Here's someone claiming to have found the responsible function in a driver.
PLEASE NOTE: ALL NAMES HAVE BEEN CHOSEN FREELY BY THE PERSON WHO MADE THE SCREENSHOT! So there's no name "BrickCLoneDevices()", it's probably called UpdateEEPromChksum or something like that in the original code, because it looks like that's what it does.
Assuming that this disassembly/decompiled code indeed is genuine, the interesting thing is explained in the 2nd comment block: A genuine FTDI device seems to be designed such that a write only to the offset that stores the PID is ignored, hence for a genuine part this code will only update the word at offset 62, and that would be matching the functionality to just update the eeprom checksum.
For comparison, here's a random mainling-list post which includes a dump of the 232 eeprom. The VID/PID is stored at word 1 and 2 of the eeprom, something that could be a checksum is down at the word with offset 0x7f (word 0x3f = 63? There's probably a off-by-one here).
Neither write has any effect on a genuine FTDI chip because both writes are to even addresses. That's also why they write to word 62 even though the checksum is in word 63 - they can't modify the checksum directly because that'd affect genuine devices, so instead they modify word 62 so that the data has the same checksum as before their changes. The entire thing has no purpose other than bricking clones.
PLEASE NOTE: ALL NAMES HAVE BEEN CHOSEN FREELY BY THE PERSON WHO MADE THE SCREENSHOT! So there's no name "BrickCLoneDevices()", it's probably called UpdateEEPromChksum or something like that in the original code, because it looks like that's what it does.
http://www.eevblog.com/forum/reviews/ftdi-driver-kills-fake-...
Assuming that this disassembly/decompiled code indeed is genuine, the interesting thing is explained in the 2nd comment block: A genuine FTDI device seems to be designed such that a write only to the offset that stores the PID is ignored, hence for a genuine part this code will only update the word at offset 62, and that would be matching the functionality to just update the eeprom checksum.
For comparison, here's a random mainling-list post which includes a dump of the 232 eeprom. The VID/PID is stored at word 1 and 2 of the eeprom, something that could be a checksum is down at the word with offset 0x7f (word 0x3f = 63? There's probably a off-by-one here).
http://developer.intra2net.com/mailarchive/html/libftdi/2009...