linux-next-20170320 break stmmac on dwmac-sunxi
Subsystems (2)
Hello
Just pushed next-20170320 to my boards and stmmac stop working on both intree dwmac-sunxi and my dev dwmac-sun8i. It seems that interrupts never fire, and transmit queue timeout.
I will try to bisect this problem but perhaps other people could try to reproduce it.
Regards
Corentin Labbe
+ Joao
please Joao, could you do a check if new changes have any impact on that?
peppe
Quoted 9 lines · Corentin Labbe
Hello
Just pushed next-20170320 to my boards and stmmac stop working on both intree dwmac-sunxi and my dev dwmac-sun8i. It seems that interrupts never fire, and transmit queue timeout.
I will try to bisect this problem but perhaps other people could try to reproduce it.Regards
Corentin Labbe
Hello Corentin
yes, bisect process is really good approach to me. Pls give us more details. Recently the multi DMA channel logic has been added so it could be that something is needed to allow your platform to manage the new code. Or we introduced some regression. If I have some other idea, I ping you.
Peppe
Quoted 9 lines · Corentin Labbe
Hello
Just pushed next-20170320 to my boards and stmmac stop working on both intree dwmac-sunxi and my dev dwmac-sun8i. It seems that interrupts never fire, and transmit queue timeout.
I will try to bisect this problem but perhaps other people could try to reproduce it.Regards
Corentin Labbe
Hi Peppe,
Às 2:39 PM de 3/21/2017, Giuseppe CAVALLARO escreveu:
Quoted 22 lines
Hello Corentin
yes, bisect process is really good approach to me. Pls give us more details. Recently the multi DMA channel logic has been added so it could be that something is needed to allow your platform to manage the new code. Or we introduced some regression. If I have some other idea, I ping you.
Peppe
Quoted 11 lines · Corentin Labbe
Hello
Just pushed next-20170320 to my boards and stmmac stop working on both intree dwmac-sunxi and my dev dwmac-sun8i.
It seems that interrupts never fire, and transmit queue timeout. I will try to bisect this problem but perhaps other people could try to reproduce it.Regards
Corentin Labbe
Corentin spotted 2 bugs (dma_op_mode was only being set for core versions >= 4.00 and rx/tx default nº of queues was being set in a place that would not be accessible if no multiqueue config is added to the DT). Thierry will send a set of patches to improve soon.
Thanks,
Joao