| Message ID | 64533952-1299-4ae2-860d-b34b97a24d98@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [RFC] powerpc: switch two fixed phy links to full duplex | expand |
On Fri, Dec 05, 2025 at 06:21:50PM +0100, Heiner Kallweit wrote: > These two fixed links are the only ones in-kernel specifying half duplex. > If these could be switched to full duplex, then half duplex handling > could be removed from phylib fixed phy, phylink, swphy. > > The SoC MAC's are capable of full duplex, fs_enet MAC driver is as well. > Anything that would keep us from switching to full duplex? What do we know about the device on the other end of the link? Maybe that is what is limiting it to 10Half? Andrew
On 12/5/2025 6:50 PM, Andrew Lunn wrote: > On Fri, Dec 05, 2025 at 06:21:50PM +0100, Heiner Kallweit wrote: >> These two fixed links are the only ones in-kernel specifying half duplex. >> If these could be switched to full duplex, then half duplex handling >> could be removed from phylib fixed phy, phylink, swphy. >> >> The SoC MAC's are capable of full duplex, fs_enet MAC driver is as well. >> Anything that would keep us from switching to full duplex? > > What do we know about the device on the other end of the link? Maybe > that is what is limiting it to 10Half? > I found no hint that anything is connected to this ethernet port on the two boards. Hard to find any information because the boards are >15yrs old. Seems this are dummy entries, just to let fs_enet load. > Andrew Heiner
diff --git a/arch/powerpc/boot/dts/mgcoge.dts b/arch/powerpc/boot/dts/mgcoge.dts index 9cefed20723..cea9bdc65dc 100644 --- a/arch/powerpc/boot/dts/mgcoge.dts +++ b/arch/powerpc/boot/dts/mgcoge.dts @@ -153,7 +153,11 @@ eth0: ethernet@11a60 { interrupt-parent = <&PIC>; linux,network-index = <0>; fsl,cpm-command = <0xce00000>; - fixed-link = <0 0 10 0 0>; + + fixed-link { + speed = <10>; + full-duplex; + }; }; i2c@11860 { diff --git a/arch/powerpc/boot/dts/tqm8xx.dts b/arch/powerpc/boot/dts/tqm8xx.dts index d16cdfd8120..e582487d5a3 100644 --- a/arch/powerpc/boot/dts/tqm8xx.dts +++ b/arch/powerpc/boot/dts/tqm8xx.dts @@ -185,7 +185,11 @@ eth0: ethernet@a00 { interrupt-parent = <&CPM_PIC>; fsl,cpm-command = <0000>; linux,network-index = <0>; - fixed-link = <0 0 10 0 0>; + + fixed-link { + speed = <10>; + full-duplex; + }; }; }; };
These two fixed links are the only ones in-kernel specifying half duplex. If these could be switched to full duplex, then half duplex handling could be removed from phylib fixed phy, phylink, swphy. The SoC MAC's are capable of full duplex, fs_enet MAC driver is as well. Anything that would keep us from switching to full duplex? Whilst at it, replace the deprecated old fixed-link binding. Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> --- arch/powerpc/boot/dts/mgcoge.dts | 6 +++++- arch/powerpc/boot/dts/tqm8xx.dts | 6 +++++- 2 files changed, 10 insertions(+), 2 deletions(-)