Examining a kobject hierarchy
Posted Nov 11, 2003 1:08 UTC (Tue) by mmarq
Parent article: Examining a kobject hierarchy
Just a ideia.
Unfortunately i cant tell by the source (i dont think so) if the I2O Linux kernel implementation follows the V2 spec defined at (http://www.intelligent-io.com/specs_resources/V2.0_spec.pdf ), but wouldn't it be grand if this, here exposed, 2.5 driver model could overlap, in the future with the I2O model at the communications level,..., that is, in a future Linux 3.0(?) spilt driver model, a "regular" driver module could "talk" with a I2O driver without any special translation layer or any other craft!
IMO whats missing in this 2.5 driver model is a communications mechanism between the various "subsystems" or " driver specific kobjects", because of multi-purposed hardware and combo peripherals...
So Why Not ?
A "special" subsystem that implements "at least" a "host MessengerInstance" type of I2O V2 made of kobjects, so that this messaging layer could be used by a kind of kobjects driver model and at the same time by the I2O model...
IMO this messaging layer rocks at the thecnical front, and also as it says in the draft spec paper:
" The architecture(MessengerInstance messaging layer) is independent of the operanting system, processor platform, and system I/O bus. This version of the specfication defines a transport interface between MessegerInstances for a shared memory environment, but does not preclude defining other transport enviroments in the future revisions."
A sure anti lock-in, and future prove model,... No wonder M$ that had backed it with Intel droped it in the recycle bin,... Could also mean a great help for stoping the lack of support from the hardware industry to the Linux project.
to post comments)