This Little Chip

This tiny chip almost had me with­out my com­puter for at least a week.

It stores the BIOS, or Basic Input / Output System, in flash mem­ory on my moth­er­board. The very first thing that hap­pens when a com­puter is booted is the decom­press­ing of the BIOS into main mem­ory, which then ini­tial­izes the com­put­ers hard­ware com­po­nents, includ­ing crit­i­cal devices such as disk dri­ves and I/O ports. This allows a user to recieve feed­back (through video), input com­mands (through a mouse or key­board), and install or run oper­at­ing sys­tems (from a hard drive).

Without a BIOS, none of this would be pos­si­ble. In the past, moth­er­board man­u­fac­tur­ers have made it a has­sle to fool around with the pro­gram burned onto the small chip, because improper steps in the repro­gram­ming process could poten­tially ren­der the chip use­less. To update the BIOS, one would have to boot to DOS with a floppy and run a flash pro­gram off the disk. Modern moth­er­boards now offer the flex­i­bil­ity to update through spe­cial soft­ware in Windows, although this process is nowhere near as sta­ble as run­ning through DOS.

Which is some­thing I had to learn the hard way last night.

Recent ran­dom reboot­ings had given me rea­son to start run­ning the lat­est BIOS ver­sion. Unfortunately, I didn’t have a floppy drive (I opted not to buy one with my new com­puter because I haven’t used a floppy in years), which meant that I was stuck with the Windows flash soft­ware. The soft­ware hap­pened to crash at a very early part of the flash­ing process, which meant that I didn’t even have basic bootup code to get a drive run­ning. At next boot — noth­ing. No BIOS POST beep, no screen sig­nal, no response from the key­board. I quickly pur­chased a floppy drive at the near­est dealer, scram­bled to find a disk, put a boot sec­tor on it, but to no avail. There wasn’t even enough code burned onto the chip to get power to the floppy drive.

Normally, when some­thing like this hap­pens, such as the power going out or the floppy being removed dur­ing a flash, the BIOS gets cor­rupted and the chip is dead. The options are to get the moth­er­board RMA’d, which means send­ing the board back to the man­u­fac­turer before they send a new one back, or pur­chas­ing a new BIOS chip with a good BIOS image on it, which means spend­ing more money and wait­ing for a replace­ment. Both choices would take at least a week, if lucky.

Neither option was sat­is­fac­tory. I couldn’t wait until who-knows-how-long for some­thing to be sent back to me. Being with­out my com­puter is like being with­out my com­fort zone, the place where I can lis­ten to music and write, play games to get away, com­mu­ni­cate with the rest of the world, or even work on my busi­ness with Aaron when I feel so inclined. I looked around the net for a faster solu­tion, and dis­cov­ered some­thing called hot flashing.

Unfortunately, faster also means riskier. Hot flash­ing involves swap­ping two BIOS chips while the com­puter is run­ning. All that’s needed is a healthy chip, an iden­ti­cal moth­er­board (which I have at work), a boot disk with appro­pri­ate flash­ing software/image, and nat­u­rally, the cor­rupted chip. A com­puter is booted to floppy with a good BIOS chip, and after get­ting to a DOS prompt where a BIOS flash can be per­formed, the cor­rupted chip is swapped and re-flashed. As a per­son who’s already squea­mish about run­ning a com­puter with just a side panel miss­ing (in case water may hap­pen to splash into the case and cause a short, or a for­eign object falls in and jams a fan), this was an extremely daunt­ing process. Playing around with chips while a com­puter is hot means that there’s the risk of elec­tro­cu­tion, or short cir­cut that could per­ma­nently dam­age the other com­po­nents. Theoretically, after the BIOS is fin­ished run­ning, the board stops sup­ply­ing power to the chip since it’s no longer needed.

I decided to my faith in such a the­ory. Going on this faith meant that I could pry the chip out with a pair of mod­i­fied paper clips with­out hav­ing to worry too much about caus­ing a short (spe­cial PLCC-socket tongs are avail­able, but rare, and would prob­a­bly take just as long to arrive after pur­chase as get­ting a new board). After a few prac­tice pulls, which, I dis­cov­ered, loosens the socket and gets pro­gres­sively eas­ier, I seated the good chip with just enough pres­sure to make the con­nec­tions in the socket. After boot­ing suc­cess­fully, I pried the chip off the board and ran the flash.

The first attempt was unsuc­cess­ful, and after try­ing to boot with a cor­rupted BIOS, some­thing unex­plain­able hap­pened. The LED on the moth­er­board that shows that there’s a con­nected power sup­ply wouldn’t go out. I pulled the power plug, turned off the ATX switch, undid both the 24-pin EATX and 4-pin 12-volt con­nec­tors, and even pulled out the CMOS bat­tery, but the light refused to turn off. My only guess was that the capac­i­tors still had enough energy stored to keep the light on. After reset­ting the CMOS, and another hot flash attempt, the com­puter booted with the cor­rupted chip run­ning the lat­est BIOS. My Windows instal­la­tion was fucked (it wouldn’t even boot into safe mode), but after a recov­ery install, every­thing was up and run­ning again.

I was down for less than 24 hours.