Profile cover photo
Profile photo
Bernhard Urban
Bernhard's posts

Post has attachment

Post has attachment

Post has shared content
"Windows tablet devices based on the established x86 platform. What could possibly go wrong? "via

Post has shared content
This is pretty cool.  I especially like the dynamic phone notification and the interactive graphs!

Post has shared content
Sometimes I'm surprised when people are surprised, as in this:

OK, to recap, Linux did something with a buggy EFI that bricked a Samsung laptop. It was, to repeat, an EFI bug (or, more properly, "a buggy implementation of an EFI function" -- customers with bricks in their hands don't actually care about the distinction that much). The fix was simple: don't do that. But that did not mean the bug went away -- it just meant that Linux people decided not to trigger it. 

So why would anyone be surprised that, when you write code to trigger the bug and run it under Windows, the bug is triggered? I don't know. The bug doesn't magically go away. It's still there.

The bigger surprise: many in our community have been saying for 10+ years that EFI has a large attack surface. "EFI attack surface" --> 30K hits on Google. Now, as the EFI laptops appear, people do trivial things with EFI and it has troubles -- why the surprise? Quote, "EFI has a large attack surface" -- there's lots of places to poke at, and this story is going to go on for a long time. The bad guys are going to have some fun. EFI is a large, closed, complicated OS on the level of complexity of Linux 2.0. What would lead anyone to think there'd be no bugs? Did I mention that EFI has a large attack surface?

Back in the day, Intel told us they could not ship coreboot as they needed to sell software they could support a warranty on. Well, I guess we're going to test the "warrantty" part of that statement. Somebody's got a warranty problem, I'll give you that. Not sure who. I'm sure there are lawyers engaged as I write this.

But, the most interesting bit about all this is that Samsung ships an x86 laptop running coreboot and x86 laptops running EFI. No reports of bricked coreboot laptops yet ... we're going to get an interesting case study of which firmware has fewer troubles. This is going to get interesting. I'm hoping the open source one will win. No guarantees, as Intel is full of very smart people, but still ... I'd like the open source one to win.

Post has shared content
I'm still annoyed about this "Linux bricks laptops" thing.
Consider this:
Take a coreboot-based chromebook, e.g. samsung. To brick it you have to
- rip off four footpads
- remove 8 screws
- crack the case open
- find a jumper
- close it
- reflash with a hardware device or flashrom

It's a not insignificant amount of work; and if you're really out to brick it you can do it just as easily with an EMP gun. Done. 

To brick the EFI-based laptop, you have to:
- boot a USB stick, with an OS that is following all the rules for EFI. 

There. Done. Problem solved. Bricked.

And they have the nerve to blame Linux? Unbelievable!

Expect more bricks. This is not the last time this is going to happen. And each time it will be blamed on Linux for not anticipating all the bugs in EFI.

Post has shared content
git tip of the day, thanks to Colin Walters:
  git log -S <STRING>

Search the history for a string in the diff itself, showing the commits involved.  I've needed that so many times in the past it's not funny, given that api changes are sometimes not always mentioned in the changelog commit text itself.

Post has shared content
Meanwhile in a Linux community

Heute das schoene Wetter genutzt und meine erste Ingress Tour bestritten. In Krems gibt es leider nicht viele Portale (drei), deswegen zum groessten Teil nur Bilder fuer potentielle Portale gemacht ;-)
19 portal submissions an Google geschickt. Mal schauen wann und ob die aktiviert werden :-)
#ingress #dataminingforgoogle

Post has shared content
Wait while more posts are being loaded