build warning: fortify memcpy #23638

<!-- This issue tracker is only for technical issues related to Bitcoin Core. General bitcoin questions and/or support requests are best directed to the Bitcoin StackExchange at https://bitcoin.stackexchange.com. For reporting security issues, please read instructions at https://bitcoincore.org/en/contact/. If the node is "stuck" during sync or giving "block checksum mismatch" errors, please ensure your hardware is stable by running memtest and observe CPU temperature with a load-test tool such as linpack before creating an issue! -->

<!-- Describe the issue -->

This is just a warning, but want to make a note here.

  CXX      libbitcoin_server_a-signet.o
In file included from /usr/include/c++/10.3.1/cstring:42,
                 from ./uint256.h:10,
                 from ./consensus/params.h:9,
                 from ./signet.h:8,
                 from signet.cpp:5:
In function 'void* memcpy(void*, const void*, size_t)',
    inlined from 'void CVectorWriter::write(const char*, size_t)' at ./streams.h:102:19,
    inlined from 'void base_blob<BITS>::Serialize(Stream&) const [with Stream = CVectorWriter; unsigned int BITS = 256]' at ./uint256.h:99:16,
    inlined from 'static std::optional<SignetTxs> SignetTxs::Create(const CBlock&, const CScript&)' at ./serialize.h:670:16:
/usr/include/fortify/string.h:53:25: warning: 'void* __builtin_memcpy(void*, const void*, long unsigned int)' writing between 1 and 32 bytes into a region of size 0 overflows the destination [-Wstringop-overflow=]
   53 |  return __builtin_memcpy(__od, __os, __n);
      |         ~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~

Expected behavior

Clean happy compilation, something like this which comes from later output:

...
  CXX      wallet/libbitcoin_wallet_a-db.o
  CXX      wallet/libbitcoin_wallet_a-dump.o
  CXX      wallet/libbitcoin_wallet_a-external_signer_scriptpubkeyman.o
  CXX      wallet/libbitcoin_wallet_a-feebumper.o
  CXX      wallet/libbitcoin_wallet_a-fees.o
  CXX      wallet/libbitcoin_wallet_a-interfaces.o
...

<!--- What behavior did you expect? -->

Actual behavior

Warnings during make.

<!--- What was the actual behavior (provide screenshots if the issue is GUI-related)? -->

To reproduce

# On Alpine Linux (musl-libc)
export CFLAGS="-pipe -s -march=native -mtune=native -O3"
export LDFLAGS="-s -no-pie"

test -x ./configure || ./autogen.sh
./configure \
  --without-gui \
  --with-incompatible-bdb \
  --disable-tests \
  --disable-bench \
  --disable-zmq \
  --disable-hardening \
  CFLAGS="$CFLAGS" \
  CXXFLAGS="$CFLAGS" \
  LDFLAGS="$LDFLAGS"
make prefix=$HOME/bitcoin-latest install

<!--- How reliably can you reproduce the issue, what are the steps to do so? -->

System information

Alpine Linux (musl-libc x86_64)

$ uname -mr
5.10.61-0-lts x86_64

<!-- What version of Bitcoin Core are you using, where did you get it (website, self-compiled, etc)? -->

Self-compiled Bitcoin Core exactly from jsarenik/bitcoin@878d8e1c0

<!-- What type of machine are you observing the error on (OS/CPU and disk type)? -->

<!-- GUI-related issue? What is your operating system and its version? If Linux, what is your desktop environment and graphical shell? -->

<!-- Any extra information that might be useful in the debugging process. -->

<!--- This is normally the contents of a `debug.log` or `config.log` file. Raw text or a link to a pastebin type site are preferred. -->

This warning was experienced while compiling version of bitcoin to test #23531 while also rebasing #22563 on top of it and also Alpine Linux locales patch.

All the rebasing was trivial and maybe the warning is showing for longer but now I noticed it for the first time.

Looks like a compiler bug

For reference, the function is:

    void write(const char* pch, size_t nSize)
    {
        assert(nPos <= vchData.size());
        size_t nOverwrite = std::min(nSize, vchData.size() - nPos);
        if (nOverwrite) {
            memcpy(vchData.data() + nPos, reinterpret_cast<const unsigned char*>(pch), nOverwrite);
        }
        if (nOverwrite < nSize) {
            vchData.insert(vchData.end(), reinterpret_cast<const unsigned char*>(pch) + nOverwrite, reinterpret_cast<const unsigned char*>(pch) + nSize);
        }
        nPos += nSize;
    }

Looks like a compiler bug

Yes I think so. Reproduced with GCC 10.3.1 (on Alpine), with -O3, no warning with -O2. Possibly https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94335 or related to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92955.

Not much we can do here, other than suggest you up/downgrade to a non-buggy compiler.

Going to close for now.