diff options
author | Jonas Devlieghere <jonas@devlieghere.com> | 2020-02-05 09:53:10 -0800 |
---|---|---|
committer | Jonas Devlieghere <jonas@devlieghere.com> | 2020-02-05 14:52:08 -0800 |
commit | 6dccad7517f83a069d9bfbf761b648b0e50870c2 (patch) | |
tree | cc7941569e591e0062ed330caf6e6618a566d6cc /llvm/lib/CodeGen/MachineBasicBlock.cpp | |
parent | 3ec28da6d6430a00b46780555a87acd43fcab790 (diff) | |
download | llvm-6dccad7517f83a069d9bfbf761b648b0e50870c2.zip llvm-6dccad7517f83a069d9bfbf761b648b0e50870c2.tar.gz llvm-6dccad7517f83a069d9bfbf761b648b0e50870c2.tar.bz2 |
[lldb/Reproducers] (De)serialize char* like const char*
The current implementation has a discrepancy between how char pointers
are serialized and deserialized. The latter treats it like a const char*
while the former serializes it as a pointer to a basic type.
Both are potentially wrong, as char pointers are mostly used in
combination with a size, and nothing guarantees that the string's length
(its first null byte to be more precise) is greater or equal to its
size. The real solution is to have a custom (de)serializer that uses
both pieces of infromation.
However, the implementation should be consistent between serialization
and deserialization and I believe treating char* as const char* is the
better alternative.
Diffstat (limited to 'llvm/lib/CodeGen/MachineBasicBlock.cpp')
0 files changed, 0 insertions, 0 deletions