Re: [PATCH] x86/syscall: Avoid memcpy() for ia32 syscall_get_arguments()

["Gustavo A. R. Silva" <gustavo@embeddedor.com>]

On 09/07/24 12:20, Mirsad Todorovac wrote:
> 
> 
> On 7/9/24 01:44, Gustavo A. R. Silva wrote:
>>
>>
>> On 7/8/24 14:22, Kees Cook wrote:
>>> Modern (fortified) memcpy() prefers to avoid writing (or reading) beyond
>>> the end of the addressed destination (or source) struct member:
>>>
>>> In function ‘fortify_memcpy_chk’,
>>>       inlined from ‘syscall_get_arguments’ at ./arch/x86/include/asm/syscall.h:85:2,
>>>       inlined from ‘populate_seccomp_data’ at kernel/seccomp.c:258:2,
>>>       inlined from ‘__seccomp_filter’ at kernel/seccomp.c:1231:3:
>>> ./include/linux/fortify-string.h:580:25: error: call to ‘__read_overflow2_field’ declared with attribute warning: detected read beyond size of field (2nd parameter); maybe use struct_group()? [-Werror=attribute-warning]
>>>     580 |                         __read_overflow2_field(q_size_field, size);
>>>         |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>
>>> As already done for x86_64 and compat mode, do not use memcpy() to
>>> extract syscall arguments from struct pt_regs but rather just perform
>>> direct assignments. Binary output differences are negligible, and actually
>>> ends up using less stack space:
>>>
>>> -       sub    $0x84,%esp
>>> +       sub    $0x6c,%esp
>>>
>>> and less text size:
>>>
>>>      text    data     bss     dec     hex filename
>>>     10794     252       0   11046    2b26 gcc-32b/kernel/seccomp.o.stock
>>>     10714     252       0   10966    2ad6 gcc-32b/kernel/seccomp.o.after
>>>
>>> Reported-by: Mirsad Todorovac <mtodorovac69@gmail.com>
>>> Closes: https://lore.kernel.org/lkml/9b69fb14-df89-4677-9c82-056ea9e706f5@gmail.com/
>>> Signed-off-by: Kees Cook <kees@kernel.org>
>>> ---
>>> Cc: Thomas Gleixner <tglx@linutronix.de>
>>> Cc: Ingo Molnar <mingo@redhat.com>
>>> Cc: Borislav Petkov <bp@alien8.de>
>>> Cc: Dave Hansen <dave.hansen@linux.intel.com>
>>> Cc: x86@kernel.org
>>> Cc: "H. Peter Anvin" <hpa@zytor.com>
>>> Cc: Daniel Sneddon <daniel.sneddon@linux.intel.com>
>>> Cc: Arnd Bergmann <arnd@arndb.de>
>>> Cc: Brian Gerst <brgerst@gmail.com>
>>> Cc: Josh Poimboeuf <jpoimboe@kernel.org>
>>> Cc: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
>>> Cc: Peter Collingbourne <pcc@google.com>
>>
>> Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
>>
>> Thanks
> 
> I can confirm that the error was fixed after applying the patch, in the same build environment.
> 
> Tested-by: Mirsad Todorovac <mtodorovac69@gmail.com>
> 
> However, why memcpy() directly from struct pt_regs doesn't work is beyond my understanding :-/
> 

This is because under CONFIG_FORTIFY_SOURCE=y, memcpy() prevents writing or reading beyond
the boundaries of dest/src objects.

--
Gustavo

> FWIW, bulk memcpy() might be replaced by a single assembler instruction? Or am I thinking still
> in 6502 mode? :-)
> 
> Best regards,
> Mirsad Todorovac
> 
>> -- 
>> Gustavo
>>
>>> ---
>>>    arch/x86/include/asm/syscall.h | 7 ++++++-
>>>    1 file changed, 6 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/arch/x86/include/asm/syscall.h b/arch/x86/include/asm/syscall.h
>>> index 2fc7bc3863ff..7c488ff0c764 100644
>>> --- a/arch/x86/include/asm/syscall.h
>>> +++ b/arch/x86/include/asm/syscall.h
>>> @@ -82,7 +82,12 @@ static inline void syscall_get_arguments(struct task_struct *task,
>>>                         struct pt_regs *regs,
>>>                         unsigned long *args)
>>>    {
>>> -    memcpy(args, &regs->bx, 6 * sizeof(args[0]));
>>> +    args[0] = regs->bx;
>>> +    args[1] = regs->cx;
>>> +    args[2] = regs->dx;
>>> +    args[3] = regs->si;
>>> +    args[4] = regs->di;
>>> +    args[5] = regs->bp;
>>>    }
>>>      static inline int syscall_get_arch(struct task_struct *task)