netdev - Re: [PATCH 2/2] netlink: export policy in extended ACK

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20201006091644.0425e0fe@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 6 Oct 2020 09:16:44 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Johannes Berg <johannes@...solutions.net>
Cc:     netdev@...r.kernel.org, dsahern@...il.com
Subject: Re: [PATCH 2/2] netlink: export policy in extended ACK

On Tue, 06 Oct 2020 17:10:44 +0200 Johannes Berg wrote:
> Sorry, hat to run out earlier and forgot to comment here.
> 
> On Tue, 2020-10-06 at 14:32 +0200, Johannes Berg wrote:
> > 
> > +	/* the max policy content is currently ~44 bytes for range min/max */
> > +	if (err && nlk_has_extack && extack && extack->policy)
> > +		tlvlen += 64;  
> 
> So I'm not really happy with this. I counted 44 bytes content (so 48
> bytes for the nested attribute) for the biggest that we have now, but if
> we ever implement e.g. dumping out the reject string for NLA_REJECT
> (though not sure anyone even uses that?) then it'd be more variable.

I wonder if we should in fact dump the reject string, in this case it
feels like an omission not to have it... although as you say, grep for
reject_message reveals it's completely unused today.

> I couldn't really come up with any better idea, but I believe we do need
> to size the skb fairly well to return the original one ...
> 
> The only solution I _could_ think of was to allocate another skb, put
> the attribute into it, check the length, and then later append it to the
> message ... but that seemed kinda ugly.
> 
> Any preferences?

It'd feel pretty idiomatic for (rt)netlink to have

	netlink_policy_dump_attr_size()

which would calculate the size. That'd cost us probably ~100 LoC?

If that's too much we could at least add a define for this constant,
and WARN_ON_ONCE() in __netlink_policy_dump_write_attr() if the dump
ends up larger?