sparse: ioctl defines and "error: bad integer constant expression"

sparse: ioctl defines and "error: bad integer constant expression"

[Hans Verkuil]

Hi!

Here is another sparse error that I get when running sparse over
drivers/media/v4l2-core/v4l2-ioctl.c:

drivers/media/v4l2-core/v4l2-ioctl.c:2043:9: error: bad integer constant expression
drivers/media/v4l2-core/v4l2-ioctl.c:2044:9: error: bad integer constant expression
drivers/media/v4l2-core/v4l2-ioctl.c:2045:9: error: bad integer constant expression
drivers/media/v4l2-core/v4l2-ioctl.c:2046:9: error: bad integer constant expression

etc.

The root cause of that turns out to be in include/asm-generic/ioctl.h:

#include <uapi/asm-generic/ioctl.h>

/* provoke compile error for invalid uses of size argument */
extern unsigned int __invalid_size_argument_for_IOC;
#define _IOC_TYPECHECK(t) \
        ((sizeof(t) == sizeof(t[1]) && \
          sizeof(t) < (1 << _IOC_SIZEBITS)) ? \
          sizeof(t) : __invalid_size_argument_for_IOC)

If it is defined as this:

#define _IOC_TYPECHECK(t) (sizeof(t))

then all is well with the world.

I can patch v4l2-ioctl.c to redefine _IOC_TYPECHECK if __CHECKER__ is defined, but
shouldn't sparse understand this instead? There was a similar situation with
ARRAY_SIZE in the past that sparse now understands.

Regards,

	Hans

Re: sparse: ioctl defines and "error: bad integer constant expression"

[Hans Verkuil]

On 03/15/2014 12:59 PM, Hans Verkuil wrote:
> Hi!
> 
> Here is another sparse error that I get when running sparse over
> drivers/media/v4l2-core/v4l2-ioctl.c:
> 
> drivers/media/v4l2-core/v4l2-ioctl.c:2043:9: error: bad integer constant expression
> drivers/media/v4l2-core/v4l2-ioctl.c:2044:9: error: bad integer constant expression
> drivers/media/v4l2-core/v4l2-ioctl.c:2045:9: error: bad integer constant expression
> drivers/media/v4l2-core/v4l2-ioctl.c:2046:9: error: bad integer constant expression
> 
> etc.
> 
> The root cause of that turns out to be in include/asm-generic/ioctl.h:
> 
> #include <uapi/asm-generic/ioctl.h>
> 
> /* provoke compile error for invalid uses of size argument */
> extern unsigned int __invalid_size_argument_for_IOC;
> #define _IOC_TYPECHECK(t) \
>         ((sizeof(t) == sizeof(t[1]) && \
>           sizeof(t) < (1 << _IOC_SIZEBITS)) ? \
>           sizeof(t) : __invalid_size_argument_for_IOC)
> 
> If it is defined as this:
> 
> #define _IOC_TYPECHECK(t) (sizeof(t))
> 
> then all is well with the world.
> 
> I can patch v4l2-ioctl.c to redefine _IOC_TYPECHECK if __CHECKER__ is defined, but
> shouldn't sparse understand this instead? There was a similar situation with
> ARRAY_SIZE in the past that sparse now understands.

Here is a small test case for this problem:

====== ioc-typecheck.c ======
extern unsigned int __invalid_size_argument_for_IOC;
#define _IOC_TYPECHECK(t) \
                ((sizeof(t) == sizeof(t[1]) && \
                    sizeof(t) < (1 << 14)) ? \
                   sizeof(t) : __invalid_size_argument_for_IOC)

#define TEST_IOCTL (50 | (_IOC_TYPECHECK(unsigned) << 8))

static unsigned iocnrs[] = {
        [TEST_IOCTL & 0xff] = 1,
};
/*
 * check-name: correct handling of _IOC_TYPECHECK
 *
 * check-error-start
 * check-error-end
 */
====== ioc-typecheck.c ======

Running sparse over this gives:

error: bad integer constant expression

Regards,

	Hans

Re: sparse: ioctl defines and "error: bad integer constant expression"

[Linus Torvalds]

On Mon, Mar 31, 2014 at 1:06 AM, Hans Verkuil <hverkuil@xs4all.nl> wrote:
>
> Running sparse over this gives:
>
> error: bad integer constant expression

So technically sparse is correct in this case. The kernel ends up
doing some questionable things that gcc allows. The index in an
assignment initializer is supposed to be a true constant expression,
not "constant expression after simplification and optimization".

So sparse warns about us playing games that just happen to work with
gcc. I'm not sure we should fix sparse for this case.

I'll think about it, but not right now (I did the previous one because
it was trivial, but now I'm in merge window mode, so I'd better go
back and look at kernel pull requests)

Chris, can you please add this to the test cases since Hans did the
work to create a nice small test-case?

            Linus

Re: sparse: ioctl defines and "error: bad integer constant expression"

[Christopher Li]

On Mon, Mar 31, 2014 at 10:22 AM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Mon, Mar 31, 2014 at 1:06 AM, Hans Verkuil <hverkuil@xs4all.nl> wrote:
>
> Chris, can you please add this to the test cases since Hans did the
> work to create a nice small test-case?

Sure, I want to add them to the test case.

Hans, May I get a signed-off for your test cases?

Chris

Re: sparse: ioctl defines and "error: bad integer constant expression"

[Hans Verkuil]

On 04/01/2014 06:48 PM, Christopher Li wrote:
> On Mon, Mar 31, 2014 at 10:22 AM, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
>> On Mon, Mar 31, 2014 at 1:06 AM, Hans Verkuil <hverkuil@xs4all.nl> wrote:
>>
>> Chris, can you please add this to the test cases since Hans did the
>> work to create a nice small test-case?
> 
> Sure, I want to add them to the test case.
> 
> Hans, May I get a signed-off for your test cases?

For all my test cases:

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>

Regards,

	Hans

Re: sparse: ioctl defines and "error: bad integer constant expression"

[Christopher Li]

On Tue, Apr 1, 2014 at 10:06 AM, Hans Verkuil <hverkuil@xs4all.nl> wrote:
>
> For all my test cases:
>
> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
>

All 3 test cases added.

Chris