Re: [PATCH v2 1/1] x86/mm/KASLR: Store pud_page_tramp into entry rather than page

[Nick Desaulniers <ndesaulniers@google.com>]

On Wed, Jun 14, 2023 at 05:38:54PM +0100, Lee Jones wrote:
> set_pgd() expects to be passed whole pages to operate on, whereas
> trampoline_pgd_entry is, as the name suggests, an entry.  The
> ramifications for using set_pgd() here are that the following thread of
> execution will not only place the suggested value into the
> trampoline_pgd_entry (8-Byte globally stored [.bss]) variable, PTI will
> also attempt to replicate that value into the non-existent neighboring
> user page (located +4k away), leading to the corruption of other global
> [.bss] stored variables.
> 
> Suggested-by: Dave Hansen <dave.hansen@linux.intel.com>
> Signed-off-by: Lee Jones <lee@kernel.org>

Nice work tracking this one down!

Fixes: 0925dda5962e ("x86/mm/KASLR: Use only one PUD entry for real mode trampoline")
Cc: <stable@vger.kernel.org>

> ---
>  arch/x86/mm/kaslr.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/mm/kaslr.c b/arch/x86/mm/kaslr.c
> index 557f0fe25dff4..37db264866b64 100644
> --- a/arch/x86/mm/kaslr.c
> +++ b/arch/x86/mm/kaslr.c
> @@ -172,10 +172,10 @@ void __meminit init_trampoline_kaslr(void)
>  		set_p4d(p4d_tramp,
>  			__p4d(_KERNPG_TABLE | __pa(pud_page_tramp)));
>  
> -		set_pgd(&trampoline_pgd_entry,
> -			__pgd(_KERNPG_TABLE | __pa(p4d_page_tramp)));
> +		trampoline_pgd_entry =
> +			__pgd(_KERNPG_TABLE | __pa(p4d_page_tramp));
>  	} else {
> -		set_pgd(&trampoline_pgd_entry,
> -			__pgd(_KERNPG_TABLE | __pa(pud_page_tramp)));
> +		trampoline_pgd_entry =
> +			__pgd(_KERNPG_TABLE | __pa(pud_page_tramp));
>  	}
>  }
> -- 
> 2.41.0.162.gfafddb0af9-goog
>