Quote:
Originally Posted by Ignitionnet
How do you propose software pushes hardware too hard during a reboot? If a router can't handle hitting 100% utilisation for a while without having a hardware failure that's a manufacturing issue.
|
Or a design issue :P Quite a few laptops these days aren't able to run at full load continuously, often having to throttle down after a few seconds to minutes - though that's slightly different as it causes no damange.
Quote:
Originally Posted by Kushan
Most devices don't spin up to 100% at reboot, they tend to stay quite underclocked until they're finished initialising, then they ramp up.
|
I agree with Ignition here - hardware that can't run at 100% without damaging itself is faulty.
Furthermore, in practice, it actually works the opposite of how you describe at least on x86 and MIPS systems (the latter being common for routers). Most devices actually operate at 100% without being capable of throttling down or underclocking until the higher level OS is loaded. Without the OS, drivers, and CPU governors loaded, by default almost all devices will run at 100% clockspeed. Further, if something crashes during startup, it usually gets stuck at 100% forever as well.
Quote:
There are a very few cases where software can harm hardware, however these usually need the software to go out of its way to damage the hardware or the hardware to have some quite provocative switches available to control it. The examples that come to mind for me are Stuxnet, though that worked on PLCs operating industrial processes so quite different from a cable modem router, and software that could kill Android devices by messing around with voltage regulation. I wouldn't have thought the router firmware would have any control at all over low level functions capable of harming hardware, that'd be looked after by whatever is in the motherboard, individual cards, and SoC's ROM/PLA? If the required buttons and switches are exposed to the router firmware that would be an 'interesting' decision.
|
I haven't investigated the SH2 but a lot of routers do in fact expose voltage and CPU controls to the higher level OS, though often you need a patched or modified kernel to access them, as most stock kernels don't have the capability baked in. So yeah, as you say software has to deliberately go a long way out of it's way to cause damage. But it is
theoretically possible and some communities of people who deliberately overclock their routers this way. TBH
most consumer devices will have
some exposed register or I/O pin somewhere that can modify voltages and/or clockspeed arbitrarily but are usually undocumented or unavailable without special modified drivers in the end product.
It's not going to happen by accident though, unless by freak occurrence RAM corruption manages to hit in such a way the kernel governor accidentally adds a '0' on to the end of the clockspeed setting while sending it to the appropriate registers...