Re: [patch] seccomp.2: Note that memory area is read-only

Re: [patch] seccomp.2: Note that memory area is read-only

[Michael Kerrisk (man-pages)]

Hi Kees,

Ping!

Cheers,

Michael



On 6 April 2015 at 17:29, Michael Kerrisk (man-pages)
<mtk.manpages@gmail.com> wrote:
> Hi Kees,
>
> I recently was asked about the point below, and had to go check the code
> to be sure, since the man page said nothing. It would be good to have
> a confirmation: the seccomp_data buffer supplied to a seccomp BPF program
> is read-only, right? (That is, one can't write to the buffer in order to
> change the arguments that a system call actually receives.)
>
> A small man page patch below.
>
> Cheers,
>
> Michael
>
> --- a/man2/seccomp.2
> +++ b/man2/seccomp.2
> @@ -232,15 +232,15 @@ struct sock_filter {            /* Filter block */
>  };
>  .fi
>  .in
>
>  When executing the instructions, the BPF program operates on the
>  system call information made available (i.e., use the
>  .BR BPF_ABS
> -addressing mode) as a buffer of the following form:
> +addressing mode) as a (read-only) buffer of the following form:
>
>  .in +4n
>  .nf
>  struct seccomp_data {
>      int   nr;                   /* System call number */
>      __u32 arch;                 /* AUDIT_ARCH_* value
>                                     (see <linux/audit.h>) */
>
>
> --
> Michael Kerrisk
> Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
> Linux/UNIX System Programming Training: http://man7.org/training/



-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: [patch] seccomp.2: Note that memory area is read-only

[Kees Cook]

On Mon, Apr 6, 2015 at 8:29 AM, Michael Kerrisk (man-pages)
<mtk.manpages@gmail.com> wrote:
> Hi Kees,
>
> I recently was asked about the point below, and had to go check the code
> to be sure, since the man page said nothing. It would be good to have
> a confirmation: the seccomp_data buffer supplied to a seccomp BPF program
> is read-only, right? (That is, one can't write to the buffer in order to
> change the arguments that a system call actually receives.)

That's correct. If BPF even allows changing the data, it's not copied
back to the syscall when it runs. Anything wanting to do things like
that would need to use ptrace to catch the call an directly modify the
registers before continuing with the call.

>
> A small man page patch below.

Looks good, thanks!

-Kees

>
> Cheers,
>
> Michael
>
> --- a/man2/seccomp.2
> +++ b/man2/seccomp.2
> @@ -232,15 +232,15 @@ struct sock_filter {            /* Filter block */
>  };
>  .fi
>  .in
>
>  When executing the instructions, the BPF program operates on the
>  system call information made available (i.e., use the
>  .BR BPF_ABS
> -addressing mode) as a buffer of the following form:
> +addressing mode) as a (read-only) buffer of the following form:
>
>  .in +4n
>  .nf
>  struct seccomp_data {
>      int   nr;                   /* System call number */
>      __u32 arch;                 /* AUDIT_ARCH_* value
>                                     (see <linux/audit.h>) */
>
>
> --
> Michael Kerrisk
> Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
> Linux/UNIX System Programming Training: http://man7.org/training/



-- 
Kees Cook
Chrome OS Security

Re: [patch] seccomp.2: Note that memory area is read-only

[Alexei Starovoitov]

On Sat, Apr 18, 2015 at 04:37:35PM -0700, Kees Cook wrote:
> On Mon, Apr 6, 2015 at 8:29 AM, Michael Kerrisk (man-pages)
> <mtk.manpages@gmail.com> wrote:
> > Hi Kees,
> >
> > I recently was asked about the point below, and had to go check the code
> > to be sure, since the man page said nothing. It would be good to have
> > a confirmation: the seccomp_data buffer supplied to a seccomp BPF program
> > is read-only, right? (That is, one can't write to the buffer in order to
> > change the arguments that a system call actually receives.)
> 
> That's correct. If BPF even allows changing the data, it's not copied
> back to the syscall when it runs. Anything wanting to do things like
> that would need to use ptrace to catch the call an directly modify the
> registers before continuing with the call.

Today BPF programs can only read from seccomp_data buffer.
I think it would be an interesting feature to allow overwrite of
syscall arguments by the program on the fly... sometime in the future.

Re: [patch] seccomp.2: Note that memory area is read-only

[Michael Kerrisk (man-pages)]

On 19 April 2015 at 01:37, Kees Cook <keescook@chromium.org> wrote:
> On Mon, Apr 6, 2015 at 8:29 AM, Michael Kerrisk (man-pages)
> <mtk.manpages@gmail.com> wrote:
>> Hi Kees,
>>
>> I recently was asked about the point below, and had to go check the code
>> to be sure, since the man page said nothing. It would be good to have
>> a confirmation: the seccomp_data buffer supplied to a seccomp BPF program
>> is read-only, right? (That is, one can't write to the buffer in order to
>> change the arguments that a system call actually receives.)
>
> That's correct. If BPF even allows changing the data, it's not copied
> back to the syscall when it runs. Anything wanting to do things like
> that would need to use ptrace to catch the call an directly modify the
> registers before continuing with the call.
>
>>
>> A small man page patch below.
>
> Looks good, thanks!

Thanks, Kees!

Cheers,

Michael


>> --- a/man2/seccomp.2
>> +++ b/man2/seccomp.2
>> @@ -232,15 +232,15 @@ struct sock_filter {            /* Filter block */
>>  };
>>  .fi
>>  .in
>>
>>  When executing the instructions, the BPF program operates on the
>>  system call information made available (i.e., use the
>>  .BR BPF_ABS
>> -addressing mode) as a buffer of the following form:
>> +addressing mode) as a (read-only) buffer of the following form:
>>
>>  .in +4n
>>  .nf
>>  struct seccomp_data {
>>      int   nr;                   /* System call number */
>>      __u32 arch;                 /* AUDIT_ARCH_* value
>>                                     (see <linux/audit.h>) */
>>
>>
>> --
>> Michael Kerrisk
>> Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
>> Linux/UNIX System Programming Training: http://man7.org/training/
>
>
>
> --
> Kees Cook
> Chrome OS Security



-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/