4.16 regression: s2ram broken on non-PAE i686

[Woody Suwalski]

There is a problem with s2ram on 4.16, and it has now been propagated
to 4.15 and 4.14 stable updates.

It originates from

commit 62c00e6122a6b5aa7b1350023967a2d7a12b54c9
Author: William Grant <william.grant@xxxxxxxxxxxxx
<mailto:william.grant@xxxxxxxxxxxxx>>
Date:   Tue Jan 30 22:22:55 2018 +1100

    x86/mm: Fix overlap of i386 CPU_ENTRY_AREA with FIX_BTMAP

s2ram works OK on PAE kernels, breaks badly on non-PAE. I do not think
that the problem can be duplicated in VMPlayer, but it is 100%
reproducible on a "real" hardware.
System goes to sleep OK, but when woken - it reboots the PC.

The issue is tracked in Bugzilla bug 198763
[https://bugzilla.kernel.org/show_bug.cgi?id=198763]

Thanks, Woody