Linux Bricks Some UEFI Samsung Laptops

[locovaca]

Oh, sorry, I was fixated on the accidental corruption of firmware, like what is occurring with the samsung-laptop module. Of course a utility that is written to support various chipsets' firmware flashing functions can make paperweights, as can a motherboard vendor's bios flash utility if the process doesn't complete properly.

(I have heard of coreboot and knew there was a flashing program but didn't clue in on the name)

Thanks

The original blanket statement I replied to, not made by you, was:

If software can brick the system, then the system is faulty to begin with.

My point is that the ability to brick a system through software doesn't make it faulty. Anyone can take flashrom code and write a virus to brick systems. Nobody knows what is at fault here. Clearly the Samsung driver is doing things it's not supposed to - this has already been proven - and there is already a solution in clearing NVRAM.

Moral of the story: MMIO is a blessing and a curse.

[DeepDayze]

Are these systems truly recoverable? If not then that's a serious problem

[curaga]

To all the people saying flashrom: no, even that is bad hw design.

The correct thing is to have a good version in ROM, able to be used if a flash fails. Including with third-party utils such as flashrom.

[hristo.venev]

To all the people saying flashrom: no, even that is bad hw design.

The correct thing is to have a good version in ROM, able to be used if a flash fails. Including with third-party utils such as flashrom.

My computer's motherboard (ASUS P8Z77-V) can be unbricked by a USB flash drive with a BIOS image on it without even using the CPU. That's an example of good design.

[DeepDayze]

My computer's motherboard (ASUS P8Z77-V) can be unbricked by a USB flash drive with a BIOS image on it without even using the CPU. That's an example of good design.

Absolutely

Some older boards generally used a floppy image to recover the BIOS in the event of brickage by user setting a jumper then booting the machine from a specially prepared floppy disk that has a recovery image. In most cases you can't see what's going on on the screen as there's no video support within the recovery firmware code but system will provide beep alerts on progress and success/failure.

[tomato]

[...]a pool of lobotomized jellyfish with typewriters that are being fed crap for spewing something that sorta works as quickly as possible[...]

"lobotomized jellyfish" is kind of redundant...

Doesn't change the fact that it sounds cool and describes a typical code monkey even better than "code monkey"

[moilami]

Are you referring to Secure Boot? UEFI was around long before secure boot, there are many machines out there that support UEFI without supporting Secure Boot, and Secure Boot is an optional part of the spec (only "mandatory" if you want Windows 8 Logo certification - a Microsoft requirement, not a UEFI one).

My Linux server uses UEFI just fine. In fact, it makes things easier with GPT partitioning on 3TB hard drives.

Yeah, I was referring to that. Should be more accurate.

[garegin]

Yeah, I was referring to that. Should be more accurate.

the fact that freebsd's uefi support sucks donkey balls kinda discredits it as the bestest big iron OS as many people in the enterprise find efi important.

[GreatEmerald]

Are you referring to Secure Boot? UEFI was around long before secure boot, there are many machines out there that support UEFI without supporting Secure Boot, and Secure Boot is an optional part of the spec (only "mandatory" if you want Windows 8 Logo certification - a Microsoft requirement, not a UEFI one).

I was very happy to read that, since at last someone pointed this obvious fact out.

My Linux server uses UEFI just fine. In fact, it makes things easier with GPT partitioning on 3TB hard drives.

And then the post ends with this... Sigh. GPT has nothing to do with UEFI, other than its support being mandatory. GPT is not UEFI-specific, and, as a matter of fact, I'm using it in a BIOS machine right now.