netdev - Re: reset value of MV88E6XXX_G1_IEEE

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:   Mon, 27 May 2019 13:02:04 +0000
From:   Rasmus Villemoes <rasmus.villemoes@...vas.dk>
To:     Vivien Didelot <vivien.didelot@...il.com>
CC:     Florian Fainelli <f.fainelli@...il.com>,
        Andrew Lunn <andrew@...n.ch>,
        Network Development <netdev@...r.kernel.org>
Subject: Re: reset value of MV88E6XXX_G1_IEEE_PRI

On 27/05/2019 14.32, Vivien Didelot wrote:
> Hi Rasmus,
> 
> On Mon, 27 May 2019 09:36:13 +0000, Rasmus Villemoes <rasmus.villemoes@...vas.dk> wrote:
>> Looking through the data sheets comparing the mv88e6240 and 6250, I
>> noticed that they have the exact same description of the G1_IEEE_PRI
>> register (global1, offset 0x18). However, the current code used by 6240 does
>>
>> int mv88e6085_g1_ieee_pri_map(struct mv88e6xxx_chip *chip)
>> {
>> 	/* Reset the IEEE Tag priorities to defaults */
>> 	return mv88e6xxx_g1_write(chip, MV88E6XXX_G1_IEEE_PRI, 0xfa41);
>> }
>>
>> while if my reading of the data sheet is correct, the reset value is
>> really 0xfa50 (fields 7:6 and 5:4 are RWS to 0x1, field 3:2 and 1:0 are
>> RWR) - and this is also the value I read from the 6250 on our old BSP
>> with an out-of-tree driver that doesn't touch that register. This seems
>> to go way back (at least 3b1588593097). Should this be left alone for
>> not risking breaking existing setups (just updating the comment), or can
>> we make the code match the comment? Or am I just reading this all wrong?
> 
> If the reset value isn't the same, the bits are certainly differently
> organized inside this register, so the proper way would be to add a
> mv88e6240_g1_ieee_pri_map function, used by both 88E6240 and 88E6250.
> 

Based on the very systematic description [ieee tags 7 and 6 are mapped
to 3, 5 and 4 to 2, 3 and 2 to 1, and 1 and 0 to 0], I strongly believe
that 0xfa50 is also the reset value for the 6085, so this is most likely
wrong for all the current chips - though I don't have a 6085 data sheet.

I can certainly add a 6250 variant that does the right thing for the
6250, and I probably will - this is more a question about the current code.

> I'm not a big fan of rewriting the default values, but that is the
> way we chose until we make actually use of these tag priority bits.

Yes, I was wondering why there's a lot of code which simply serves to
set default values - but I guess it makes sense to force the switch into
a known state in case the bootloader did something odd.

Rasmus