[PATCH] x86/efi: Free efi_pgd with free_pages()

[PATCH] x86/efi: Free efi_pgd with free_pages()

[Lennart Glauer]

This patch fixes another occurrence of free_page() that was missed
in 06ace26.
The efi_pgd is allocated as PGD_ALLOCATION_ORDER pages and therefore must
also be freed as PGD_ALLOCATION_ORDER pages with free_pages().

Signed-off-by: Lennart Glauer <mail@lennart-glauer.de>
---
 arch/x86/platform/efi/efi_64.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
index 08ce8177c3af..acad22a44774 100644
--- a/arch/x86/platform/efi/efi_64.c
+++ b/arch/x86/platform/efi/efi_64.c
@@ -222,7 +222,7 @@ int __init efi_alloc_page_tables(void)
 	pgd = efi_pgd + pgd_index(EFI_VA_END);
 	p4d = p4d_alloc(&init_mm, pgd, EFI_VA_END);
 	if (!p4d) {
-		free_page((unsigned long)efi_pgd);
+		free_pages((unsigned long)efi_pgd, PGD_ALLOCATION_ORDER);
 		return -ENOMEM;
 	}
 
-- 
2.17.1

Re: [PATCH] x86/efi: Free efi_pgd with free_pages()

[Ard Biesheuvel]

Hi Lennart,

Thanks for the patch.

On Tue, 28 May 2019 at 14:58, Lennart Glauer <mail@lennart-glauer.de> wrote:
>
> This patch fixes another occurrence of free_page() that was missed
> in 06ace26.

Please don't trim commit IDs like that. I think 12 digits is our
current minimum?

error: short SHA1 06ace26 is ambiguous
hint: The candidates are:
hint:   06ace26edc14 commit 2016-02-12 - staging: lustre: fix all NULL
comparisons in LNet layer
hint:   06ace26f4e6f commit 2018-03-22 - x86/efi: Free efi_pgd with free_pages()

> The efi_pgd is allocated as PGD_ALLOCATION_ORDER pages and therefore must
> also be freed as PGD_ALLOCATION_ORDER pages with free_pages().
>
> Signed-off-by: Lennart Glauer <mail@lennart-glauer.de>

Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>

Ingo, can you take this directly? I can't really test this anyway.


> ---
>  arch/x86/platform/efi/efi_64.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
> index 08ce8177c3af..acad22a44774 100644
> --- a/arch/x86/platform/efi/efi_64.c
> +++ b/arch/x86/platform/efi/efi_64.c
> @@ -222,7 +222,7 @@ int __init efi_alloc_page_tables(void)
>         pgd = efi_pgd + pgd_index(EFI_VA_END);
>         p4d = p4d_alloc(&init_mm, pgd, EFI_VA_END);
>         if (!p4d) {
> -               free_page((unsigned long)efi_pgd);
> +               free_pages((unsigned long)efi_pgd, PGD_ALLOCATION_ORDER);
>                 return -ENOMEM;
>         }
>
> --
> 2.17.1

[PATCH] x86/efi: Free efi_pgd with free_pages()

[Waiman Long]

The efi_pgd is allocated as PGD_ALLOCATION_ORDER pages and so should
also be freed as PGD_ALLOCATION_ORDER pages with free_pages().

Fixes: d9e9a6418065 ("x86/mm/pti: Allocate a separate user PGD")
Signed-off-by: Waiman Long <longman@redhat.com>
---
 arch/x86/platform/efi/efi_64.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
index c310a82..f9cfbc0 100644
--- a/arch/x86/platform/efi/efi_64.c
+++ b/arch/x86/platform/efi/efi_64.c
@@ -227,7 +227,7 @@ int __init efi_alloc_page_tables(void)
 	if (!pud) {
 		if (CONFIG_PGTABLE_LEVELS > 4)
 			free_page((unsigned long) pgd_page_vaddr(*pgd));
-		free_page((unsigned long)efi_pgd);
+		free_pages((unsigned long)efi_pgd, PGD_ALLOCATION_ORDER);
 		return -ENOMEM;
 	}
 
-- 
1.8.3.1