Showing posts with label driver. Show all posts
Showing posts with label driver. Show all posts

Monday, December 29, 2008

Fear of the Driver Disk

The problem of bloatware/crapware on retail PCs is well known -- to the extent that Apple makes fun of it, pointing out the absence of such software on new Macs, while PC tools exist just to clean it.

But bloatware has a less-famous, equally annoying sibling: all the garbage that brand-name hardware devices install off their driver or utility disk.

Pick up a peripheral -- printer, web cam, DVD drive -- from a major brand, and if you follow the automagical installer on the drive disk, you'll get a half-dozen apps that you may not need or like. In some cases, they're just bad apps (that have a habit of arranging to start at boot), while in other cases they can destabilize a perfectly-running system.

The problems are that

  1. in some cases you do need these apps, because some hardware features require "support software" to be present, and don't fully leverage the many built-in drivers and control panels available for free in Windows ...
  2. most hardware companies internally view the driver/utility software as an afterthought, writing it hastily, testing it inadequately, and staffing it with ... well ... whomever they can find.

There are two main remedies.

In many cases, getting an unbranded or "strange-branded" device is a smart idea (provided you know what you're getting). I've found these devices have straight-forward, minimalist support apps, make great use of built-in Windows drivers, and don't put any garbage on your system -- for the simple reason that they don't have the resources to write a bunch of half-baked apps, or to form "distribution partnerships" with people who do.

If I do have a brand-name product, I generally attempt to install it without its own driver disk, no matter what the instructions say. In many cases, the device is fully supported by Windows out of the box; in other cases, some features may not be available -- but I may not need them. (E.g., If I wanted to use my digital camera as a webcam, that would have required the vendor driver disk... but I have never wanted to use that feature of the device.)

And if that latter approach fails, it's pretty easy to uninstall the device or otherwise convince the PC it's never seen the device before -- so that you can go the RTFM route and use the supplied disk.

Sunday, April 27, 2008

Don't Feel Bad, It's Also Amateur Hour at Analog Devices

Analog Devices, whose D/A, A/D, and DSP circuitry power on-board sound in everything from ThinkPads to ASUS' main line of motherboards, makes a wicked bad driver suite.

Last summer, I was troubleshooting a situation where laptops were exhibiting lots of really short freezes at a close-to-the-metal level ... keystroke input being delayed, or mouse cursors getting jittery. Turned out to be DPC storms resulting from the Analog Devices SoundMAX driver and usermode controls.

Today I was struggling with the latest flavor of this driver set, which attempts to detect what you've plugged in to which port, and then auto-configure. Which it does wrong, forcing you to come up with tricks to get basic things like a headset and microphone working. Guys, we've had the "green" and "pink" sockets on sound cards for like 15 years now, give me some credit here.

But that's not what got me really annoyed. No, these guys have gotten themselves really confused about how multi-user logon and fast user switching works on XP. They've only had seven years to get that right. They pop up some of their controls and wizards in the wrong user session ... when I tried to close them and switch back to the account I needed to be on, their driver blue-screened my box.

As a developer, I install all kinds of stuff on my machines and I often torture them in unseemly ways. But, since 2001 when I started using Windows XP, this is the first time I've had a blue screen during a regular old user session, after a successful boot.

This is a WHQL certified driver too. For a long time there were problems where developers didn't get their drivers WHQL certified, which led to users clicking the infamous "continue anyway" button, and companies ignoring WHQL; I'd hate to suppose Microsoft solved that problem by letting junk like this pass the test.