Re: 2.1.31 doesn't build on Intel with glibc

David Hinds (dhinds@hyper.stanford.edu)
Mon, 14 Apr 1997 16:13:09 -0700

Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Remco van de Meent: "[2.0.30] transparant proxy code broken?"
Previous message: yuri mironoff: "Linux on Apple PowerPC???"

Actually, the problem with the new <asm/io.h> stuff is at least a bit
more severe. User programs are welcome to use this file as well, as
far as I know. But when a user program, as a side effect, ends up
including <linux/vmalloc.h>, bad things happen.

I don't really understand what the ioremap stuff is doing in this
header anyway. Wouldn't this be better positioned in <linux/mm.h> or
some similar place? At a minimum, I think the ioremap stuff should be
enclosed in an #ifdef KERNEL ... #endif block (along with the #include
of vmalloc.h).

-- Dave Hinds

Next message: Remco van de Meent: "[2.0.30] transparant proxy code broken?"
Previous message: yuri mironoff: "Linux on Apple PowerPC???"