linux-kernel - Re: [PATCH net-next] net: phy: qcom: qca808x: Add .get_rate

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: <aMf/e5GfO9HyU+n0@oss.qualcomm.com>
Date: Mon, 15 Sep 2025 17:28:51 +0530
From: Mohd Ayaan Anwar <mohd.anwar@....qualcomm.com>
To: "Russell King (Oracle)" <linux@...linux.org.uk>, andrew@...n.ch
Cc: Heiner Kallweit <hkallweit1@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>, netdev@...r.kernel.org,
        linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH net-next] net: phy: qcom: qca808x: Add .get_rate_matching
 support

On Mon, Sep 15, 2025 at 02:25:42PM +0530, Mohd Ayaan Anwar wrote:
> On Sun, Sep 14, 2025 at 07:10:36PM +0100, Russell King (Oracle) wrote:
> > On Sun, Sep 14, 2025 at 08:36:48PM +0530, Mohd Ayaan Anwar wrote:
> > > Add support for rate matching to the QCA8081 PHY driver to correctly
> > > report its capabilities. Some boards[0][1] with this PHY currently
> > > report support only for 2.5G.
> > > 
> > > Implement the .get_rate_matching callback to allow phylink to determine
> > > the actual PHY capabilities and report them accurately.
> > 
> > Sorry, but this is incorrect.
> > 
> > The PHY does not support rate matching, but switches between SGMII
> > and 2500BASE-X depending on the negotiated speed according to the code:
> > 
> > static void qca808x_fill_possible_interfaces(struct phy_device *phydev)
> > {
> >         unsigned long *possible = phydev->possible_interfaces;
> > 
> >         __set_bit(PHY_INTERFACE_MODE_SGMII, possible);
> > 
> >         if (!qca808x_is_1g_only(phydev))
> >                 __set_bit(PHY_INTERFACE_MODE_2500BASEX, possible);
> > }
> > 
> > static int qca808x_read_status(struct phy_device *phydev)
> > {
> > ...
> >         if (phydev->link) {
> >                 if (phydev->speed == SPEED_2500)
> >                         phydev->interface = PHY_INTERFACE_MODE_2500BASEX;
> >                 else
> >                         phydev->interface = PHY_INTERFACE_MODE_SGMII;
> >         } else {
> > 
> > The driver certainly does not support rate-matching, even if the PHY
> > can support it, and even with your patch. All you are doing is making
> > ethtool suggest that other speeds are supported, but I think you'll
> > find that if the PHY negotiates those speeds, it won't work.
> > 
> 
> Weirdly, I was able to test both 1G and 2.5G with my patch. Could this
> be because the driver is already deviating from the standard in other
> areas?
> 
> > So, the bug is likely elsewhere, or your ethernet MAC doesn't support
> > SGMII and you need to add complete support for  rate-matching to the
> > driver.
> > 
> 
> I tried setting phy-mode=sgmii in the Devicetree and I am able to get 1G
> and lower speeds to work.
> 
> > Please enable phylink debugging and send the kernel messages so I can
> > see what's going on.
> > 
> 
> Filtered logs (without my patch):
> [    7.937871] qcom-ethqos 23040000.ethernet: IRQ eth_wake_irq not found
> [    7.944581] qcom-ethqos 23040000.ethernet: IRQ eth_lpi not found
> [    7.953753] qcom-ethqos 23040000.ethernet: User ID: 0x20, Synopsys ID: 0x52
> [    7.960927] qcom-ethqos 23040000.ethernet:   DWMAC4/5
> [    7.966049] qcom-ethqos 23040000.ethernet: DMA HW capability register supported
> [    7.973564] qcom-ethqos 23040000.ethernet: RX Checksum Offload Engine supported
> [    7.981073] qcom-ethqos 23040000.ethernet: TX Checksum insertion supported
> [    7.988139] qcom-ethqos 23040000.ethernet: TSO supported
> [    7.993603] qcom-ethqos 23040000.ethernet: Enable RX Mitigation via HW Watchdog Timer
