Re: [AVaRICE-user] info io_registers hangs on 2.13 with m164p

[Joerg W. <j...@ur...>]

As Oliver Večerník wrote:

> > Here how it looked for me (AVR-GDB 7.3.1, AVaRICE running on an
> > ATmega1284P which has a few more IO registers than your ATmega164P):
> 
> Unfortunately I can't get 7.3.1 compiled on my system.

Why not?

>  I can confirm
> the same behaviour with the latest binutils-gdb (GNU gdb (GDB)
> 7.7.50.20140120-cvs) code and Ubuntu 13.10:

Yes, I can also confirm the bug for GDB 7.6.2 (which is the latest
official release).

Apparently, someone broke target_read_alloc() (or its API changed?).
The infinite loop happens in avr-tdep.c, function
avr_io_reg_read_command():

  /* Find out how many io registers the target has.  */
  bufsiz = target_read_alloc (&current_target, TARGET_OBJECT_AVR,
                              "avr.io_reg", &buf);

This target_read_alloc() is supposed to return the number of IO
registers (the value passed from AVaRICE), but it never returns.

Please write a GDB bug report for this.

> BTW: is there any specific reason (besides it works) for using such an
> ancient version on your system?

I wouldn't consider it "ancient" (it's about a couple of years old),
but the reason is: yes, because it works.  With minor bugs, yes, like
GDB's unability to correctly handle the AVR's Harvard architecture,
and always getting a mixup between code and data pointers when
specifying an address.  But in general, it is usable.

I simply never blindly upgraded it without giving it some testing
before.

> Is FreeBSD, I suppose, preferred over
> (Ubuntu) Linux for AVR work?

I don't think so.  Currently, I lag quite behind in upgrading the
FreeBSD ports, and in other areas (like GCC), it's really due to
upgrade because many things changed to the better recently.
-- 
cheers, Joerg               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/
Never trust an operating system you don't have sources for. ;-)