You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
From the "how the fuck did this hardware bug happen" department: Sandy Bridge GPUs cannot access the low 1MB of RAM, plus five pages. The pages at 0x20050000, 0x20110000, 0x20130000, 0x20138000, 0x40004000. Specifically.pic.twitter.com/rMN20Lt3xX
Also this only works for main RAM, if you have a PCIe device IO range mapped somewhere in there that the GPU has to hit (e.g. an nvidia card exposing a framebuffer for sharing or something), expect things to explode and there is no fix.
Also nobody seems to know how this interacts with the IOMMU ¯\_(ツ)_/¯
*tinfoil hat* These almost look like magic values, maybe some production debug/verification ranges with special decoder behaviour? 2005, 2011, 2013, 2013(.5) all look like years. 40004000 is too pretty to be accidental.
20000000-201fffff and 40004000-40004fff are "used by 'pnp 00:08'" on my system, which is either related to the culprit, or a BIOS workaround (Ivy Bridge here). Trying to find out what pnp 00:08 is.
The email associated with another patch addressing this issue says the information on the random pages came from someone at Intel, with no further explanation, when they sought confirmation on the <1MB issue. https://www.mail-archive.com/dri-devel@lists.freedesktop.org/msg22275.html …
Maybe it's something about this: https://github.com/coreboot/coreboot/blob/master/src/northbridge/intel/sandybridge/northbridge.c#L107 … Basically, some weird hardware bug that has no explanation available unless you like NDAs
@_Splitspace_
For anyone else researching the mystery, here's the Linux discussion. "SNB graphics devices have a bug that prevent them from accessing certain memory ranges" https://lore.kernel.org/patchwork/patch/340647/ …
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
