On Mon, 2002-05-20 at 11:56, Borsenkow Andrej wrote: > В Пнд, 20.05.2002, в 15:00, Roger написал: > > It looks like the "/etc/rc.d/init.d/pcmcia stop" script is trying to > > rmmod the pcmcia_core/ds related modules, but it hangs because it hasn't > > unloaded the usb-ohci & ehci-hcd modules prior (which are in use by the > > pcmcia_core/ds modules). > > > > It hangs _in_ ehci-hcd if kernel output you have shown is correct. > > > > > To note (you reminded me of something), the module ehci-hcd is not being > > shown as being used by the usb-ohci modules. > > It is not used _by_ this module.
Correction. it is ;-) i even tested cd-rw burn rates to make sure. Burn rates for a without the ehci-hcd module being loaded was ~8x's. Burn rates with the ehci-hcd module loaded showed ~25x's The ehci-hcd will register itself as "not being used by any other module" and it will work transperantly. The following file has more info: /usr/src/linux/Documentation/usb/ehci.txt In brief, if loaded, it will take over usb packet transmissions, but if it isn't present or the bus doesn't support the higher transfer rates of the USB 2.0 spec, transmission then defaults back to the old usb-ohci/usb-uhci module (the USB 1.x spec) > > > mmmm...my bet too is that > > when the "/etc/rc.d/init.d/pcmcia stop" is executed, it's not realizing > > that ehci-hcd.o is being used by the usb-ohci module so it doesn't > > unload this, as such, everything hangs. I'll test my theory when i get > > my next chance to reboot (later this evening sometime). Ah Ha! my theory worked. the /etc/rc.d/init.d/pcmcia script is checking & unloading all drivers that are listed to be registered with the pcmcia_core/ds modules. (i believe the $PCIC variable within the script is responsable for unloading these modules.) Since ehci-hcd is not listed (by lsmod) as being registered with pcmcia_core/ds modules (or even being used for that fact), the pcmcia script simply ignores the ehci-hcd module. But, in reality, it is being used transperantly as i mentioned above. I'm thinking the reason the maintainer of the ehci-hcd choose "not to show the module as being registered" with any other usb module" as it may have caused a conflict with being able to fall-back to the older usb-ohci/usb-uhci driver. Just a complete guess here tho. Or, it's simply an issue that was over-looked. > > You may of course try to test every wild guess ... really, recompile > kernel, show me the output and be prepared to recompile a couple of > times more :-) (but second time it will be much faster). > > -andrej well, I've finished recompiling but i don't think i'll need to do that unless we want to recompile the ehci-hcd module to show itself as being registered and see what happens form there. > -- Roger ----- Verify my pgp/gnupg signature on my HomePage: http://www.alltel.net/~rogerx/about/index.html -- Roger ----- Verify my pgp/gnupg signature on my HomePage: http://www.alltel.net/~rogerx/about/index.html
signature.asc
Description: This is a digitally signed message part
