Suspending Tasks before they are finished
Hi, I've been developing some tests and observed some task behaviour that i can't understand and couldn't find any documentation about it. I have tests where I create 2 extra tasks (Workers) and then those tasks are blocked and unlocked using semaphores and barriers. However, whenever main task (Runner) is about to get blocked, the test crashes (just finishes instantly). In some configurations is seems to "deadlock" on a semaphore instead, while it should be able to move forward. I realised that if I suspend the Worker tasks before they are about to finish, with for e.g: `rtems_event_receive( RTEMS_ALL_EVENTS, RTEMS_DEFAULT_OPTIONS, 0, &events );` line at the very end of worker execution, it all works as expected. Do you maybe know what is the reason for such behaviour? I'm also not sure in which state is the task after it finishes it's execution but before it is deleted, is it 'blocked' or is it 'dormant'? Thanks and all the best, Jerzy ___ users mailing list users@rtems.org http://lists.rtems.org/mailman/listinfo/users
Re: Split BSP in RTEMS 4.11 - autotools newbie question
On Mon, Mar 14, 2022 at 7:37 AM Dariusz.sabala wrote: > Hello > > I am developing SAMRH71 BSP in RTEMS 4.11, I started editing atsam BSP and > accomplished some success, but the decision was made to support both BSP > (atsam and samrh71) in separate folders. I reverted my changes, copied and > renamed atsam folder to samrh71 and tried to build this renamed BSP, but > unfortunately nothing is being built - make just enters the folder and > leaves it with no artifact build > > make[1]: Entering directory > '/home/dsa/projects/p1/repo/rtems/samrh71/arm-rtems4.11/c' > Making all in . > make[2]: Entering directory > '/home/dsa/projects/p1/repo/rtems/samrh71/arm-rtems4.11/c' > make[2]: Nothing to be done for 'all-am'. > > Can you give me advice on what and where to change in the buildsystem to > make it build? Honestly I am cmake kid :) and knowing that autotools is no > longer supported in the newest RTEMS does not encourage learning this > powerful but complex tool. > Assuming there aren't any mistakes lurking. :) >From the top directory: bootstrap -c bootstrap -p bootstrap When you add, move, and delete directories, it is possible that the generated fragments which include lists of BSPs could get out of sync. Overall, I'd recommend you move to 6. --joel > > Best regards > dsa > ___ > users mailing list > users@rtems.org > http://lists.rtems.org/mailman/listinfo/users > ___ users mailing list users@rtems.org http://lists.rtems.org/mailman/listinfo/users
Re: Suspending Tasks before they are finished
On Mon, Mar 14, 2022 at 6:44 AM Jerzy J wrote: > Hi, > > I've been developing some tests and observed some task behaviour that i > can't understand and couldn't find any documentation about it. > > I have tests where I create 2 extra tasks (Workers) and then those tasks > are blocked and unlocked using semaphores and barriers. However, whenever > main task (Runner) is about to get blocked, the test crashes (just finishes > instantly). In some configurations is seems to "deadlock" on a semaphore > instead, while it should be able to move forward. > > I realised that if I suspend the Worker tasks before they are about to > finish, with for e.g: > `rtems_event_receive( RTEMS_ALL_EVENTS, RTEMS_DEFAULT_OPTIONS, 0, &events > );` > line at the very end of worker execution, it all works as expected. > If these are Classic API tasks (e.g. rtems_task_create) and they are falling off the bottom of the function that is the task body, that's a fatal error. If it is a POSIX thread, it silently goes away. > > Do you maybe know what is the reason for such behaviour? I'm also not sure > in which state is the task after it finishes it's execution but before it > is deleted, is it 'blocked' or is it 'dormant'? > Depends on the API and attributes on a pthread. --joel > > Thanks and all the best, > Jerzy > ___ > users mailing list > users@rtems.org > http://lists.rtems.org/mailman/listinfo/users > ___ users mailing list users@rtems.org http://lists.rtems.org/mailman/listinfo/users
Re: Suspending Tasks before they are finished
Hi Joel, Thanks for the explanation! It was in the context of Classic API. All the best, Jerzy pon., 14 mar 2022 o 13:12 Joel Sherrill napisał(a): > > > On Mon, Mar 14, 2022 at 6:44 AM Jerzy J wrote: > >> Hi, >> >> I've been developing some tests and observed some task behaviour that i >> can't understand and couldn't find any documentation about it. >> >> I have tests where I create 2 extra tasks (Workers) and then those tasks >> are blocked and unlocked using semaphores and barriers. However, whenever >> main task (Runner) is about to get blocked, the test crashes (just >> finishes >> instantly). In some configurations is seems to "deadlock" on a semaphore >> instead, while it should be able to move forward. >> >> I realised that if I suspend the Worker tasks before they are about to >> finish, with for e.g: >> `rtems_event_receive( RTEMS_ALL_EVENTS, RTEMS_DEFAULT_OPTIONS, 0, &events >> );` >> line at the very end of worker execution, it all works as expected. >> > > If these are Classic API tasks (e.g. rtems_task_create) and they are > falling off > the bottom of the function that is the task body, that's a fatal error. If > it is a POSIX > thread, it silently goes away. > >> >> Do you maybe know what is the reason for such behaviour? I'm also not sure >> in which state is the task after it finishes it's execution but before it >> is deleted, is it 'blocked' or is it 'dormant'? >> > > Depends on the API and attributes on a pthread. > > --joel > >> >> Thanks and all the best, >> Jerzy >> ___ >> users mailing list >> users@rtems.org >> http://lists.rtems.org/mailman/listinfo/users >> > ___ users mailing list users@rtems.org http://lists.rtems.org/mailman/listinfo/users
Re: Split BSP in RTEMS 4.11 - autotools newbie question
On 14/03/2022 15:12, dsa93 wrote: I tried to change it to "rtems-c-src-lib-libbsp-arm-samrh71" (same as bsp root folder), but doing this kind of autotools voodo didnt helped. The result is still the same: Make does not see any target inside BSP folder. I guess I will continue work on two GIT branches and maybe colleagues will fix it some day :) I would also recommend you try it with RTEMS 6 which uses a new build system. -- embedded brains GmbH Herr Sebastian HUBER Dornierstr. 4 82178 Puchheim Germany email: sebastian.hu...@embedded-brains.de phone: +49-89-18 94 741 - 16 fax: +49-89-18 94 741 - 08 Registergericht: Amtsgericht München Registernummer: HRB 157899 Vertretungsberechtigte Geschäftsführer: Peter Rasmussen, Thomas Dörfler Unsere Datenschutzerklärung finden Sie hier: https://embedded-brains.de/datenschutzerklaerung/ ___ users mailing list users@rtems.org http://lists.rtems.org/mailman/listinfo/users
