On Tue, May 30, 2023 at 06:17:35PM +0200, Ard Biesheuvel wrote: > On Tue, 30 May 2023 at 18:15, Maximilian Luz <luzmaximilian@xxxxxxxxx> wrote: > > > > On 5/30/23 17:25, Kees Cook wrote: > > > On Mon, May 29, 2023 at 01:03:48AM +0200, Maximilian Luz wrote: > > >> Add a ucs2_strlcpy() function for UCS-2 strings. The behavior is > > >> equivalent to the standard strlcpy() function, just for 16-bit character > > >> UCS-2 strings. > > > > > > Eek, no. strlcpy() is dangerous in multiple ways[1]. Please implement > > > strscpy() (i.e. use strnlen(), negative error on truncation, etc). > > > > Right, make sense, thanks. Somehow I missed that the kernel has a better > > function than the C stdlib for that... > > > > > Additionally, it'd be nice of the ucs2 helpers here also implemented the > > > rest of the CONFIG_FORTIFY_SOURCE mitigations (i.e. checking for source > > > and destination buffer size overflows at compile-time and run-time with > > > __builtin_object_size() and __builtin_dynamoc_object_size() respectively). > > > > I can certainly try that, but I think this might be better suited for a > > follow-up series, given that we then should also add those to the other > > helpers. > > > > Agreed. Let's log the followup work as a kspp work item, no need to > make that part of this series. Yeah, that's fine. Can you please open a KSSP issue for it so we don't forget? :) -- Kees Cook
