netdev - Re: [PATCH 03/10] stmmac: sanitize the rx coe and add the type-1 csum

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 03 Apr 2012 08:49:34 +0200
From:	Giuseppe CAVALLARO <peppe.cavallaro@...com>
To:	Deepak SIKRI <deepak.sikri@...com>
Cc:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"davem@...emloft.net" <davem@...emloft.net>,
	Srinivas KANDAGATLA <srinivas.kandagatla@...com>,
	spear-devel <spear-devel@...t.st.com>,
	Shiraz HASHIM <shiraz.hashim@...com>,
	Viresh KUMAR <viresh.kumar@...com>,
	"bhutchings@...arflare.com" <bhutchings@...arflare.com>
Subject: Re: [PATCH 03/10] stmmac: sanitize the rx coe and add the type-1
 csum

Hello Deepak,

On 4/2/2012 6:18 PM, Deepak SIKRI wrote:
> On 4/2/2012 4:37 PM, Giuseppe CAVALLARO wrote:
[snip]
>> I've not clear at all your question.
>> The driver well uses the rx_coe as briefly described below:
>>
>> probe funct
>>    |__ hw_init
>>           |_ check the RX type from HW cap reg
>>                   |__ Override the rx_coe if required
>>
>> After that the rx_coe is used and passed to the core as expected.
>> In case of there is no HW cap register so the rx_coe from platform will
>> be used.
>>
>> Peppe
> 
> In the same patch, this portion of the code has been removed.
> 
> -static int dwmac1000_rx_coe_supported(void __iomem *ioaddr)
> -{
> -    u32 value = readl(ioaddr + GMAC_CONTROL);
> -
> -    value |= GMAC_CONTROL_IPC;
> -    writel(value, ioaddr + GMAC_CONTROL);
> -
> -    value = readl(ioaddr + GMAC_CONTROL);
> -
> -    return !!(value&  GMAC_CONTROL_IPC);
> -}
> 
> Earlier this was taking care of setting the IP Checksum offloading feature
> in case its available. This code has to be present, as I do not see any
> other location where the IPC bit is being programmed.
> 
> Also, the location of setting the IPC should be post the mac has been
> reset.

Previously, the stmmac called the dwmac1000_rx_coe_supported to verify
it could do the CSUM in Hw. If true the driver used the type 2 by default.

I've voluntarily removed this function because not necessary anymore.
In fact, YOU improved the rx_coe from the platform. If it is passed as
STMMAC_RX_COE_NONE then it means the driver is not able to perform any
csum for the incoming frames. This is actually used on old gmac/mac
cores. In new cores, the HW cap register will be used to manage and fix
this logic.
I could restore the core you are mentioning  but just to do another
safety check at run-time in case of the user provided a broken setting
from the platform and there is not the HW cap register. Hmm, I do not
know if this actually could help indeed... just an extra check IMHO.

> 
> I hope this clears the things a bit. Sorry for the miscommunication.

No problem for the miscommunication ;-)

Let me know
Ciao
Peppe

> 
> Rgds
> Deepak
> 
> 
> 
> 
> 
> 
>>> Regards
>>> Deepak
>>>
>> .
>>
> 
> 

--
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