netdev - Re: [PATCH v4] net: bnx2x: convert to hw

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:	Tue, 12 Apr 2011 16:49:40 +0200
From:	Michał Mirosław <mirq-linux@...e.qmqm.pl>
To:	Vladislav Zolotarov <vladz@...adcom.com>
Cc:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	Eilon Greenstein <eilong@...adcom.com>
Subject: Re: [PATCH v4] net: bnx2x: convert to hw_features

On Tue, Apr 12, 2011 at 05:36:52PM +0300, Vladislav Zolotarov wrote:
> On Tue, 2011-04-12 at 07:07 -0700, Michał Mirosław wrote:
> > On Tue, Apr 12, 2011 at 03:10:28PM +0300, Vladislav Zolotarov wrote:
> > > On Mon, 2011-04-11 at 13:26 -0700, Michał Mirosław wrote:
> > > > Since ndo_fix_features callback is postponing features change when
> > > > bp->recovery_state != BNX2X_RECOVERY_DONE, netdev_update_features()
> > > > has to be called again when this condition changes. Previously,
> > > > ethtool_ops->set_flags callback returned -EBUSY in that case
> > > > (it's not possible in the new model).
> > > ACK (with reservations). ;)
> > > Could u, pls., just add this comment I've asked for in the previous
> > > e-mail? 
> > The one about vlan_features? 
> Yeah, this one.
> > I'll just fix the comment in netdevice.h
> > instead, since it might be not clear enough.
> Could u still add this comment to bnx2x in your patch as well. It'll
> just make these code section more clear regardless the netdevice.h
> contents... ;)

This would be duplicating information easily found elsewhere.
I don't like it. :)

> > > The things I first thought to comment on are:
> > > 	- Removing TPA_ENABLED_FLAG the similar way u've removed the
> > > bp->rx_csum.
> > > 	- Merging the code handling 'features' in bnx2x_init_bp() with the
> > > similar code in bnx2x_init_dev().
> > > 
> > > However I think it would be right if we clear our mess by ourselves and
> > > that u have already done much enough... ;) 
> > > 
> > > I've run our standard test suite (which in particular heavily tests the
> > > RX_CSUM and LRO flags toggling) on this patch and it passed it.
> > It might be possible to get rid of ndo_set_features, since it looks like
> > the reset/recovery handler is doing unload/load itself. This needs more
> > digging into the driver than this simple conversion.
> Hmm... I don't get u here. Although the recovery handler does
> unload/load itself if there has been an attempt to change features
> during the recovery it won't be able to get applied until the whole
> recovery process ends. So, this patch added the extra call for
> netdev_update_features() right after we know that the recovery process
> has successfully ended. So, could u, pls., explain exactly what do u
> mean here by saying "It might be possible to get rid of
> ndo_set_features"?

Hmm. I thought one, and wrote another.

Since bnx2x_parity_recover() runs with rtnl_lock(), as should
netdev_update_features(), then in case the recovery is in progress
it should be enough to not call bnx2x_reload_if_running() then
and just change the flags --- changes will be picked up on bnx2x_nic_load()
after recovery is complete. This removes the need for netdev_update_features()
calls.

Best Regards,
Michał Mirosław
--
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