minicom bug and linux kernel 2.6.25.4

[Adam Lackorzynski]

On Mon Jun 09, 2008 at 22:55:33 -0500, R.L. Horn wrote:
> On Mon, 9 Jun 2008, Adam Lackorzynski wrote:
>
>> Looks like I'm seeing the same with my Courier now too but somehow your 
>> patch doesn't help (same non-working behaviour with it and without).  
>> Need to further investigate this...
>
> Try this one (against the 2.3 sources).  It adds two functions, 
> m_setdtr() and, conditionally, m_cleardtr(), that complement m_setrts() 
> (using TIOCMOD[GS] instead of TIOCMBI[CS]).  m_setdtr() is called after  
> m_setrts() and is also used by m_dtrtoggle(), so RTS and DTR should both  
> be positively set.
>
> If that doesn't work, you might check that you have all the serial ioctls 
> used by minicom (at least TIOCM_DTR, TIOCM_RTS, TIOCMGET, and TIOCMSET)  
> defined.  A quick way to do so, after applying the included patch, is:
>
>   gcc -DHAVE_CONFIG_H -I. -I.. -I../lib -E sysdep1.c | grep m_cleardtr
>
> On my system, (glibc 2.4 built against linux 2.6.18), TIOCM(GET|SET) are  
> defined in <asm/iotcls.h> (kernel header included from <bits/ioctl.h>  
> included from <sys/ioctl.h>) and TIOCM_(DTR|RTS) are defined in  
> <bits/ioctl-types.h> (glibc header included from <sys/ioctl.h>).
>
> In particular, TIOCMGET and TIOCMSET may not be defined depending on the  
> state of your userspace kernel headers (i.e. <linux/*> and <asm/*>).  
> With the 2.5 and early 2.6 kernels, there was a lot of BS from the linux  
> developers about not including kernel headers in userspace programs (they 
> maintained that distributors should "sanitize" all the headers) and, for  
> some time, the kernel headers were practically unusable.  As a result, a  
> lot of screwed-up headers ended up in the wild.
>
>
>
> Or, the serial drivers could just be buggered...

Thanks for your effort. I found my problem... I just used the wrong
cable (uuups...). So when I try everything works now, without any
modifications in minicom and on 2.6.25. So I can suspect something very
special. On the other hand the standard PC stuff should all use the same
serial driver, weird...
So I suspect there's nothing strange in your dmesg about ttyS? (non-16550A
type e.g.)


Adam
-- 
Adam                 adam at os.inf.tu-dresden.de
  Lackorzynski         http://os.inf.tu-dresden.de/~adam/