I always run badblocks on new drives when I get them, out of paranoia that there’s a chunk of bad disk just waiting to eat my data. Now that drives are starting to get into ludicrous sizes, the amount of time it takes to do this is going up dramatically. I just bought two 1TB drives. It turned out to be slightly cheaper than buying one 2TB drive, but had the other advantage that I could run two badblocks instances in parallel.

I invoke badblocks via mkfs’s -cc command line, which does a destructive read/write of patterns 0xaa, 0×55, 0xff, and 0×00 one 4k sector at a time. So how long does it take ?

Two days, and twenty minutes.

I’m extremely fortunate in that my job exposes me to a lot of shiny new hardware. A lot of it before it’s even released to the general public. Every so often though, I get something shiny and new that’s actual production hardware. This sounds like a geeks wet dream. Free shiny new toys? Who wouldn’t want that? But It’s not always as great as it sounds.

Take for example, the current laptop I’m using. A Sony Vaio Z540. They don’t get much shinier. It’s incredibly light, and it looks stunning. A huge step up from my previous laptop I lugged around when travelling (A giant HP ‘desktop replacement’ monster with I kid you not.. two internal hard disks).

Unfortunately, whilst it’s shiny on the surface, it quickly became apparent after using it for a while, that Sony hates you.
(Well, me in this case, but you get the idea).

When I first got the thing, the number of things that didn’t work in Linux were numerous.

• It has one of those lovely switchable graphics switches to toggle between Intel graphics (Stamina) or Nvidia graphics (Speed). Whilst it’s nice to have a SUCCESS <--> FAIL switch on a laptop, it is utterly useless in Linux, because we don’t speak WMI or whatever magical incantations Sony have dreamed up to a) get notifications when the switch is toggled, and b) do something about it like switch chipsets. Oh, and X will probably shit its pants in such a scenario right now even if it did do something useful. Back to the story: When I first got this thing, X would freak out at startup. The paraphrased logs went something like

  • Oh, an Intel chipset. I know how to drive this.

  • oh wow, I found an nvidia chipset. I have no idea what to do with this.
  • Umm, what was that first chipset again? I forgot. Sorry. How about I just fail to start X ?

  This got fixed up pretty quickly thankfully. Because whilst I love me some tty action, it really was kind of miserable.

• Rebooting didn’t work. Linux supports about a half dozen or so methods of rebooting. From triple faulting the CPU, to calling ACPI methods, to other strange actions that have traditionally caused a computer to reboot. None of them worked on the Vaio. This magically started working recently, I’m still not entirely sure what fixed it.
• Fun with sound. You’d expect that when you plug in some headphones and start grooving to your tunes that you’d stop annoying everyone else in the room. But no! Sony decided to make disabling of the internal speakers a software driven thing, so now Alsa needs to poke magical bits somewhere when it detects you plugged in headphones. Except it doesn’t. So we still fail. I keep meaning to get around to poking at this. It’s probably something trivial like yet another quirk that needs adding to the hda-intel driver.
• Virtualisation.
  Remember how I said above that Sony hates you? You have a shiny Core 2 Duo P8600, which has VMX. Sony opt to disable it in the BIOS, and not even give you an option to turn it back on. There are some truly heroic efforts to reenable it on some other models of the Vaio, but they involve all kinds of madness that wouldn’t exist if Sony weren’t being complete dicks.

• Insyde BIOS.
  This deserves it’s own entry because it’s the most shockingly godawful BIOS known to man. I never thought I’d say I missed an Award BIOS.

• The keyboard.
  This is my number one reason right now for advising people not to buy one of these. All the previous problems are at least fixable (or work-around-able). When I first got it, it was actually remarkably nice to type on. In fact all the reviews you’ll find online talk up how nice it is to type on. Indeed it was, when it was brand new. Unfortunately after two months, it seems I got the James Bond self-destructing laptop. If you imagine looking at a key side-on, it’s supposed to be horizontal, and remain horizontal as you push it down. My ctrl and alt keys now look more like \ and they don’t seem to be easily fixable. Attempts at removing them have been aborted when they feel like they’re just going to snap off. I don’t have a particularly aggressive typing action. I actually prefer ’softer’ keyboards to clunky horrors like the Model M (heresy I know, bite me). So I don’t think I’m a special case here. If I hammered on the keys I could perhaps forgive them for failing so soon. I’d love to know Sonys return rate for failed keyboards.

That’s just the more obvious pet peeves about this machine. Remarkably, the things that usually plague laptops (wireless, suspend/resume) actually ‘just worked’ in Fedora 10. I was actually really surprised by this.

Andrew Morton terrorised kernel developers for years with his vaio of doom. There’s just something about them that makes them a complete pain in the ass to deal with. At least one of the reasons is that Sony go out of their way to do things differently to everyone else, even when there’s no really good reason to.

So yeah, don’t buy a Vaio. You’ll be thankful.

I spent way too much time over the last few days chasing bugs which turned out to have nothing to do with Linux.
I bought a SATA controller which arrived just before the weekend. It seems there is a fundamental flaw with the Silicon Image 3114 chips. Or to be more precise, with the firmware on some of the boards using this chip.

This thread is a summary of all manner of problems with it, but in short, it corrupts data past a certain block number. This took a lot of tracking down. (And badblocks takes forever to run when in destructive mode).

There is mention in that thread that a firmware update fixes the problem. Unfortunately, the DOS based flasher program seems completely unable to even write to my card.

I guess I’ll only use this controller for smaller disks, unless someone comes up with a workaround.

This is curious. I’ve had a PCI port-80 card for quite a while, and it’s come in handy from time to time for debugging. What intrigues me about this one though is the addition of the USB & parallel port. Not seen that before. Suppose it saves you needing to have the case open to see the numeric readout.

Too bad it isn’t PCI express. The last few boxes I’ve had no longer have 32-bit PCI.

That site has all manner of bizarre junk for sale. It also seems to be the cheapest source I’ve found for things like SATA cables.