Hi Denis,
On Tue, 2011-03-15 at 14:28 -0500, Denis Kenzior wrote:
Since we're dealing with MTUs of ~1500, our frames don't really ever
exceed it. So 256 byte overhead is probably ok, but might be worth
checking if making this dynamic buys us anything.
Sounds like a good idea. We can expect
Hi Patrick,
On 03/08/2011 09:58 AM, Patrick Porlan wrote:
Extend the write buffer handling in gathdlc.c to minimize stalling and
process switching during large PPP transfers. The single write buffer
is replaced by a queue of buffers, allowing for much larger emission
windows without hugely
Extend the write buffer handling in gathdlc.c to minimize stalling and
process switching during large PPP transfers. The single write buffer
is replaced by a queue of buffers, allowing for much larger emission
windows without hugely impacting memory consumption. This reduces the
time required to
Hi Denis,
On Wed, 2011-03-02 at 09:28 -0600, Denis Kenzior wrote:
Is there a particular reason why you chose to use a ring buffer of ring
buffers? A simple GQueue might be much easier to understand. If you are
worried about 'infinite queuing' then a simple counter might help to
Hi Denis,
On Tue, 2011-03-01 at 21:47 -0600, Denis Kenzior wrote:
This is a great description, but right after reading it one should
realize that this patch is better broken down into two. The first patch
addressing the ringbuffer performance improvements and the second one
dealing with
Hi Patrick,
Is there a particular reason why you chose to use a ring buffer of ring
buffers? A simple GQueue might be much easier to understand. If you are
worried about 'infinite queuing' then a simple counter might help to
alleviate that.
Well, this implementation opens the door to
Replace modulo operations in ringbuffer.c by masking operations.
That's possible because the size of the ring buffers is always
a power of two, and yields a small performance improvement.
More importantly, extend the write buffer handling in gathdlc.c
to minimize stalling and process switching
Hi Patrick,
On 03/01/2011 09:07 AM, Patrick Porlan wrote:
Replace modulo operations in ringbuffer.c by masking operations.
That's possible because the size of the ring buffers is always
a power of two, and yields a small performance improvement.
More importantly, extend the write buffer