Funnily enough I just started looking into xfsdump warnings a few
minutes ago..
> --- a/common/global.c
> +++ b/common/global.c
> @@ -82,7 +82,7 @@ global_hdr_alloc(int argc, char *argv[])
>
> /* fill in the magic number
> */
> - strncpy(ghdrp->gh_magic, GLOBAL_HDR_MAGIC, GLOBAL_HDR_MAGIC_SZ);
> + memcpy(ghdrp->gh_magic, GLOBAL_HDR_MAGIC, GLOBAL_HDR_MAGIC_SZ);
This chunk and all the other ones switching to memcpy where we have
a fixed size look good and impossible to improve on to me.
> - sprintf(question,
> + snprintf(question, sizeof(question),
> "pre-erase (-%c) option specified "
> "and non-blank media encountered:\n"
> "please confirm media erase "
> diff --git a/invutil/fstab.c b/invutil/fstab.c
For this and a few others that just s(n)printf I wonder if just
switching to asprintf and dynamically allocating the buffer is
the right thing to do. That's a GNU/BSD extension, but we probably
don't care about anything else.
> index 88d849e..56132e1 100644
> --- a/invutil/fstab.c
> +++ b/invutil/fstab.c
> @@ -149,7 +149,7 @@ fstab_select(WINDOW *win, node_t *current, node_t *list)
> int
> fstab_highlight(WINDOW *win, node_t *current, node_t *list)
> {
> - static char txt[256];
> + static char txt[512];
And for put_info_line/put_line I suspect just passing a format
string is the best thing to do, as this avoids the extra
snprintf and buffer entirely. That's in fact what I had just started
on.
