On Wed, 9 Aug 2000, Todd Vierling wrote:
> I will have to add some more Linuxisms to /proc soon, such as /proc/cpuinfo
> and /proc/meminfo.  Since procfs is rather low-use compared to other
> filesystems, I could rip out the COMPAT_LINUX comparisons, and just check
> (via strcmp()) the emulation's e_name to choose the appropriate behavior.  
> This would make the procfs LKM happy....

The problem I see with entries being only visible to "linux" processes is
that it makes debugging things harder. Imagine you ktrace a linux binary,
see it accesses /proc/.../exe. Then look with your "netbsd" shell, and get
a surprise. Followed by confusion.

I'd rather see these entries either there all the time, or switched on
conditionally (so that one can mount -t proc -o linux foo
/emul/linux/proc).


 - Hubert

-- 
NetBSD - because Unix isn't just #include <linux.h>, i386, ILP32, ELF, ...!