RHash 1.4.6 crash

Windows 7 x64, Rhash 1.4.6 (Full or English only, 32- or 64-bit).

$ rhash_1.4.6.exe
Usage: rhash [OPTION...] <FILE>...
Run `rhash --help' for more help.

$ rhash_1.4.6.exe --help

However, 1.4.5 works as expected.

$ rhash_1.4.5.exe --help
...
      --template=<file> Load a printf-like template from the <file>
  -p, --printf=<format string>  Format and print message digests.
                   See the RHash manual for details.

Jun 11 '25 23:06 sergeevabc

Hello?

Oct 20 '25 12:10 sergeevabc

It seems RHash, since v1.4.6, started to use some function inaccessible in Windows 7. I'm currently unable to investigate this due to other commitments. To help resolve this, can you run it under a debugger to identify the problematic function?

Oct 28 '25 16:10 rhash

Алексей, я с радостью предоставлю вам детали по этому кейсу, но как выполнить то, о чём вы просите, не знаю. Очевидно. мне нужен дебаггер, о котором вы говорите. Если он есть под рукой и мало весит, киньте ссылку, пожалуйста.

Oct 29 '25 20:10 sergeevabc

RHash 1.4.6 introduced native SHA-1 support that depends on CPU Intel instruction SHA-1 and may not run on older CPUs without this instruction (crash with SIGILL, Illegal Instruction). I have encountered with same issue after upgrading MSYS2, running CMake (which uses RHash) will crash with that said error code. Make sure that your CPU actually supports it.

Nov 09 '25 21:11 thehatkid

@thehatkid My CPU has no native SHA1 support, but I don't calculate SHA1 in this case, just run it with --help switch.

Nov 09 '25 21:11 sergeevabc

Can't reproduce the crash for 32-bit RHash (rhash-1.4.6-win32.zip and rhash-1.4.6-english-win32.zip) on Win7 (Ver 6.1 Build 7601).

For 64-bit RHash there is another error:

Regarding SHA-NI (Intel native SHA1), RHash must check for CPU support before using it.

Nov 23 '25 12:11 rhash

According to Release notes at https://sourceforge.net/projects/rhash/files/rhash/1.4.6/

Win32 version requires: Win XP/2K/Vista/7/8/10.
Win64 version requires: Win 8/10 64-bit.

So for 64-bit RHash it is not a bug.

Nov 23 '25 12:11 rhash

My initial bug report clearly says “Rhash 1.4.6 (Full or English only, 32- or 64-bit)”. It means this very bug is reproduced with 32-bit version as well.

Is there really so much new code between 1.4.5 and 1.4.6 that you can't go through it and find the problem? Or try to compile RHash using the previous toolchain to check if that's the case? RHash calculated hashes fine for YEARS, so what's happened now?

Nov 23 '25 12:11 sergeevabc