What I discovered after further investigation into
my own BSOD is this:
1. A backup restore will not revive the OS.
2. This was an Intel Turbo Memory (Robson NAND) issue. The TM was confused about where the cache is located. If you actually go into the TM Console installed on the HDD or OS volume that's unbootable, you'll find that it's a standalone executable and can be opened from inside any other OS. When I looked at the Console, it told me that the TM cache location was the HDD, not the TM. (TM does not have a 100GB cache!)
There may be underlying fault with the Intel ICH8M and ICH8M-E I/O controller hub architecture (southbridge chip), since many PCs I've looked up that have this problem use this ICH model. The design problems probably have to do with the IDE emulation mode (in BIOS, the SATA 'compatibility mode'), which Intel dropped in subsequent I/O Controller Hub models -- going with all AHCI / SATA.
3. In my case, this problem is with a Lenovo ThinkPad T61p. The T61p has a swappable HDD bay (the Ultrabay Slim) that takes a SATA connector, but only runs (Lenovo does not acknowledge this) in IDE mode. (See
this thread.) This is conjecture, but it may be that this is confusing the ICH. Lenovo has not come forward with any explanation of the sources of this BSOD.
4. It is, however, recognized that the Intel Matrix Storage Manager (iaNvStor.sys), which is a driver necessary for running the Turbo Memory, and which is a key component of the ICH8M and ICH8M-E causes a lot of different problems on the T61p.
This includes problems with freezes due to MSM interference with the 'drop guard' (the Active Protection System accelerometer). See
this thread. These freezes are especially prominent with hybrid HDDs, where Windows Vista has a low-power write-caching setting called Windows Hybrid Hard Disk Power Saving. When the PC gets bumped, the APS kicks in, and the WHHDPS wants to write everything to disc before the HDD is locked, and MSM can't handle both requests.
5. I have not heard a final update from Intel, but I will quote an Intel engineer who helped a little bit with pinpointing the issue:
The issue is with the system seeing the ReadyDrive Cache in Robson (Mobile) but it is not loading properly in the boot sequence. I have seen this issue when the module is moved from one system to another or if the backing HDD is removed / duplicated with the cache blocks in Robson not matching the disk location.
I would have the customer go into the laptop’s BIOS and turn off Robson (which will then keep the system from trying to boot from the ReadyDrive cache….)
After the system boots go into control panel and uninstall the Intel Turbo Memory Driver and reboot. After the reboot completes successfully, reboot and stop in BIOS to re-enable the Robson feature After the system boots, scan for hardware changes and the driver should re-install.
The preceeding should clean / invalidate the cache on ReadyDrive and the system should come back fine.
In other words, swapping the HDD from the internal slot to the Ultrabay set this off. As a matter of fact, TechNet's Keith Combs mentions an unspecified problem seemingly along these exact lines in a blog post from November 2007. (See this link.) Again, Lenovo has not commented on this at all.
Although I am recommending the AHCI setting above, please note there is a bug associated with that setting if using an Ultrabay hard drive adaptor and second hard drive. In that situation until the bug is fixed, you must flip the Serial ATA setting to Compatibility and add the second hard drive to the boot order of the machine. If you don't, you will see one or more bugs we've already identified and reported to Lenovo. They are working on BIOS fixes at present.
There are two unresolved mysteries about Intel's explanation. First, the HDD that BSODed has no ReadyDrive cache -- it's an older model without a hybrid NAND cache. Second, the T61p BIOS has no means to turn off Robson. Eventually I will go in and physically unplug the Robson chip, but I haven't gotten that drastic yet. I'm waiting to see what Intel says, if they're going to bother to reply.