Re: Coverity: iopt_map_user_pages(): Null pointer dereferences

[Jason Gunthorpe <jgg@nvidia.com>]

On Wed, Nov 02, 2022 at 12:57:05PM -0700, coverity-bot wrote:
> Hello!
> 
> This is an experimental semi-automated report about issues detected by
> Coverity from a scan of next-20221102 as part of the linux-next scan project:
> https://scan.coverity.com/projects/linux-next-weekly-scan
> 
> You're getting this email because you were associated with the identified
> lines of code (noted below) that were touched by commits:
> 
>   Mon Oct 31 13:37:57 2022 -0300
>     91b4be750274 ("iommufd: Data structure to provide IOVA to PFN mapping")
> 
> Coverity reported the following:
> 
> *** CID 1527090:  Null pointer dereferences  (REVERSE_INULL)
> /drivers/iommu/iommufd/io_pagetable.c: 417 in iopt_map_user_pages()
> 411     	list_add(&elm.next, &pages_list);
> 412
> 413     	rc = iopt_map_pages(iopt, &pages_list, length, iova, iommu_prot, flags);
> 414     	if (rc) {
> 415     		if (elm.area)
> 416     			iopt_abort_area(elm.area);
> vvv     CID 1527090:  Null pointer dereferences  (REVERSE_INULL)
> vvv     Null-checking "elm.pages" suggests that it may be null, but it has already been dereferenced on all paths leading to the check.
> 417     		if (elm.pages)
> 418     			iopt_put_pages(elm.pages);
> 419     		return rc;
> 420     	}
> 421     	return 0;
> 422     }
> 
> If this is a false positive, please let us know so we can mark it as
> such, or teach the Coverity rules to be smarter. If not, please make
> sure fixes get into linux-next. :) For patches fixing this, please
> include these lines (but double-check the "Fixes" first):
> 
> Reported-by: coverity-bot <keescook+coverity-bot@chromium.org>
> Addresses-Coverity-ID: 1527090 ("Null pointer dereferences")
> Fixes: 91b4be750274 ("iommufd: Data structure to provide IOVA to PFN mapping")
> 
> This looks like the earlier "if (IS_ERR(elm.pages))" should use
> IS_ERR_OR_NULL() ?

No, that exits the function, can't get to this code with a NULL or ERR
from iopt_alloc_pages()

It becomes NULL because one of the things iopt_map_pages() is to make
it NULL.

I think it is complaining that on the rc!=0 path of iopt_map_pages()
it never makes it NULL.

But I think I will leave this alone as the protocol to destroy an elm
is to do all these tests, and an open coded stack creation of a elm
shouldn't be asymetric with the heap allocated elms.

Thanks,
Jason