Re: [PATCH net-next] net/devlink: Use strscpy() to copy strings into arrays

[David Laight <david.laight.linux@gmail.com>]

On Wed, 10 Jun 2026 13:58:51 -0700
Kees Cook <kees@kernel.org> wrote:

> On Tue, Jun 09, 2026 at 03:39:35PM +0200, Paolo Abeni wrote:
> > On 6/8/26 11:54 AM, david.laight.linux@gmail.com wrote:  
> > > From: David Laight <david.laight.linux@gmail.com>
> > > 
> > > Replacing strcpy() with strscpy() ensures that overflow of the target
> > > buffer cannot happen.
> > > 
> > > Signed-off-by: David Laight <david.laight.linux@gmail.com>
> > > ---
> > > This is one of a group of patches that remove potentially unbounded
> > > strcpy() calls.
> > > 
> > > They are mostly replaced by strscpy() or, when strlen() has just been
> > > called, with memcpy() (usually including the '\0').
> > > 
> > > Calls with copy string literals into arrays are left unchanged.
> > > They are safe and easily detected as such.
> > > 
> > > The changes were made by getting the compiler to detect the calls and
> > > then fixing the code by hand.
> > > 
> > > Note that all the changes are only compile tested.
> > > 
> > > Some Makefiles were changed to allow files to contain strcpy().
> > > As well as 'difficult to fix' files, this included 'show' functions
> > > as they really need to use sysfs_emit() or seq_printf().
> > > 
> > > All the patches are being sent individually to avoid very long cc lists.
> > > Apologies for the terse commit messages and likely unexpected tags.
> > > (There are about 100 patches in total.)
> > > 
> > >  net/devlink/port.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/net/devlink/port.c b/net/devlink/port.c
> > > index 485029d43428..108926d3f899 100644
> > > --- a/net/devlink/port.c
> > > +++ b/net/devlink/port.c
> > > @@ -1222,7 +1222,7 @@ static void __devlink_port_type_set(struct devlink_port *devlink_port,
> > >  			devlink_port->type_eth.ifindex = netdev->ifindex;
> > >  			BUILD_BUG_ON(sizeof(devlink_port->type_eth.ifname) !=
> > >  				     sizeof(netdev->name));
> > > -			strcpy(devlink_port->type_eth.ifname, netdev->name);
> > > +			strscpy(devlink_port->type_eth.ifname, netdev->name);  
> > 
> > Given the above BUILD_BUG, I don't see how this change can help?!?
> > 
> > Generally speaking, I suggest restricting this kind of tool-assisted
> > changes to real problems (if any).  
> 
> How did this patch get generated? This isn't an "unbounded" case: both
> devlink_port->type_eth.ifname and netdev->name have known sizes. I think
> a way to focus these kinds of refactors would be to replace strcpy()
> with a macro wrapper that fails a build when the sizes aren't known, but
> then otherwise calls strscpy instead. With that in place, you can find
> all the actually unbounded cases and work through each of them, and once
> they have all been purged from the kernel, that strcpy macro can land,
> which will keep new unbounded instances from showing up.
> 

My current wrapper (which will allow the above is):
#if !defined(ALLOW_STRCPY)
#define strcpy(dst, src) ({ \ 
        size_t _dst_size, _src_size; \
        char *const _dst = dst; \
        const char *_src = src; \
        bool _use_memcpy = true; \
\
        _dst_size = __is_array(dst) ? sizeof(dst) : __builtin_object_size(dst, 1); \
        BUILD_BUG_ON_MSG(!_dst_size, "strcpy() dst size zero"); \
        BUILD_BUG_ON_MSG(_dst_size == SIZE_MAX, "strcpy() dst size unknown"); \
\
        if (__builtin_constant_p(_src && _src[0])) \
                _src_size = __builtin_strlen(_src) + 1; \
        else { \
                BUILD_BUG_ON_MSG(!__is_array(src), "strcpy() src unbounded"); \
                _src_size = sizeof(src); \
                BUILD_BUG_ON_MSG(!_src_size, "strcpy() src is zero length array"); \
                _use_memcpy = _src_size <= 32; \
        } \
        BUILD_BUG_ON_MSG(!statically_true(_src_size <= _dst_size), "strcpy() src too big"); \
        if (_use_memcpy) \
                __underlying_memcpy(_dst, _src, _src_size); \
        else \
                __underlying_strcpy(_dst, _src); \
        _dst; \
})

__FORTIFY_INLINE __diagnose_as(__builtin_strcpy, 1, 2)
char *(strcpy)(char * const POS p, const char * const POS q)
{
        BUILD_BUG_ON_MSG(1, "Direct call to strcpy");
        return p;
}
#else
...
#endif

The one I was using before required the source be a constant string.

Note that this is just intended to find the unbounded copies at compile time,
rather than being a suggested patch.
The ALLOW_STRCPY is needed for things like the acpi code which are out of tree
(and other files I didn't want to fix).

The acpi code is stunningly bad, it saves the string lengths for the kmalloc()
calls and then uses strcpy() to copy the data.
If anyone manages to gen them out of sync it will overwrite memory.

	David