Re: [PATCH v2 13/19] x86/signal: Check if vdso_image_32 is mapped before trying to land on it

[Dmitry Safonov]

On 11/24/20 11:43 PM, Andy Lutomirski wrote:
> On Mon, Nov 23, 2020 at 4:29 PM Dmitry Safonov <d...@arista.com> wrote:
>>
>> Provide current_has_vdso_image_32() helper and check it apriory landing
>> attempt on vdso vma.
>> The helper is a macro, not a static inline funciton to avoid
>> linux/sched/task_stack.h inclusion in asm/vdso.h.
> 
> Can you make this more general?  For example:
> 
> current_has_vdso(&vdso_image_whatever)

Sounds good, will do.

> 
> also:
> 
>> diff --git a/arch/x86/entry/common.c b/arch/x86/entry/common.c
>> index 18d8f17f755c..d9ab58cc765b 100644
>> --- a/arch/x86/entry/common.c
>> +++ b/arch/x86/entry/common.c
>> @@ -142,11 +142,16 @@ static noinstr bool __do_fast_syscall_32(struct 
>> pt_regs *regs)
>>  /* Returns 0 to return using IRET or 1 to return using SYSEXIT/SYSRETL. */
>>  __visible noinstr long do_fast_syscall_32(struct pt_regs *regs)
>>  {
>> +       unsigned long landing_pad;
>> +
>> +       if (!current_has_vdso_image_32())
>> +               force_sigsegv(SIGSEGV);
> 
> Falling through seems incorrect here.  I would instead write 0 to rip,
> do the force_sigsegv(), and exit, making sure that exiting goes
> through the correct path.
> 

Something like this?

:       if (!current_has_vdso(&vdso_image_32)) {
:               regs->ax = -EFAULT;
:               regs->ip = 0;
:               force_sigsegv(SIGSEGV);
:               syscall_exit_to_user_mode(regs);
:       }

Thanks,
          Dmitry