On Fri, 2006-03-31 at 16:35 +0200, Jurgen Kramer wrote:
On Fri, 2006-03-31 at 17:09 +0300, Gilboa Davara wrote:
> On Fri, 2006-03-31 at 12:56 +0200, Jurgen Kramer wrote:
> > On my Intel system (Xeon with EMT64 support) which runs the 32-bit
> > version of FC5 I can see that NX support is active (from dmesg):
> >
> > NX (Execute Disable) protection: active
> >
> > On my other system, a AMD64 system running the 64-bit version of FC4,
> > there is no message regarding NX support being enabled of disabled at
> > all.
> >
> > Checking init.c in the kernel sources it seems that NX support is
> > depended on PAE support and NX support from the processor. As the AMD
> > processor supports both I aspect to see a NX procection active message
> > on my AMD system as well. Does NX support suppose to work on 64-bit
> > kernels ?
> >
> > Both systems run kernel 2.6.16.
> >
> > Jurgen
> >
>
> AFAICS, It's activated quietly if present on iAMD64 systems.
> No printk code.
>
> So, if /proc/cpuinfo includes nx, you're in the clear.
>
Yep, nx is there. Odd that it is only displayed on Intel systems.
Actually the message will only get displayed on i386 kernels, not on
x86_64 kernels (running on both AMD and Intel platforms)
I'd assume that the reasoning for dropping the printk's on x86_64 is
simple:
NX on i386 is pretty much limited to rather recent machines; all/most
pure i386 CPU core do not support it. OTOH, most x86_64 CPU core support
NX out of the box.
To cut a long story short... if you're on x86_64, you most likely have
NX. No need to say/display/printk anything about it.
Thanks.
Jurgen
Gilboa