Update ReadDir::next in std::sys::pal::unix::fs to use &raw const (*p).field instead of p.byte_offset().cast()
#134678
Conversation
…ptr).field` instead of `ptr.offset(...).cast()`. Also, the macro is only called three times, and all with the same local variable entry_ptr, so just use the local variable directly, and rename the macro to entry_field_ptr.
rustbot
added
O-unix
| d_ino: *offset_ptr!(entry_ptr, d_ino) as u64, | ||
| d_ino: *entry_field_ptr!(d_ino) as u64, | ||
| #[cfg(not(any( | ||
| target_os = "solaris", | ||
| target_os = "illumos", | ||
| target_os = "aix", | ||
| target_os = "nto", | ||
| )))] | ||
| d_type: *offset_ptr!(entry_ptr, d_type) as u8, | ||
| d_type: *entry_field_ptr!(d_type) as u8, |
There was a problem hiding this comment.
Any idea why the as casts are here? Looking at libc the ino types are just typedefs so it seems they wouldn't be needed. https://github.com/rust-lang/libc/blob/1867bf30eb9d22d3f8e86c2a8c25cf92fe5580da/src/unix/linux_like/linux/mod.rs#L1474-L1480, https://github.com/rust-lang/libc/blob/1867bf30eb9d22d3f8e86c2a8c25cf92fe5580da/src/unix/linux_like/linux/mod.rs#L11, https://github.com/rust-lang/libc/blob/1867bf30eb9d22d3f8e86c2a8c25cf92fe5580da/src/unix/mod.rs#L9
There was a problem hiding this comment.
I don't know why the as casts are there, no. dirent64 is imported as aliases from various different platforms, so maybe the as casts were added just to make sure it worked (or at least compiled) on any weird platforms where the fields were different type?
Removing the as casts doesn't make any platforms I've tried with ./x.py check library/std complain though1, so maybe they could be removed?
Footnotes
-
x86_64-unknown-linux-gnu,aarch64-unknown-linux-gnu,aarch64-unknown-freebsd,arm-linux-androideabi,armv7-linux-androideabi,i686-linux-android↩
library/std/src/sys/pal/unix/fs.rs
Outdated
| // To make sure the field actually exists and is visible, | ||
| // and we aren't silently doing any Deref coercion. | ||
| const _: usize = mem::offset_of!(dirent64, $field); | ||
| &raw const (*entry_ptr).$field |
There was a problem hiding this comment.
The deref concern makes sense to me (and is scary in general), but what happens if the first clause happens i.e. the field doesn't exist or isn't visible? Afaik, &raw doesn't allow bypassing field visibility.
There was a problem hiding this comment.
I think that is true, yes, any error about a field being missing or too private should cause an error on both the offset_of! and the &raw equally. I could remove the the field actually exists and is visible, and part of the comment if that would be clearer.
Edit: I've updated the comment to only talk about Deref coercion.
Since rust-lang/reference#1387 and #117572,
&raw mut (*p).field/addr_of!((*p).field)is defined to have the same inbounds preconditions asptr::offset/ptr::byte_offset. I.e.&raw const (*p).fielddoes not require thatp: *const Tpoint to a fullsize_of::<T>()bytes of memory, only thatp.byte_add(offset_of!(T, field))is defined.The old comment "[...] we don't even get to use
&raw const (*entry_ptr).d_namebecause that operation requires the full extent of *entry_ptr to be in bounds of the same allocation, which is not necessarily the case here [...]" is now outdated, and the code can be simplified to use&raw const (*entry_ptr).field.There should be no behavior differences from this PR.
The
: *const dirent64on line 716 and theconst _: usize = mem::offset_of!(dirent64, $field);and comment on lines 749-751 are just sanity checks and should not affect semantics.Since the
offset_ptr!macro is only called three times, and all with the same local variable entry_ptr, I just used the local variable directly in the macro instead of taking it as an input, and renamed the macro toentry_field_ptr!.The whole macro could also be removed and replaced with just using
&raw const (*entry_ptr).fieldin the three places, but the comments on the macro seemed worthwhile to keep.