Patchwork [1/4] arm64: memblock: don't permit memblock resizing until linear mapping is up

login
register
mail settings
Submitter Ard Biesheuvel
Date Nov. 6, 2018, 11:37 a.m.
Message ID <20181106113732.16351-2-ard.biesheuvel@linaro.org>
Download mbox | patch
Permalink /patch/650347/
State New
Headers show

Comments

Ard Biesheuvel - Nov. 6, 2018, 11:37 a.m.
Bhupesh reports that having numerous memblock reservations at early
boot may result in the following crash:

  Unable to handle kernel paging request at virtual address ffff80003ffe0000
  ...
  Call trace:
   __memcpy+0x110/0x180
   memblock_add_range+0x134/0x2e8
   memblock_reserve+0x70/0xb8
   memblock_alloc_base_nid+0x6c/0x88
   __memblock_alloc_base+0x3c/0x4c
   memblock_alloc_base+0x28/0x4c
   memblock_alloc+0x2c/0x38
   early_pgtable_alloc+0x20/0xb0
   paging_init+0x28/0x7f8

This is caused by the fact that we permit memblock resizing before the
linear mapping is up, and so the memblock_reserved() array is moved
into memory that is not mapped yet.

So let's ensure that this crash can no longer occur, by deferring to
call to memblock_allow_resize() to after the linear mapping has been
created.

Reported-by: Bhupesh Sharma <bhsharma@redhat.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---
 arch/arm64/mm/init.c | 2 --
 arch/arm64/mm/mmu.c  | 2 ++
 2 files changed, 2 insertions(+), 2 deletions(-)
Will Deacon - Nov. 6, 2018, 9:22 p.m.
On Tue, Nov 06, 2018 at 12:37:29PM +0100, Ard Biesheuvel wrote:
> Bhupesh reports that having numerous memblock reservations at early
> boot may result in the following crash:
> 
>   Unable to handle kernel paging request at virtual address ffff80003ffe0000
>   ...
>   Call trace:
>    __memcpy+0x110/0x180
>    memblock_add_range+0x134/0x2e8
>    memblock_reserve+0x70/0xb8
>    memblock_alloc_base_nid+0x6c/0x88
>    __memblock_alloc_base+0x3c/0x4c
>    memblock_alloc_base+0x28/0x4c
>    memblock_alloc+0x2c/0x38
>    early_pgtable_alloc+0x20/0xb0
>    paging_init+0x28/0x7f8
> 
> This is caused by the fact that we permit memblock resizing before the
> linear mapping is up, and so the memblock_reserved() array is moved
> into memory that is not mapped yet.
> 
> So let's ensure that this crash can no longer occur, by deferring to
> call to memblock_allow_resize() to after the linear mapping has been
> created.
> 
> Reported-by: Bhupesh Sharma <bhsharma@redhat.com>
> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> ---
>  arch/arm64/mm/init.c | 2 --
>  arch/arm64/mm/mmu.c  | 2 ++
>  2 files changed, 2 insertions(+), 2 deletions(-)

Thanks for posting this so quickly.

Acked-by: Will Deacon <will.deacon@arm.com>

Bhupesh -- please can you give this series a spin and confirm that it fixes
the problem you were seeing?

Thanks,

Will

Patch

diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
index 9d9582cac6c4..9b432d9fcada 100644
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c
@@ -483,8 +483,6 @@  void __init arm64_memblock_init(void)
 	high_memory = __va(memblock_end_of_DRAM() - 1) + 1;
 
 	dma_contiguous_reserve(arm64_dma_phys_limit);
-
-	memblock_allow_resize();
 }
 
 void __init bootmem_init(void)
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
index 394b8d554def..d1d6601b385d 100644
--- a/arch/arm64/mm/mmu.c
+++ b/arch/arm64/mm/mmu.c
@@ -659,6 +659,8 @@  void __init paging_init(void)
 
 	memblock_free(__pa_symbol(init_pg_dir),
 		      __pa_symbol(init_pg_end) - __pa_symbol(init_pg_dir));
+
+	memblock_allow_resize();
 }
 
 /*