Gregory Haerr

> I thought the driver is vgaplan4.c and the ega part of that can be rewritten Yes, that will work. vgaplan4.c is a sub-driver of scr_bios.c. > It uses Intel...

> We need to rethink this because open initializes the device and close (virtually) disconnects it Yes, I see. > So we're either back to counting the opens Ok, that...

> although that's quite dangerous for the OS stability, since you could (in theory) modify any part of the memory. Yes, I suppose so. I wonder what real use POKE...

> Probably that was the case, yes. It's been a long time since we begun working on the SD driver! I'll add this to my list of things to look...

Hello @Mellvik, > Enable loading of more than one ethernet driver in the kernel in order to be able to switch fast between them Rather than the way I initially...

Regarding PR #1338: > the first step would be to move most of the code from the init routine, and instead call it at open time; Only the `request_irq` code...

Hello @Mellvik, > I'm having reservations about the suggested initialization scheme though. I think I miscommunicated the issue with IRQs - they won't need to be shared between NICs. The...

> I don't think this is beneficial. If we conclude that the ability to assign/deassign the IRQ isn't useful, because a particular IRQ would only be used by one piece...

> I was pondering if expanding the eth struct would be good for placing the NIC params You mean netif_stat? Good idea, except the kernel init code will need three...

> I don't have the big picture at this time. The bigger picture is that instead of having each NIC driver register itself to be /dev/eth (at 9, 0), a...