netdev - Re: [PATCH net-next] net: make for_each_netdev

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]

Date: Thu, 13 Jun 2024 18:57:28 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Przemek Kitszel <przemyslaw.kitszel@...el.com>
Cc: <davem@...emloft.net>, <netdev@...r.kernel.org>, <edumazet@...gle.com>,
 <pabeni@...hat.com>
Subject: Re: [PATCH net-next] net: make for_each_netdev_dump() a little more
 bug-proof

On Fri, 14 Jun 2024 03:45:47 +0200 Przemek Kitszel wrote:
> you are right, it would be easier to go one step [of macros] up, we have
>   453│ #define xa_for_each_range(xa, index, entry, start, last)         \
>   454│         for (index = start,                                      \
>   455│              entry = xa_find(xa, &index, last, XA_PRESENT);      \
>   456│              entry;                                              \
>   457│              entry = xa_find_after(xa, &index, last, XA_PRESENT))
> 
> You could simply change L456 to:
> entry || (index = 0);
> to achieve what you want; but that would slow down a little lot's of
> places, only to change value of the index that should not be used :P
> 
> For me a proper solution would be to fast forward into C11 era, and move
> @index allocation into the loop, so value could not be abused.

I think we're already in C11 era for that exact reason 🤐️
But please don't take this as an invitation to do crazy things!

In netlink dumps, tho, we keep the index in persistent socket context,
because the iteration is split into multiple recvmsg() calls,
each one continuing where the previous one left off.