netdev - Re: [PATCH net-next 2/4] igb: do not clobber the TSAUXC bits on reset.

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]

Message-ID: <1416266522.21363.47.camel@jekeller-desk1.amr.corp.intel.com>
Date:	Mon, 17 Nov 2014 23:22:02 +0000
From:	"Keller, Jacob E" <jacob.e.keller@...el.com>
To:	"richardcochran@...il.com" <richardcochran@...il.com>
CC:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"davem@...emloft.net" <davem@...emloft.net>,
	"Allan, Bruce W" <bruce.w.allan@...el.com>,
	"Ronciak, John" <john.ronciak@...el.com>,
	"Kirsher, Jeffrey T" <jeffrey.t.kirsher@...el.com>,
	"Vick, Matthew" <matthew.vick@...el.com>
Subject: Re: [PATCH net-next 2/4] igb: do not clobber the TSAUXC bits on
 reset.

On Tue, 2014-11-18 at 00:06 +0100, Richard Cochran wrote:
> The TSAUXC register has a number of different bits, one of which disables
> the main clock function. Previously, the clock was re-enabled by clearing
> the entire register. This patch changes the code to preserve the values
> of the other bits in that register.
> 

This is a step in the right direction, but won't fully work due to this
being called after a MAC reset.

> Signed-off-by: Richard Cochran <richardcochran@...il.com>
> ---
>  drivers/net/ethernet/intel/igb/igb_ptp.c |    8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/intel/igb/igb_ptp.c b/drivers/net/ethernet/intel/igb/igb_ptp.c
> index 794c139..ce57128 100644
> --- a/drivers/net/ethernet/intel/igb/igb_ptp.c
> +++ b/drivers/net/ethernet/intel/igb/igb_ptp.c
> @@ -905,6 +905,8 @@ void igb_ptp_stop(struct igb_adapter *adapter)
>  void igb_ptp_reset(struct igb_adapter *adapter)
>  {
>  	struct e1000_hw *hw = &adapter->hw;
> +	unsigned long flags;
> +	u32 tsauxc;
>  
>  	if (!(adapter->flags & IGB_FLAG_PTP))
>  		return;
> @@ -923,7 +925,11 @@ void igb_ptp_reset(struct igb_adapter *adapter)
>  	case e1000_i210:
>  	case e1000_i211:
>  		/* Enable the timer functions and interrupts. */
> -		wr32(E1000_TSAUXC, 0x0);
> +		spin_lock_irqsave(&adapter->tmreg_lock, flags);
> +		tsauxc = rd32(E1000_TSAUXC);
> +		tsauxc &= ~TSAUXC_DISABLE;
> +		wr32(E1000_TSAUXC, tsauxc);
> +		spin_unlock_irqrestore(&adapter->tmreg_lock, flags);

This won't work. We call igb_ptp_reset directly after a HW reset (MAC
reset). TSAUXC data is not maintained. We probably need to store driver
state of the TSAUXC bits, and then re-enable them in this reset path.
This code today is fine, because there was no other state we cared about
in the path.

That being said, I don't think this change breaks anything. I just think
it will not solve the problem.

Thanks,
Jake

>  		wr32(E1000_TSIM, TSYNC_INTERRUPTS);
>  		wr32(E1000_IMS, E1000_IMS_TS);
>  		break;