Re: [uml-devel] patch uml_net_rx reactivate_fd

[Jeff D. <jd...@ka...>]

ste...@ho... said:
> The attached patch moves reactivate_fd so that it gets called only
> after the  fd has been drained, and no error condition has been
> reported on the fd.

What problem are solving with this patch?  The first chunk is definitely
wrong.  It can permanently shut down a device if a memory allocation fails.

> Is there a way to ensure a fast writer can't keep the UML stuck in
> uml_net_rx indefinitly? 

No.  Physical boxes have the same problem, BTW.

				Jeff