Hi maxime, I actually didn't use experimental-3.10 (sorry if I said it wrongly), I only took some patches which relate to sun4i-emac and applied it against mainline 3.10.
Two days ago I gave up emac and continued on stmmac. I picked early stmmac patches from arm mailing list and applied it on mainline 3.10. It behaved the same way as sun4i-emac, the driver got probed, but can't send or receive packet (thus can't udhcpc or ping). I figured out it was because I built the kernel on big endian mode. When I built it on little endian mode, dhcp could work. However, I needed big endian. I don't know why the driver does not work accordingly on big endian kernel.. (I wasn't trying to understand how these driver work, so maybe I missed something) Yes, I absolutely need 3.10 since I am trying to port an already established system that has user applications compiled on big endian mode. I only need the rootfs to get working either via NFS or MMC. Isa On Fri, Dec 12, 2014 at 6:09 PM, Maxime Ripard < [email protected]> wrote: > Hi, > > On Mon, Dec 08, 2014 at 03:00:20AM -0800, [email protected] wrote: > > Hi, > > > > I am trying to run kernel 3.10 (I know, it's old and experimental, > > but I need it) on Cubieboard2. I referenced linux-sunxi's > > "experimental/sunxi-3.10" branch, and took some patch out of it > > concerning pinctrl, clk, and dts. > > > > I gave up trying to backport mmc support from 3.16 since it's too > > far, so I am trying to enable emac to boot from NFS. I saw EMAC > > "enable" patch on experimental/sunxi-3.10, and tried it on. Since > > nothing yet under driver/net/ethernet/allwinner on 3.10, I copied > > the code for sun4i-emac driver from 3.11. > > > > Running on initramfs, the emac driver loaded and the link is on > > (eth0 shown on ifconfig), but dhcp cannot retreive ip address. I > > tried setting the ip manually, but I can't seem to connect to the > > network. No ping reached. > > > > So I tried debugging sun4i-emac, I figured that : > > 1. emac_tx_done is called after emac_start_xmit, but not sure if it's > sent out > > of PHY > > 2. emac_rx is called from ISR, however the packet is always removed from > > FIFO after failing to check the EMAC_UNDOCUMENTED_MAGIC on the packet > header. > > It always show the same value: 7fbb7dbe > > > > Has anyone tried experimental-3.10 on cubieboard2 and ran emac before? > > I would seriously advise against using the experimental-3.10 > kernel. On top of being experimental, it's not (and never has been) > actively maintained, is mostly untested, and you'll encounter such > things. > > Why do you absolutely need such an old kernel? > > Do you have the same behaviour if you take a mainline 3.10 and > backport the patches? > > Maxime > > -- > Maxime Ripard, Free Electrons > Embedded Linux, Kernel and Android engineering > http://free-electrons.com > > -- > You received this message because you are subscribed to a topic in the > Google Groups "linux-sunxi" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/linux-sunxi/3H_cAHr2apc/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > For more options, visit https://groups.google.com/d/optout. > -- Isa Ansharullah aLab, inc. Tokyo, Shibuya-ku, Yoyogi 1-42-10 Brooklyn Court 2F -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
