Testing of the legacy Intel driver work done by Chris to integrate back a user-space mode-setting path for this old hardware (and the UMS path also works with new Intel chipsets too) along EXA 2D acceleration has been somewhat positive, but it hasn't addressed all problems reported by Intel Linux users. This also just provides an older code path for users rather than addressing the actual problems with GEM/KMS itself for these customers.
Chris Wilson has submitted his patch yesterday afternoon to the Intel mailing list to re-integrate legacy user-space mode-setting support into the open-source driver. This would add over 48,000 lines of code back to the Intel driver as it takes all of their 2D non-GEM EXA acceleration code from their xf86-video-intel 2.6 driver with updates through their xf86-video-intel 2.9 driver (right before EXA/UMS was dropped).
Though in order for this code to be pulled back into the mainline Linux 2D driver, Chris/Intel would like a commitment from distribution vendors to supporting and using this older code. "I'd like a show of commitment from the distributions that this is how we should proceed. Our goal, and our focus, is on providing KMS and GEM for Intel chipsets. We do not want legacy/ums to hinder that, however we also do not want to leave machines behind (i8xx, the current bane of our existence) and force each distributions to fork their own legacy driver."
Those wanting to try this legacy UMS driver can find the code in the legacy branch of Chris Wilson's xf86-video-intel driver. This work could be integrated and released with xf86-video-intel 2.13, which will be Intel's next quarterly driver update and will be released within a month or two.