Size mismatch in write() call (in test code)

[rgmiller]

System information

Type	Version/Name
Operating System	Ubuntu
OS Version	22.04
Architecture	x86_64
UnifyFS Version	latest from git

Describe the problem you're observing

https://github.com/LLNL/UnifyFS/blob/f95abed7c1aa91864d047b7c0ad38ac05e5719a7/t/sys/write-read.c#L379

The call to write() on this line specifies a count of 300, but the data to be written is only 2 bytes long. Newer versions of gcc (at least 11.3.0, possibly somewhat earlier) will issue a warning about this. And by default, we build with -Werror enabled, so the warning turns into an error and the build stops.

Temporary work-around is to compile with -Wno-error=stringop-overread.

Describe how to reproduce the problem

Just look at the code.

Include any warning or errors or releveant debugging data

The exact error from GCC:

./../t/sys/write-read.c: In function ‘write_pre_existing_file_test’:
../../t/sys/write-read.c:379:16: error: ‘write’ reading 300 bytes from a region of size 2 [-Werror=stringop-overread]
  379 |     rc = (int) write(fd, "a", 300);
      |                ^~~~~~~~~~~~~~~~~~~

[adammoody]

Good catch. I think maybe that buf was intended for this parameter instead of "a"?

https://github.com/LLNL/UnifyFS/blob/f95abed7c1aa91864d047b7c0ad38ac05e5719a7/t/sys/write-read.c#L364

[rgmiller]

That buffer is initialized to all zeros. Is that sufficient for this test? Or should we have different values in that buffer?

Also, the code uses the literal 300 in a couple of different places. Should we replace that with a #define or const int?

[MichaelBrim]

@rgmiller Based on what it's testing, I don't think the zeroes are a problem. I agree that we should use a constant or #define for the buffer size.