On Mon, Jan 08, 2024 at 09:51:18PM -0800, Darrick J. Wong wrote:
> From: Darrick J. Wong <djwong@xxxxxxxxxx>
>
> A recent refactoring to xfs_idata_realloc in the kernel made it depend
> on krealloc returning NULL if the new size is zero. The xfsprogs
> wrapper instead aborts, so we need to make it follow the kernel
> behavior.
>
> Signed-off-by: Darrick J. Wong <djwong@xxxxxxxxxx>
Reviewed-by: Carlos Maiolino <cmaiolino@xxxxxxxxxx>
> ---
> libxfs/kmem.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/libxfs/kmem.c b/libxfs/kmem.c
> index 42d813088d6a..c264be018bdc 100644
> --- a/libxfs/kmem.c
> +++ b/libxfs/kmem.c
> @@ -98,6 +98,16 @@ kmem_zalloc(size_t size, int flags)
> void *
> krealloc(void *ptr, size_t new_size, int flags)
> {
> + /*
> + * If @new_size is zero, Linux krealloc will free the memory and return
> + * NULL, so force that behavior here. The return value of realloc with
> + * a zero size is implementation dependent, so we cannot use that.
> + */
> + if (!new_size) {
> + free(ptr);
> + return NULL;
> + }
> +
> ptr = realloc(ptr, new_size);
> if (ptr == NULL) {
> fprintf(stderr, _("%s: realloc failed (%d bytes): %s\n"),
