fix(android/aarch64): use `libc::ucontext_t` from Android NDK by StackOverflowExcept1on · Pull Request #5189 · rust-lang/libc

StackOverflowExcept1on · 2026-06-21T10:53:41Z

Resolves #5159

https://android.googlesource.com/platform/bionic/+/refs/tags/ndk-r29/libc/include/sys/ucontext.h#102

It would be good to backport this to libc-0.2

r? @tgross35

rustbot · 2026-06-21T10:53:44Z

Some changes occurred in an Android module

cc @maurer

teor2345 · 2026-06-25T12:41:58Z

+        pub uc_sigmask__c_anonymous_union: __c_anonymous_uc_sigmask,
+        /* The kernel adds extra padding after uc_sigmask to match
+         * glibc sigset_t on ARM64. */
+        __padding: Padding<[c_char; 128 - size_of::<crate::sigset_t>()]>,


Why is this subtracting sigset_t (32 bit?), but the comment refers to ARM64?

This calculation seems weird for either 32-bit or 64-bit, and might need further comments:

the union is the size of sigset64_t (assuming it's bigger) but we subtract sigset_t (the 32-bit size) from the padding

View changes since the review

https://android.googlesource.com/platform/bionic/+/refs/tags/ndk-r29/libc/include/sys/ucontext.h#102

teor2345

I'm not an expert here, but I have coded a lot of C, and some parts of this seem weird without further context

View changes since this review

teor2345 · 2026-06-25T12:44:44Z

+    if #[cfg(feature = "extra_traits")] {
+        impl PartialEq for __c_anonymous_uc_sigmask {
+            fn eq(&self, other: &__c_anonymous_uc_sigmask) -> bool {
+                unsafe { self.uc_sigmask == other.uc_sigmask }


Is this sound and correct if the 64-bit sigmask was initialised in the union?

if the 64-bit sigmask is larger and has no padding, then the equality check misses some bytes

if the 32-bit sigmask is larger (unlikely), then we're comparing uninitialised bytes

if there's padding, then does it need to be ignored?

teor2345 · 2026-06-25T12:45:06Z

+        impl Eq for __c_anonymous_uc_sigmask {}
+        impl hash::Hash for __c_anonymous_uc_sigmask {
+            fn hash<H: hash::Hasher>(&self, state: &mut H) {
+                unsafe { self.uc_sigmask.hash(state) }


Same here, not sure we can do this with unions, without some guarantee they are the same size.

This code was simply borrowed from the ARM module:

libc/src/unix/linux_like/android/b32/arm.rs

Lines 59 to 73 in a8e2958

cfg_if! {

if #[cfg(feature = "extra_traits")] {

impl PartialEq for __c_anonymous_uc_sigmask {

fn eq(&self, other: &__c_anonymous_uc_sigmask) -> bool {

unsafe { self.uc_sigmask == other.uc_sigmask }

}

}

impl Eq for __c_anonymous_uc_sigmask {}

impl hash::Hash for __c_anonymous_uc_sigmask {

fn hash<H: hash::Hasher>(&self, state: &mut H) {

unsafe { self.uc_sigmask.hash(state) }

}

}

}

}

x86_64 has same layout:

libc/src/unix/linux_like/android/b64/x86_64/mod.rs

Lines 170 to 184 in a8e2958

cfg_if! {

if #[cfg(feature = "extra_traits")] {

impl PartialEq for __c_anonymous_uc_sigmask {

fn eq(&self, other: &__c_anonymous_uc_sigmask) -> bool {

unsafe { self.uc_sigmask == other.uc_sigmask }

}

}

impl Eq for __c_anonymous_uc_sigmask {}

impl hash::Hash for __c_anonymous_uc_sigmask {

fn hash<H: hash::Hasher>(&self, state: &mut H) {

unsafe { self.uc_sigmask.hash(state) }

}

}

}

}

StackOverflowExcept1on · 2026-06-25T13:08:26Z

@teor2345 I created test repository demonstrating that the offsets are correct (Android NDK C++). It also has passed CI.

