[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4249c911-91c8-5d56-78c2-460e8711456d@seco.com>
Date: Tue, 7 Mar 2023 11:41:04 -0500
From: Sean Anderson <sean.anderson@...o.com>
To: Andrew Lunn <andrew@...n.ch>,
"Russell King (Oracle)" <linux@...linux.org.uk>
Cc: Heiner Kallweit <hkallweit1@...il.com>, netdev@...r.kernel.org,
"David S . Miller" <davem@...emloft.net>,
Vladimir Oltean <olteanv@...il.com>,
linux-kernel@...r.kernel.org, Paolo Abeni <pabeni@...hat.com>,
Eric Dumazet <edumazet@...gle.com>,
Florian Fainelli <f.fainelli@...il.com>,
Tobias Waldekranz <tobias@...dekranz.com>,
Jakub Kicinski <kuba@...nel.org>
Subject: Re: [PATCH net-next] net: mdio: Add netlink interface
On 3/7/23 08:47, Andrew Lunn wrote:
> On Mon, Mar 06, 2023 at 10:48:48PM +0000, Russell King (Oracle) wrote:
>> On Mon, Mar 06, 2023 at 03:45:16PM -0500, Sean Anderson wrote:
>> > +static int mdio_nl_eval(struct mdio_nl_xfer *xfer)
>> > +{
>> > + struct mdio_nl_insn *insn;
>> > + unsigned long timeout;
>> > + u16 regs[8] = { 0 };
>> > + int pc, ret = 0;
>>
>> So "pc" is signed.
>>
>> > + int phy_id, reg, prtad, devad, val;
>> > +
>> > + timeout = jiffies + msecs_to_jiffies(xfer->timeout_ms);
>> > +
>> > + mutex_lock(&xfer->mdio->mdio_lock);
>> > +
>> > + for (insn = xfer->prog, pc = 0;
>> > + pc < xfer->prog_len;
>>
>> xfer->prog_len is signed, so this is a signed comparison.
>>
>> > + case MDIO_NL_OP_JEQ:
>> > + if (__arg_ri(insn->arg0, regs) ==
>> > + __arg_ri(insn->arg1, regs))
>> > + pc += (s16)__arg_i(insn->arg2);
>>
>> This adds a signed 16-bit integer to pc, which can make pc negative.
>>
>> And so the question becomes... what prevents pc becoming negative
>> and then trying to use a negative number as an index?
>
> I don't know ebpf very well, but would it of caught this? I know the
> aim of this is to be simple, but due to its simplicity, we are loosing
> out on all the inherent safety of eBPF. Is a eBPF interface all that
> complex? I assume you just need to add some way to identify MDIO
> busses and kfunc to perform a read on the bus?
Regarding eBPF over netlink, the last time this was discussed, Tobias said
> - Why not use BPF?
>
> That could absolutely be one way forward, but the GENL approach was
> easy to build out-of-tree to prove the idea. Its not obvious how it
> would work though as BPF programs typically run async on some event
> (probe hit, packet received etc.) whereas this is a single execution
> on behalf of a user. So to what would the program be attached? The
> output path is also not straight forward, but it could be done with
> perf events i suppose.
I'm not familiar enough with eBPF to comment further.
--Sean
Powered by blists - more mailing lists