> The test of the PCI clock is not intended to prevent user from stupidly
> breaking their hardware, but some reality test to prevent the driver from
> being wrong about its estimation of the SCSI clock of the chip. Applying
> the algorithm to the PCI bus and checking the result against the expected
> reality (33Mhz), allows to make sure the algorithm is correct. This
> algorithm is based on the udelay() stuff of Linux that sometimes may give
> weird results, especially on new hardware, and then break the driver
> calculation of the SCSI clock.
In which case your chosen 37Mhz seems a very wide tolerance.
>
> I will NOT remove the reality test of the PCI clock, but allow user to
> tell the driver about the reality they choose for the PCI clock frequency.
> This will be a boot command option that will be defaulted to 33 MHz.
> Probably something like : pciclk=41 to inform the driver about the PCI BUS
> being clocked to 41 MHz.
OK - can you also provide some documentation for this in the error
message.
>
> I actually donnot care of people who overclock their hardware. I just ask
> them for not wasting our time with problems due to their decision to
> overclock. These people must know that hardware that have been overclocked
> can be definitely damaged and experience far more error than specs said.
I can certainly understand your frustration with people reporting
problems with drivers which subsequently turn out to be because
they've overclocked the system. I will not usually report any problem
as software unless it persists when I've reset the system to the
correct speed.
>
> Overcloking a CPU that is in fact sold with an underclocked specification
> is one thing. Overclocking a BUS and then everything that connect to it by
> more than 10% was not imaginable by me. By the way, I think that people
> that overclock PCI 33 Mhz buses up to 41MHz or more are just foolish and
> plain idiot, but I don't care of them.
You're entitled to your view and I'd agree if I had any expectations
of reliability for this machine - but it's my home machine and so I
can tolerate more problems than, say, my machine at work or a
server. Personally I think that not having ECC memory is more "stupid"
if you need reliability.
It's about probability anyway - machines are clocked so fast these
days that I don't think you can claim 100% absolute reliability even
at the "correct" speed - yes I'm eating into the safety margin but
I've hammered this machine fairly hard at 83 MHz FSB (which BTW is
still _underclocking_ the memory) without problems so I'm fairly
confident that 75Mhz FSB is within reasonably safe limits for the
expectations that I have of this hardware.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/