openssh bugs - Oct 2022 - [Bug 3482] New: Compilation fails with clang

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

            Bug ID: 3482
           Summary: Compilation fails with clang
           Product: Portable OpenSSH
           Version: 9.1p1
          Hardware: amd64
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P5
         Component: Miscellaneous
          Assignee: unassigned-bugs at mindrot.org
          Reporter: julien at julien-bigot.fr

This issue is similar to
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98733 or
https://bugs.llvm.org/show_bug.cgi?id=23562 

Clang fails to compile `bsd-asprintf.c` that's part of openssh. It
seems to turn the `asprintf` symbol into a macro when using `-O2`
`-D_FORTIFY_SOURCE=2` + `#include includes.h`.

More detail here: https://github.com/spack/spack/issues/33131

```
/opt/spack/lib/spack/env/clang/clang -g -O2 -pipe
-Wunknown-warning-option -Qunused-arguments -Wall -Wpointer-arith
-Wuninitialized -Wsign-compare -Wformat-security
-Wsizeof-pointer-memaccess -Wno-pointer-sign -Wno-unused-result
-Wmisleading-indentation -Wbitwise-instead-of-logical
-fno-strict-aliasing -mretpoline -D_FORTIFY_SOURCE=2 -ftrapv
-fzero-call-used-regs=all -fno-builtin-memset -fstack-protector-strong 
 -fPIC -I. -I.. -I. -I./..  -D_XOPEN_SOURCE=600 -D_BSD_SOURCE
-D_DEFAULT_SOURCE
-I/opt/install/linux-ubuntu18.04-x86_64/clang-15.0.0/krb5-1.19.3-riaeukuk7bbucxf5gmp6jeoh2ubhkrul/include
-I/opt/install/linux-ubuntu18.04-x86_64/clang-15.0.0/krb5-1.19.3-riaeukuk7bbucxf5gmp6jeoh2ubhkrul/include
-DHAVE_CONFIG_H -c bsd-asprintf.c


bsd-asprintf.c:86:5: error: expected parameter declarator
int asprintf(char **str, const char *fmt, ...)
    ^
/usr/include/x86_64-linux-gnu/bits/stdio2.h:199:24: note: expanded from
macro 'asprintf'
  __asprintf_chk (ptr, __USE_FORTIFY_LEVEL - 1, VA_ARGS)
                       ^
/usr/include/features.h:382:31: note: expanded from macro
'__USE_FORTIFY_LEVEL'
#  define __USE_FORTIFY_LEVEL 2
                              ^
bsd-asprintf.c:86:5: error: expected ')'
/usr/include/x86_64-linux-gnu/bits/stdio2.h:199:24: note: expanded from
macro 'asprintf'
  __asprintf_chk (ptr, __USE_FORTIFY_LEVEL - 1, VA_ARGS)
                       ^
/usr/include/features.h:382:31: note: expanded from macro
'__USE_FORTIFY_LEVEL'
#  define __USE_FORTIFY_LEVEL 2
                              ^
bsd-asprintf.c:86:5: note: to match this '('
/usr/include/x86_64-linux-gnu/bits/stdio2.h:199:18: note: expanded from
macro 'asprintf'
  __asprintf_chk (ptr, __USE_FORTIFY_LEVEL - 1, VA_ARGS)
                 ^
bsd-asprintf.c:86:5: error: conflicting types for '__asprintf_chk'
int asprintf(char **str, const char *fmt, ...)
    ^
/usr/include/x86_64-linux-gnu/bits/stdio2.h:199:3: note: expanded from
macro 'asprintf'
  __asprintf_chk (ptr, __USE_FORTIFY_LEVEL - 1, VA_ARGS)
  ^
/usr/include/x86_64-linux-gnu/bits/stdio2.h:158:12: note: previous
declaration is here
extern int __asprintf_chk (char **__restrict __ptr, int __flag,
           ^
bsd-asprintf.c:92:15: error: use of undeclared identifier 'fmt'
        va_start(ap, fmt);
                     ^
bsd-asprintf.c:93:23: error: use of undeclared identifier 'fmt'
        ret = vasprintf(str, fmt, ap);
                             ^
5 errors generated.
```

-- 
You are receiving this mail because:
You are watching the assignee of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

Damien Miller <djm at mindrot.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |djm at mindrot.org

--- Comment #1 from Damien Miller <djm at mindrot.org> ---
This code should only be compiled when libc fails to provide
asprintf(), which most modern ones do (inc. glibc). So the problem is
more likely configure failing to detect its presence.

Can you search though config.log to see what went wrong?

