Skip to content

tests: Switch one of the Travis jobs to an unsigned char environment (-funsigned-char) - #15134

Merged
laanwj merged 1 commit into
bitcoin:masterfrom
practicalswift:unsigned-char
Jul 30, 2019
Merged

laanwj merged 1 commit into
bitcoin:masterfrom
practicalswift:unsigned-char

Conversation

@practicalswift

@practicalswift practicalswift commented Jan 9, 2019 •

Copy link
Copy Markdown
Contributor

Switch one of the Travis jobs to an unsigned char environment (-funsigned-char).

This will help us catch errors due to code written under the assumption that char has the same value range as signed char.

The signedness of char is implementation-defined.

Example:

$ uname -a
Linux […] x86_64 x86_64 x86_64 GNU/Linux
$ cat foo.cpp
#include <iostream>

int main() {
    char c;
    std::cin >> c;
    int i = (unsigned char)c;
    std::cout << i << "\n";
}
$ clang++ -o foo foo.cpp
$ echo -e "\xff" | ./foo
255
$ clang++ -fsigned-char -o foo foo.cpp
$ echo -e "\xff" | ./foo
255
$ clang++ -funsigned-char -o foo foo.cpp
$ echo -e "\xff" | ./foo
255
$ cat bar.cpp
#include <iostream>

int main() {
    char c;
    std::cin >> c;
    int i = c;
    std::cout << i << "\n";
}
$ clang++ -o bar bar.cpp
$ echo -e "\xff" | ./bar
-1
$ clang++ -fsigned-char -o bar bar.cpp
$ echo -e "\xff" | ./bar
-1
$ clang++ -funsigned-char -o bar bar.cpp
$ echo -e "\xff" | ./bar
255

gcc chars:

  • signed: alpha, hppa, ia64, m68k, mips, sh, sparc, x86
  • unsigned: arm, powerpc, s390

About -funsigned-char:

Let the type "char" be unsigned, like "unsigned char".

Each kind of machine has a default for what "char" should be. It is either like "unsigned char" by default or like "signed char" by default.

Ideally, a portable program should always use "signed char" or "unsigned char" when it depends on the signedness of an object. But many programs have been written to use plain "char" and expect it to be signed, or expect it to be unsigned, depending on the machines they were written for.

This option, and its inverse, let you make such a program work with the opposite default. The type "char" is always a distinct type from each of "signed char" or "unsigned char", even though its behavior is always just like one of those two.

@fanquake fanquake added the Tests label Jan 9, 2019
@laanwj

laanwj commented Jan 9, 2019

Copy link
Copy Markdown
Member

Concept ACK, mistakes with char are quite common in C use, without this I guess things suddenly might start to break on ARM or RISC-V without Travis noticing.

@practicalswift practicalswift changed the title build: Add a Travis ASan/LSan/UBSan job testing in a unsigned char environment (-funsigned-char) tests: Add a Travis ASan/LSan/UBSan job testing in a unsigned char environment (-funsigned-char) Jan 9, 2019
@laanwj

laanwj commented Jan 16, 2019

Copy link
Copy Markdown
Member

@theuni can you take a look at this please

@DrahtBot

DrahtBot commented Jan 30, 2019 •

Copy link
Copy Markdown
Contributor

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Conflicts

Reviewers, this pull request conflicts with the following ones:

  • #16320 (ci: Add Travis check to make sure unit test coverage reports stay deterministic by practicalswift)

If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

@Rspigler

Rspigler commented Feb 1, 2019

Copy link
Copy Markdown
Contributor

Would this cover Power ISA? (#14066)

@laanwj

laanwj commented Feb 16, 2019

Copy link
Copy Markdown
Member

Would this cover Power ISA? (#14066)

PowerPC has unsigned chars afaik so in a sense, yea.

@practicalswift

Copy link
Copy Markdown
Contributor Author

Can we move forward with this one? :-)

Should hopefully be trivial to review since it only touches .travis.yml.

Perhaps you could take a look too @MarcoFalke? :-)

Comment thread .travis.yml Resolved
@practicalswift

Copy link
Copy Markdown
Contributor Author

@MarcoFalke Thanks for reviewing! Now rebased and updated. Please re-review :-)

Comment thread .travis.yml Resolved
@practicalswift practicalswift changed the title tests: Add a Travis ASan/LSan/UBSan job testing in a unsigned char environment (-funsigned-char) tests: Switch one of the Travis jobs to an unsigned char environment (-funsigned-char) Mar 13, 2019
@practicalswift

Copy link
Copy Markdown
Contributor Author

@MarcoFalke Updated version looks good?

@maflcko

maflcko commented Mar 14, 2019

Copy link
Copy Markdown
Member

ACK

@practicalswift

Copy link
Copy Markdown
Contributor Author

@laanwj Would you mind reviewing?

