Description of problem:
Mono 4.2.1 fails to build on Koji for armv7hl.

we get an assertion with stacktrace.

How reproducible:
always.

Steps to Reproduce:
build Mono from git master http://pkgs.fedoraproject.org/cgit/mono.git/ on Koji

Actual results:

see scratch build http://koji.fedoraproject.org/koji/taskinfo?taskID=12289721

mkdir -p -- ../../class/lib/build/
MCS     [build] mscorlib.dll
* Assertion at sgen-alloc.c:558, condition `tlab_next_addr_offset != -1' not met
Native stacktrace:
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0xab634) [0x7f63c634]
	/lib/libc.so.6(__default_rt_sa_restorer+0) [0x4017f150]
	/lib/libc.so.6(gsignal+0x30) [0x4017dce0]
	/lib/libc.so.6(abort+0x12c) [0x4017f2b8]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x22f274) [0x7f7c0274]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x22f494) [0x7f7c0494]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x22f580) [0x7f7c0580]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x1d9634) [0x7f76a634]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x1e605c) [0x7f77705c]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x1ca404) [0x7f75b404]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x19862c) [0x7f72962c]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(+0x2d718) [0x7f5be718]
	/builddir/build/BUILD/mono-4.2.1/mono/mini/mono(mono_main+0x314) [0x7f60e908]
	/lib/libc.so.6(__libc_start_main+0x114) [0x40165df8]



Expected results:
should build fine, as it does for i686 and x86_64.

Additional info:
this bug at upstream looks the same, although people report it on different architecture: https://bugzilla.xamarin.com/show_bug.cgi?id=30594