2.4: BOOTPC /proc info.
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 2 lines · Craig
Marcelo,
This patch is against 2.4.19-pre5, please apply.
This is unbelievable ugly. Can't you just save the packet as a binary buffer, output it as binary in /proc and parse and format it in user space ?
Better would be to not use bootpc at all in kernel, but run it in initrd (that is the long term plan anyways, removing dhcp/bootp completely and only supporting them from initrd)
-Andi
Quoted 8 lines · Andi Kleen
Quoted 2 lines · Craig
Marcelo,
This patch is against 2.4.19-pre5, please apply.This is unbelievable ugly. Can't you just save the packet as a binary buffer, output it as binary in /proc and parse and format it in user space ?
Sure, we *could* have done that. We chose not to because some of the user space programs were having problems during bootup times. Instead, we did it in the kernel for our specific application as that was the better place where we had more control (for *our* application).
Quoted 4 lines
Better would be to not use bootpc at all in kernel, but run it in initrd (that is the long term plan anyways, removing dhcp/bootp completely and only supporting them from initrd)
Yes, Alan mentions the same thing.
We didn't realize that was the long term plan. Is that documented anywhere, or was it discussed on this list eons ago and 'decided'? ;)
--
Craig.
+------------------------------------------------------+ http://www.wombat.ca Why? Why not. http://www.washington.edu/pine/ Pine @ the U of Wash. +-------------=*sent via Pine4.42*=--------------------+
Quoted 3 lines · Craig
Yes, Alan mentions the same thing.
We didn't realize that was the long term plan. Is that documented anywhere, or was it discussed on this list eons ago and 'decided'? ;)
It has been discussed on this list several times. In fact every time this has come up. There are a lot of policy decisions the in kernel dhcp/bootpc code that are generally best left to user space.
Al Viro has gone so far as proposed moving a lot of other parts into user space as well. But initramfs needs to be in the kernel for that.
Eric