#include <ucontext.h>

static_assert(offsetof(ucontext_t, uc_flags) == 0);
static_assert(offsetof(ucontext_t, uc_link) == 8);
static_assert(offsetof(ucontext_t, uc_stack) == 16);
static_assert(offsetof(ucontext_t, uc_sigmask) == 40);
static_assert(offsetof(ucontext_t, uc_mcontext) == 176);

static_assert(sizeof(ucontext_t) == 4560);

git clone https://github.com/StackOverflowExcept1on/library-android.git
cd library-android
PATH_TO_NDK=~/Android/Sdk/ndk/27.2.12479018
./build_android_arm64_v8a.sh $PATH_TO_NDK

+ '[' -z /home/.../Android/Sdk/ndk/27.2.12479018 ']'
+ CMAKE_BUILD_TYPE=Release
+ PATH_TO_NDK=/home/.../Android/Sdk/ndk/27.2.12479018
+ mkdir -p build
+ cd build
+ cmake -DCMAKE_TOOLCHAIN_FILE=/home/.../Android/Sdk/ndk/27.2.12479018/build/cmake/android.toolchain.cmake -DANDROID_ABI=arm64-v8a -DANDROID_PLATFORM=35 -DANDROID_STL=c++_shared -DANDROID_USE_LEGACY_TOOLCHAIN_FILE=OFF -DCMAKE_BUILD_TYPE=Release ..
-- Android: Targeting API '35' with architecture 'arm64', ABI 'arm64-v8a', and processor 'aarch64'
-- Android: Selected unified Clang toolchain
-- The C compiler identification is Clang 18.0.3
-- The CXX compiler identification is Clang 18.0.3
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /home/.../Android/Sdk/ndk/27.2.12479018/toolchains/llvm/prebuilt/linux-x86_64/bin/clang - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /home/.../Android/Sdk/ndk/27.2.12479018/toolchains/llvm/prebuilt/linux-x86_64/bin/clang++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Configuring done (0.9s)
-- Generating done (0.0s)
-- Build files have been written to: /mnt/tmpfs/library-android/build
+ cmake --build .
[ 50%] Building CXX object CMakeFiles/library.dir/library.cpp.o
[100%] Linking CXX shared library liblibrary.so
[100%] Built target library
+ cmake --install .
-- Install configuration: "Release"
-- Installing: /mnt/tmpfs/library-android/build/bin/liblibrary.so

I also cloned my branch with the patch in rust-lang/libc and built it on my phone in Termux.

cargo new app
cd app
cargo add libc --git https://github.com/StackOverflowExcept1on/libc.git --branch fix-android

src/main.rs

fn main() {
    const _: () = {
        assert!(std::mem::offset_of!(libc::ucontext_t, uc_flags) == 0);
        assert!(std::mem::offset_of!(libc::ucontext_t, uc_link) == 8);
        assert!(std::mem::offset_of!(libc::ucontext_t, uc_stack) == 16);
        assert!(std::mem::offset_of!(libc::ucontext_t, uc_sigmask__c_anonymous_union) == 40);
        assert!(std::mem::offset_of!(libc::ucontext_t, uc_mcontext) == 176);

        assert!(std::mem::size_of::<libc::ucontext_t>() == 4560);
    };
}

rustc --version --verbose
rustc 1.96.0 (ac68faa20 2026-05-25) (built from a source tarball)
binary: rustc
commit-hash: ac68faa20c58cbccd01ee7208bf3b6e93a7d7f96
commit-date: 2026-05-25
host: aarch64-linux-android
release: 1.96.0
LLVM version: 21.1.8
cargo build --release
   Compiling libc v1.0.0-alpha.3 (https://github.com/StackOverflowExcept1on/libc.git?branch=fix-android#f674b086)
   Compiling app v0.1.0 (/data/data/com.termux/files/home/app)
    Finished `release` profile [optimized] target(s) in 2.41s

tgross35

We can do the padding change in 0.2 for sure, but I'm not sure about the sigmask portion. It's technically more accurate to source but is there any difference?

