Thread (21 messages) flat view 21 messages, 3 authors, 2021-03-26

Re: [PATCH net-next v2 11/12] net: phy: marvell10g: print exact model

From: Marek Behún <kabel@kernel.org>
Date: 2021-03-25 20:55:11

On Thu, 25 Mar 2021 21:44:21 +0100
Heiner Kallweit [off-list ref] wrote:
On 25.03.2021 21:29, Marek Behún wrote:
quoted
On Thu, 25 Mar 2021 15:54:52 +0000
Russell King - ARM Linux admin [off-list ref] wrote:
  
quoted
The 88X3310 and 88X3340 can be differentiated by bit 3 in the revision.
In other words, 88X3310 is 0x09a0..0x09a7, and 88X3340 is
0x09a8..0x09af. We could add a separate driver structure, which would
then allow the kernel to print a more specific string via standard
methods, like we do for other PHYs. Not sure whether that would work
for the 88X21x0 family though.  
According to release notes it seems that we can also differentiate
88E211X from 88E218X (via bit 3 in register 1.3):
 88E211X has 0x09B9
 88E218X has 0x09B1

but not 88E2110 from 88E2111
    nor 88E2180 from 88E2181.

These can be differentiated via register
  3.0004.7
(bit 7 of MDIO_MMD_PCS.MDIO_SPEED., which says whether device is capable
 of 5g speed)
  
If the PHY ID's are the same but you can use this register to
differentiate the two versions, then you could implement the
match_phy_device callback. This would allow you to have separate
PHY drivers. This is just meant to say you have this option, I don't
know the context good enough to state whether it's the better one.
Nice, didn't know about that. But I fear whether this would always work
for the 88X3310 vs 88X3310P, it is possible that this feature is only
recognizable if the firmware in the PHY is already running.

I shall look into this.

Marek
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help