[llvm] [JITLink][AArch32] Fix Unaligned Data Symbol Address Resolution (PR #97030)

[Eymen Ünay via llvm-commits]

================
@@ -0,0 +1,72 @@
+# RUN: llvm-mc -triple=armv7-linux-gnueabi -arm-add-build-attributes -filetype=obj -o %t_armv7.o %s
+# RUN: llvm-objdump -s --section=.rodata %t_armv7.o | FileCheck --check-prefix=CHECK-OBJ %s
+# RUN: llvm-jitlink -noexec -slab-address 0x76ff0000 -slab-allocate 10Kb \
+# RUN:              -slab-page-size 4096 %t_armv7.o -debug-only=jitlink 2>&1 \
+# RUN:              | FileCheck --check-prefix=CHECK-LG %s
+# RUN: llvm-jitlink -noexec -slab-address 0x76ff0000 -slab-allocate 10Kb \
+# RUN:              -slab-page-size 4096 %t_armv7.o -check %s
+
+# RUN: llvm-mc -triple=thumbv7-linux-gnueabi -arm-add-build-attributes -filetype=obj -o %t_thumbv7.o %s
+# RUN: llvm-objdump -s --section=.rodata %t_thumbv7.o | FileCheck --check-prefix=CHECK-OBJ %s
+# RUN: llvm-jitlink -noexec -slab-address 0x76ff0000 -slab-allocate 10Kb \
+# RUN:              -slab-page-size 4096 %t_thumbv7.o -debug-only=jitlink 2>&1 \
+# RUN:              | FileCheck --check-prefix=CHECK-LG %s
+# RUN: llvm-jitlink -noexec -slab-address 0x76ff0000 -slab-allocate 10Kb \
+# RUN:              -slab-page-size 4096 %t_thumbv7.o -check %s
+
+# The strings of "H1\00", "H2\00" and "H3\00" are encoded as
+#               0x483100, 0x483200 and 0x483300 .rodata section.
+# CHECK-OBJ: Contents of section .rodata:
+# CHECK-OBJ: 0000 48310048 32004833 00                 H1.H2.H3.
+
+# CHECK-LG: Starting link phase 1 for graph
+# CHECK-LG: section .rodata:
+
+# CHECK-LG:       block 0x0 size = 0x00000009, align = 1, alignment-offset = 0
+# CHECK-LG-NEXT:    symbols:
+# CHECK-LG-NEXT:      0x0 (block + 0x00000000): size: 0x00000003, linkage: strong, scope: default, live  -   Lstr.H1
+# CHECK-LG-NEXT:      0x3 (block + 0x00000003): size: 0x00000003, linkage: strong, scope: default, live  -   Lstr.H2
+# CHECK-LG-NOT:       0x2 (block + 0x00000002): size: 0x00000003, linkage: strong, scope: default, live  -   Lstr.H2
+# CHECK-LG-NEXT:      0x6 (block + 0x00000006): size: 0x00000003, linkage: strong, scope: default, live  -   Lstr.H3
+
+# FIXME: The expression we want is either *{3}(Lstr.H1) = ...
+#                                  or *{4}(Lstr.H1) & 0x00ffffff = ...
+#        The first is not supported and the latter segfaults.
+#        Also, whitespaces are not recognized and not consumed by the checker.
----------------
eymay wrote:

It does! Thanks for the tip.

https://github.com/llvm/llvm-project/pull/97030