Hacker News Reader

This is an important point. It used to be that there really wasn't a whole lot of damage you could do to a machine with software, sure there was the 'set the monitor horizontal refresh to zero and burn up the flyback transistor' but that was about it. These days however, in an age of tightly managing clock speeds, heat, voltages, etc all under BIOS/driver control to maximize battery life, it is possible to not only damage but to completely brick a machine if it has either broken or malicious software running in 'ring 0'. And that puts people like NewEgg in a tight spot, having already to deal pretty evil stuff [1].

Then if you combine that with there is no way to 'restore' a laptop to its original OS install if you've re-partitioned and overwritten the 'recovery partition' on the hard drive. Nobody bothers to send recovery CDs any more, that is $3.85 in plastic they can't (or won't) put into the box. So now if you install Linux you've made it effectively impossible for the vendor to even attempt to start from a 'known good OS' and determine suitability. It does kinda suck.

That said, I prefer Amazon's policy as well and as more folks move there NewEgg will either adapt or die.

[1] Evil doer activity (documented) - get some bad memory (generally for free from some scrap pile, 'buy' a laptop, take its good memory out, put dead memory in, 'return' the laptop. Sad really. All the engineering effort that goes into thwarting the petty schemes of evil people.

Am I the only person that images my manufacturer's HDD when I first get a new box? I thought that would be fairly standard operating procedure for any geek who wants to avoid this sort of crap.

Or pull the original HDD and replace it with a nice SSD. Should something happen, pop in the original HDD and away you go.

"away you go" sans all your data.

Away the computer goes to the RMA processor, sans all your data. This is bad how?

No you are not, like others I typically swap the drive with an SSD so I can return to 'factory' easily. That generally however involves breaking the 'factory seal' on the bottom which for NewEgg is one of their deal breakers.

As a data point, the last 3 laptops I've bought (Toshiba, Toshiba, ASUS) did not require breaking any 'factory seal' to swap out the hard drive.

This is why you buy Thinkpads. They've got a great warranty system, post the whole maintenance manual online, and don't void the whole warranty for single item changes (basically you change a hardware subsystem and you can't get warranty on that bit but the whole bit is still eligible)

I'd double check that next time you buy a new ThinkPad. I broke the screen on mine recently, and chose to replace it myself after being quoted an eyewatering (nearly cost of laptop) amount to have it repaired. The tech support person told me in no uncertain terms this would totally void my warranty.

Just worth checking.

If there are no warranty stickers, and you don't make any obvious changes or damage to the machine, you sometimes get away with 'breaking' the warranty. I've had this occur for a laptop that required opening to change the RAM (voiding warranty by all supporting documentation). I restored the original RAM before warranty, and they fixed it all the same.

Maybe it's just for minor changes, simple user upgrades instead of huge replacements like the screen? shrug I've never had to warranty my Thinkpad anyways thanks to them being tanks.

Clonezilla was the first word that popped into my head when I read the original post. Also helps resale value to put the 'factory' OS back on with the recovery partitions &c.

Hell, I do this twice when dealing with in-laws. The first time to make sure I can restore the OEM partition, and the second time so I can restore the thoroughly sanitized install when they filthy it up with malware and browser bars.

I am a Linux user for years now, and I have never even come close to wrecking anything with it with 'improper configs' or otherwise, on a laptop or desktop.

The problem you describe is not a matter of an 'improper config' wrecking a laptop due to the awful user-unfriendliness of Linux. It is a matter of you messing with something you didn't understand, jamming the machine in a bag to overheat, and then having to replace the battery.

There does not exist an operating system which is impervious to this kind of nonsense.

I recently discovered that bad software can really cause hardware to fail. I switched from using the proprietary AMD fglrx drivers to the open source radeon driver a few years back. For some time, my notebook has been running pretty hot after being powered on for a few hours, even when idle. I always put this down to a dust-blocked heat sink. When the problem persisted after cleaning out the fan and heatsink and replacing the thermal grease, I started to research other causes. Turns out that clock gating, a technique that clocks down the gpu when it is idle in order to save power and reduce heat output, is disabled by default in the open source radeon driver [1]. However, I discovered this too late - the constant overheating has already seems to have damaged the gpu to the point where random color artifacts appear on the screen and the system will freeze after being powered on for a few hours.