> [    8.001654] qcom-ethqos 23040000.ethernet: Enabled L3L4 Flow TC (entries=8)
> [    8.008817] qcom-ethqos 23040000.ethernet: Enabled RFS Flow TC (entries=10)
> [    8.008819] qcom-ethqos 23040000.ethernet: Enabling HW TC (entries=128, max_off=64)
> [    8.008821] qcom-ethqos 23040000.ethernet: TSO feature enabled
> [    8.008822] qcom-ethqos 23040000.ethernet: SPH feature enabled
> [    8.008824] qcom-ethqos 23040000.ethernet: Using 36/40 bits DMA host/device width
> [    8.243500] qcom-ethqos 23040000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-0
> [    8.253778] qcom-ethqos 23040000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-1
> [    8.261991] qcom-ethqos 23040000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-2
> [    8.262527] qcom-ethqos 23040000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-3
> [    8.348697] qcom-ethqos 23040000.ethernet eth0: PHY stmmac-0:1c uses interfaces 4,23, validating 23
> [    8.358304] qcom-ethqos 23040000.ethernet eth0:  interface 23 (2500base-x) rate match none supports 6,13-14,47
> [    8.368589] qcom-ethqos 23040000.ethernet eth0: PHY [stmmac-0:1c] driver [Qualcomm QCA8081] (irq=POLL)
> [    8.368595] qcom-ethqos 23040000.ethernet eth0: phy: 2500base-x setting supported 0000000,00000000,00008000,00006040 advertising 0000000,00000000,00008000,00006040
> [    8.381057] qcom-ethqos 23040000.ethernet eth0: Enabling Safety Features
> [    8.416398] qcom-ethqos 23040000.ethernet eth0: IEEE 1588-2008 Advanced Timestamp supported
> [    8.425541] qcom-ethqos 23040000.ethernet eth0: registered PTP clock
> [    8.434778] qcom-ethqos 23040000.ethernet eth0: configuring for phy/2500base-x link mode
> [    8.446169] qcom-ethqos 23040000.ethernet eth0: major config, requested phy/2500base-x
> [    8.454323] qcom-ethqos 23040000.ethernet eth0: interface 2500base-x inband modes: pcs=00 phy=00
> [    8.463353] qcom-ethqos 23040000.ethernet eth0: major config, active phy/outband/2500base-x
> [    8.471939] qcom-ethqos 23040000.ethernet eth0: phylink_mac_config: mode=phy/2500base-x/none adv=0000000,00000000,00000000,00000000 pause=00
> [    8.485780] 8021q: adding VLAN 0 to HW filter on device eth0
> [    8.489653] qcom-ethqos 23040000.ethernet eth0: phy link down 2500base-x/Unknown/Unknown/none/off/nolpi
> [   13.615848] qcom-ethqos 23040000.ethernet eth0: phy link up 2500base-x/2.5Gbps/Full/none/rx/tx/nolpi
> [   13.617924] qcom-ethqos 23040000.ethernet eth0: Link is Up - 2.5Gbps/Full - flow control rx/tx
> 
> // I changed the link partner speed to 1G here:
> [   74.031182] qcom-ethqos 23040000.ethernet eth0: phy link down 2500base-x/Unknown/Unknown/none/off/nolpi
> [   74.031773] qcom-ethqos 23040000.ethernet eth0: Link is Down
> 
> For reference, this board is using the same MAC as [0] which works
> perfectly fine with the AQR115C PHY. I got the (wrong) idea to add
> .get_rate_matching after comparing the two PHY drivers. The MAC driver
> is stmmac/dwmac-qcom-ethqos.c
> 
> 	Ayaan
> ---
> [0] https://elixir.bootlin.com/linux/v6.17-rc5/source/arch/arm64/boot/dts/qcom/sa8775p-ride-r3.dts

Sorry, I missed adding the following.

After checking the phylink logs, I am now wondering if we need to
register a `.get_interfaces` callback in `dwmac-qcom-ethqos.c` and
do something like:

case PHY_INTERFACE_MODE_2500BASEX:
	__set_bit(PHY_INTERFACE_MODE_2500BASEX, interfaces);
	fallthrough;
case PHY_INTERFACE_MODE_SGMII:
	__set_bit(PHY_INTERFACE_MODE_SGMII, interfaces);
	break;
...

This is based on my limited understanding, so I’d love to hear your
thoughts.

	Ayaan