Re: [PATCH] x86: vdso32/syscall.S: do not load __USER32_DS to %ss

[Andy Lutomirski]

On Wed, Mar 25, 2015 at 7:55 AM, Denys Vlasenko <dvlasenk@xxxxxxxxxx> wrote:
> On 03/24/2015 10:40 PM, Andy Lutomirski wrote:
>> The syscall and sysenter stuff is IMO really nasty.  Here's how I'd
>> like it to work:
>>
>> When you do "call __kernel_vsyscall", I want the net effect to be that
>> your eax, ebx, ecx, edx, esi, edi, and ebp at the time of the call end
>> up *verbatim* in pt_regs.  Your eip and rsp should be such that, if we
>> iret normally using pt_regs, we end up returning correctly to
>> userspace.  I want this to be true *regardless* of whether we're doing
>> a fast-path or slow-path system call.
>>
>> This means that we have, literally (see below for why ret $4):
>>
>> int $0x80
>> ret $4  <-- regs->eip points here
>>
>> Then we add an opportunistic return trampoline: if a special ti flag
>> is set (which we set on entry here) and the return eip and regs are
>> appropriate, then we change the return at the last minute to vdso code
>> that looks like:
>>
>> popl $ecx
>> popl $edx
>> ret
>
> I don't fully understand your intent.

The idea would be that syscall and sysenter would each be exactly
equivalent to a different sequence of instructions, each culminating
in a jmp to an int80, except that they would enable opportunistic exit
optimizations.  I think I made some mistakes below, though.

>
>> The vdso code would be something like (so untested it's not even funny):
>>
>> __kernel_vsyscall:
>>   ALTERNATIVE_2(something or other)
>>
>> __kernel_vsyscall_for_intel:
>>   pushl $edx
>>   pushl $ecx
>>   sysenter
>>   hlt  <-- just for clarity
>>
>> __kernel_vsyscall_for_amd:
>>   pushl $ecx
>>   syscall
>> __vsyscall_after_syscall_insn:
>>  ret $4 <-- for binary tracers only
>
> This ret would use former ecx value as return address?
>

Nope.  The idea is that syscall32 would be close enough to equivalent
to "mov (%esp),%ecx; int $0x80" that binary tracers would do the right
thing.  But I could easily be off a bit.  If I were to implement it,
that ret instruction would be the very last thing I added, since it
would depend on everything else.

>
>> __kernel_vsyscall_for_int80:
>>   int $0x80  <-- regs->eip points here during *all* vsyscalls
>>
>> __kernel_vsyscall_slow_ret:
>>   ret $4
>
> After returning, this will pop an extra word from __kernel_vsyscall() caller.
> They don't expect that.
>

Whoops.  I did say this was completely untested :).  I guess it would
be more like:

__kernel_vsyscall_for_int80:
  pushl $eax  /* dummy */
  pushl $eax  /* dummy */
  int $0x80  <-- regs->eip points here during *all* vsyscalls

__kernel_vsyscall_slow_ret:
  ret $8

since having the amount of extra scratch stack space vary by entry
type is probably unnecessarily confusing.

>
>> __kernel_vsyscall_sysretl_target:
>>   popl $ecx
>>   ret
>>
>> There is no sysexit.  Take that, Intel.
>>
>> On sysenter, we copy regs->cx and regs->dx from user memory and then
>> we increment regs->sp by 4 and point regs->eip to
>> __kernel_vsyscall_for_int80.  On syscall, we copy regs->cx from user
>> memory and point regs->eip to __kernel_vsyscall_for_int80.
>>
>> On opportunistic sysretl, we do:
>>
>> *regs->sp = regs->cx;  /* put_user or whatever */
>> regs->eip = __kernel_vsyscall_sysretl_target
>> ...
>> sysretl
>>
>> We never do sysexit or sysretl in any other code path.  That is, there
>> is no really fast path anymore.
>
> I still don't understand the purpose those "ret 4" insns.
> They don't look right.

They were wrong.  It's the idea that counts, I hope :)

--Andy
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/