Ultimate Boot CD

Hello everyone who uses CPUstress image in UBCD for stress testing.

I'm planning to build a next version of this image, with an updated Linux kernel and stress programs. For now, I face a small question about the power button behavior in it. Currently, pressing power button in CPUstress does nothing. It won't do software shutdown or hard shutdown either. I would like to fix it in the next version, but I'm not sure which behavior is better. So I would like to see people who does use these stress tools to give me some opinions.

Do you prefer software shutdown or hard shutdown when you press the power button in CPUstress?

Hard shutdown:
- Kernel is slightly smaller. No need to compile ACPI power button driver or acpid.
- Same behavior as in original, CPUstress image from Icecube.
- Con: I'm not sure if pressing power button during middle of mprime test is good for CPU... (There won't be hard disk damage, though.)

Software shutdown:
- Will need ACPI power button driver, acpid, and configuration.
- Will be same as typing poweroff in the shell.
- If a PC has a reset button, it will still be a hard reset, though.

Or, keep the current behavior (do nothing).

Opinions, please.

I am not a "heavy" user of CPUStress. I have only used it a handful of times helping friends burn-in their newly acquired desktops.

I actually prefer the current no-action behaviour. Reason is because it is very easy to accidentally press the power button, especially if you have kids in the house. I feel the reset button or the switch on the PSU are good enough options if something gets stuck.

That's just my opinion.

Victor Chew wrote:I actually prefer the current no-action behaviour. Reason is because it is very easy to accidentally press the power button, especially if you have kids in the house. I feel the reset button or the switch on the PSU are good enough options if something gets stuck.

That's just my opinion.

I actually thinks the current no-action behavior is a bug. It's because I have the power button driver compiled in the kernel, but don't have any ACPI event handling for it. (I need ACPI support simply because the multiprocessing of my laptop wouldn't work without it.)

For now I am thinking about a case when someone leaves his computer on stress-testing while he's away, then his mother wants the computer off so she presses the power button which does nothing. I guess after that she'll unplug the power cord directly.

I simply believe the power button have to do something; there's no reason to keep a PC on when someone presses it. Ideally most operating systems will do a soft shutdown, suspend, or "wait for N seconds for user to cancel before soft shutdown". It should be one of the three. I didn't think of any use case of kids accidentally pressing it that can matter. Kid can accidentally press Reset (which is always a hard reset) the same way, or accidentally unplug the power cord.

I'm waiting for more opinions before I make the decision here.

If your going to enable it, either way works for me but I do usually shut down after doing burn-in testing. Could you enable a popup that offered both options and a cancel? To be honest, I haven't used CPUstress in quite a while.

The Piney wrote:If your going to enable it, either way works for me but I do usually shut down after doing burn-in testing. Could you enable a popup that offered both options and a cancel? To be honest, I haven't used CPUstress in quite a while.

Providing options and a way for users to cancel will be an overkill. The user can stop any test at any time with a Ctrl+C, and it won't take her seconds to turn on the monitor, type 'poweroff' and press Enter. I would prefer simplicity here. Either soft poweroff, hard poweroff, or do nothing at all.

And I don't believe popping up a message during middle of a test is a good thing. Imagine when you run CPU Burn-in:
The message can get ignored easily.

Like I said, it's just me. I actually set the power button on my laptop to no-op. No choice. Three little monsters around the house.

But if you feel it absolutely has to do something, I think a proper power-down is the correct thing to do, since that's what a normal user would expect.

Just for your information, the FDUBCD doesn't support ACPI power button, and on my laptop, pressing the button in FDUBCD will go hard poweroff.

It seems that most BIOSes allows you to configure the power button behavior to either go suspend or instant-off (look at the screenshots I google'd). Too bad my laptop doesn't have this option. And interestingly there's no "do nothing" choice that I previously expect.