[PATCH] Longhaul - Broken again

From: RafaÅ Bilski
Date: Sun May 06 2007 - 16:25:02 EST


Not so long ago Longhaul was marked broken, because to protect system
from lockup it was clearing BMDMA master bit on each device. Later I came
with patches which make use of hardware support present in VIA north and
south bridges. Unfortunately after some time this support seems to be broken
for many cases. First report was for older model of Epia mainboard and seemed
to be only case. There is no explanation for it. Hardware is exactly the same
as for my M10000, with exception of one additional NIC. There is no way to
detect broken chipsets. All hardware revisions are exactly the same as on
perfectly working system.

Yesterday Longhaul was reported as broken for VIA Eden ESP 7000 CPU
(http://lkml.org/lkml/2007/5/4/115). Transition is successful, but lock up is
only question of short time. CPU has Nehemiah core and it is claiming that it
is supporting Longhaul MSR. And it is true. Info in Longhaul MSR is correct.

Yesterday Longhaul was reported to cause lockup. Looks like chipset is blocking
AGP DMA, internal PCI DMA and processor access to PCI bus, but it is granting
DMA request to additional PCI card (Hauppauge PVR150). Reported by Wander
Winkelhorst.

Yesterday Longhaul was reported to cause lockup after 2 or 3 hours of use
(http://lkml.org/lkml/2007/5/4/513). It may sound great on Windows(tm) based
systems, but for Linux it is unacceptable short time.

Probably there are many more systems on which Longhaul is causing trouble
directly, by lockup during transition, or triggering bug in other hardware.
Simply it wasn't reported as broken so far. Even if there are systems on
which Longhaul is perfectly OK (I'm writting from such system now) there is
no place for it in the kernel because there is no way to know how to setup
hardware to correct state (looks like it is more than one bit or register)
and there is no way to detect broken hardware (looks like there is more then
one revision).

Signed-off-by: Rafal Bilski <rafalbilski@xxxxxxxxxx>
---
arch/i386/kernel/cpu/cpufreq/Kconfig | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/i386/kernel/cpu/cpufreq/Kconfig b/arch/i386/kernel/cpu/cpufreq/Kconfig
--- a/arch/i386/kernel/cpu/cpufreq/Kconfig
+++ b/arch/i386/kernel/cpu/cpufreq/Kconfig
@@ -206,7 +206,7 @@ config X86_LONGRUN
config X86_LONGHAUL
tristate "VIA Cyrix III Longhaul"
select CPU_FREQ_TABLE
- depends on ACPI_PROCESSOR
+ depends on ACPI_PROCESSOR && BROKEN
help
This adds the CPUFreq driver for VIA Samuel/CyrixIII,
VIA Cyrix Samuel/C3, VIA Cyrix Ezra and VIA Cyrix Ezra-T
--



----------------------------------------------------------------------
Kasia Cichopek eksponuje biust
>>> http://link.interia.pl/f1a6f

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/