From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bart Veer To: tadams@extremeeng.com Cc: ecos-discuss@sources.redhat.com Subject: Re: [ECOS] cyg_io_lookup () and handles Date: Thu, 14 Jun 2001 04:16:00 -0000 Message-id: <200106141116.f5EBGLJ27034@sheesh.cambridge.redhat.com> References: <001101c0f438$edbe9510$090110ac@BURN> X-SW-Source: 2001-06/msg00498.html >>>>> "Trenton" == Trenton D Adams writes: Trenton> Ok, this is fine. But does this mean I'm going to have to Trenton> go looking through the sources just to find what handles Trenton> I can use? After all, isn't a handle just an unsigned Trenton> integer identifier? Or is it just the "/dev/ser1" kind of Trenton> thing? In which case I could call mine Trenton> "/dev/mySuperWirelessDeviceThingy"! LOL If you are really worried, how about just running something like the following (untested): #include int main(int argc, char** argv) { cyg_devtab_entry_t *dev; for (dev = &(__DEVTAB__[0]); dev != &__DEVTAB_END__; dev++) { printf("Device %s\n", dev->name); } } The real point is to remember the typical kind of target system that eCos is intended for. There will only be a small number of devices in the system and you should have a pretty good idea what they are. Admittedly the documentation could be more explicit about this sort of thing, e.g. the documentation for a given target might have a section listing all of the devices that will be present plus details of relevant configuration options etc. Now in theory we could add complexity to the build system to check that all devtab entries are unique, or add code bloat by doing a run-time check, but in practice there are many more important things to work on. Bart Trenton> -----Original Message----- Trenton> From: jlarmour@cambridge.redhat.com Trenton> [ mailto:jlarmour@cambridge.redhat.com ] On Behalf Of Jonathan Larmour Trenton> Sent: Wednesday, June 13, 2001 12:24 PM Trenton> To: Trenton D. Adams Trenton> Cc: 'eCos' Trenton> Subject: Re: [ECOS] cyg_io_lookup () and handles Trenton> "Trenton D. Adams" wrote: >> >> I think this is a simple easy to answer question. >> When creating drivers, does the driver developer select their own >> handle for their device? If so, how does one pick a handle that >> doesn't conflict with any of the system device handles? Jifl> The developer is in control of the code they are using so Jifl> it's up to them to only have unique names in their sources Jifl> :-). Just name new ones unambiguously, preferably based Jifl> around what is distinctive about their driver. Jifl> There's no standard as such because the only way to make names truly Jifl> unique is to have a central registration authority, and that's just too Jifl> much overhead for something small like this :-).