Re: issue with _Static_assert and __builtin()s

On 09/02/2026 16:25, Jeff Johnson wrote:
> On 1/7/26 21:41, Dan Carpenter wrote:
>> On Wed, Jan 07, 2026 at 07:41:45PM +0000, Ben Dooks wrote:
>>> So I am looking at why the Linux kernel's check for nul characters
>>> in strings is causing errors out of sparse.
>>>
>>> EG:
>>> drivers/md/dm.c:3813:1: error: bad constant expression
>>> drivers/md/dm.c:3814:1: error: bad constant expression
>>> drivers/md/dm.c:3816:1: error: bad constant expression
>>> drivers/md/dm.c:3817:1: error: bad constant expression
>>>
>>> I've tracked it down to the sizeof(str) - 1 == __builtin_strlen(str)
>>> failing to be a good constant expression...
>>>
>>> This is an example of the assert which isn't working:
>>>
>>> _Static_assert(sizeof("moo") - 1 == __builtin_strlen("moo"), "nul!");
>>>
>>> This does at least get past w/o warnings
>>> _Static_assert(__builtin_types_compatible_p(int, int), "doh!");
>>>
>>> I've had a go at updating builtin.c to deal with __builtin_strlen()
>>> for a string constant (attached) but that's just changing the
>>> output to "error: bad integer constant expression" so not sure
>>> what I've missed here.
>>>
>>
>> Al has a fix for that.
>> https://git.kernel.org/pub/scm/linux/kernel/git/viro/sparse.git/commit/?id=2634e39bf02697a18fece057208150362c985992
> 
> Al: Will you be contributing this fix (and perhaps the rest of your
> accumulated changes) to linux-sparse?
> 
> /jeff

And are there any other goodies in that tree?

-- 
Ben Dooks				http://www.codethink.co.uk/
Senior Engineer				Codethink - Providing Genius

https://www.codethink.co.uk/privacy.html