So as it turns out, Linux may really be bad for your hardware. Still, there's no way I'm going to stick with Windows on my new notebook.

[1] https://wiki.archlinux.org/index.php/ATI#Powersaving

Linux is not bad for your hardware, misconfiguration is bad for your hardware.

Annecdotally, heavy graphic card use on any laptop (yes even running windows) will cause a heat death in a shorter ammuont of time than normally expected.

No, in this case, Linux really is bad for your hardware. Reading gp's wiki link, it would seem the feature to clock down the GPU is disabled by default because it's still a work in progress. That would mean that although it works in a lot of cases, it's not yet truly dependable. And if it's not yet truly dependable, then it's not able to dependably care for the hardware in the best way possible. And as such, Linux can be bad for your hardware, though it isn't always.

Any rebuttal you might throw at this reality would make me to look at you more and more like this. :)

http://xkcd.com/644/

Per your claim, some driver for some piece of hardware has a disabled feature which you can use to mess up your hardware.

So why don't you just freaking avoid enabling that feature?

Is it Linux's responsibility to absolutely prevent you from shooting yourself in the foot after taking very special effort to research the best way to blow it off?

Question: does OS X fully support every possible bit of hardware which you can technically make to run with it, with every feature you want? Is it absolutely impossible to shoot yourself in the foot with OS X?

You misunderstood. The lack of the feature (of clocking down the GPU when it's not used) caused his issue. He did not enable it.

What you say is completely incorrect.

Linux distributions (keyword: distributions) have often shipped in a default configuration which can result in damage from overheating. There's no need to accuse the OP of "messing with something he doesn't understand."

Vendors often perform QA on hardware to ensure that it operates properly in conjunction with an operating system. When hardware is designed such that it requires particular operating system behavior to prevent damage then it is absolutely reasonable to require that operating system be used to maintain the warranty. Closing the lid of a laptop is completely normal behavior -- and it's a fact that many linux distributions ship with a configuration that will not properly suspend the device, leading to potential damage.

This is not a matter of "user unfriendliness." This is primarily a matter of hardware vendors limiting their testing to the behavior of certain operating systems (and thus limiting their warranty -- you can't warranty what you haven't tested; what isn't well-specified). I wouldn't be surprised if Apple refused to warranty a device which had Windows installed on it.

There's no need to get religious here. It's merely an issue of vendor support, and hardware which relies on particular OS behavior to operate safely.

Just to be clear, the issue has more to do with the vendor-supplied OS vs consumer-installed OS. My biggest overheating-when-should-be-sleeping issues have almost always been Windows on a Dell. But I have had it happen on a Cr-48 and MacBook as well.

The point he is making is Newegg shouldn't have to warranty a defect in a user-installed OS, not whether or not Linux is user friendly.

Do you understand the meaning of the word "anecdotal"?

I think almost everyone carries a laptop in a bag, and most don't have ventilation. The problem was that even though my linux configs were set (correctly) to sleep when closing the lid, it would often get prevented from sleeping by a program or driver that didn't respond properly to the sleep request. I did mention the case where Windows had a similar issue.

I was only trying to show that the OS and Software can play a big role in the health of the hardware. Software bugs can cause hardware failure, and a notebook not going to sleep when the lid is closed is usually a bug unless it is specifically configured to do that.

Wrecking things with a default build is still an option with binary blobs and sneaky variant hardware. Our dev team standardized on the x120e's a while back; the first one or two were great, but the second batch had a serious flaw in about half of them -- the fan speed reported to the kernel is inverted. (How you invert a PWM count, I've got no clue.)

Same distro, same config, and half the batch has a death wish, shutting down the fan in response to increased CPU temp.

>It is a matter of you messing with something you didn't understand, jamming the machine in a bag to overheat, and then having to replace the battery.

