Re: [PATCH][next] mm/execmem: Remove logically deadcode in execmem.c

[Shuah Khan]

On 10/10/24 12:10, Everest K.C. wrote:
> NULL check of variable `area` within the `mas_for_each` loop is
> unnecessary, as the varialbe `area` can never be NULL. So, the
> `continue` statement inside the if block is never reached.
>
> Remove the if block that performs the NULL check.
>
> This was reported by Coverity Scan:
> https://scan7.scan.coverity.com/#/project-view/51525/11354?selectedIssue=1600362
>
> Fixes: d44c3485820e ("execmem: add support for cache of large ROX pages")
> Signed-off-by: Everest K.C. <everestkc@xxxxxxxxxxxxxxxx>
> ---
>   mm/execmem.c | 3 ---
>   1 file changed, 3 deletions(-)
>
> diff --git a/mm/execmem.c b/mm/execmem.c
> index 9c6ff9687860..97706d8ed720 100644
> --- a/mm/execmem.c
> +++ b/mm/execmem.c
> @@ -75,9 +75,6 @@ static void execmem_cache_clean(struct work_struct *work)
>   	mas_for_each(&mas, area, ULONG_MAX) {
>   		size_t size;
>   
> -		if (!area)
> -			continue;
> -

This more of a question than comment:
mas_for_each() says:
Note: may return the zero entry.

Does that mean mas_range_len() can be zero? Does that
need to be handled?


>   		size = mas_range_len(&mas);
>   
>   		if (IS_ALIGNED(size, PMD_SIZE) &&


thanks,
-- Shuah