On Wed, Jun 21, 2023 at 11:08:46AM +0300, Dmitry Antipov wrote:
> On 6/20/23 19:08, Brian Norris wrote:
>
> > I'm not sure how a compile-time constant makes this "unsafe" at all, but
> > if you feel the need to change this, then sure, this works too.
>
> The only reason is to avoid strlcpy() which is now considered deprecated.
Sure, OK.
> > FWIW, this 'firmware' pointer is all derived from compile-time constants
> > too. So the commit messages seems misleading ("all other calls [...]
> > should be safe" --> well, *all* calls are safe).
>
> Indeed. So I think we can stay with strcpy() everywhere except strlcpy() to strscpy() replacement
> (just to follow https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy rather than
> to fix something).
That works too. It's cool to drop stcrpy() anyway though, since it's
really just a feature of a poor language (C) that we have to reason
about whether any given string operation is "safe" or not. I was just
noting that your commit message reasoning was slightly off.
Brian
