Hi François, Gerd, et al,
Francois Romieu wrote:
> Realtek has specified that the post 8168c gigabit chips and the post
> 8105e fast ethernet chips recover automatically from a Rx FIFO overflow.
> The driver does not need to clear the RxFIFOOver bit of IntrStatus and
> it should rather avoid messing it.
>
> The implementation deserves some explanation:
I would be interested in some subset of these fixes for 3.0.y and
2.6.32.y. In particular:
> 1. events outside of the intr_event bit mask are now ignored.
> a no-processing policy for the events that either should not be there
> or should be ignored.
This seems like a valuable and unrisky change.
> 2. RxFIFOOver was already ignored in rtl_cfg_infos[RTL_CFG_1] for the
> whole 8168 line of chips with two exceptions:
> - RTL_GIGA_MAC_VER_22 since b5ba6d12bdac21bc0620a5089e0f24e362645efd
> ("use RxFIFO overflow workaround for 8168c chipset.").
> This one should now be correctly handled.
This seems useful if we can test it.
[...]
> 3. RxFIFOOver is masked for post 8105e 810x chips, namely the sole 8105e
> (RTL_GIGA_MAC_VER_30) itself.
This also seems useful if we can test it.
What do you think? Is there any way I can help? (E.g., given rough
guidelines about what approach is acceptable I'd be happy to work with
Gerd to produce a tested patch that does (1) and (2) but not (3) for
3.0.y.)
Thanks for your hard work,
Jonathan
Jonathan Nieder <[email protected]> :
[...]
> What do you think? Is there any way I can help? (E.g., given rough
> guidelines about what approach is acceptable I'd be happy to work with
> Gerd to produce a tested patch that does (1) and (2) but not (3) for
> 3.0.y.)
Few guidelines are needed: go for it.
I would trust the manufacturer for (3) though.
If you still have time to help, you should really try to break davem-next
branch. I won't claim it's bug-free but the IRQ / napi / Tx processing in
the r8169 driver should now be easier to read and maintain.
--
Ueimor