Hi Daniel,

I think it's the same reason you seen two IO_CONNECT operations - the 
first is the server resolve to procnto, the second is the actual operation.

In this case it's the same server twice, I guess, since it's like the 
pathmgr_link() embedded in resmgr_attach()

On 10-05-19 11:04 AM, Daniel Huenig wrote:
>
> We are working on reports generated from Kerneltraces and need more 
> information as provided in the docs (surprise, surprise). Recently we 
> don't know how to interpret the link() function calls in context of 
> device creation.
>
> In Kerneltraces the following pattern can be observed, when a 
> ressourcemanager provides a device:
>
> Event, Time, Owner, Type, Data, CPU, Cycle
> 1510, 52ms 644us, devc-ser8250-jacinto Thread 1, MsgSendvnc Enter, 
> coid 0x40000004 sparts 4 rparts 2 msg0 0x70100 msg1 0x2 msg2 0x100a18 
> function link(), 1, 0x006D28CB
>
> Event, Time, Owner, Type, Data, CPU, Cycle
> 1516, 52ms 756us, procnto-instr Thread 13, Path Manager, pid 5 tid 1 
> path_name /dev/ser1 process devc-ser8250-jacinto, 1, 0x006D334C
>
> Event, Time, Owner, Type, Data, CPU, Cycle
> 1524, 52ms 960us, devc-ser8250-jacinto Thread 1, MsgSendvnc Exit, 
> ret_val 0x800b0000 rmsg 0x70100, 1, 0x006D4671
>
>
> Event, Time, Owner, Type, Data, CPU, Cycle
> 1529, 53ms 35us, devc-ser8250-jacinto Thread 1, MsgSendvnc Enter, coid 
> 0x40000004 sparts 4 rparts 2 msg0 0x70100 msg1 0x2 msg2 0xf09d8 
> function link(), 1, 0x006D4D6D
>
>  NO PATH_MANAGER_EVENT !!!
>
> Event, Time, Owner, Type, Data, CPU, Cycle
> 1544, 53ms 324us, devc-ser8250-jacinto Thread 1, MsgSendvnc Exit, 
> ret_val 0x0 rmsg 0x70100, 1, 0x006D6896
>
> Why are there 2 successive send calls with function link() and what 
> does the return value mean ?
>
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post55161
>

-- 
cburgess@qnx.com