>> I think I know why the TiPB dies. Looks like the kernel doesn't properly >> recognize the 7410 CPU as having an L2 cache, and so we don't properly >> flush it & invalidate it. (We don't catch the altivec neither). I'm >> working on a fix. > >Do you regard sleep support on Pismo and the Ti sufficiently stable so I >can remove all these sick hacks from the pmud package? I was preparing a >new release (right now I'm hunting for a static version of apt-get so my >upgrade to testing won't end in disaster).
Not yet. I also plan to add some new features to /dev/pmu API for pmud. On the core99 'books, you don't need to "poll" for enviroment changes (like the lid status). The PMU already sends them periodically to the driver. I plan to add a way for pmud to "ask" for those, they will then be transmitted as messages you can read() from /dev/pmu when you are not waiting for a command reply. The driver will also take care of doing the polling on older hardware to keep things consistent. (and eventually provide battery status via /proc or build an APM "emulation" in the kernel). it's not ready yet. You can already test for the new ioctl I added that tells you if the machine supports sleep or not. For now, it will return -ENOSYS if not supported, but I will change that asap so that it returns -ENODEV or whatever other result code. This way, you can differenciate between "sleep not supported bu new driver present" and "old driver" cases. But don't hurry too much. All of this is still a work in progress. I'll let you know when things are stable enough API-wise. Ben.
