[lldb][AArch64] Correct type of 32 bit GPR RegisterValues when using core files #70054
Conversation
…core files As ReadRegister always read into a uint64_t, when it called operator= with uint64_t it was setting the RegisterValue's type to eTypeUInt64 regardless of its size. This mostly works because most registers are 64 bit, and very few bits of code rely on the type being correct. However, cpsr, fpsr and fpcr are in fact 32 bit, and my upcoming register fields code relies on this type being correct. Which is how I found this bug and unfortunately is the only way to test it. As RegisterValue::Type never makes it out via the API anywhere. So this change will be tested once I start adding register field information.
|
@llvm/pr-subscribers-lldb Author: David Spickett (DavidSpickett) ChangesAs ReadRegister always read into a uint64_t, when it called operator= with uint64_t it was setting the RegisterValue's type to eTypeUInt64 regardless of its size. This mostly works because most registers are 64 bit, and very few bits of code rely on the type being correct. However, cpsr, fpsr and fpcr are in fact 32 bit, and my upcoming register fields code relies on this type being correct. Which is how I found this bug and unfortunately is the only way to test it. As RegisterValue::Type never makes it out via the API anywhere. So this change will be tested once I start adding register field information. Full diff: https://github.com/llvm/llvm-project/pull/70054.diff 1 Files Affected:
diff --git a/lldb/source/Plugins/Process/elf-core/RegisterContextPOSIXCore_arm64.cpp b/lldb/source/Plugins/Process/elf-core/RegisterContextPOSIXCore_arm64.cpp
index 6f6c6d073939a61..f3e763b0475ceb1 100644
--- a/lldb/source/Plugins/Process/elf-core/RegisterContextPOSIXCore_arm64.cpp
+++ b/lldb/source/Plugins/Process/elf-core/RegisterContextPOSIXCore_arm64.cpp
@@ -184,11 +184,9 @@ bool RegisterContextCorePOSIX_arm64::ReadRegister(const RegisterInfo *reg_info,
offset = reg_info->byte_offset;
if (offset + reg_info->byte_size <= GetGPRSize()) {
- uint64_t v = m_gpr_data.GetMaxU64(&offset, reg_info->byte_size);
- if (offset == reg_info->byte_offset + reg_info->byte_size) {
- value = v;
- return true;
- }
+ value.SetFromMemoryData(*reg_info, m_gpr_data.GetDataStart() + offset,
+ reg_info->byte_size, lldb::eByteOrderLittle, error);
+ return error.Success();
}
const uint32_t reg = reg_info->kinds[lldb::eRegisterKindLLDB];
|
This fixes a crash when trying to use cpsr field info with a core file, so this will be covered in those tests later. |
As ReadRegister always read into a uint64_t, when it called operator= with uint64_t it was setting the RegisterValue's type to eTypeUInt64 regardless of its size.
This mostly works because most registers are 64 bit, and very few bits of code rely on the type being correct. However, cpsr, fpsr and fpcr are in fact 32 bit, and my upcoming register fields code relies on this type being correct.
Which is how I found this bug and unfortunately is the only way to test it. As RegisterValue::Type never makes it out via the API anywhere. So this change will be tested once I start adding register field information.