Utilize emmc HPI to make way for an higher priority request

Meta:
Headline: Utilize emmc HPI to interrupt an ongoing request to make way for an higher priority request.
Acceptance: test that the interrupted command is interrupted and can resume correctly
Roadmap id: CARD-143

[jakub-pavelek 2012-11-09] This is slowly ongoing, Micron folks doing some work, svenkatr following up. Slow progress.
[jakub-pavelek 2013-01-02] Blocked, no developer. Can go deferred if not resolved
[jakub-pavelek 2013-02-26] Backlogged

Work items for 12.11:
[svenkatr] Test custom implementation on 2.6.35 with Android GB on Pandaboard: DONE
[svenkatr] Forward port the MMC abort interface: DONE
[svenkatr] Send draft v0.3 patches for internal review: DONE
[svenkatr] Test abort with eMMC device from Manufacturer-A: DONE
[svenkatr] Test abort with eMMC device from Manufacturer-B: DONE
[svenkatr] Implement sysfs for tuning the timing of abort wait time: TODO
[svenkatr] Test abort while STOP is in progress: DONE
[svenkatr] Post 1st version to lkml for review: DONE
[svenkatr] : Implement omap_hsmmc abort to execute even if the transfer is in progress: DONE
[svenkatr] Test abort with eMMC device from Manufacturer-C: DONE
[svenkatr] : Create sysfs entry for delay threshold: DONE
[svenkatr] : Create documentation for sysfs entries at Documentation/ABI: DONE
[svenkatr] : Post completed version 2 to lkml: DONE

Work items for backlog:
[svenkatr] Follow-up to get it accepted: INPROGRESS
[svenkatr] Investigate policy implementation - don't do HPI for readaheads, but only for sync reads: TODO
[svenkatr] Check with Memory part vendors if benchmarks can be published based on this patchset, to get wider acceptance (Oct 23): INPROGRESS
[svenkatr] If not, it can be blocked: TODO
[valiberti] Provide performance test benchmarks: TODO