rust-lang / rust-lang/rust

Bounds check not elided for `slice::split_at` after loop

Open
#143,027 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

C-optimization T-compiler
Dominant language
Rust
Stars
119k
Forks
16.1k
PR merge metrics
PR metrics pending

Description

split_at panics when mid > len.

In the following code for trimming the start of a slice, i will only be less than or equal to len but the information seems lost after the loop.

#[inline(never)]
pub fn src(s: &[u8]) -> &[u8] {
    let mut i = 0;
    while i < s.len() {
        if s[i] != 0 {
            break;
        }
        i += 1;
    }
    s.split_at(i).1
}

#[inline(never)]
pub const fn tgt(s: &[u8]) -> &[u8] {
    let mut i = 0;
    while i < s.len() {
        if s[i] != 0 {
            return s.split_at(i).1;
        }
        i += 1;
    }
    &[]
}

https://rust.godbolt.org/z/EWPbf4Pdd

Assembly outputsrc function

example::src::he3649242b6e21a77:
        xor     eax, eax
        test    rsi, rsi
        je      .LBB0_6
.LBB0_1:
        cmp     byte ptr [rdi + rax], 0
        jne     .LBB0_5
        inc     rax
        cmp     rsi, rax
        jne     .LBB0_1
        mov     rax, rsi
        jmp     .LBB0_6
.LBB0_5:
        cmp     rax, rsi
        ja      .LBB0_4
.LBB0_6:
        add     rdi, rax
        sub     rsi, rax
        mov     rax, rdi
        mov     rdx, rsi
        ret
.LBB0_4:
        sub     rsp, 56
        lea     rax, [rip + .Lanon.7d93f8663e2093497ca272392f600a6e.1]
        mov     qword ptr [rsp + 8], rax
        mov     qword ptr [rsp + 16], 1
        mov     qword ptr [rsp + 24], 8
        xorps   xmm0, xmm0
        movups  xmmword ptr [rsp + 32], xmm0
        lea     rsi, [rip + .Lanon.7d93f8663e2093497ca272392f600a6e.3]
        lea     rdi, [rsp + 8]
        call    qword ptr [rip + core::panicking::panic_fmt::hf1310915e65e5ab4@GOTPCREL]

.Lanon.7d93f8663e2093497ca272392f600a6e.0:
        .ascii  "mid > len"

.Lanon.7d93f8663e2093497ca272392f600a6e.1:
        .quad   .Lanon.7d93f8663e2093497ca272392f600a6e.0
        .asciz  "\t\000\000\000\000\000\000"

.Lanon.7d93f8663e2093497ca272392f600a6e.2:
        .asciz  "/app/example.rs"

.Lanon.7d93f8663e2093497ca272392f600a6e.3:
        .quad   .Lanon.7d93f8663e2093497ca272392f600a6e.2
        .asciz  "\020\000\000\000\000\000\000\000\n\000\000\000\007\000\000"

tgt function

example::tgt::hfc7b3a5f713674dc:
        mov     eax, 1
        test    rsi, rsi
        je      .LBB0_1
        mov     rdx, rsi
.LBB0_4:
        cmp     byte ptr [rdi], 0
        jne     .LBB0_5
        inc     rdi
        dec     rdx
        jne     .LBB0_4
.LBB0_1:
        xor     edx, edx
        ret
.LBB0_5:
        mov     rax, rdi
        ret

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the provided src and tgt reproducer and compare their assembly output using the linked Rust Compiler Explorer example. Investigate how slice::split_at handles the loop's range information, and consider the work complete when the equivalent src case no longer emits the unnecessary bounds check while preserving its behavior.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
compilers, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.