*** 21/04/2001 : pppd blocked, status = resolved ***

- if pppd is killed with signal 9, pppoa2 remains blocked in read()
thus preventing another pppd to launch a new pppoa2 since the first one
still "own" all USB endpoints.

I thought that normally the pipe between pppd & pppoa2 should be broken.
However here, pppd is run on a slave tty and pppoa2 on the master side of
the tty. What's wrong?

See HDLC_BUG for the solution.

*** 25-04-2001 : pppoa2 hanging, status : resolved ***

- if the pppd connection is broken (by a LCP Terminate), one of
the pppoa2 process does not exit since it is blocked, reading from the USB bus.

*** 30-05-2001 : wrong speed, status : resolved ***

The upstream/downstream speed values are incorrect in the UK. It looks like
an endian pb, but it is not