Re: [PATCH] x86/platform/uv: refactor deprecated strcpy and strncpy

[Dimitri Sivanich]

On Wed, Aug 23, 2023 at 04:00:13PM -0700, Kees Cook wrote:
> On Wed, Aug 23, 2023 at 03:49:34PM -0700, Justin Stitt wrote:
> > On Wed, Aug 23, 2023 at 4:07 AM Andy Shevchenko
> > <andy.shevchenko@xxxxxxxxx> wrote:
> > >
> > > On Wed, Aug 23, 2023 at 1:32 AM Justin Stitt <justinstitt@xxxxxxxxxx> wrote:
> > > >
> > > > Both `strncpy` and `strcpy` are deprecated for use on NUL-terminated
> > > > destination strings [1].
> > > >
> > > > A suitable replacement is `strscpy` [2] due to the fact that it
> > > > guarantees NUL-termination on its destination buffer argument which is
> > > > _not_ the case for `strncpy` or `strcpy`!
> > > >
> > > > In this case, we can drop both the forced NUL-termination and the `... -1` from:
> > > > |       strncpy(arg, val, ACTION_LEN - 1);
> > > > as `strscpy` implicitly has this behavior.
> > >
> > > ...
> > >
> > > >         char arg[ACTION_LEN], *p;
> > > >
> > > >         /* (remove possible '\n') */
> > > > -       strncpy(arg, val, ACTION_LEN - 1);
> > > > -       arg[ACTION_LEN - 1] = '\0';
> > > > +       strscpy(arg, val, ACTION_LEN);
> > > >         p = strchr(arg, '\n');
> > > >         if (p)
> > > >                 *p = '\0';
> > >
> > > https://lore.kernel.org/all/202212091545310085328@xxxxxxxxxx/
> > >
> > > ...
> > >
> > > > +               strscpy(uv_nmi_action, arg, strlen(uv_nmi_action));
> > >
> > > strlen() on the destination?!

The original code for the above (strcpy()), copies strlen(arg) assuming null
termination, so strlen(uv_nmi_action) is not correct for this case.  You
probably want to use sizeof of the destination.

> > >
> > > ...
> > >
> > > > -                       strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action));
> > > > +                       strscpy(uv_nmi_action, "dump", strlen(uv_nmi_action));
> > >
> > > Again, this is weird.
> > 
> > This is a common pattern with `strxcpy` and `sizeof` if you `$ rg
> > "strncpy\(.*sizeof"`. Do you recommend I switch the strlen(dest) to
> > strlen(src)? I only kept as-is because that's what was there
> > originally and I assumed some greater purpose of it.
> 
> It's best to avoid any assumptions. If it can't be answered through code
> inspection, the next best thing would be to ask for clarification. In
> looking I see uv_nmi_action is a string:
> 
> arch/x86/platform/uv/uv_nmi.c:193:typedef char action_t[ACTION_LEN];
> 
> arch/x86/platform/uv/uv_nmi.c:          strcpy(uv_nmi_action, arg);
> arch/x86/platform/uv/uv_nmi.c:module_param_named(action, uv_nmi_action, action, 0644);
> arch/x86/platform/uv/uv_nmi.c:  return (strncmp(uv_nmi_action, action, strlen(action)) == 0);
> arch/x86/platform/uv/uv_nmi.c:                  strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action));
> 
> using strlen() here seems "accidentally safe", as it's overwriting
> "kdump":
> 
>         if (uv_nmi_action_is("kdump")) {
>                 uv_nmi_kdump(cpu, master, regs);
> 
>                 /* Unexpected return, revert action to "dump" */
>                 if (master)
>                         strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action));
> 
> anyway, a simple "sizeof" should be used AFAICT.
>

I agree.