Messages in this thread | | | Subject | Re: /proc/pci design idea | Date | Wed, 17 Sep 1997 08:41:48 +0100 | From | David Howells <> |
| |
Martin `MJ' Mares wrote: > Well, doing this via /proc is much slower. And how do you want to separate > byte, word and longword accesses?
I'd have thought that if you want speed, then doing it through any sort of system call interface is going to be too slow, be it a purpose made system call, or be it a write() on a special /proc file.
By the way... I'm currently adding such support to my Configuration Manager and can now read arbitrary ranges of vendor registers from PCI or PnP devices through the usual mechanism. I'm also intending to add a facility to do the be able to set arbitrary ranges of registers.
Furthermore, I'm looking at making it possible to have the configuration 'file' open multiple times, thus meaning someone can open it and keep it open for as long as they wish to use it. I'm also thinking of reinstating the old ioctl() interface to it.
Watch this space...
David Howells
| |