On 2016-08-13 11:34, Theo de Raadt wrote:
I have to admit, it's a bit shocking that VOLTAGE REGULATORS have to
be
exposed to the software in the first place.
Just imagine a bug in some OS or firmware causing the voltages to
spike
up and fry the hell out of a device.
I guess that's modern-day hardware design for you.
No, that is the arm ecosystem.
There are three choices:
1. A vendor-locked stable platform that has only one, ok maybe two,
ok maybe a couple of slow-moving shifts, over a 20 year sequence
of time, because the market is gigantic:
a. i386
b. macppc, for a while
c. vax, but then time moved on.
c. sparc64 but then time moved on, and uhm oracle and $$$.
2. A stable hardware platform where operating systems don't need to
change because the investment is big but market is narrow:
a. sparc
b. sgi
c. loongson
3. An unstable platform with a description / machine-dependent
handling
a. non-mainstream arm platforms
b. non-mainstrain mips platform
c. vendor-de-jour die-next-year
Pick a place you spend your money, and understand the consequences.
It really is that simple. If you don't like the shock, don't spend
the money.
Or spend the money, and understand the shock.
Below the covers it is all the same hardware.
The long-term vendor-locked platforms do a better job of investing in
abstracting it away (ACPI or OpenFirmware), because exposing something
too strictly causes them harm because they have to expose it forever.
Which might cause them to lose their position, and they didn't become
tier 1 by being sloppy.
You -- by buying an arm platform de-jour -- on the other hand are
doing a great job being a tier 3 customer. Not that it will ever
become a tier 2 or tier 3 platform, because you will eagerly buy the
next board of theirs when it hits the market... which is probably
before you finish reading this mail...
IBM Power architecture is in category 1?
