linux-kernel - Re: [PATCH -next] x86/efi_64: fix a user-memory-access in runtime

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAKv+Gu8WBSsG2e8bVpARcwNBrGtMLzUA+bbikHymrZsNQE6wvw@mail.gmail.com>
Date:   Sat, 18 Jan 2020 09:00:04 +0100
From:   Ard Biesheuvel <ard.biesheuvel@...aro.org>
To:     Qian Cai <cai@....pw>
Cc:     Ard Biesheuvel <ardb@...nel.org>, Ingo Molnar <mingo@...hat.com>,
        kasan-dev <kasan-dev@...glegroups.com>,
        linux-efi <linux-efi@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH -next] x86/efi_64: fix a user-memory-access in runtime

On Sat, 18 Jan 2020 at 07:30, Qian Cai <cai@....pw> wrote:
>
> The commit 698294704573 ("efi/x86: Split SetVirtualAddresMap() wrappers
> into 32 and 64 bit versions") introduced a KASAN error during boot,
>
>  BUG: KASAN: user-memory-access in efi_set_virtual_address_map+0x4d3/0x574
>  Read of size 8 at addr 00000000788fee50 by task swapper/0/0
>
>  Hardware name: HP ProLiant XL450 Gen9 Server/ProLiant XL450 Gen9
>  Server, BIOS U21 05/05/2016
>  Call Trace:
>   dump_stack+0xa0/0xea
>   __kasan_report.cold.8+0xb0/0xc0
>   kasan_report+0x12/0x20
>   __asan_load8+0x71/0xa0
>   efi_set_virtual_address_map+0x4d3/0x574
>   efi_enter_virtual_mode+0x5f3/0x64e
>   start_kernel+0x53a/0x5dc
>   x86_64_start_reservations+0x24/0x26
>   x86_64_start_kernel+0xf4/0xfb
>   secondary_startup_64+0xb6/0xc0
>
> It points to this line,
>
> status = efi_call(efi.systab->runtime->set_virtual_address_map,
>
> efi.systab->runtime's address is 00000000788fee18 which is an address in
> EFI runtime service and does not have a KASAN shadow page. Fix it by
> doing a copy_from_user() first instead.
>

Can't we just use READ_ONCE_NOCHECK() instead?

> Fixes: 698294704573 ("efi/x86: Split SetVirtualAddresMap() wrappers into 32 and 64 bit versions")
> Signed-off-by: Qian Cai <cai@....pw>
> ---
>  arch/x86/platform/efi/efi_64.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
> index 515eab388b56..d6712c9cb9d8 100644
> --- a/arch/x86/platform/efi/efi_64.c
> +++ b/arch/x86/platform/efi/efi_64.c
> @@ -1023,6 +1023,7 @@ efi_status_t __init efi_set_virtual_address_map(unsigned long memory_map_size,
>                                                 u32 descriptor_version,
>                                                 efi_memory_desc_t *virtual_map)
>  {
> +       efi_runtime_services_t runtime;
>         efi_status_t status;
>         unsigned long flags;
>         pgd_t *save_pgd = NULL;
> @@ -1041,13 +1042,15 @@ efi_status_t __init efi_set_virtual_address_map(unsigned long memory_map_size,
>                 efi_switch_mm(&efi_mm);
>         }
>
> +       if (copy_from_user(&runtime, efi.systab->runtime, sizeof(runtime)))
> +               return EFI_ABORTED;
> +
>         kernel_fpu_begin();
>
>         /* Disable interrupts around EFI calls: */
>         local_irq_save(flags);
> -       status = efi_call(efi.systab->runtime->set_virtual_address_map,
> -                         memory_map_size, descriptor_size,
> -                         descriptor_version, virtual_map);
> +       status = efi_call(runtime.set_virtual_address_map, memory_map_size,
> +                         descriptor_size, descriptor_version, virtual_map);
>         local_irq_restore(flags);
>
>         kernel_fpu_end();
> --
> 2.21.0 (Apple Git-122.2)
>