Hi, Greg
When did I try to remove all procfs stuff from pci_hotplug_core.c,
I found I could only cut little codes off. So I suggest:
How about to get rid of all procfs stuff for PCI subsystem?
It could reduce about 700 lines codes from the kernel.
I think we could get all information from sysfs, right?
But it may break some user mode utilities.
What do you think about it?
Regards,
-Stan
--
Opinions expressed are those of the author and do not represent Intel
Corporation
On Tue, Jan 28, 2003 at 05:59:28PM +0800, Stanley Wang wrote:
> Hi, Greg
> When did I try to remove all procfs stuff from pci_hotplug_core.c,
> I found I could only cut little codes off.
But you cut out everything that was there, right? There wasn't much.
> So I suggest:
> How about to get rid of all procfs stuff for PCI subsystem?
> It could reduce about 700 lines codes from the kernel.
> I think we could get all information from sysfs, right?
> But it may break some user mode utilities.
If you look, a lot of it is now under a config option to just not enable
it at all, which helps out a lot.
thanks,
greg k-h
On Tue, 28 Jan 2003, Greg KH wrote:
> On Tue, Jan 28, 2003 at 05:59:28PM +0800, Stanley Wang wrote:
> > Hi, Greg
> > When did I try to remove all procfs stuff from pci_hotplug_core.c,
> > I found I could only cut little codes off.
>
> But you cut out everything that was there, right? There wasn't much.
>
> > So I suggest:
> > How about to get rid of all procfs stuff for PCI subsystem?
> > It could reduce about 700 lines codes from the kernel.
> > I think we could get all information from sysfs, right?
> > But it may break some user mode utilities.
>
> If you look, a lot of it is now under a config option to just not enable
> it at all, which helps out a lot.
Is there a plan to update pci-utils to work with sysfs? lspci is a pretty
valuable debugging tool, it would be a shame to lose the use of it in 2.6.
Scott
--
Scott Murray
SOMA Networks, Inc.
Toronto, Ontario
e-mail: [email protected]
On Tue, Jan 28, 2003 at 05:21:37PM -0500, Scott Murray wrote:
>
> Is there a plan to update pci-utils to work with sysfs? lspci is a pretty
> valuable debugging tool, it would be a shame to lose the use of it in 2.6.
I'm pretty sure it works without the /proc pci stuff already (well the
big tables in there, I think it still needs the individual pci device
entries.)
And yes, sysfs support would be nice, hint, hint, hint... :)
thanks,
greg k-h