... in the socket flags. This is causing Edge Triggered > EPOLLOUT events to be missed for ... tcp_sendmsg detects lack of sndbuf space > Case 2: tcp_poll returns not writable > > This is ... not consume it (say a tcp_sendmsg that re-fill the buffer), you can't ...

linux-kernel - Davide Libenzi - Sep 20 2007 - 13:42

... ), if you do not consume it (say a > tcp_sendmsg that re-fill the buffer), you can't see other OUT event > ... happens) more usable. Also the epoll man page says so "... Edge Triggered event distribution delivers events only when events happens on ...

linux-kernel - Nagendra Tomar - Sep 20 2007 - 18:09

... ), if you do not consume it (say a > > tcp_sendmsg that re-fill the buffer), you can't see other OUT event > ... ) more usable. > Also the epoll man page says so > > "... Edge Triggered event distribution delivers events only when > events happens on ...

linux-kernel - Davide Libenzi - Sep 20 2007 - 18:37

... free the buffers. >>>> Note that this patch assumes that the SOCK_NOSPACE check in ... . >>> I already replied to your patch posting explaining that whatever is >>> ... >>> Please address that, thanks. >> You're not planning of putting the notion ...

linux-kernel - Eric Dumazet - Sep 20 2007 - 02:11

... the ET EPOLLOUT event with the SOCK_NOSPACE check in place ? 2. How much effect will removing the check have (if we ... the ET EPOLLOUT notification w/ the check in place) ? Regding (2), IMHO for a "fast sender" the SOCK_NOSPACE check will almost ...

linux-kernel - Nagendra Tomar - Sep 19 2007 - 19:32

... free the buffers. > >>>> Note that this patch assumes that the SOCK_NOSPACE check in > ... > >>> I already replied to your patch posting explaining that whatever is > ... >>> Please address that, thanks. > >> You're not planning of putting the notion ...

linux-kernel - Nagendra Tomar - Sep 20 2007 - 04:02

On Thu, 20 Sep 2007, Eric Dumazet wrote: > Does it means that with your patch each ACK on a ET managed socket will > trigger an epoll event ? > > Maybe your very sensitive high throuput appication needs to set a flag or > something at socket ...

linux-kernel - Davide Libenzi - Sep 20 2007 - 13:56

... what traditional Unix poll referred to. Again, events here are "readability" and "writeability" and the fact that the lower network layer freed a buffer is not, per se, an event (unless before, "writeability" was tested and reported as unavailable ...

linux-kernel - Davide Libenzi - Sep 21 2007 - 13:45

... the ep_poll_callback is not getting called and > hence the socket fd is not getting added to the ready list. I know, I saw the patch. I was just commenting the point where DaveM was heading to ;) This things needs to be looked at a little bit more ...

linux-kernel - Davide Libenzi - Sep 20 2007 - 01:43

... acks free the buffers. > > > Note that this patch assumes that the SOCK_NOSPACE check in > ... > > I already replied to your patch posting explaining that whatever is > ... > Please address that, thanks. > > You're not planning of putting the notion ...

linux-kernel - Nagendra Tomar - Sep 19 2007 - 19:50

--- Davide Libenzi wrote: > > Unfortunately f_op->poll() does not let the caller to specify the events > it's interested in, that would allow to split send/recevie wait queues and > better detect read/write cases. >

linux-kernel - Nagendra Tomar - Sep 20 2007 - 18:24

--- Davide Libenzi wrote: > That's not what POLLOUT means in the Unix meaning. POLLOUT indicates the > ability to write, and it is not meant as to signal every time a packet > (skb) is sent on the wire (and the buffer

linux-kernel - Nagendra Tomar - Sep 20 2007 - 18:58

... the incoming acks free the buffers. > Note that this patch assumes that the SOCK_NOSPACE check in > tcp_check_space is ... can be safely removed. I already replied to your patch posting explaining that whatever is not setting SOCK_NOSPACE should ...

linux-kernel - David Miller - Sep 19 2007 - 18:44

... when the incoming acks free the buffers. > > Note that this patch assumes that the SOCK_NOSPACE check in > > tcp_check_space is ... safely removed. > > I already replied to your patch posting explaining that whatever is > not setting SOCK_NOSPACE ...

linux-kernel - Nagendra Tomar - Sep 19 2007 - 18:55

From: Nagendra Tomar Date: Wed, 19 Sep 2007 15:55:58 -0700 (PDT) > I agree that setting SOCK_NOSPACE would have been a more elegant > fix. Infact I thought a lot about that before deciding on this fix. I guess this

linux-kernel - David Miller - Sep 19 2007 - 19:10

... acks free the buffers. > > Note that this patch assumes that the SOCK_NOSPACE check in > ... > I already replied to your patch posting explaining that whatever is > not ... > Please address that, thanks. You're not planning of putting the notion ...

linux-kernel - Davide Libenzi - Sep 19 2007 - 19:11