-- 
You are receiving this mail because:
You are watching someone on the CC list of the bug.
You are watching the assignee of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

Darren Tucker <dtucker at dtucker.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dtucker at dtucker.net

--- Comment #2 from Darren Tucker <dtucker at dtucker.net> ---
The log in the referenced bug does in fact detect asprintf and set
"#define HAVE_ASPRINTF 1" so that code should not even be compiled on
that configuration:

configure:11975: /opt/spack/lib/spack/env/clang/clang -o conftest -g
-O2 -pipe -Wunknown-warning-option -Qunused-arguments -Wall
-Wpointer-arith -Wuninitialized -Wsign-compare -Wformat-security
-Wsizeof-pointer-memaccess -Wno-pointer-sign -Wno-unused-result
-Wmisleading-indentation -Wbitwise-instead-of-logical
-fno-strict-aliasing -mretpoline -D_FORTIFY_SOURCE=2 -ftrapv
-fzero-call-used-regs=all -fno-builtin-memset -fstack-protector-strong
-fPIE  -D_XOPEN_SOURCE=600 -D_BSD_SOURCE -D_DEFAULT_SOURCE 
-Wl,-z,relro -Wl,-z,now -Wl,-z,noexecstack -fstack-protector-strong
-pie conftest.c -lutil -lz  >&5
conftest.c:151:18: warning: format string missing [-Wformat]
return asprintf ();
       ~~~~~~~~  ^
1 warning generated.
configure:11975: $? = 0
configure:11975: result: yes
[...]
| #define HAVE_ASPRINTF 1

I am also not able to reproduce on Fedora 36 with a locally built clang
15:

$ /opt/clang-15.0.1/bin/clang --version
clang version 15.0.1 (https://github.com/llvm/llvm-project.git
b73d2c8c720a8c8e6e73b11be4e27afa6cb75bdf)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /opt/clang-15.0.1/bin

Is something modifying config.h, or is it picking up a config.h from
some other location?

-- 
You are receiving this mail because:
You are watching someone on the CC list of the bug.
You are watching the assignee of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

--- Comment #3 from Julien Bigot <julien at julien-bigot.fr> ---
Hi, I am not the expert of spack packaging of openssh. I only reported
the issue there and the spack packager pointed me here as he considered
the issue to be upstream.

The config.log is attached to the spack issue:
https://github.com/spack/spack/files/9741543/config.log
in there, I see ac_cv_func_asprintf=yes

As you can see in this issue
https://github.com/spack/spack/issues/33131 , I provided docker image
you can use to reproduce the bug:
ghcr.io/pdidev/test_env/builder:latest-clang-2e69267
In this image, you can just run:
```
spack -e pdienv add -l mpi openssh
spack -e pdienv install --show-log-on-error --fail-fast
```

I don't see why the config.h would be tampered with and looking at the
spack openssh "recipe",
https://github.com/spack/spack/blob/develop/var/spack/repos/builtin/packages/openssh/package.py
I don't see anything pointing to it.

-- 
You are receiving this mail because:
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

haampie <me at harmenstoppels.nl> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |me at harmenstoppels.nl

--- Comment #4 from haampie <me at harmenstoppels.nl> ---
openssh undefs it here:

/*
 * Don't let systems with broken printf(3) avoid our replacements
 * via asprintf(3)/vasprintf(3) calling libc internally.
 */
#if defined(BROKEN_SNPRINTF)
# undef HAVE_VASPRINTF
# undef HAVE_ASPRINTF
#endif

given BROKEN_SNPRINTF=1 in the config.log.

And it seems like the relevant configure test is broken?

| #include <stdio.h>
| #include <stdlib.h>
| #include <string.h>
| #ifdef HAVE_SNPRINTF
| main()
| {
|       char buf[50];
|       char expected_out[50];
|       int mazsize = 50 ;
| #if (SIZEOF_LONG_INT == 8)
|       long int num = 0x7fffffffffffffff;
| #else
|       long long num = 0x7fffffffffffffffll;
| #endif
|       strcpy(expected_out, "9223372036854775807");
|       snprintf(buf, mazsize, "%lld", num);
|       if(strcmp(buf, expected_out) != 0)
|               exit(1);
|       exit(0);
| }
| #else
| main() { exit(0); }
| #endif

there's no `int main` and it errors with 

conftest.c:367:1: error: type specifier missing, defaults to 'int'; ISO
C99 and later do not support implicit int [-Wimplicit-int]

-- 
You are receiving this mail because:
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

--- Comment #5 from haampie <me at harmenstoppels.nl> ---
It's clang 15, see
https://github.com/llvm/llvm-project/commit/7d644e1215b376ec5e915df9ea2eeb56e2d94626
or
https://releases.llvm.org/15.0.0/tools/clang/docs/ReleaseNotes.html#improvements-to-clang-s-diagnostics

-- 
You are receiving this mail because:
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

Darren Tucker <dtucker at dtucker.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |3480
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #6 from Darren Tucker <dtucker at dtucker.net> ---
Applied https://github.com/openssh/openssh-portable/pull/349, thanks.


Referenced Bugs:

https://bugzilla.mindrot.org/show_bug.cgi?id=3480
[Bug 3480] tracking bug for openssh-9.2
-- 
You are receiving this mail because:
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.

https://bugzilla.mindrot.org/show_bug.cgi?id=3482

Damien Miller <djm at mindrot.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #7 from Damien Miller <djm at mindrot.org> ---
OpenSSH 9.3 has been released. Close resolved bugs

-- 
You are receiving this mail because:
You are watching the assignee of the bug.
You are watching someone on the CC list of the bug.