Phil Sutter <phil@xxxxxx> wrote:
> On Thu, Nov 24, 2022 at 02:49:39PM +0100, Florian Westphal wrote:
> > Previous patches cause minor test breakage, e.g:
> > exp: nft 'add rule ip mangle INPUT iifname "iifname" tcp dport 80 ct state invalid,untracked counter synproxy sack-perm timestamp wscale 7 mss 1460'
> > res: nft 'add rule ip mangle INPUT iifname "iifname" tcp dport 80 ct state invalid,untracked counter synproxy sack-perm timestamp wscale 7 mss 1460 '
> >
> > So fix up the ->xlate callbacks of the affected modules to not print a
> > tailing space character.
>
> I am considering something like this:
>
> | diff --git a/libxtables/xtables.c b/libxtables/xtables.c
> | index 479dbae078156..367eefaba8e74 100644
> | --- a/libxtables/xtables.c
> | +++ b/libxtables/xtables.c
> | @@ -2496,6 +2496,11 @@ static void __xt_xlate_add(struct xt_xlate *xl, enum xt_xlate_type type,
> | struct xt_xlate_buf *buf = &xl->buf[type];
> | int len;
> |
> | + if (buf->off && !isspace(buf->data[buf->off - 1])) {
> | + buf->data[buf->off] = ' ';
> | + buf->off++;
> | + buf->rem--;
> | + }
> | len = vsnprintf(buf->data + buf->off, buf->rem, fmt, ap);
> | if (len < 0 || len >= buf->rem)
> | xtables_error(RESOURCE_PROBLEM, "OOM");
>
> Maybe it needs an extra xt_xlate_add_nospc() but it should eliminate the
> majority of the "whitespace here or there or not at all" fiddling all over the
> code. WDYT?
I think its a good idea.
