netdev - Re: [PATCH v2 net 1/1] amd-xgbe: fix active cable

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: <64edbded-51af-f055-9c2f-c1f81b0d3698@kupper.org>
Date:   Sat, 12 Nov 2022 20:12:24 +0100
From:   Thomas Kupper <thomas@...per.org>
To:     Tom Lendacky <thomas.lendacky@....com>, netdev@...r.kernel.org
Cc:     Jakub Kicinski <kuba@...nel.org>,
        Raju Rangoju <Raju.Rangoju@....com>
Subject: Re: [PATCH v2 net 1/1] amd-xgbe: fix active cable



On 11/11/22 17:00, Thomas Kupper wrote:
> 
> On 11/11/22 15:18, Tom Lendacky wrote:
>> On 11/11/22 02:46, Thomas Kupper wrote:
>>> When determine the type of SFP, active cables were not handled.
>>>
>>> Add the check for active cables as an extension to the passive cable check.
>>
>> Is this fixing a particular problem? What SFP is this failing for? A more descriptive commit message would be good.
>>
>> Also, since an active cable is supposed to be advertising it's capabilities in the eeprom, maybe this gets fixed via a quirk and not a general check this field.

Tom,

are you sure that an active cable has to advertising it's speed? Searching for details about it I read in "SFF-8472 Rev 12.4", 5.4.2, Table 5-5 Transceiver Identification Examples:

Transceiver Type Transceiver Description	Byte	Byte	Byte	Byte	Byte	Byte	Byte	Byte
						3	4	5	6	7 	8	9	10
...
		10GE Active cable with SFP(3,4)	 00h	00h	00h	00h	00h	08h	00h	00h

And footnotes:
3) See A0h Bytes 60 and 61 for compliance of these media to industry electrical specifications
4) For Ethernet and SONET applications, rate capability of a link is identified in A0h Byte 12 [nominal signaling
rate identifier]. This is due to no formal IEEE designation for passive and active cable interconnects, and lack
of corresponding identifiers in Table 5-3.

Wouldn't that suggest that byte 3 to 10 are all zero, except byte 8?


/Thomas

> 
> It is fixing a problem regarding a Mikrotik S+AO0005 AOC cable (we were in contact back in Feb to May). And your right I should have been more descriptive in the commit message.
> 
>>>
>>> Fixes: abf0a1c2b26a ("amd-xgbe: Add support for SFP+ modules")
>>> Signed-off-by: Thomas Kupper <thomas.kupper@...il.com>
>>> ---
>>>   drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c | 5 +++--
>>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
>>> index 4064c3e3dd49..1ba550d5c52d 100644
>>> --- a/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
>>> +++ b/drivers/net/ethernet/amd/xgbe/xgbe-phy-v2.c
>>> @@ -1158,8 +1158,9 @@ static void xgbe_phy_sfp_parse_eeprom(struct xgbe_prv_data *pdata)
>>>       }
>>>
>>>       /* Determine the type of SFP */
>>> -    if (phy_data->sfp_cable == XGBE_SFP_CABLE_PASSIVE &&
>>> -        xgbe_phy_sfp_bit_rate(sfp_eeprom, XGBE_SFP_SPEED_10000))
>>> +    if ((phy_data->sfp_cable == XGBE_SFP_CABLE_PASSIVE ||
>>> +         phy_data->sfp_cable == XGBE_SFP_CABLE_ACTIVE) &&
>>> +         xgbe_phy_sfp_bit_rate(sfp_eeprom, XGBE_SFP_SPEED_10000))
>>
>> This is just the same as saying:
>>
>>     if (xgbe_phy_sfp_bit_rate(sfp_eeprom, XGBE_SFP_SPEED_10000))
>>
>> since the sfp_cable value is either PASSIVE or ACTIVE.
>>
>> I'm not sure I like fixing whatever issue you have in this way, though. If anything, I would prefer this to be a last case scenario and be placed at the end of the if-then-else block. But it may come down to applying a quirk for your situation.
> 
> I see now that this cable is probably indeed not advertising its capabilities correctly, I didn't understand what Shyam did refer to in his mail from June 6.
> 
> Unfortunately I haven't hear back from you guys after June 6 so I tried to fix it myself ... but do lack the knowledge in that area.
> 
> A quirk seems a good option.
> 
> From my point of view this patch can be cancelled/aborted/deleted.
> I'll look into how to fix it using a quirk but maybe I'm not the hest suited candidate to do it.
> 
> /Thomas
> 
>>
>> Thanks,
>> Tom
>>
>>>           phy_data->sfp_base = XGBE_SFP_BASE_10000_CR;
>>>       else if (sfp_base[XGBE_SFP_BASE_10GBE_CC] & XGBE_SFP_BASE_10GBE_CC_SR)
>>>           phy_data->sfp_base = XGBE_SFP_BASE_10000_SR;
>>> -- 
>>> 2.34.1
>>>