB(32)
+#define FRAMETABLE_VIRT_START  (SLOT0(4) + GB(32))
 #define FRAMETABLE_SIZE        GB(32)
 #define FRAMETABLE_NR          (FRAMETABLE_SIZE / sizeof(*frame_table))
 
diff --git a/xen/arch/arm/mm.c b/xen/arch/arm/mm.c
index b99806af996c..1d09d61dd922 100644
--- a/xen/arch/arm/mm.c
+++ b/xen/arch/arm/mm.c
@@ -153,7 +153,19 @@  static void __init __maybe_unused build_assertions(void)
 #endif
     /* Page table structure constraints */
 #ifdef CONFIG_ARM_64
-    BUILD_BUG_ON(zeroeth_table_offset(XEN_VIRT_START));
+    /*
+     * The first few slots of the L0 table is reserved for the identity
+     * mapping. Check that none of the other regions are overlapping
+     * with it.
+     */
+#define CHECK_OVERLAP_WITH_IDMAP(virt) \
+    BUILD_BUG_ON(zeroeth_table_offset(virt) < IDENTITY_MAPPING_AREA_NR_L0)
+
+    CHECK_OVERLAP_WITH_IDMAP(XEN_VIRT_START);
+    CHECK_OVERLAP_WITH_IDMAP(VMAP_VIRT_START);
+    CHECK_OVERLAP_WITH_IDMAP(FRAMETABLE_VIRT_START);
+    CHECK_OVERLAP_WITH_IDMAP(DIRECTMAP_VIRT_START);
+#undef CHECK_OVERLAP_WITH_IDMAP
 #endif
     BUILD_BUG_ON(first_table_offset(XEN_VIRT_START));
 #ifdef CONFIG_ARCH_MAP_DOMAIN_PAGE
@@ -496,10 +508,11 @@  void __init setup_pagetables(unsigned long boot_phys_offset)
     phys_offset = boot_phys_offset;
 
 #ifdef CONFIG_ARM_64
-    p = (void *) xen_pgtable;
-    p[0] = pte_of_xenaddr((uintptr_t)xen_first);
-    p[0].pt.table = 1;
-    p[0].pt.xn = 0;
+    pte = pte_of_xenaddr((uintptr_t)xen_first);
+    pte.pt.table = 1;
+    pte.pt.xn = 0;
+    xen_pgtable[zeroeth_table_offset(XEN_VIRT_START)] = pte;
+
     p = (void *) xen_first;
 #else
     p = (void *) cpu0_pgtable;