Wrong.

https://help.ubuntu.com/community/AsusZenbook#Suspend

>If this workaround not used, there have been reports that the laptop's memory controller setting may be screwed. After an incorrect suspend, if the corruption happens, many memory blocks starting with ~1G will be corrupt. Good way to see it is to use "memtest86+". The only way to fix controller setting is to open the case and plug off the battery. Please note that on this ultrabook the battery is not a user serviceable part and this could mean that by opening you can get your warranty void! If you start Windows or Linux with a corrupted controller, you will get system crashes or/and damaged file system. More info here: https://bugzilla.kernel.org/show_bug.cgi?id=42728 and https://bugs.launchpad.net/ubuntu/+source/linux/+bug/962798

She should contact Lenovo directly. I've fried my Lenovo T61 with overclocking and they still repaired it immediately. Additionally, they expressly stated to REMOVE THE HDD. This is obviously a cop out by Newegg, since any technician worth his salt could just run hardware diagnostics on it. According to the buyer, she did that AND booted a normal Windows and the errors persisted.

Didn't we just have an article about how great Amazons UX is? Well, i guess this is the Newegg version of that, just inverted.

For every warranty repair I've ever done with a manufacturer they have asked for the HDD to be removed before sending the unit in. As a retailer, or maybe even just a drop shipper in some cases, I just don't see NewEgg doing anything other than replacements.

I certainly wouldn't expect, nor trust, Amazon to do this.

Shouldn't it be the firmware's responsibility to prevent overcharging, overclocking, overheating, etc. - unless explicitly overridden?

Yes it should. Blaming the OS is like blaming a userspace program for a kernel panic.

In the brave new world of ACPI, the OS tells the firmware "I'll take care of that, trust me, I know what I'm doing". Your computer starts with fans at full speed for this reason, but once the OS assumes control, it's in control.

Something that also happens often is that Linux has problems sending the hard drive to sleep properly and you end up with a "I'll stop and start it 3 times a minute" scenario which is able to wreck a HDD pretty quickly compared to normal usage

A bit of a tangent but this is one of the top reasons I buy Apple laptops. Sleep/Wake is pretty much flawless. I occasionally had some hangs on wake back in Tiger or earlier versions of Mac OS, but never with Leopard or later.

Weird, because every now and again my Macbook will just die after waking up. I also find that while initially it was really fast and worked great, it's now really slow and takes up to 10 seconds, while my linux laptop (which doesn't even have an SSD) wakes up pretty much instantly

Is there anything suspicious in the system log? I'm guessing that you might have a kext (like for VMWare Fusion of VirtualBox) that's misbehaving when waking up.

My apps crash frequently after I wake up on Lion.

Both me and my father have had "sleep" incidents with our macbooks where the computer ran at full blast with the lid closed in our bags until they were too hot to touch. I'm on Lion now and I still occasionally have this problem. Apples are not immune, but I agree that they are the best at sleep and wake I've had so far.

Happens to me all the time. I also have often the problem with a black screen after sleep :/

Some of the newer PCs aren't too bad with this. My desktop sucks with sleep, but the two laptops I use are pretty flawless (though one's battery is terrible). I haven't shut one down in 3 weeks, and the other in a week. They're pretty standard, too; the good battery life one was around $600.

Haven't used a Macbook, though, so can't compare.

I'm also a long-time Linux user and while I have seen Linux destroy hardware (X11 was misconfigured and fried an LCD back in the day), it is certainly not common and pretty difficult to do these days. I don't think the risk is high enough to justify voiding the hardware warranties completely.

I've seen that happen with windows as well. Opened program wants to close, but needs to save or some such, laptop overheats. Sometimes it's just a lesson learned, I wouldn't say it was Linux that wrecked the laptops though.

I do find it interesting that newegg isn't using a CD loaded testing suite anyways, using windows PE or a DOS based environment. (like Eurosoft's PC-Check software.)

There's also the fact that it should take about, ten to fifteen minutes just to slip the hard drive out, hook it to an imaging workstation, and image the default OS over to it. (My imaging station takes 5 - 10 minutes). To the best of my understanding it's pretty much assumed that when you send stuff in you are waiving your right to have the same data sent back anyways.

I have had this happen when I was running a XP mode virtual machine in a Windows 7 laptop. The Virtual Machine prevented the main OS from going to sleep.

It was surprising my backpack didn't catch on fire.

Is it really that hard to engineer in hardware or bios-based failsafes? It seems like bad design to leave basic hardware safety issues to the OS.