Re: [PATCH net-next v5 6/8] net: bcmgenet: pad transmit frames out of the packet ready window

[Nicolai Buchwitz <nb@tipi-net.de>]

On 17.10.2026 03:25, Jakub Kicinski wrote:
> On Tue, 13 Oct 2026 16:20:35 +0200 Nicolai Buchwitz wrote:
>> A frame that ends a few bytes past the transmit packet ready threshold
>> stops the transmitter as soon as a shorter frame follows. Tx DMA then
>> refuses to halt, so every later bcmgenet_init_dma() fails and the 
>> interface
>> cannot be opened again. IP fragmentation generates that pattern on 
>> every
>> datagram, full frames and a short tail.
> 
> This is really odd. Sounds like the IP can't really support high MTU
> properly. Should the user be somehow informed that the driver will
> append out-of-spec bytes to their frames or better still opt into
> this hack?

Fair, I don't like it either. In the last days I learned, that the bad
window is exactly one 64 byte DMA unit wide, so it smells like packet
ready and end of packet landing in the same transfer. Splitting the
frame over two descriptors probably avoids it and leaves the wire bytes 
alone.

In v6 I will do that and drop the padding.

Thanks,
Nicolai

--
pw-bot: cr