[PATCH] D71174: [clang-tidy] new check: bugprone-signed-char-misuse

[Tamás Zolnai via Phabricator via cfe-commits]

ztamas added a comment.

I run the new check on LLVM codebase with the option CharTypdefsToIgnore = "int8_t".
The check produced 12 findings.

All findings seem valid use cases, where a char -> integer conversion happens.
I had a closer look at some of the catches.

Somewhere I see only type mismatch. Char is used instead of an integer type.
For example, in the finding above, trailingBytesForUTF8 is defined as char array, but it is used
everywhere as integer:

  /home/zolnai/clang/llvm-project/llvm/lib/Support/ConvertUTF.cpp:623:43: warning: singed char -> integer ('unsigned short') conversion; consider to cast to unsigned char first. [bugprone-signed-char-misuse]
          unsigned short extraBytesToRead = trailingBytesForUTF8[*source];

Another use case seems suspicious to me. Here we have some parsing code, if I'm right, which uses the EOF from cstdio header, which is -1,
so might be a similar use case which is mentioned by the CERT specification.
/home/zolnai/clang/llvm-project/llvm/lib/TableGen/TGLexer.cpp:596:20: warning: singed char -> integer ('int') conversion; consider to cast to unsigned char first. [bugprone-signed-char-misuse]

  int NextChar = *CurPtr;

At the location of the third use case, there is already a comment about signed chars ("/* xxx what about signed chars here... */"),
so I interpret it as a validation of the check.
/home/zolnai/clang/llvm-project/llvm/lib/Support/regcomp.c:929:12: warning: singed char -> integer ('int') conversion; consider to cast to unsigned char first. [bugprone-signed-char-misuse]

  for (i = start; i <= finish; i++)

The whole output can be found here:
https://gist.github.com/tzolnai/b1cbe3f021b53a7ca1a6ecbffc1a9bf6


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D71174/new/

https://reviews.llvm.org/D71174