uplcom / ucom problems on RELENG_5

Bernd Walter ticso at cicely12.cicely.de
Wed Mar 30 04:50:07 PST 2005


On Wed, Mar 30, 2005 at 07:22:07AM -0500, Mike Tancsa wrote:
> At 05:41 AM 30/03/2005, Bernd Walter wrote:
> >> putc to a clist with no reserved cblocks
> >> putc to a clist with no reserved cblocks
> >> putc to a clist with no reserved cblocks
> >> putc to a clist with no reserved cblocks
> >> putc to a clist with no reserved cblocks
> >> putc to a clist with no reserved cblocks
> >> putc to a clist with no reserved cblocks
> >>
> >> fill up the console.  At this point, ppp is totally hung, and I cannot 
> >kill
> >> ppp
> >
> >You need a way to notice a connection failure.
> >In case of modem or line failure you should see DCD loss.
> >Does your cabling release DCD, or do you use LQR in your configuration?
> >Have you compared the same configuration with a classical 16550 based
> >serial?
> 
> Hi,
>         PPP does see the loss, the problem is that its totally wedged and 
> cant do anything else.  In the case where there is not too much data going 
> across the serial line, the connection dropping has no adverse effect.  I 
> have LCP keep alives enbled and it does detect the link going away and does 
> properly redial.  However, it seems where there is too much outstanding 
> data things get into trouble.

That's possible - currently we can't control the transmit/receive
buffers inside the USB device.

> Like I said, at this point I cant even kill -9 ppp.

That fits - the device buffers even survive a reopen of the tty.

> What more information can I provide to help track this problem down ?

We need to implement complete flushing support for uplcom.

-- 
B.Walter                   BWCT                http://www.bwct.de
bernd at bwct.de                                  info at bwct.de



More information about the freebsd-usb mailing list