Andreson: A practical advice regarding the radio stack code: compile with -DTASKLET_IS_TASK to avoid the tasklet code being compiled in.
The issue is, btw, with tos-ramsize, as it blindly traverses the call graph without looking at data dependencies or higher level usage semantics. It tells you the worst case stack usage. The best fix for this problem would be defining some sort of code annotations as a means to tell the stack depth checker that an apparent recursion is not a true recursion in practice. Janos On Tue, Apr 10, 2012 at 9:48 AM, Anderson Nascimento <[email protected]> wrote: > Hello, > > I have been recently having some problems that I think may be caused by > stack overflow. I found tos-ramsize (and stack-check make option) but came > up with some errors. > > When testing apps/RadioCountToLeds with the iris platform, I get the > following: > >> cannot estimate stack depth due to recursive loop of length 3: >> TaskletC__Tasklet__schedule @ 1862 -> TaskletC__doit @ f18 >> TaskletC__doit @ f18 -> RF230DriverLayerP__Tasklet__schedule @ 1886 >> RF230DriverLayerP__Tasklet__schedule @ 1886 -> >> TaskletC__Tasklet__schedule @ 1862 >> tos-ramsize FAIL at /usr/bin/tos-ramsize line 1543. >> make: [exe0] Error 255 (ignored) > > > When testing apps/tests/storage/Config with the iris platform, I get the > following: > >> cannot estimate stack depth due to recursive loop of length 2: >> At45dbP__handleRWRequest @ 187e -> At45dbP__syncOrFlushAll @ 1d06 >> At45dbP__syncOrFlushAll @ 1d06 -> At45dbP__handleRWRequest @ 187e >> tos-ramsize FAIL at /usr/bin/tos-ramsize line 1543. >> make: [exe0] Error 255 (ignored) > > > Is it possible to remove these recursive loops, or is it something that I > just have to live with? Also, should I expect to see this error for any > other driver or protocol modules? > > PS: I know that these are not the source of my problem, but I cannot run > stack-check on my application if parts other than my application are failing > stack-check. > > Thanks, > -- > Anderson Nascimento > Embedded Systems Engineer > Spensa Technologies, Inc > > > _______________________________________________ > Tinyos-help mailing list > [email protected] > https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help _______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
