On Wed, May 06, 2020 at 10:44:15AM -0700, Linus Torvalds wrote:
> So while I think using a consistent convention is good, and it's true
> that there is a difference in the convention between the two cases
> ("unsafe" at the beginning vs end), one of them is actually about the
> safety and security of the operation (and we have automated logic
> these days to verify it on x86), the other has nothing to do with
> "safety", really.
>
> Would it be better to standardize around a "probe_xyz()" naming?
So:
probe_strncpy, probe_strncpy_user, probe_strnlen_user?
Sounds weird, but at least it is consistent.
> Or perhaps a "xyz_nofault()" naming?
That sounds a little better:
strncpy_nofault, strncpy_user_nofault, strnlen_user_nofault
> I realize this is nit-picky, and I think the patch series as-is is
> already an improvement, but I do think our naming in this area is
> really quite bad.
Always open for improvements :)
> The fact that we have "probe_kernel_read()" but then
> "strncpy_from_user_unsafe()" for the _same_ conceptual difference
> really tells me how inconsistent the naming for these kinds of "we
> can't take page faults" is. No?
True. If we wanted to do _nofaul, what would the basic read/write
versions be?
