>There is a gld_reset entry point in the driver, which will initialize the
>hardware and enable the interrupt.

I'm not sure where gld_reset() comes into play.

The iprbreset() function is called on panic/reboot and it's a undocumented
driver interface which is required for some devices to survive warm boot.

(Both Windows and Solaris hang on warm boot from Solaris when iprb did not
disable the receiver and interrupt before going down)

>Do you mean I should disable the interrupt in the gld_reset() and enable
>it in the gld_start()?

No.  You need to look at the drv_reset() entry point; not gld_reset.

(I'm not sure were and how gld_reset() is called but I expect it to be
more like iprb_reset() than iprbreset().)

Casper


Reply via email to