`cast_ptr_alignment` doesn't see alignment check that follows right after

[nazar-pc]

Summary

The lint is all about alignment, but doesn't take manual alignment checks into consideration

Lint Name

cast_ptr_alignment

Reproducer

I tried this code:

let bytes_ptr = bytes.as_ptr().cast::<u128>();
if !bytes_ptr.is_aligned()
    || bytes.len()
        < size_of::<TransactionHeader>() + size_of::<SerializedTransactionLengths>()
{
    return None;
}

I saw this happen:

warning: casting from `*const u8` to a more-strictly-aligned pointer (`*const u128`) (1 < 16 bytes)
   --> crates/shared/ab-core-primitives/src/transaction.rs:136:25
    |
136 |         let bytes_ptr = bytes.as_ptr().cast::<u128>();
    |                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/main/index.html#cast_ptr_alignment

I tried this code:

if !bytes.as_ptr().cast::<u128>().is_aligned()
    || bytes.len()
        < size_of::<TransactionHeader>() + size_of::<SerializedTransactionLengths>()
{
    return None;
}

I saw this happen:

   --> crates/shared/ab-core-primitives/src/transaction.rs:136:13
    |
136 |         if !bytes.as_ptr().cast::<u128>().is_aligned()
    |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    |
    = help: for further information visit https://rust-lang.github.io/rust-clippy/main/index.html#cast_ptr_alignment
    = note: `-W clippy::cast-ptr-alignment` implied by `-W clippy::pedantic`
    = help: to override `-W clippy::pedantic` add `#[allow(clippy::cast_ptr_alignment)]`

I expected to see this happen: No warning since the code explicitly checks pointer alignment right after casting

Version

rustc 1.100.0-nightly (e7769602a 2026-08-24)
binary: rustc
commit-hash: e7769602aca3770e8d8ea55716becb22e839a579
commit-date: 2026-08-24
host: x86_64-unknown-linux-gnu
release: 1.100.0-nightly
LLVM version: 23.1.0

Additional Labels

No response

[CommanderStorm]

Yes, is one of the known problems with this lint.
We don't do any complex controll flow logic, which is fairly hard to figure out.
Feel free to investigate how to make it smarter.
Is likely fun to make smarter, but might not be as simple 😉

Known problems

Using std::ptr::read_unaligned and std::ptr::write_unaligned or similar on the resulting pointer is fine. Is over-zealous: casts with manual alignment checks or casts like u64 -> u8 -> u16 can be fine. Miri is able to do a more in-depth analysis.

There were a bunch of fixes that went into this lint like #2881 so I am not 100% if the problem 100% exist.

[nazar-pc]

I saw that, I do use unaligned reads/writes too, which I had to suppress in some cases. This example seemed like the most direct and possibly the easiest to fix. While I don't expect it to be fixed any time soon, having an actionable issue to track this narrow case might be useful.

[CommanderStorm]

@rustbot label +E-hard +L-pedantic

[saberoueslati]

@rustbot claim