netdev - Re: KSZ8795 not detected at start to boot from NFS

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: <a578b29f-53f0-4e33-91a4-3932fa759cd1@lunn.ch>
Date: Sat, 7 Dec 2024 21:47:31 +0100
From: Andrew Lunn <andrew@...n.ch>
To: Jörg Sommer <joerg@...so.de>
Cc: netdev@...r.kernel.org, Christian Eggers <ceggers@...i.de>
Subject: Re: KSZ8795 not detected at start to boot from NFS

On Sat, Dec 07, 2024 at 08:53:38AM +0100, Jörg Sommer wrote:
> Hi,
> 
> I'm trying to load the root filesystem via NFS with a
> NET_DSA_MICROCHIP_KSZ8795_SPI attached to an TI_DAVINCI_EMAC. With 5.10 it
> works, but with later versions the kernel fails to detect the switch. It is
> since
> 
> commit 8c4599f49841dd663402ec52325dc2233add1d32
> Author: Christian Eggers <ceggers@...i.de>
> Date:   Fri Nov 20 12:21:07 2020 +0100
> 
>     net: dsa: microchip: ksz8795: setup SPI mode
> 
>     This should be done in the device driver instead of the device tree.
> 
>     Signed-off-by: Christian Eggers <ceggers@...i.de>
>     Signed-off-by: Jakub Kicinski <kuba@...nel.org>
> 
> diff --git drivers/net/dsa/microchip/ksz8795_spi.c drivers/net/dsa/microchip/ksz8795_spi.c
> index 8b00f8e6c02f..f98432a3e2b5 100644
> --- drivers/net/dsa/microchip/ksz8795_spi.c
> +++ drivers/net/dsa/microchip/ksz8795_spi.c
> @@ -49,6 +49,12 @@ static int ksz8795_spi_probe(struct spi_device *spi)
>         if (spi->dev.platform_data)
>                 dev->pdata = spi->dev.platform_data;
> 
> +       /* setup spi */
> +       spi->mode = SPI_MODE_3;
> +       ret = spi_setup(spi);
> +       if (ret)
> +               return ret;
> +
>         ret = ksz8795_switch_register(dev);
> 
>         /* Main DSA driver may not be started yet. */
> 
> The kernel reports
> 
> [    1.912756] ksz8795-switch: probe of spi0.1 failed with error -22
> …
> [    2.048054] davinci_emac 1e20000.ethernet: incompatible machine/device type for reading mac address
> [    2.062352] davinci_emac davinci_emac.1: failed to get EMAC clock
> [    2.068632] davinci_emac: probe of davinci_emac.1 failed with error -16
> 
> It tries to initialize the switch before the ethernet of the SoC is ready.
> 
> Before this commit the kernel returned EPROBE_DEFER instead of EINVAL (or
> ENODEV) as a quick

This is often true for DSA switches, that the first probe fails with
EPROBE_DEFER, because the conduit ethernet is not ready.

What i don't understand from your description is why:

> +       /* setup spi */
> +       spi->mode = SPI_MODE_3;
> +       ret = spi_setup(spi);
> +       if (ret)
> +               return ret;
> +

is causing this issue. Is spi_setup() failing?

	Andrew