[llvm-objcopy] Apply encryptable offset to first segment, not section

[drodriguez]

Bug introduced #120995. The LLD code calculates the "size" of the Mach-O headers, and then uses that size to place the segments, but the __TEXT section stays at fileoff zero. When I wrote the code into llvm-objcopy I calculated the extra space into the initial offset, which moved all the sections back 1 page.

Besides the modified test checking for the right fileoff values of the sections and the segments, I also manually checked the generated binaries after llvm-objcopy using dyld_info, as the bug report suggested.

Fixes #130472

[llvmbot]

@llvm/pr-subscribers-llvm-binary-utilities

Author: Daniel Rodríguez Troitiño (drodriguez)

Changes

Bug introduced #120995. The LLD code calculates the "size" of the Mach-O headers, and then uses that size to place the segments, but the __TEXT section stays at fileoff zero. When I wrote the code into llvm-objcopy I calculated the extra space into the initial offset, which moved all the sections back 1 page.

Besides the modified test checking for the right fileoff values of the sections and the segments, I also manually checked the generated binaries after llvm-objcopy using dyld_info, as the bug report suggested.

Fixes #130472

---

Full diff: https://github.com/llvm/llvm-project/pull/130517.diff

2 Files Affected:

• (modified) llvm/lib/ObjCopy/MachO/MachOLayoutBuilder.cpp (+7-5)
• (modified) llvm/test/tools/llvm-objcopy/MachO/strip-with-encryption-info.test (+9-3)

diff --git a/llvm/lib/ObjCopy/MachO/MachOLayoutBuilder.cpp b/llvm/lib/ObjCopy/MachO/MachOLayoutBuilder.cpp
index d4eb6a9b9fc0b..91f73b59db1bd 100644
--- a/llvm/lib/ObjCopy/MachO/MachOLayoutBuilder.cpp
+++ b/llvm/lib/ObjCopy/MachO/MachOLayoutBuilder.cpp
@@ -116,11 +116,9 @@ uint64_t MachOLayoutBuilder::layoutSegments() {
   const bool IsObjectFile =
       O.Header.FileType == MachO::HeaderFileType::MH_OBJECT;
   uint64_t Offset = IsObjectFile ? (HeaderSize + O.Header.SizeOfCmds) : 0;
-  if (O.EncryptionInfoCommandIndex) {
-    // If we are emitting an encryptable binary, our load commands must have a
-    // separate (non-encrypted) page to themselves.
-    Offset = alignToPowerOf2(HeaderSize + O.Header.SizeOfCmds, PageSize);
-  }
+  // If we are emitting an encryptable binary, our load commands must have a
+  // separate (non-encrypted) page to themselves.
+  bool FirstSectExtraEncryptableOffset = O.EncryptionInfoCommandIndex.has_value();
   for (LoadCommand &LC : O.LoadCommands) {
     auto &MLC = LC.MachOLoadCommand;
     StringRef Segname;
@@ -174,6 +172,10 @@ uint64_t MachOLayoutBuilder::layoutSegments() {
         if (!Sec->hasValidOffset()) {
           Sec->Offset = 0;
         } else {
+          if (FirstSectExtraEncryptableOffset) {
+            SectOffset =  alignToPowerOf2(SectOffset, PageSize);
+            FirstSectExtraEncryptableOffset = false;
+          }
           Sec->Offset = SegOffset + SectOffset;
           Sec->Size = Sec->Content.size();
           SegFileSize = std::max(SegFileSize, SectOffset + Sec->Size);
diff --git a/llvm/test/tools/llvm-objcopy/MachO/strip-with-encryption-info.test b/llvm/test/tools/llvm-objcopy/MachO/strip-with-encryption-info.test
index 19b06b1ec02c8..2b2bd670613de 100644
--- a/llvm/test/tools/llvm-objcopy/MachO/strip-with-encryption-info.test
+++ b/llvm/test/tools/llvm-objcopy/MachO/strip-with-encryption-info.test
@@ -1,13 +1,19 @@
 # RUN: rm -rf %t && mkdir %t
 # RUN: yaml2obj %s -o %t/original
 # RUN: llvm-strip --strip-all %t/original -o %t/stripped
-# RUN: llvm-readobj --macho-segment %t/stripped | FileCheck %s
+# RUN: llvm-readobj --macho-segment --section-headers %t/stripped | FileCheck %s
+
+# CHECK-LABEL: Sections [
+# CHECK:      Index: 0
+# CHECK-NEXT: Name: __text
+# CHECK-NEXT: Segment: __TEXT
+# CHECK:      Offset: 16384
 
 # CHECK-LABEL: Name: __PAGEZERO
-# CHECK:       fileoff: 16384
+# CHECK:       fileoff: 0
 
 # CHECK-LABEL: Name: __TEXT
-# CHECK:       fileoff: 16384
+# CHECK:       fileoff: 0
 
 # The YAML below is the following code
 # int main(int argc, char **argv) { return 0; }

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[jh7370]

Unfortunately, this is outside my minimal knowledge of Mach-O, so I'll leave it to another reviewer to look into. If nobody else shows up, let me know and I'll try adding some other possible reviewers.

[drodriguez]

Sorry James. I though you were the other commenter in the other thread. My bad.

@glandium do you mind checking if this was what was affecting your binary from #120995 ? For some reason the fileoff being a page into the binary was not crashing anything for me, but the dyld_info checker obviously find it incorrect.

[jh7370]

Sorry James. I though you were the other commenter in the other thread. My bad.

Not a problem. I'm involved in pretty much every change in llvm-objcopy ELF and generic code, but have also been heavily involved in some aspects of the other file formats, so confusion is entirely understandable.

[glandium]

Applying this patch fixes the problem I had indeed.

[alx32]

LGTM