An almost catastrophic OpenZFS bug and the humans that made it (despairlabs.com)

At the same time, the tooling has gotten much better in the last years.

Clang-analyzer is fast enough to run as part of the CI. Newer gcc also give quite a few more warnings for unused results.

My recommendation to the project is to

- Remove all compiler warnings and enable warning-as-error

- Increase the coverage of unit tests to >80%

That is a lot of work. But that's what is required for high-criticality systems engineering.

    typedef struct { size_t size; } PhysicalSize;
    typedef struct { size_t size; } AllocatedSize;

    PhysicalSize psize = { 123 }; // or { .size = 123 }
    AllocatedSize asize = { 234 };

    psize = asize; // error

IME overly strong typing is a double-edged sword though, on one hand it makes the code more robust, but on the other hand also more 'change resistant'.

I still would like to see a strong typedef variant in C so that the struct wrapping wouldn't be needed, e.g.:

https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3320.htm

(yes this is very much an armchair opinion, I mostly do front-end JS for a living)

It's the kind of bug that makes you stop breathing for a brief moment. So I ran this function through Gemini 2.5 Pro, ChatGPT o3 and Grok 3. No context other than the source code fragment was provided. All three also clearly identified the problem.

There is no longer a reason for trivial flaws such as this surviving to release code. We're past that now, and that's an astonishing thing.

These are the questions I ponder: Should we consider the possibility that the ongoing, incomplete and decades long pursuit of "better" languages is now misguided? Or, that perhaps "better" might now mean whatever properties make code easier for LLMs to analyze and validate against specifications, as opposed to merely preventing the flaws humans are prone to make?

    struct TypeA { int inner; };
    struct TypeB { int inner; };

    struct TypeA a1, a2;
    // whoops
    TypeB result = {.inner = a1.inner + a2.inner};

The use case I felt worked best was seperating out UTC datetime vs localised datetime structs - mainly since it's something you already had to use function calls to add/subtract from, you couldn't play with raw values.

If you want free software I recommend IKOS - a is a sound static analyzer for C/C++ developed at NASA. Checks: https://github.com/NASA-SW-VnV/ikos/blob/master/analyzer/REA... Numerical abstract domains: https://github.com/NASA-SW-VnV/ikos/blob/master/analyzer/REA...

Commercial tool like Astree https://www.absint.com/astree/index.htm if you have money.

It seems you are dooming your project the minute you start ignoring your first warning.

I would have expected that the compiler should complain that psize is computed but unused.

Why isn't -Wunused-variable enabled in OpenZFS?

>The provided code snippet is a function that converts an allocation size (asize) into the largest physical size (psize) that can be safely written to an allocation for a given virtual device (vdev). Here are some observations and potential issues in the code:

Return Value: The function currently returns asize instead of psize. This seems to be a mistake, as the purpose of the function is to compute and return the psize. The correct return statement should be:

    return psize;

Variable Initialization: The variable psize is calculated based on asize, ashift, cols, and nparity. Ensure that these values are correctly initialized and that vdev_raidz_get_logical_width returns a valid value for cols. If cols is zero, it could lead to division by zero in the DIV_ROUND_UP macro.

Assertions: The assertion ASSERT0(asize % (1 << ashift)); checks that asize is a multiple of 2^ashift. If this assertion fails, it will cause the program to terminate. Ensure that this condition is always met before calling this function, or consider handling this case more gracefully.

Macro Usage: The DIV_ROUND_UP macro should be checked to ensure it handles edge cases correctly, such as when psize is zero. If psize is zero, the division could lead to unexpected behavior.

Code Readability: Consider adding more comments to explain the logic behind the calculations, especially for someone who may not be familiar with the RAID-Z implementation.

Error Handling: There is no error handling in the function. Depending on the context in which this function is used, it might be beneficial to add checks for invalid input values and return an error code or handle exceptions accordingly.

Here is a revised version of the function with the return value corrected:

.....(Snipped for formatting issues)

Make sure to test the function thoroughly to ensure it behaves as expected under various conditions.

I have been playing with NetBSD for a bit and I discovered their FS has a log parameter, using that on mount to me makes their file system almost as good as ext4. Interesting to me it is not advertised too much.