RE: randomize_kstack: To init or not to init?

[David Laight <David.Laight@ACULAB.COM>]

From: Jann Horn
> Sent: 09 December 2021 21:16
...
> This doesn't just affect alloca(), right? According to godbolt.org
> (https://godbolt.org/z/jYrWEx7o8):
> 
> void bar(char *p);
> void foo() {
>   char arr[512];
>   bar(arr);
> }
> 
> when compiled with "-ftrivial-auto-var-init=pattern -O2 -mno-sse"
> gives this result:
> 
> foo:                                    # @foo
>         push    rbx
>         sub     rsp, 512
>         mov     rbx, rsp
>         mov     edx, 512
>         mov     rdi, rbx
>         mov     esi, 170
>         call    memset@PLT
>         mov     rdi, rbx
>         call    bar
>         add     rsp, 512
>         pop     rbx
>         ret

Jeepers - I don't ever want that to happen not ever...

There is plenty of userspace code that allocates large arrays on stack
(I bet some get into MB sizes) that are correctly bound-checked but
the expense of initialising them will be horrid.

So you end up with horrid, complex, more likely to be buggy, code
that tries to allocate things that are 'just big enough' rather
than just a sanity check on a large buffer.

Typical examples are char path[MAXPATH].
You know the path will almost certainly be < 100 bytes.
MAXPATH is overkill - but can be tested for.
But you don't want path[] initialised.
So you cane to pick a shorter length - and then it all goes 'TITSUP'
when the actual path is a bit longer than you allowed for.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)