Messages in this thread | | | Subject | Re: [PATCH net-next] net: dsa: qca8k: Initialize the switch with correct number of ports | From | Michal Vokáč <> | Date | Fri, 25 Oct 2019 12:50:13 +0200 |
| |
On 24. 10. 19 18:33, Florian Fainelli wrote: > On 10/24/19 7:12 AM, Andrew Lunn wrote: >> On Thu, Oct 24, 2019 at 03:46:58PM +0200, Michal Vokáč wrote: >>> Since commit 0394a63acfe2 ("net: dsa: enable and disable all ports") >>> the dsa core disables all unused ports of a switch. In this case >>> disabling ports with numbers higher than QCA8K_NUM_PORTS causes that >>> some switch registers are overwritten with incorrect content. >> >> Humm. >> >> The same problem might exist in other drivers: >> >> linux/drivers/net/dsa$ grep -r "ds->num_ports = DSA_MAX_PORTS" >> qca8k.c: priv->ds->num_ports = DSA_MAX_PORTS; >> b53/b53_common.c: ds->num_ports = DSA_MAX_PORTS; > > Not for b53 because this later gets clamped with dev->num_ports in > b53_setup().
I quickly checked the code and I think there is still an issue in the b53_enable_port and b53_disable_port functions which are called from the dsa_port_setup().
>> mt7530.c: priv->ds->num_ports = DSA_MAX_PORTS; >> microchip/ksz_common.c: ds->num_ports = DSA_MAX_PORTS;
At first glance it looks like mt7530 and microchip has the same problem.
>> dsa_loop.c: ds->num_ports = DSA_MAX_PORTS; >> >> dsa_loop.c looks O.K, it does support DSA_MAX_PORTS ports. >> >> But the others? >> >> Andrew
I can respin and fix those drivers as well. Or a separate patch for each one?
Michal
| |