I think we should be ready for merge :-)

Comment thread .travis.yml Resolved
@practicalswift

practicalswift commented Jun 28, 2019 •

Copy link
Copy Markdown
Contributor Author

@MarcoFalke @laanwj Is this one ready for merge? Having one -funsigned-char testing job would make me slightly less worried about mistakes due to incorrect signedness assumptions for char :-)

Let me know if any further changes are needed!

@practicalswift

Copy link
Copy Markdown
Contributor Author

@MarcoFalke @laanwj Friendly ping: is this PR still of interest? Let me know if it should closed :-)

@maflcko

maflcko commented Jul 30, 2019

Copy link
Copy Markdown
Member

Concept ACK, but I am lacking the background to properly review

@laanwj

laanwj commented Jul 30, 2019

Copy link
Copy Markdown
Member

ACK 0c78e49

@laanwj
laanwj merged commit 0c78e49 into bitcoin:master Jul 30, 2019
laanwj added a commit that referenced this pull request Jul 30, 2019
…r environment (-funsigned-char)

0c78e49 tests: Switch one of the Travis jobs to an unsigned char environment (-funsigned-char) (practicalswift)

Pull request description:

  Switch one of the Travis jobs to an unsigned char environment (`-funsigned-char`).

  This will help us catch errors due to code written under the assumption that `char` has the same value range as `signed char`.

  The signedness of `char` is implementation-defined.

  Example:

  ```
  $ uname -a
  Linux […] x86_64 x86_64 x86_64 GNU/Linux
  $ cat foo.cpp
  #include <iostream>

  int main() {
      char c;
      std::cin >> c;
      int i = (unsigned char)c;
      std::cout << i << "\n";
  }
  $ clang++ -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ clang++ -fsigned-char -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ clang++ -funsigned-char -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ cat bar.cpp
  #include <iostream>

  int main() {
      char c;
      std::cin >> c;
      int i = c;
      std::cout << i << "\n";
  }
  $ clang++ -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  -1
  $ clang++ -fsigned-char -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  -1
  $ clang++ -funsigned-char -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  255
  ```

  `gcc` chars:
  * signed: alpha, hppa, ia64, m68k, mips, sh, sparc, x86
  * unsigned: arm, powerpc, s390

  About `-funsigned-char`:

  > Let the type "char" be unsigned, like "unsigned char".
  >
  > Each kind of machine has a default for what "char" should be.  It is either like "unsigned char" by default or like "signed char" by default.
  >
  > Ideally, a portable program should always use "signed char" or "unsigned char" when it depends on the signedness of an object.  But many programs have been written to use plain "char" and expect it to be signed, or expect it to be unsigned, depending on the machines they were written for.
  >
  > This option, and its inverse, let you make such a program work with the opposite default. The type "char" is always a distinct type from each of "signed char" or "unsigned char", even though its behavior is always just like one of those two.

ACKs for top commit:
  laanwj:
    ACK 0c78e49

Tree-SHA512: ba04590415c0bb9a0bbd348623e57068f75274f53da7247d5c5ecad82e365a5b45893a4a491d318e82a8feb6a25f019d46e01990afb33162e2c9740d33a343d7
maflcko pushed a commit that referenced this pull request Jul 2, 2020
…se of uninitialized memory

870f0cd build: Add MemorySanitizer (MSan) in Travis to detect use of uninitialized memory (practicalswift)

Pull request description:

  Add MemorySanitizer (MSan) in Travis to detect use of uninitialized memory.

  First UBSan, then ASan followed by TSan... and now: yes, the wait is over -- **MSan is finally here!** :)

  Some historical context:
  * 2017: Continuous compilation with Clang Thread Safety analysis enabled (#10866, #10923)
  * 2018: Continuous testing with trapping on signed integer overflows (`-ftrapv`) (#12686)
  * 2018: Continuous testing of use of locale dependent functions (#13041)
  * 2018: Continuous testing of format strings (#13705)
  * 2018: Continuous compilation with MSVC `TreatWarningAsError` (#14151)
  * 2018: Continuous testing under UndefinedBehaviorSanitizer – UBSan (#14252, #14673, #17006)
  * 2018: Continuous testing under AddressSanitizer – ASan (#14794, #17205, #17674)
  * 2018: Continuous testing under ThreadSanitizer – TSan (#14829)
  * 2019: Continuous testing in an unsigned char environment (`-funsigned-char`) (#15134)
  * 2019: Continuous compile-time testing of assumptions we're making (#15391)
  * 2019: Continuous testing of fuzz test cases under Valgrind (#17633, #18159, #18166)
  * 2020: Finally... MemorySanitizer – MSAN! :)

  What is the next step? What tools should we add to CI to keep bugs from entering `master`? :)

ACKs for top commit:
  MarcoFalke:
    ACK 870f0cd

Tree-SHA512: 38327c8b75679d97d469fe42e704cacd1217447a5a603701dd8a58ee50b3be2c10248f8d68a479ed081c0c4b254589d3081c9183f991640b06ef689061f75578
@practicalswift
practicalswift deleted the unsigned-char branch April 10, 2021 19:38
PastaPastaPasta pushed a commit to PastaPastaPasta/dash that referenced this pull request Sep 11, 2021
…ned char environment (-funsigned-char)

0c78e49 tests: Switch one of the Travis jobs to an unsigned char environment (-funsigned-char) (practicalswift)

Pull request description:

  Switch one of the Travis jobs to an unsigned char environment (`-funsigned-char`).

