[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
On Tue, 19 Jun 2012 22:18:14 +0200 Lambrecht Jürgen <J.Lambrecht@xxxxxxxxxxx> wrote: > On 06/19/2012 03:16 PM, Jesper K. Pedersen wrote: > > On Sun, 17 Jun 2012 16:43:02 +0200 > > "Jesper K. Pedersen"<linux@xxxxxxxxx> wrote: > > > >> I am at my wits end getting the twothreads example working on my > >> Olimex P256 board (AT91SAM7S256 based). > >> > >> Calling the cyg_mutex_init does not return. Also I notice that the > >> call to the printf line before the call to cyg_mutex_init prints > >> extremely slow on my terminal. > >> > >> I have set the default console up to tty0 going to ser0. This part > >> works fine with my other software running on the board (normal > >> expected transmission speed). > >> > >> Are there any "gotcha'" involving threading on the AT91 series > >> microcontrollers? > >> > > Problem was solved. > > > > For some reason if you put any output function (printf's) in the > > cyg_user_start the threading will hang. > > > > Removing all output will allow for threading and the example works > > as expected. > > > > I wonder if this is intended or if its a bug in the A91SAM7 eCos. > > > > Best regards > > JesperKP > > > I have lots of diag_printf's in my cyg_user_start() running on an > AT91 ARM7 and a ARM9. > Maybe because you use printf? > Be carefull: Code running in initialization context runs with > interrupts disabled and the scheduler locked. Only after > cyg_user_start, the scheduler starts and resumes all threads. > Maybe (if I remember well; search the list for the diff between both > printfs) printf uses interrupts (to know when the hardware has sent a > byte), and diag_printf doesn't. > > You must also realize that printf's can take a lot of CPU. > > And no, it is certainly not intended or a bug in the ecos CVS code > (although I don't know the twothreads test): it is your bug ;-) . > If you look in the ecos-3.0-user-guide.pdf on page 52 "A Sample Program with Two Threads" it is shown with printf's in the cyg_user_start. If that is not a correct way to do it - perhaps that example should be changed? :-D Best regards JesperKP -- Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss