On Wed, Aug 19, 2020 at 09:50:06AM -0500, Josh Poimboeuf wrote:
> The x86 uaccess code uses barrier_nospec() in various places to prevent
> speculative dereferencing of user-controlled pointers (which might be
> combined with further gadgets or CPU bugs to leak data).
>
> There are some issues with the current implementation:
>
> - The barrier_nospec() in copy_from_user() was inadvertently removed
> with: 4b842e4e25b1 ("x86: get rid of small constant size cases in
> raw_copy_{to,from}_user()")
>
> - copy_to_user() and friends should also have a speculation barrier,
> because a speculative write to a user-controlled address can still
> populate the cache line with the original data.
>
> - The LFENCE in barrier_nospec() is overkill, when more lightweight user
> pointer masking can be used instead.
>
> Remove all existing barrier_nospec() usage, and instead do user pointer
> masking, throughout the x86 uaccess code. This is similar to what arm64
> is already doing.
>
> barrier_nospec() is now unused, and can be removed.
One thing to consider is whether you need a speculation barrier after
set_fs(). Otherwise for code like:
| fs = get_fs();
| if (cond)
| set_fs(KERNEL_DS);
| copy_to_user(...)
| set_fs(fs)
... the set_fs() can occur speculatively, and may be able to satisfy
the masking logic if forwarded within the cpu.
See arm64 commit:
c2f0ad4fc089cff8 ("arm64: uaccess: Prevent speculative use of the current
addr_limit")
Thanks,
Mark.
