On Thu, 12 Nov 2020 00:49:03 +0100 Tobias Waldekranz wrote:
> > I'm unclear what this fixes. What functionality is broken on 6097?
>
> VLAN configuration. As soon as you add the second port to a VLAN, all
> other port membership configuration is overwritten with zeroes. The HW
> interprets this as all ports being "unmodified members" of the VLAN.
>
> I suspect that is why it has not been discovered. In the simple case
> when all ports belong to the same VLAN, switching will still work. But
> using multiple VLANs or trying to set multiple ports as tagged members
> will not work.
I see, this info would be good to include in the commit message.
User impact is what backporters care about the most.
> At the lowest level, the current implementation assumes that it can
> perform two consecutive operations where each op will load half of a
> register, and then read out the union of the information. This is true
> for some devices (6352), but not for others (6097).
>
> 6352 pseudo-hdl-in-c:
>
> stu_get_next()
> {
> *data |= stu_data & 0xf0f0;
> }
>
> vtu_get_next()
> {
> *data |= vtu_data & 0x0f0f;
> }
>
> 6097 pseudo-hdl-in-c:
>
> stu_get_next()
> {
> *data = stu_data;
> }
>
> vtu_get_next()
> {
> *data = vtu_data;
> }
>
> > Can we identify the commit for a fixes tag?
>
> I will try to pinpoint it tomorrow. I suppose I should also rebase it
> against "net" since it is a bug.
Indeed, thanks :)
