Re: Strange behaviour after cross-compiling
On Wed, Dec 13, 2000 at 07:59:37AM +0100, Farid Hajji wrote:
> 1. gnumach from CVS can't do color display (that's okay, but
> I'd like to know how the freeX gnumach.gz was generated!).
>From the debian source package (see the debian ftp archive).
> 2. passwords are displayed on screen; every command is
> also echoed after being submitted.
> 3. commands can be used as usual (e.g. ls works etc...),
> but as soon as non-dumb screen capabilities are needed
> (including pipes!), I get the following error message:
> bash [36: 1] tcsetattr: (ipc/mig) server type check failure
That means the message that was send by the user is rejected by the mig
generated code on the server side. This seems to point to a broken cross-mig
or something like that. Make sure everything is allright with your cross mig
(build it from current CVS), and the *.defs files in the include/hurd
directory of the build tree are allright. Then recompile glibc and the Hurd
> The error is not only from bash, but from other programs as well,
> e.g. vi:
> ex/vi: Error: tcsetattr: (ipc/mig) server type check failure
You will get this from all ncurses using binaries, and probably more.
> Every time this error occurs, the program is stopped by bash
> and resuming it with 'fg' results in the same error again.
I have seen this recently, but only with a very hacked glibc I tried
locally. So I know what you see, but don't know what happens in your case.
> I tried to set TERM from mach-color to dumb, but the error
This has nothing to do with the TERM variable. The TIOCSETA ioctl is
failing, because the generated message (sysdepds/mach/hurd/ioctl.c) is
rejected by the Hurd term server.
`Rhubarb is no Egyptian god.' Debian http://www.debian.org email@example.com
Marcus Brinkmann GNU http://www.gnu.org firstname.lastname@example.org