[ECOS] Proper Thread Environment (was Re: [ECOS] SPI usage)

Gary Thomas gary@mlbassoc.com
Tue Feb 27 12:06:00 GMT 2007


Chuck McManis wrote:
> At 06:56 AM 2/26/2007, Gary Thomas wrote:
>  > Try putting these functions into 'main()', which is called
>  > with a proper thread environment and the kernel+interrupts
>  > running.
> 
> I read this and it reminded me of a question I had, if you have a 
> pointer I'd be grateful for it.
> 
> Basically when I got the HTTP server to run I did so by loading and 
> running the httpd1 test program and noting that it worked fine. Then I 
> tried to run a "main" program and the web server simultaneously. 
> Basically I added:
> 
> So I had this simple bit of code:
> 
> ----------------------- bit of lame included code ------------------------
> 
> #include <stdio.h>
> #include <network.h>
> 
> /*
>  * Note this is for backward compatibility we don't really
>  * think that ECOS will call main, it calls cyg_start and
>  * that function will call main.
>  */
> int
> main(int argc, char *argv[]) {
>         char buf[2048];
> 
>         printf("Welcome to the ECOS world!\n");
>         while (1) {
>                 printf("Enter some text: ");
>                 gets(buf);
>                 printf("'%s'\n", buf);
>         }
>         exit (0);
> }
> /*
>  * This was mostly stolen from the HTTPD test
>  */
> 
> #define STACK_SIZE (8192 + 0x1000)
> static char stack[STACK_SIZE];
> static cyg_thread thread_data;
> static cyg_handle_t thread_handle;
> 
> void
> user_thread(cyg_addrword_t p) {
> 
>         diag_printf("Startup code for ECOS is running...\n");
>         init_all_network_interfaces();
> /*
>         main(1, aa);
> */
> }
> 
> void
> cyg_user_start(void) {
>     cyg_thread_create(10,               // Priority - just a number
>                       user_thread,      // entry
>                       0,                // entry parameter
>                       "simple httpd",   // Name
>                       &stack[0],         // Stack
>                       STACK_SIZE,        // Size
>                       &thread_handle,    // Handle
>                       &thread_data       // Thread data structure
>             );
>     cyg_thread_resume(thread_handle);  // Start it
>         cyg_scheduler_start();
> }
> ----------------------- end of lame included code 
> ---------------------------
> 
> And it has a really weird way of working. Basically what happens is that 
> somehow, main() gets called anyway and runs. And frankly I can't figure 
> out how it gets called, but even stranger, its interaction with the 
> console blocks the networking stack's attempt to print the result of 
> running DHCP which keeps the web server from running. (it nicely 
> demonstrates the Rhine "feature" of resetting CR0 however if you don't 
> service received packets ;-)
> 
> So what I was looking for, and could not find, was sort of a crt0 type 
> file which would provide "the proper thread environment + kernel and 
> interrupts running" which would include starting a thread by calling 
> 'main()'. That way I could keep eCos code fairly segregated from my 
> "application" which normally would run as a process on a UNIX or Linux 
> machine.
> 
> Thoughts? Should I just create two threads? Will they still interfere 
> with each other's use of the console?

Your problem is not with the form of your code; it's fine.
The problem has to do with the console device.  The default console
device is the debug device, normally provided by RedBoot via the
virtual vector interface.  This interface is not interrupt driven,
thus when you ask to read from it, everything comes to a grinding
halt.  To avoid this, you need to set up your console stream to
an interrupt driven device, which will require enabling the appropriate
serial driver.



-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss



More information about the Ecos-discuss mailing list