Messages in this thread | | | Date | Thu, 19 Jun 2008 03:20:05 -0500 (CDT) | From | "R.L. Horn" <> | Subject | Re: 2.6.25.3: serial problem (minicom) |
| |
On Wed, 18 Jun 2008, Alan Cox wrote:
> I don't generally work via git so I don't offhand know the changeset id. > I guess it could do. If its causing this many actual problem cases it > might also want to go into stable.
I vote for nipping it in the bud ASAP. There are currently eight stable kernel versions that exhibit this bug (which I believe is fairly serious, in breadth, if not depth), and it already has Adam Lackorzynski and me scratching our heads trying to figure out what, if anything, to do with minicom to accomodate it.
I would suggest one change to the 26-rc version. Rather than bypassing tty_termios_encode_baud_rate() entirely in the B0 case, why not do something like:
if (tty_termios_baud_rate(termios)) tty_termios_encode_baud_rate(termios, baud, baud); else tty_termios_encode_baud_rate(termios, 0, 0);
to ensure that c_ispeed and c_ospeed are set (just for the sake of consistency)?
| |