|
Posted by ctacke/ on November 9, 2007, 11:54 am
Please log in for more thread options
I never recommend INFINITE as it's a system block. I'd at least do a
timeout loop so the thread can be terminated if necessary.
--
Chris Tacke, eMVP
Join the Embedded Developer Community
http://community.opennetcf.com
> On Nov 9, 11:03 am, "<ctacke/>" <ctacke[at]opennetcf[dot]com> wrote:
>> I'm thinking you'd need a service that exposes a set of IOCTLs for
>> registration, deregistration, etc. When an app wants to register, it
>> would
>> call an IOCTL passing in its own process handle, and the service would
>> spawn
>> a thread specifically for that process. It could then use a wait on that
>> process handle, which will unblock when the process terminates. The apps
>> would *not* link to the service (as they really can't in any meaningful
>> way).
>>
>> To make it easier, I'd probably create some form of SDK that provides a
>> lib
>> that apps link to and call. That lib would extract the PID fromt he
>> running
>> process and in turn call the IOCTL behind the scenes.
>>
>> --
>>
> Exactly what I'm looking to do. That was the whole reference to the
> RIL implementation (RIL proxy is compared to my library clients link
> to, RIL driver is compared to my service DLL that processes the IOCTL
> calls from the proxy). The APIs will be something like Initialize,
> Deinitialize, GetThisItem, SetThisItem, etc
>
> So the library can get the calling PID from the Initialize call. Then
> it can just set some wait on that PID with WaitForSingleObject in a
> new thread in the service DLL. That would allow the main thread in
> service DLL to still continue to process, while still waiting for the
> client to terminate. Can I just set the wait time as INFINITE and I
> will be notified just as the client exits?
>
| 
XML Sitemap