Messages in this thread | | | From | "Ulrich Windl" <> | Date | Fri, 29 Mar 1996 12:58:32 +0100 | Subject | Re: Linux-1.3.80 |
| |
On 29 Mar 96 at 11:24, Hans Lermen wrote:
> On Fri, 29 Mar 1996, Linus Torvalds wrote: > > > > It does indeed fix a latency problem, but only on alpha machines (which > > called that routine with interrupts disabled for simpler assembly code). > > I changed the x86 asm sources to do it that way too (see the setup.S part > > of the 80-patch) > > > > Linus > > > > Ok, I noticed the changes to setup.S: > > +++ linux/arch/i386/boot/setup.S Wed Mar 27 16:41:05 1996 > @@ -114,6 +114,11 @@ > mov dl,#0x81 > int 0x13 > > +! Reset the disk controller. > + mov ax,#0x0000 > + mov dl,#0x80 > + int 0x13 > + > > [ no other changes in setup.S of 1.3.80 ]
I guess it's just to make sure the controllers are initialized (reset) properly by their BIOS...
> > However, I thought they were intended to cancel any pending IO on the > disk subsystem _before_ booting the kernel. > How does it effect the latency problem _in_ the kernel ? > Are you suspecting IRQs from the pre-boot-phase to lock the interrupt > controler ? > ( ... just to understand what is happening ) > > Hans > <lermen@elserv.ffm.fgan.de> > > > >
| |