* Matthias.Buehler@xxxxxxxxxxxxx | 2015-05-28 07:55:16 [+0000]: >Hi, Hi, >I am running two real-time processes on a 3.12.13.rt21. >The first process receives start commands from an external control >via network and relays these starts to the second process. This second >process does the actual job and tells the result back to the first. >This cycle is repeated several times a second. > >After some time (half an hour or so) the system stalls. There is no >network connection available. A parallel running GUI is also frozen. > >I tried this scenario with 3.12.40-rt55 and 3.18.13-rt9 with the same >result. > >Running the second process as SCHED_RR does not lead to this situation. What do you mean by "second process as SCHED_RR". What was it before? SCHED_FIFO? Can you tell me the what your realtime applications are and their priorities? Is any of the applications pinned to a CPU? >On the serial console I can see the following output: There is a RCU stall on CPU0. CPU1 looks idle all the time. >[ 4012.800040] >[ 4012.800040] runnable tasks: >[ 4012.800040] task PID tree-key switches prio exec-runtime sum-exec sum-sleep >[ 4012.800040] ---------------------------------------------------------------------------------------------------------- >[ 4012.825647] ksoftirqd/0 3 -3.000000 655617 98 -3.000000 8702.785784 0.000000 >[ 4012.836376] rcuc/0 15 -3.000000 278716 98 -3.000000 5810.323312 0.000000 >[ 4012.847109] systemd-journal 196 1513493.594716 2449 120 1513493.594716 273.461656 2623283.352271 >[ 4012.857844] VisionLineGui 1622 0.000000 2388301 39 0.000000 384277.128155 1916449.569945 >[ 4012.868572] VisionLineGui 1724 1513493.594716 102610 120 1513493.594716 1919.879611 2587477.132774 >[ 4012.879296] screen 3572 1513493.594716 45227 120 1513493.594716 1808.874962 769084.691362 >[ 4012.890018] RVisionLineDaemo 3637 0.000000 4339 39 0.000000 1405501.394654 0.000000 >[ 4012.900740] QThread 4224 0.000000 151421 59 0.000000 15053.995468 0.000000 VisionLineDaemo seems to run. VisionLineGui (1622) would also run if it could, it is likely next in line. Are those in two the tasks to which you refer that the stall can be avoided by setting a task SCHED_RR? A task traceback could show what VisionLineDaemo is doing (in case it stalls RCU and waits for something). > >Matthias Bühler Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
