I knew that I had to write three versions of the ACPI HAL, one for the
PIC, one for the UP-APIC case and one for the MP-APIC case. At that
time (it was early 1997) Intel was promising us that by the time NT 5
(as it was known then) shipped, there would be no new machines shipping
without full APIC support.
Thus it seemed that HALACPI was almost a throwaway. It would support a
few laptops and not much else. And laptops are performance constrained
by their I/O subsystems, not by their processors or their interrupt
controllers.
The timer code and the interrupt controller code in the HAL are pretty
tightly integrated. When I wrote HALACPI, I had to re-write the timer
code so that it would tolerate the processor changing its frequency on
the fly. Given that we thought that HALACPI would only run on a few
laptops, with even the next generation of laptops moving to APIC, I just
took the simplest 8259 PIC code that existed at the time for a starting
point, which was the Strict IRQL code from the old Microchannel HAL.
When Windows XP came around, Intel was again promising us that every
machine would be APIC-based in the very near future, which made
re-writing HALACPI seem like a waste of time.
Today, it’s mostly true that all first-tier desktop and server machines
support the APIC. There are still no laptops that support APIC.
Jake Oshins
Windows Kernel Group
This posting is provided “AS IS” with no warranties, and confers no
rights.
-----Original Message-----
Subject: RE: Device Interrupt priority - Reviewing Jose Flores
From: “Maxim S. Shatskih”
Date: Tue, 10 Dec 2002 16:46:10 +0300
X-Message-Number: 24
>does many fewer writes to the interrupt controller. At present, we
only
>do Lazy IRQL in hal.dll, which is the default single-processor
non-ACPI
>HAL. All the other HALs use Strict IRQL.
Sorry, but why Lazy IRQL was not used on UP ACPI HAL (HALACPI)?
Max