Thanks for your help and prompt reply Michel, I'm just trying to compact the BSP on our device, although our device is already headless and has already quite a basic config (standard shell is not present for instance). Regarding the fix, I simply ported the code present in the last WEC2013 QFE hotfix back to WEC7 (the stuff contained in 181031_KB4466833 vs the default WEC2013 state) - the rest is almost the same (well not fully but the functionality seems not to be affected).
Anyway, as I'm testing I'd get your app to break often by just going a bit over the default stall timeout (set to 1ms):
...
High priority reader thread 0 stall of 100 us took maximum 848 us.
High priority reader thread 0 stall of 100 us took maximum 891 us.
High priority reader thread 0 stall of 100 us took maximum 928 us.
High priority reader thread 0 stall of 100 us took maximum 968 us.
High priority reader thread 0 stall of 100 us took maximum 981 us.
High priority reader thread 0 stall of 100 us took maximum 982 us.
High priority reader thread 0 stall of 100 us took maximum 1036 us.
which makes me believe that the stall timeout is set a bit low (at least it seems to be like this for our default system). As I said I'll compact it further, but the current CPU usage when used stock is not going much over 1.5% when idling. Aside from compacting I'll increase the stall timeout to 1.5ms and see how that works, might be that someone else finds it useful as well.
Thanks for your help and
Thanks for your help and prompt reply Michel, I'm just trying to compact the BSP on our device, although our device is already headless and has already quite a basic config (standard shell is not present for instance). Regarding the fix, I simply ported the code present in the last WEC2013 QFE hotfix back to WEC7 (the stuff contained in 181031_KB4466833 vs the default WEC2013 state) - the rest is almost the same (well not fully but the functionality seems not to be affected).
Anyway, as I'm testing I'd get your app to break often by just going a bit over the default stall timeout (set to 1ms):
...
High priority reader thread 0 stall of 100 us took maximum 848 us.
High priority reader thread 0 stall of 100 us took maximum 891 us.
High priority reader thread 0 stall of 100 us took maximum 928 us.
High priority reader thread 0 stall of 100 us took maximum 968 us.
High priority reader thread 0 stall of 100 us took maximum 981 us.
High priority reader thread 0 stall of 100 us took maximum 982 us.
High priority reader thread 0 stall of 100 us took maximum 1036 us.
which makes me believe that the stall timeout is set a bit low (at least it seems to be like this for our default system). As I said I'll compact it further, but the current CPU usage when used stock is not going much over 1.5% when idling. Aside from compacting I'll increase the stall timeout to 1.5ms and see how that works, might be that someone else finds it useful as well.