  This will help us catch errors due to code written under the assumption that `char` has the same value range as `signed char`.

  The signedness of `char` is implementation-defined.

  Example:

  ```
  $ uname -a
  Linux […] x86_64 x86_64 x86_64 GNU/Linux
  $ cat foo.cpp
  #include <iostream>

  int main() {
      char c;
      std::cin >> c;
      int i = (unsigned char)c;
      std::cout << i << "\n";
  }
  $ clang++ -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ clang++ -fsigned-char -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ clang++ -funsigned-char -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ cat bar.cpp
  #include <iostream>

  int main() {
      char c;
      std::cin >> c;
      int i = c;
      std::cout << i << "\n";
  }
  $ clang++ -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  -1
  $ clang++ -fsigned-char -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  -1
  $ clang++ -funsigned-char -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  255
  ```

  `gcc` chars:
  * signed: alpha, hppa, ia64, m68k, mips, sh, sparc, x86
  * unsigned: arm, powerpc, s390

  About `-funsigned-char`:

  > Let the type "char" be unsigned, like "unsigned char".
  >
  > Each kind of machine has a default for what "char" should be.  It is either like "unsigned char" by default or like "signed char" by default.
  >
  > Ideally, a portable program should always use "signed char" or "unsigned char" when it depends on the signedness of an object.  But many programs have been written to use plain "char" and expect it to be signed, or expect it to be unsigned, depending on the machines they were written for.
  >
  > This option, and its inverse, let you make such a program work with the opposite default. The type "char" is always a distinct type from each of "signed char" or "unsigned char", even though its behavior is always just like one of those two.

ACKs for top commit:
  laanwj:
    ACK 0c78e49

Tree-SHA512: ba04590415c0bb9a0bbd348623e57068f75274f53da7247d5c5ecad82e365a5b45893a4a491d318e82a8feb6a25f019d46e01990afb33162e2c9740d33a343d7
PastaPastaPasta pushed a commit to PastaPastaPasta/dash that referenced this pull request Sep 12, 2021
…ned char environment (-funsigned-char)

0c78e49 tests: Switch one of the Travis jobs to an unsigned char environment (-funsigned-char) (practicalswift)

Pull request description:

  Switch one of the Travis jobs to an unsigned char environment (`-funsigned-char`).

  This will help us catch errors due to code written under the assumption that `char` has the same value range as `signed char`.

  The signedness of `char` is implementation-defined.

  Example:

  ```
  $ uname -a
  Linux […] x86_64 x86_64 x86_64 GNU/Linux
  $ cat foo.cpp
  #include <iostream>

  int main() {
      char c;
      std::cin >> c;
      int i = (unsigned char)c;
      std::cout << i << "\n";
  }
  $ clang++ -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ clang++ -fsigned-char -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ clang++ -funsigned-char -o foo foo.cpp
  $ echo -e "\xff" | ./foo
  255
  $ cat bar.cpp
  #include <iostream>

  int main() {
      char c;
      std::cin >> c;
      int i = c;
      std::cout << i << "\n";
  }
  $ clang++ -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  -1
  $ clang++ -fsigned-char -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  -1
  $ clang++ -funsigned-char -o bar bar.cpp
  $ echo -e "\xff" | ./bar
  255
  ```

  `gcc` chars:
  * signed: alpha, hppa, ia64, m68k, mips, sh, sparc, x86
  * unsigned: arm, powerpc, s390

  About `-funsigned-char`:

  > Let the type "char" be unsigned, like "unsigned char".
  >
  > Each kind of machine has a default for what "char" should be.  It is either like "unsigned char" by default or like "signed char" by default.
  >
  > Ideally, a portable program should always use "signed char" or "unsigned char" when it depends on the signedness of an object.  But many programs have been written to use plain "char" and expect it to be signed, or expect it to be unsigned, depending on the machines they were written for.
  >
  > This option, and its inverse, let you make such a program work with the opposite default. The type "char" is always a distinct type from each of "signed char" or "unsigned char", even though its behavior is always just like one of those two.

ACKs for top commit:
  laanwj:
    ACK 0c78e49

Tree-SHA512: ba04590415c0bb9a0bbd348623e57068f75274f53da7247d5c5ecad82e365a5b45893a4a491d318e82a8feb6a25f019d46e01990afb33162e2c9740d33a343d7
@bitcoin bitcoin locked as resolved and limited conversation to collaborators Aug 18, 2022
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

8 participants