Re: [PATCH v4 1/5] selftests/mm: make file helpers return errors

[Mike Rapoport]

On Wed, 27 May 2026 19:54:28 +0530, Sarthak Sharma <sarthak.sharma@xxxxxxx> wrote:

Hi Sarthak,

>
> diff --git a/tools/testing/selftests/mm/hugepage_settings.c b/tools/testing/selftests/mm/hugepage_settings.c
> index 2eab2110ac6a..c2f97fe97e58 100644
> --- a/tools/testing/selftests/mm/hugepage_settings.c
> +++ b/tools/testing/selftests/mm/hugepage_settings.c
> @@ -61,7 +62,9 @@ int thp_read_string(const char *name, const char * const strings[])
>  		exit(EXIT_FAILURE);
>  	}
>  
> -	if (!read_file(path, buf, sizeof(buf))) {
> +	ret = read_file(path, buf, sizeof(buf));
> +	if (ret < 0) {
> +		errno = -ret;
>  		perror(path);

Hmm, looks like those slipped ksft conversion :/

For kselftest compatibility this should be

	printf("# %s: %s (%d)\n", path, strerror(errno), errno);

> @@ -103,12 +106,18 @@ void thp_write_string(const char *name, const char *val)
>  		printf("%s: Pathname is too long\n", __func__);
>  		exit(EXIT_FAILURE);
>  	}
> -	write_file(path, val, strlen(val) + 1);
> +	ret = write_file(path, val, strlen(val) + 1);
> +	if (ret < 0) {
> +		errno = -ret;
> +		perror(path);
> +		exit(EXIT_FAILURE);

These seem to repeat themself, how about we move prints to read/write
helpers?

Then it wouldn't matter for the callers what the exact error was, and no
need for errno games.

>
> diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
> index 311fc5b4513e..290e5c29123e 100644
> --- a/tools/testing/selftests/mm/vm_util.c
> +++ b/tools/testing/selftests/mm/vm_util.c
> @@ -703,62 +703,72 @@ int read_file(const char *path, char *buf, size_t buflen)
>  	int fd;
>  	ssize_t numread;
>  
> +	if (buflen < 2)
> +		return -EINVAL;

heh, this feels overly protective :)

-- 
Sincerely yours,
Mike.