On Thu, Nov 07, 2002 at 12:23:50PM -0800, David Brownell wrote:
> >>As well as updating some internal docs so they're current,
> >>this patch adds the physical path to the data available
> >>through hotplugging.  This feature was requested at the
> >>LinuxKongress hotplug BOF, and was also on my 2.5 list.
> >
> >
> >And is already present with the DEVPATH environment variable, right?
> 
> No, DEVPATH isn't as stable as the physical path. And will
> not be available on 2.4 ... which will be important for a
> few years yet, particularly for layers above the kernel.

Ok, but this was a patch for 2.5, right?  We have DEVPATH, and don't
need to have USB_PATH for 2.5.

> Plus, you were at pains to explain to me how DEVPATH should
> not be perceived as a physical path.  They don't even try to
> address the same problems.

Sure, but what is the problem?  It's as stable as what you are trying to
use, right?

> Here's part of what I got from hotplugging one keyboard:
> 
>   DEVICE=/proc/bus/usb/001/006
>   ACTION=add
>   TYPE=9/0/0
>   USB_PATH=usb-02:00.0-1.4
>   DEVPATH=root/pci0/00:03.0/02:00.0/usb1/1-1/1-1.4/1-1.4:0
>   PRODUCT=58f/9213/101
> 
> At best, DEVPATH needs a _lot_ of mangling to turn into
> the physical ID.  And using long IDs is always error prone.

But the last file in DEVPATH is equal to what you wanted to use in
USB_PATH, right?   And if you really wanted the pci id, it's there too.

> At worst, remember that identifiers like "pci0" and "usb1"
> change a lot more readily than "02:00.0".  (Even on cardbus
> and similar hotplug pci, since most people have control over
> which slots they use.)  Using DEVPATH would ensure that
> many more "should-be-minor" changes in system config will
> need sysadmin and tool attention.  That's generally not-good,
> particularly when needs for routine tool use are so easy to
> avoid ... save tool work for the important stuff!

But I don't want to drag around two naming schemes for devices, let's
stick to the one the rest of the kernel is going to use, DEVPATH.

> If nothing else, think of it as a useful stable short
> identifier ... you were the one who disliked the very
> long driverfs paths, yes?  :)

Yes, but it's necessary :)

I do like your comment cleanups, if you want to send me that part of the
patch again.

thanks,

greg k-h


-------------------------------------------------------
This sf.net email is sponsored by: See the NEW Palm 
Tungsten T handheld. Power & Color in a compact size!
http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0001en
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to