Comment 5 for bug 752870

Looking at the code, it seems that this commit removed the setting/restore of mmu_cr4_features, and the crash happens when it loads probably an invalid mmu_cr4_features

And indeed that's the case, today I saw this commit coming in in Linus tree:
commit 4da9484bdece39ab0b098fa711e095e3e9fc8684
Author: H. Peter Anvin <email address hidden>
Date: Wed Apr 6 13:10:02 2011 -0700

    x86, hibernate: Initialize mmu_cr4_features during boot

    Restore the initialization of mmu_cr4_features during boot, which was
    removed without comment in checkin e5f15b45ddf3afa2bbbb10c7ea34fb32b6de0a0e

    x86: Cleanup highmap after brk is concluded

    thereby breaking resume from hibernate. This restores previous
    functionality in approximately the same place, and corrects the
    reading of %cr4 on pre-CPUID hardware (%cr4 exists if and only if
    CPUID is supported.)

    However, part of the problem is that the hibernate suspend/resume
    sequence should manage the save/restore of %cr4 explicitly.

    Signed-off-by: H. Peter Anvin <email address hidden>
    Cc: Rafael J. Wysocki <email address hidden>
    Cc: Stefano Stabellini <email address hidden>
    Cc: Yinghai Lu <email address hidden>
    LKML-Reference: <email address hidden>

and it fixes the bug for me too, testing here