regression: unaligned references to packed fields are now an error

[Mark-Simulacrum]

This is an expected strengthening of a future compatibility lint, mostly filing an issue for tracking purposes (i.e., release notes). I believe no action is expected.

• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/Adam-Gleave.multiboot2-rs/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/ColonelPhantom.fe_chess/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/NilSet.dnsclient/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/Olangu.bindgen-bug/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/SevenCubedLabs._64_gfx/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/SheSnake.game_client/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/TomBebb.looa/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/chengzhycn.rself/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/coeuvre.handmade-rs/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/ctsrc.persistence/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/evan-brass.osu-graphics/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/focus-zhaos.UringForStratoVirt/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/ican2002.stratorvirt/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/jthornber.lvm-rust/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/mars-research.binunpack-rs/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/mataha.gunz-replay/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/mickvangelderen.fins-rust/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/mrandri19.opengl-compute-shaders/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/nishalkulkarni.btrfs-walk-tut/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/numerodix.pcitools/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/project-serum.governance-registry/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/pudh4418.udfstd/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/rbreslow.jpi-parser/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/svenslaggare.stackjit-rs/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/tlent.learn-opengl-rs/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/tyler569.tcp/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/gh/vbendeb.exifgeo/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/SOEM-sys-0.2.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/aeron-rs-0.1.3/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/apricot-client-0.15.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/autd3-soem-link-1.10.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/bebop-2.4.9/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/blake2b-rs-0.2.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/brash-0.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/btrfs-1.2.2/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/btrfs2-1.2.2/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/dbs-utils-0.2.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/ejdb-sys-0.3.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/elrond-1.0.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/wavetable-0.1.2/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/vuk-0.0.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/vec-2-10-10-10-0.1.2/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/v4l2-sys-1.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/v4l2-sys-mit-0.2.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/usrsctp-sys-0.2.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/usb-host-0.1.3/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/uhid-virt-0.0.5/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/uhid-sys-1.0.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/uhid-fs-0.0.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/uefi-ffi-0.1.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/typeinfo-0.1.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/tokio-linux-uhid-0.3.2/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/tmux_plugin-0.0.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/sprint-dir-0.0.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/solo5-sys-0.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/simple_bitfield-0.1.8/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/rplidar_drv-0.6.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/rdisk_shared-0.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/pfctl-0.4.5/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/elrond-1.0.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/enet-sys-1.0.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/ethercat-soem-sys-0.2.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/evoke-core-0.2.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/freeimage-0.2.2/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/gjio-0.1.3/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/hactool-sys-0.4.4/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/hdf5-derive-0.8.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/hydrogen-0.1.5/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/iwlib_sys-0.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/j2534-0.3.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/libdvb-0.3.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/lightway-sys-0.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/linux-loader-0.8.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/linux-nvme-sys-0.3.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/lwext4-sys-0.1.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/nfq-0.2.5/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/pe-0.1.1/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/persistence-0.0.5/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.69-3/beta-2023-03-18/reg/pfctl-0.4.5/log.txt

[apiraino]

ok, then I will remove the I-prioritize flag, too. IIUC, these changes will be in the release notes and crates will need to adapt to the new lints

@rustbot label -I-prioritize

[glandium]

Having hit this one in https://github.com/mozilla/authenticator-rs/blob/abde4645a4711e29f1a481802776d5079c559eb1/src/windows/winapi.rs#L224, I'm wondering if this new error is not too restrictive.
Here is a smaller testcase that is not x86 windows specific:

#[repr(packed)]
pub struct Foo {
    a: usize,
    b: [u16; 1],
}

pub fn foo(f: &Foo) {
    bar(f.b.as_ptr());
}

pub fn bar(_ptr: *const u16) {}

How can accessing the b field ever be unaligned?

[Mark-Simulacrum]

Cc @RalfJung @rust-lang/lang (also nominated for lang discussion; this will hit stable in ~2 weeks)

FWIW it seems like that example is a little misleading, the as_ptr call takes &self, so just &f.b as the argument is equivalent.

repr(packed) doesn't alter field reordering by itself I think (and the reference seems to agree), so the [u16; 1] could be at the front of the struct.

But adding repr(C) doesn't prevent the error, and I think with that this would only be unaligned on 8-bit platforms (which I'm not sure we support; I'm also not sure if this error is target specific, I assume so). My guess is that the implementation isn't trying to be smart about this, just always errors on any field with lowered alignment requirements as a result of packed.

[glandium]

so just &f.b as the argument is equivalent.

it's actually not, because bar takes a *const u16, and &f.b is a [u16; 1]. .as_ptr() returns a *const u16 (per https://doc.rust-lang.org/std/primitive.slice.html#method.as_ptr)

FWIW the original error happens on x86 windows only because that's windows specific code, and only x86 has repr(packed) on the relevant struct. It also does have a #[repr(C)] that was hidden in a macro in winapi, which I had missed.

[Mark-Simulacrum]

The return type doesn't seem relevant? as_ptr has &self, so for the purposes of the error what happens after that doesn't matter; bar isn't relevant. (It would change the type of the expression to move to &, and stop compiling if that's the only change made)

[scottmcm]

How can accessing the b field ever be unaligned?

I'm not following what you mean by this example.

I switched the & to addr_of to make it compile, put a read of the pointer in bar as a way to test alignment, then wrote some basic entirely safe code to demonstrate that it absolutely can end up unaligned:

error: Undefined Behavior: accessing memory with alignment 1, but alignment 2 is required
  --> src/main.rs:13:14
   |
13 |     unsafe { ptr.read() };
   |              ^^^^^^^^^^ accessing memory with alignment 1, but alignment 2 is required

https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=246744a5d3066b205afeb097bbcf053e

[glandium]

The return type doesn't seem relevant? as_ptr has &self, so for the purposes of the error what happens after that doesn't matter; bar isn't relevant. (It would change the type of the expression to move to &, and stop compiling if that's the only change made)

Sure, for E0793, the return type is irrelevant.

How can accessing the b field ever be unaligned?

I'm not following what you mean by this example.

I switched the & to addr_of to make it compile, put a read of the pointer in bar as a way to test alignment, then wrote some basic entirely safe code to demonstrate that it absolutely can end up unaligned:

error: Undefined Behavior: accessing memory with alignment 1, but alignment 2 is required
  --> src/main.rs:13:14
   |
13 |     unsafe { ptr.read() };
   |              ^^^^^^^^^^ accessing memory with alignment 1, but alignment 2 is required

https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=246744a5d3066b205afeb097bbcf053e

Add repr(C) to the struct as mentioned in the more recent messages.

[glandium]

Interestingly, miri doesn't barf if you use addr_of!(f.b[0]).

Scrap that, it does.