libc/src/unix/linux_like/android/b64/mod.rs

Lines 11 to 13 in e59daae

    
           pub struct sigset_t { 
        
               __val: [c_ulong; 1], 
        
           }

and

libc/src/unix/linux_like/android/b64/mod.rs

Lines 138 to 140 in e59daae

    
           pub struct sigset64_t { 
        
               __bits: [c_ulong; 1], 
        
           }

(sigset_t and sigset64_t) appear to have the same definition.

I'm fine with or without it, but will defer to @maurer

View changes since this review

rustbot added the S-waiting-on-review label Jun 21, 2026

rustbot assigned tgross35 Jun 21, 2026

fix(android/aarch64): use libc::ucontext_t from Android NDK

f674b08

StackOverflowExcept1on force-pushed the fix-android branch from 43ad692 to f674b08 Compare June 21, 2026 11:17

StackOverflowExcept1on mentioned this pull request Jun 23, 2026

refactor(protocol/lazy-pages): improve #[cfg(...)] attributes, add support Android x86_64 gear-tech/gear#5599

Merged

teor2345 reviewed Jun 25, 2026

View reviewed changes

tgross35 reviewed Jun 26, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(android/aarch64): use `libc::ucontext_t` from Android NDK#5189

fix(android/aarch64): use `libc::ucontext_t` from Android NDK#5189
StackOverflowExcept1on wants to merge 1 commit into
rust-lang:mainfrom
StackOverflowExcept1on:fix-android

StackOverflowExcept1on commented Jun 21, 2026 •

edited

Loading

Uh oh!

rustbot commented Jun 21, 2026

Uh oh!

teor2345 Jun 25, 2026 •

edited by rustbot

Loading

Uh oh!

StackOverflowExcept1on Jun 25, 2026

Uh oh!

teor2345 left a comment •

edited by rustbot

Loading

Uh oh!

teor2345 Jun 25, 2026

Uh oh!

teor2345 Jun 25, 2026

Uh oh!

StackOverflowExcept1on Jun 25, 2026 •

edited

Loading

Uh oh!

StackOverflowExcept1on Jun 25, 2026

Uh oh!

StackOverflowExcept1on commented Jun 25, 2026 •

edited

Loading

Uh oh!

tgross35 left a comment •

edited by rustbot

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

	cfg_if! {
	if #[cfg(feature = "extra_traits")] {
	impl PartialEq for __c_anonymous_uc_sigmask {
	fn eq(&self, other: &__c_anonymous_uc_sigmask) -> bool {
	unsafe { self.uc_sigmask == other.uc_sigmask }
	}
	}
	impl Eq for __c_anonymous_uc_sigmask {}
	impl hash::Hash for __c_anonymous_uc_sigmask {
	fn hash<H: hash::Hasher>(&self, state: &mut H) {
	unsafe { self.uc_sigmask.hash(state) }
	}
	}
	}
	}

Uh oh!

Conversation

StackOverflowExcept1on commented Jun 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Jun 21, 2026

Uh oh!

teor2345 Jun 25, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

StackOverflowExcept1on Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

teor2345 left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

teor2345 Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

teor2345 Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

StackOverflowExcept1on Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

StackOverflowExcept1on Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

StackOverflowExcept1on commented Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tgross35 left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

StackOverflowExcept1on commented Jun 21, 2026 •

edited

Loading

teor2345 Jun 25, 2026 •

edited by rustbot

Loading

teor2345 left a comment •

edited by rustbot

Loading

StackOverflowExcept1on Jun 25, 2026 •

edited

Loading

StackOverflowExcept1on commented Jun 25, 2026 •

edited

Loading

tgross35 left a comment •

edited by rustbot

Loading