On 8/11/2023 4:29 PM, Justin Chen wrote:
> The 54810 does not support c45. The mmd_phy_indirect accesses return
> arbirtary values leading to odd behavior like saying it supports EEE
> when it doesn't. We also see that reading/writing these non-existent
> MMD registers leads to phy instability in some cases.
>
> Signed-off-by: Justin Chen <[email protected]>
Thanks for submitting this fix, I would be tempted to slap a:
Fixes: b14995ac2527 ("net: phy: broadcom: Add BCM54810 PHY entry")
so we get it back ported to stable trees where appropriate. It is not
clear whether we should return -EINVAL vs. -EOPNOTSUPP which may more
clearly indicate the inability to support MMD registers?
--
Florian
On 8/11/2023 4:47 PM, Justin Chen wrote:
> On Fri, Aug 11, 2023 at 4:40 PM Florian Fainelli
> <[email protected]> wrote:
>>
>>
>>
>> On 8/11/2023 4:29 PM, Justin Chen wrote:
>>> The 54810 does not support c45. The mmd_phy_indirect accesses return
>>> arbirtary values leading to odd behavior like saying it supports EEE
>>> when it doesn't. We also see that reading/writing these non-existent
>>> MMD registers leads to phy instability in some cases.
>>>
>>> Signed-off-by: Justin Chen <[email protected]>
>>
>> Thanks for submitting this fix, I would be tempted to slap a:
>>
>> Fixes: b14995ac2527 ("net: phy: broadcom: Add BCM54810 PHY entry")
>>
>> so we get it back ported to stable trees where appropriate. It is not
>> clear whether we should return -EINVAL vs. -EOPNOTSUPP which may more
>> clearly indicate the inability to support MMD registers?
>
> Hmm agreed EOPNOTSUPP seems better here. Will submit v2 with fixes tag
> if there are no objections to this patch.
Since this is a fix, you will want to use [PATCH net] as the subject
prefix per:
https://www.kernel.org/doc/Documentation/networking/netdev-FAQ.txt
and wait 24hrs in case someone provides additional review that you can
incorporate in your v2.
--
Florian
On Fri, Aug 11, 2023 at 4:40 PM Florian Fainelli
<[email protected]> wrote:
>
>
>
> On 8/11/2023 4:29 PM, Justin Chen wrote:
> > The 54810 does not support c45. The mmd_phy_indirect accesses return
> > arbirtary values leading to odd behavior like saying it supports EEE
> > when it doesn't. We also see that reading/writing these non-existent
> > MMD registers leads to phy instability in some cases.
> >
> > Signed-off-by: Justin Chen <[email protected]>
>
> Thanks for submitting this fix, I would be tempted to slap a:
>
> Fixes: b14995ac2527 ("net: phy: broadcom: Add BCM54810 PHY entry")
>
> so we get it back ported to stable trees where appropriate. It is not
> clear whether we should return -EINVAL vs. -EOPNOTSUPP which may more
> clearly indicate the inability to support MMD registers?
Hmm agreed EOPNOTSUPP seems better here. Will submit v2 with fixes tag
if there are no objections to this patch.
Thanks,
Justin
> --
> Florian
> Hmm agreed EOPNOTSUPP seems better here.
I was thinking the same.
Andrew