[GDBRemote] Fix processing of comma-separated memory region entries

[felipepiovezan]

The existing algorithm was performing the following comparisons for an aaa,bbb,ccc,ddd:

aaa\0bbb,ccc,ddd == "stack"
aaa\0bbb\0ccc,ddd == "stack"
aaa\0bbb\0ccc\0ddd == "stack"

Which wouldn't work. This commit just dispatches to a known algorithm implementation.

[llvmbot]

@llvm/pr-subscribers-lldb

Author: Felipe de Azevedo Piovezan (felipepiovezan)

Changes

The existing algorithm was performing the following comparisons for an aaa,bbb,ccc,ddd:

aaa\0bbb,ccc,ddd == "stack"
aaa\0bbb\0ccc,ddd == "stack"
aaa\0bbb\0ccc\0ddd == "stack"

Which wouldn't work. This commit just dispatches to a known algorithm implementation.

---

Full diff: https://github.com/llvm/llvm-project/pull/105873.diff

2 Files Affected:

• (modified) lldb/source/Plugins/Process/gdb-remote/GDBRemoteCommunicationClient.cpp (+2-10)
• (modified) lldb/unittests/Process/gdb-remote/GDBRemoteCommunicationClientTest.cpp (+5-2)

diff --git a/lldb/source/Plugins/Process/gdb-remote/GDBRemoteCommunicationClient.cpp b/lldb/source/Plugins/Process/gdb-remote/GDBRemoteCommunicationClient.cpp
index 83ba27783da471..d80ccae0518088 100644
--- a/lldb/source/Plugins/Process/gdb-remote/GDBRemoteCommunicationClient.cpp
+++ b/lldb/source/Plugins/Process/gdb-remote/GDBRemoteCommunicationClient.cpp
@@ -1632,17 +1632,9 @@ Status GDBRemoteCommunicationClient::GetMemoryRegionInfo(
             }
           }
         } else if (name == "type") {
-          std::string comma_sep_str = value.str();
-          size_t comma_pos;
-          while ((comma_pos = comma_sep_str.find(',')) != std::string::npos) {
-            comma_sep_str[comma_pos] = '\0';
-            if (comma_sep_str == "stack") {
+          for (llvm::StringRef entry: llvm::split(value, ',')) {
+            if (entry == "stack")
               region_info.SetIsStackMemory(MemoryRegionInfo::eYes);
-            }
-          }
-          // handle final (or only) type of "stack"
-          if (comma_sep_str == "stack") {
-            region_info.SetIsStackMemory(MemoryRegionInfo::eYes);
           }
         } else if (name == "error") {
           StringExtractorGDBRemote error_extractor(value);
diff --git a/lldb/unittests/Process/gdb-remote/GDBRemoteCommunicationClientTest.cpp b/lldb/unittests/Process/gdb-remote/GDBRemoteCommunicationClientTest.cpp
index 11e14f9472164d..18020c8e43fe06 100644
--- a/lldb/unittests/Process/gdb-remote/GDBRemoteCommunicationClientTest.cpp
+++ b/lldb/unittests/Process/gdb-remote/GDBRemoteCommunicationClientTest.cpp
@@ -343,24 +343,27 @@ TEST_F(GDBRemoteCommunicationClientTest, GetMemoryRegionInfo) {
   EXPECT_EQ(MemoryRegionInfo::eYes, region_info.GetExecutable());
   EXPECT_EQ("/foo/bar.so", region_info.GetName().GetStringRef());
   EXPECT_EQ(MemoryRegionInfo::eDontKnow, region_info.GetMemoryTagged());
+  EXPECT_EQ(MemoryRegionInfo::eDontKnow, region_info.IsStackMemory());
 
   result = std::async(std::launch::async, [&] {
     return client.GetMemoryRegionInfo(addr, region_info);
   });
 
   HandlePacket(server, "qMemoryRegionInfo:a000",
-               "start:a000;size:2000;flags:;");
+               "start:a000;size:2000;flags:;type:stack;");
   EXPECT_TRUE(result.get().Success());
   EXPECT_EQ(MemoryRegionInfo::eNo, region_info.GetMemoryTagged());
+  EXPECT_EQ(MemoryRegionInfo::eYes, region_info.IsStackMemory());
 
   result = std::async(std::launch::async, [&] {
     return client.GetMemoryRegionInfo(addr, region_info);
   });
 
   HandlePacket(server, "qMemoryRegionInfo:a000",
-               "start:a000;size:2000;flags: mt  zz mt  ;");
+               "start:a000;size:2000;flags: mt  zz mt  ;type:ha,ha,stack;");
   EXPECT_TRUE(result.get().Success());
   EXPECT_EQ(MemoryRegionInfo::eYes, region_info.GetMemoryTagged());
+  EXPECT_EQ(MemoryRegionInfo::eYes, region_info.IsStackMemory());
 }
 
 TEST_F(GDBRemoteCommunicationClientTest, GetMemoryRegionInfoInvalidResponse) {

[jasonmolenda]

Thanks for fixing this, looks perfect.

[github-actions]

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

[felipepiovezan]

Since this is mostly uncontroversial, I am going to go ahead and merge it now. Happy to address feedback though!

[felipepiovezan]

Fixed clang format issue