Work around a bug in the interaction between newer dyld's and older simulator dyld's

[jimingham]

There's a bad interaction between the macOS 14 dyld and the "dyld_sim" shim that comes from older (iOS 15) simulator downloads that results in dyld reporting some modules twice in the return from the dyld callback to list modules. The records were identical, but lldb wasn't happy with seeing the duplicates...

Since it's not possible to load two different modules at the same address, this change just picks the first instance of any entries that have the same load address.

There really isn't a good way to test this patch.

[llvmbot]

@llvm/pr-subscribers-lldb

Author: None (jimingham)

Changes

There's a bad interaction between the macOS 14 dyld and the "dyld_sim" shim that comes from older (iOS 15) simulator downloads that results in dyld reporting some modules twice in the return from the dyld callback to list modules. The records were identical, but lldb wasn't happy with seeing the duplicates...

Since it's not possible to load two different modules at the same address, this change just picks the first instance of any entries that have the same load address.

There really isn't a good way to test this patch.

---

Full diff: https://github.com/llvm/llvm-project/pull/78004.diff

1 Files Affected:

• (modified) lldb/tools/debugserver/source/MacOSX/MachProcess.mm (+8)

diff --git a/lldb/tools/debugserver/source/MacOSX/MachProcess.mm b/lldb/tools/debugserver/source/MacOSX/MachProcess.mm
index 29fabd087ecbe1..93b976127ba45c 100644
--- a/lldb/tools/debugserver/source/MacOSX/MachProcess.mm
+++ b/lldb/tools/debugserver/source/MacOSX/MachProcess.mm
@@ -33,6 +33,7 @@
 #include <algorithm>
 #include <chrono>
 #include <map>
+#include <unordered_set>
 
 #include <TargetConditionals.h>
 #import <Foundation/Foundation.h>
@@ -1053,9 +1054,16 @@ static bool mach_header_validity_test(uint32_t magic, uint32_t cputype) {
     dyld_process_info info =
         m_dyld_process_info_create(m_task.TaskPort(), 0, &kern_ret);
     if (info) {
+      // There's a bug in the interaction between dyld and older dyld_sim's
+      // (e.g. from the iOS 15 simulator) that causes dyld to report the same
+      // binary twice.  We use this set to eliminate the duplicates.
+      __block std::unordered_set<uint64_t> seen_header_addrs;
       m_dyld_process_info_for_each_image(
           info,
           ^(uint64_t mach_header_addr, const uuid_t uuid, const char *path) {
+            auto res_pair = seen_header_addrs.insert(mach_header_addr);
+            if (!res_pair.second)
+              return;
             struct binary_image_information image;
             image.filename = path;
             uuid_copy(image.macho_info.uuid, uuid);

[jasonmolenda]

LGTM. debugserver never expects libdyld's API (to iterate over the binary images in a process) to return the same binary multiple times, and lldb won't handle that well if we pass that behavior along. Eliminating the incorrect duplicate here when we are getting the list of binaries from libdyld is the best place for it.