fix: (Util.cpp) use static_cast<int> on FICLONE, port downstream patch form alpine linux

[MaxPeal]

port downstream patch form alpine linux
enable support for file cloning on ppc64le
"musl uses an int instead of a unsigend long for the ioctl function
prototype, contrary to glibc, since POSIX mandates the former. This
causes a spurious error on ppc64le which can be silenced by casting to
int explicitly.

See https://www.openwall.com/lists/musl/2020/01/20/2"
https://gitlab.alpinelinux.org/alpine/aports/-/blob/de8bce08376a189ad14809d0272bb4c02e74cdf0/main/ccache/ioctl.patch

[jrosdahl]

Looks good to me. We're generally using static_cast<int>(...) rather than (int)..., though.

[MaxPeal]

@jrosdahl Thank you, PR updated as suggested.
must i squash this PR?

[jrosdahl]

must i squash this PR?

No, that's not necessary.

---

I had a closer look now and I'm not yet convinced that the cast is appropriate. The original patch description says "spurious error". Does that mean a real error or a compile-time warning?

On all glibc-based Linux dists the second argument to ioctl should be unsigned long, not int. Is there a risk of errors or warnings in a non-musl case if the value is cast to int?

If there is need to cast to int for musl it spontaneously feels like it should be done only when compiling with musl.

[andypost]

btw Linux using unsigned int https://github.com/torvalds/linux/blob/master/fs/ioctl.c#L823

[ccache-bot]

This issue has been automatically closed due to no response to our request for more information from the original author. Please feel free to leave a comment to reopen the issue if you have more information.