Hacker News Reader

Comment by CyberRabbi | Hacker News Reader

> I won't avoid Unix-ism when writing code (e.g. I'll probably use fork(2)).

It’s probably fine to not avoid most Unix APIs but fork() is truly an exception here. Fork() is not friendly to any third party library you may use because their state may become invalidated after a fork but the library has no way to know if the process has been forked. This is especially bad if the library uses multi threading. The best way to avoid a random unintended bug later on is to only use fork() when you plan to execve() right after.

View on HN

Author here. Good point. I ended up not using fork() without exec(), so that should be fine now, but here is my original plan to use fork():

I wanted to keep a linker process running as a daemon so that it doesn't read the same files over and over again. After loading input files, the linker becomes a daemon and calls fork() to create a worker process. Then the worker process does the rest of linking. In other word, a daemon is a "clean" copy of a linker process image, and each child is specialized for each actual linker invocation.

It turned out that the linker runs much slower with fork() because of the overhead of copy-on-write. You cannot keep a fresh copy of a process just by calling fork() for free. There's a tax associated with it. I tried to workaround, but in the end I had to give up with the fork()-based worker process design.

I wonder if the tax could be reduced with huge pages. Much of the cost of COW is the large number of page faults, but with huge pages you could reduce the number of faults to 1.

> in the end I had to give up with the fork()-based worker process design.

Honestly that's a good thing, or your program just fundamentally couldn't possibly ever work on native Windows!

Or mainframe OSes.

Not my area. Does compiling happen on mainframes these days?

For sure, IBM i, z/OS, and Unisys ClearPath have kept their own programming languages, in addition to Java, C and C++, both on their native environments as well as POSIX application layer.

Modern mainframe development environments are browser based.

Not possibly working on Windows is, for some of us, a feature. I have no wish to support that ecosystem due to the past ethical violations that founded it, and making it impossible to do so would avoid having to deal with it.

What about posix_spawn() with POSIX_SPAWN_USEVFORK? That saves some of the overhead. See eg https://github.com/rtomayko/posix-spawn#benchmarks

posix_spawn is just a wrapper that takes care of setting common parameters for newly forked instances (eg pgrp) and prevents you from doing things that might be overly unsafe or could break vfork from being used in its optimized form. It’s implanted at the libc level, so it’s not a magic syscall that moves the burden of process spawning to the kernel.

This doesn’t change your core point but just to be thorough, it’s actually a system call on a few systems, notably macOS.

Yes, but with that define it wraps vfork instead of fork.

I can't think of any reason a linker needs to use many libraries, especially not one that start threads. A linker reads files and writes a file. It's a single-purpose tool, not a monolithic app with many dependencies.

Either way, threads should be controlled by the application and not libraries. Well-written libraries like sqlite and Lua are parameterized by I/O and concurrency. They don't read files and start threads behind your back.

I agree with you on both points but when building large complex applications (or tools!) this stuff tends to unintentionally creep up over time and it’s not fun to debug non-deterministic bugs due to an opaquely broken implicit contract between you and a poorly written library that got pulled in transitively. Sometimes those libraries are unavoidable closed source system libraries (https://bugs.python.org/issue33725) and the issue ends up being not easily fixable. It’s just risk minimization.

It's a moot point since the author already said he's not using fork(), but C and C++ apps don't transitively pull in dependencies, especially in the open source world.

That kind of thinking comes from node.js and Rust as far as I can tell.

There is never ever a situation where a program like GCC or Clang will acquire a third party dependency without a commit that explicitly changes the build system.

The Python example isn't really relevant because Python modules are shared libraries, but a linker doesn't have or need shared library plugins (i.e. dynamically linking against arbitrary machine code at runtime).

Of course C and C++ apps transitively pull in dependencies.

First, pkg-config exists and most projects on Unixy OSes use it these days. It doesn't really make much sense to argue "it all has to be specified on the command-line" when the command-line the developer cares about looks like this:

gcc `pkg-config --cflags gtk+-3.0` -o test test.c `pkg-config --libs gtk+-3.0`

(That adds things like `-lcairo` which aren't at all visible at first glance)

Second, why do you think things like libwayland-client, libfreetype, and libexpat show up in the ldd output? They're certainly not in the pkg-config output.

Likewise, nothing says a library has to bump its major version number when it adds a new backend dependency, which means that a distro may add a transitive dependency through a shared library update that doesn't correspond to you bumping your version.

(You can gain new backend dependencies without even recompiling anything you're responsible for.)

Heck, this post explicitly touches on that disconnect between what you add to your build plan and what ldd lists for C and C++ codebases.

https://wiki.alopex.li/LetsBeRealAboutDependencies

That is a good point... I overstated the case, although I would say the GCC and Clang examples in particular are true. I guess the stronger argument is that compilers are more of a special case low in the dependency stack, as opposed to apps (and especially GUI apps). Also, I really hope none of those dependencies start long-lived threads without the app's knowledge, though there's nothing stopping them from doing so.

In my example, using Apple’s system carbon libraries at the C level transitively pulls in their multithreading framework. It does happen. Everything that can break will break, that attitude is necessary to build robust software.

Forget libraries, a static linker (back in the day) only needed something like half-a-dozen different syscalls, not even a full userland or anything as bougie as libraries.

(if it's unclear, this is strong agreement with chubot. If you make the problem harder than it has to be, you have only yourself to blame...)

Reply on news.ycombinator.com