Snapshot 20000104 failed under CygWin 20.1 --- follow-up

[Joel Sherrill]

Rosimildo da Silva wrote:
> 
> I have finally got my application running with this snapshot. Beyond the
> problems stated before to get CDTEST up and running, there are more:
> 
>   + POSIX API behaves strange ( probably bugs or configuration ) when
> ITRON API is turned off.
> 
> One in particular that fails to me is:
> 
>    pthread_key_create()
> 
> It returns ENOMEM error, even though the machine has 128M, I've tried to
> give 8M to the
> workspace without any luck. :-)

Is this with a standard test or one of your own?  You have to configure
a maximum number of keys.  I recall that this is the error when you 
run out of the keys.
 
> I did not debug as much as could, but this snapshot gave me
> a lot of fun. :-).

No reason to debug in this 

> If I turn ITRON API on, ENABLE_RTEMS_ITRON=yes, everythings seems to
> work a lot more reliable.
> 
> Is there any special configuration setting to turn ITRON off ?

It is off by default when explicitly using configure.  The bit script 
user.cfg may accidentally have it turned on.

> The good news is that the "build environment" is getting better for
> CygWin.
> Except for rpcgen, I had no problems building RTEMS under CygWin this
> time.

Great.

> Rosimildo.
> 
> Rosimildo da Silva wrote:
> >
> > I need more coffee to continue tonight...
> >
> > I guess I found the problems.
> >
> > There are 2 problems:
> >
> >  pc386/start/start16.S
> > -- somebody left a comment open in the routine delay() "eating"
> > the return. This drove me nuts for while.
> >
> > pc386/startup/ldsegs.S
> > --  the jump to _load_segments() from start() never returns with
> > the code in 20000104 snapshot. I guess this new delay() is devil. -:).
> >
> > I have replaced these two files with my old ones from rtems-19991117
> > and I have got cdtest to work so far. What a pain, jesus !!!.
> >
> > I guess somedoby was poking around with the "delay()" routine
> > and caused a few problems.
> >
> > Do you know what was being solved here ?
> > None of the changes made sense to me.
> >
> > Rosimildo.
> >
> >
> >
> > Rosimildo da Silva wrote:
> > >
> > > BSP === pc386
> > > -------------
> > >
> > > Turning remote debug off, I've got everything configuring and
> > > compiling ok. This seems to be a new option that I never used
> > > before.
> > >
> > > ENABLE_RTEMS_RDBG=no
> > >
> > > If we support this option for PC386, it is broken. It does not
> > > pass the configuration phase.
> > >
> > > Now, the bad news. This snapshot does not work.
> > >
> > > GRUB    -- locks up.
> > > NETBOOT -- keeps rebooting.
> > >
> > > What the hell did change so drastically ?
> > >
> > > NOTE: I did not compiled gcc/newlib.
> > >
> > > Rosimildo.

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985