netdev - Re: ixp4xx_hss modifications for 2x4 HDLC

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:	Wed, 25 Feb 2009 00:38:42 +0100
From:	Krzysztof Halasa <khc@...waw.pl>
To:	Miguel Ángel Álvarez 
	<gotzoncabanes@...il.com>
Cc:	linux-arm-kernel <linux-arm-kernel@...ts.arm.linux.org.uk>,
	netdev@...r.kernel.org
Subject: Re: ixp4xx_hss modifications for 2x4 HDLC

Hi,

Miguel Ángel Álvarez <gotzoncabanes@...il.com> writes:

> I would appreciate your comments very much, because I think this
> functionallity is quite usefull and I would like to add it to the main
> kernel stream.

Well, I guess it's a bit like a proof of concept rather than
an applicable patch? :-)

And it's based on "chan" branch. This isn't good, the "chan" isn't yet
ready to go upstream, the interface isn't yet stable. It's really
experimental, based on undocumented behaviour etc. Do you need the
"channelized" feature? If not, perhaps you should stick to HDLC-only
David's net-next?

Ok. The following doesn't IMHO make sense:

> @@ -349,12 +294,58 @@
>  		.set_clock	= hss_set_clock,
>  		.open		= hss_open,
>  		.close		= hss_close,
> -		.txreadyq	= 2,
> +		.txreadyq	= 32,
> +		.hss        = 0,
> +		.hdlc       = 0,
>  	}, {
>  		.set_clock	= hss_set_clock,
>  		.open		= hss_open,
>  		.close		= hss_close,
> -		.txreadyq	= 19,
> +		.txreadyq	= 36,
> +		.hss        = 1,
> +		.hdlc       = 0,
> +	}, {
> +		.set_clock	= hss_set_clock,
> +		.open		= hss_open,
> +		.close		= hss_close,

There are only two HSS devices on the chip. Not 8 :-)

> @@ -367,6 +358,30 @@
>  		.name			= "ixp4xx_hss",
>  		.id			= 1,
>  		.dev.platform_data	= hss_plat + 1,
> +	}, {
> +		.name			= "ixp4xx_hss",
> +		.id			= 2,
> +		.dev.platform_data	= hss_plat + 2,
> +	}, {
> +		.name			= "ixp4xx_hss",
> +		.id			= 3,
> +		.dev.platform_data	= hss_plat + 3,

Same here.

This only needs a tiny change, increasing the number of HDLC devices
to 2 or 4 (for each HSS), adding the extra queue numbers etc.

This is completely orthogonal to:
a) clocks
b) channelized "services"
c) sync options (excluding 1024-bit max length).

The clock generator is a different thing and needs to be investigated.
Copying Intel's table doesn't make sense here (especially given the
internal clocks are rarely used). Do you need internal clocks?

Channelized thing needs a stable userspace interface first, it really
have to be discussed on netdev.

Sync options are simply independent.

I know I promised you I'll look at this stuff soon. Unfortunately, that
"soon" is a bit less "soon" than I though. Will look when able,
currently I'm really overwhelmed.
-- 
Krzysztof Halasa
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html