> -Original Message-
> From: Tuong Lien Tong
> Sent: 26-Aug-19 07:44
> To: Jon Maloy ; 'Jon Maloy'
>
> Cc: Mohan Krishna Ghanta Krishnamurthy
> ;
> parthasarathy.bhuvara...@gmail.com; Tung Quang Nguyen
> ; Hoang Huu Le
> ; Gordan Mihaljevic
> ; ying@windriver.com; tipc-
>
Hi Jon,
Yes, you are right, my previous patch was not complete (sorry, I have not
verified it but just wanted to give a general idea...). Actually, we could
still preserve the necessary data/header of the original message for
building the wakeup message later as needed (look, just the message
Hi Tuong,
While experimenting with byte-oriented flow control I realized that this is a
very real problem that has to be fixed.
I first tried your suggestion with putting the congestion test at the end of
tipc_link_xmit(), but realized that we need access to the original message
header when we
From: Jon Maloy
Date: Tue, 30 Jul 2019 16:23:18 +0200
> In commit 365ad353c256 ("tipc: reduce risk of user starvation during
> link congestion") we allowed senders to add exactly one list of extra
> buffers to the link backlog queues during link congestion (aka
> "oversubscription"). However,
In commit 365ad353c256 ("tipc: reduce risk of user starvation during
link congestion") we allowed senders to add exactly one list of extra
buffers to the link backlog queues during link congestion (aka
"oversubscription"). However, the criteria for when to stop adding
wakeup messages to the input