Rewrite unions such that all members are of the same size

In the C front-end, rewrite union components with types smaller than the
union's size to anonymous structs. Each such struct contains the
original union component plus padding. Assignments to union members thus
always assign all bytes that make up the object representation of a
union. The use of an anonymous struct ensures that member accesses can
still be resolved.

As this is a change in the semantics of goto programs, the goto binary
version is incremented. rewrite_union is no longer necessary, but bugs
(hidden by rewrite_union) in handling endianness in simplify_expr_member
and convert_member surfaced and had to be fixed.

Author: tautschnig

jbmc/src/jdiff/jdiff_parse_options.cpp

@@ -40,7 +40,6 @@ Author: Peter Schrammel
 #include <goto-programs/remove_unused_functions.h>
 #include <goto-programs/remove_vector.h>
 #include <goto-programs/remove_virtual_functions.h>
-#include <goto-programs/rewrite_union.h>
 #include <goto-programs/set_properties.h>
 #include <goto-programs/show_properties.h>
 #include <goto-programs/string_abstraction.h>
@@ -296,7 +295,6 @@ bool jdiff_parse_optionst::process_goto_program(
     remove_returns(goto_model);
     remove_vector(goto_model);
     remove_complex(goto_model);
-    rewrite_union(goto_model);
 
     // add generic checks
     log.status() << "Generic Property Instrumentation" << messaget::eom;

regression/ansi-c/arch_flags_mcpu_bad/object.intel

@@ -11,7 +11,7 @@ This tests the -mcpu=cortex=a15 flag that should activate ARM-32 mode.
 The object file 'object.intel' was compiled from 'source.c' with goto-cc
 on a 64-bit platform:
 
-  goto-cc -c source.c
+  goto-cc -c source.c -o object.intel
 
 preproc.i is already pre-processed so that it can be linked in without
 needing to invoke a pre-processor from a cross-compile toolchain on your

regression/ansi-c/arch_flags_mcpu_bad/test.desc

@@ -11,7 +11,7 @@ This tests the -mthumb flag that should activate ARM-32 mode. The object
 file 'object.intel' was compiled from 'source.c' with goto-cc on a
 64-bit platform:
 
-  goto-cc -c source.c
+  goto-cc -c source.c -o object.intel
 
 preproc.i is already pre-processed so that it can be linked in without
 needing to invoke a pre-processor from a cross-compile toolchain on your

regression/ansi-c/arch_flags_mcpu_good/object.arm

@@ -1,15 +1,15 @@
 CORE
 main.c
 --nan-check
-\[main.NaN.1\] line \d+ NaN on \+ in byte_extract_little_endian\(c, (0|0l|0ll), float\) \+ myzero: SUCCESS
+\[main.NaN.1\] line \d+ NaN on \+ in c\.f \+ myzero: SUCCESS
 \[main.NaN.2\] line \d+ NaN on / in myzero / myzero: FAILURE
 \[main.NaN.3\] line \d+ NaN on / in \(float\)n / myinf: FAILURE
 \[main.NaN.4\] line \d+ NaN on \* in myinf \* myzero: FAILURE
 \[main.NaN.5\] line \d+ NaN on \+ in -myinf \+ myinf: FAILURE
 \[main.NaN.6\] line \d+ NaN on - in myinf - myinf: FAILURE
 \[main.NaN.7\] line \d+ NaN on - in -myinf - -myinf: FAILURE
-\[main.NaN.8\] line \d+ NaN on \+ in byte_extract_little_endian\(c, (0|0l|0ll), float\) \+ byte_extract_little_endian\(c, (0|0l|0ll), float\): SUCCESS
-\[main.NaN.9\] line \d+ NaN on / in byte_extract_little_endian\(c, (0|0l|0ll), float\) / myzero: SUCCESS
+\[main.NaN.8\] line \d+ NaN on \+ in c\.f \+ c\.f: SUCCESS
+\[main.NaN.9\] line \d+ NaN on / in c\.f / myzero: SUCCESS
 \[main.NaN.10\] line \d+ NaN on \* in mynan \* mynan: SUCCESS
 \[main.NaN.11\] line \d+ NaN on \+ in mynan \+ mynan: SUCCESS
 \[main.NaN.12\] line \d+ NaN on - in mynan - mynan: SUCCESS

regression/ansi-c/arch_flags_mthumb_bad/object.intel

@@ -0,0 +1,11 @@
+CORE broken-smt-backend
+main.c
+--big-endian --no-simplify
+^EXIT=0$
+^SIGNAL=0$
+^VERIFICATION SUCCESSFUL$
+--
+^warning: ignoring
+--
+Although this test can now be fully solved via constant propagation and
+simplification, it should also succeed without simplification.

regression/ansi-c/arch_flags_mthumb_bad/test.desc

@@ -1,11 +1,11 @@
-CORE broken-smt-backend
+CORE
 main.c
 --big-endian
+Generated 3 VCC\(s\), 0 remaining after simplification
 ^EXIT=0$
 ^SIGNAL=0$
 ^VERIFICATION SUCCESSFUL$
 --
 ^warning: ignoring
 --
-This test reports a VERIFICATION ERROR when running with the SMT2 solver on
-Windows only.
+This test can now be fully solved via constant propagation and simplification.