rustc segfault #70117
Comments
haraldh
added
C-bug
|
Centril
added
I-crash
|
oops, corrected backtrace of nightly.. pasted the stable output again. |
Centril
added
I-nominated
A-LLVM
|
@haraldh can you share the code that causes this issue?. |
|
triage: we couldn't decide the priority of the issue due to the lack of code. Left nomination so we can check again once we have the code that triggers the issue. |
$ sudo apt-get install musl-tools
$ git clone --single-branch -b rust_segfault https://github.com/haraldh/enarx_sev_kvm_demo.git
$ cd enarx_sev_kvm_demo/kernel/
$ while cargo build -vvv --release ; do cargo clean; done; |
|
@rustbot ping cleanup-crew |
|
Error: This team ( Please let |
|
@rustbot ping cleanup |
rustbot
added
the
ICEBreaker-Cleanup-Crew
|
Hey Cleanup Crew ICE-breakers! This bug has been identified as a good cc @AminArria @chrissimpkins @contrun @DutchGhost @elshize @ethanboxx @h-michael @HallerPatrick @hdhoang @hellow554 @imtsuki @jakevossen5 @kanru @KarlK90 @LeSeulArtichaut @MAdrianMattocks @matheus-consoli @mental32 @nmccarty @Noah-Kennedy @pard68 @PeytonT @pierreN @Redblueflame @RobbieClarken @RobertoSnap @robjtede @SarthakSingh31 @senden9 @shekohex @sinato @spastorino @turboladen @woshilapin @yerke |
|
Discussed in T-compiler meeting. Prioritizing as P-high for the initial investigation into what's going on. (It may or may not deserve a higher or lower priority once we figure out where the segfault is coming from...) |
could be replaced with |
|
updated the branch again, so the assembler code compiles with more toolchains |
|
are you guys able to reproduce this, or is this only happening with my kernel+glibc+whatever combination? |
|
Backtrace: #68663 looks suspicious and I sort of confirmed it: |
|
@rustbot modify labels: +regression-from-stable-to-stable |
|
Error: Label regression-from-stable-to-stable can only be set by Rust team members Please let |
Mark-Simulacrum
added
the
regression-from-stable-to-stable
|
@mati865 that looks like a blocked thread, not a segfaulting threading. You'll likely want to get the backtrace of all threads if you're seeing a crash, and the ones that aren't blocked in pthreads are likely the ones segfaulting. Or if it is indeed segfaulting in pthreads, it would be good to confirm that by inspecting the instruction to see what the faulting instruction was, since that could hint to possible data corruption via jobserver stuff, but I'm pretty doubtful that's the cause here given the original backtrace in the OP. This seems unrelated to jobserver stuff from what I'm reading, so I'm going to unsubscribe myself. Feel free to ping me again if it's more confirmed though that jobserver stuff is at fault. |
|
Still an issue.. this time with released "stable" |
|
$ sudo apt-get install musl-tools |
|
And it's segfaulting way more often, than succeeding. |
Thread 2 "rustc" received signal SIGBUS, Bus error.
[Switching to Thread 0x7fffef1ff700 (LWP 1351394)]
0x00007ffff15391bd in llvm::AttributeList::addAttributes(llvm::LLVMContext&, unsigned int, llvm::AttrBuilder const&) const ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/../lib/libLLVM-9-rust-1.43.0-stable.so
(gdb) bt
#0 0x00007ffff15391bd in llvm::AttributeList::addAttributes(llvm::LLVMContext&, unsigned int, llvm::AttrBuilder const&) const ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/../lib/libLLVM-9-rust-1.43.0-stable.so
#1 0x00007ffff1616226 in llvm::Function::addAttributes(unsigned int, llvm::AttrBuilder const&) ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/../lib/libLLVM-9-rust-1.43.0-stable.so
#2 0x00007ffff4ef2c90 in LLVMRustAddFunctionAttribute ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#3 0x00007ffff4ee1813 in rustc_codegen_llvm::declare::declare_raw_fn ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#4 0x00007ffff4dbfff2 in <rustc_codegen_llvm::context::CodegenCx as rustc_codegen_ssa::traits::misc::MiscMethods>::eh_personality ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#5 0x00007ffff4e2ee84 in rustc_codegen_ssa::mir::codegen_mir ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#6 0x00007ffff4df0bac in <rustc::mir::mono::MonoItem as rustc_codegen_ssa::mono_item::MonoItemExt>::define ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#7 0x00007ffff4d3657a in rustc_codegen_llvm::base::compile_codegen_unit::module_codegen ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#8 0x00007ffff4e1ad81 in rustc::dep_graph::graph::DepGraph::with_task ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#9 0x00007ffff4d36144 in rustc_codegen_llvm::base::compile_codegen_unit ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#10 0x00007ffff4e4fc89 in rustc_codegen_ssa::base::codegen_crate ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#11 0x00007ffff4e02895 in <rustc_codegen_llvm::LlvmCodegenBackend as rustc_codegen_utils::codegen_backend::CodegenBackend>::codegen_crate ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#12 0x00007ffff4cb3436 in rustc_interface::passes::start_codegen ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#13 0x00007ffff4cc6a93 in rustc::ty::context::tls::enter_global ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#14 0x00007ffff4bd8846 in rustc_interface::queries::Queries::ongoing_codegen ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#15 0x00007ffff4a875e6 in rustc_interface::interface::run_compiler_in_existing_thread_pool ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#16 0x00007ffff4a3aacc in std::sys_common::backtrace::__rust_begin_short_backtrace ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#17 0x00007ffff43eb627 in __rust_maybe_catch_panic () at src/libpanic_unwind/lib.rs:86
#18 0x00007ffff4a89eb6 in core::ops::function::FnOnce::call_once{{vtable-shim}} ()
from /home/harald/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/bin/../lib/librustc_driver-7e244282122ae6e4.so
#19 0x00007ffff43b65bf in <alloc::boxed::Box<F> as core::ops::function::FnOnce<A>>::call_once ()
at /rustc/4fb7144ed159f94491249e86d5bbd033b5d60550/src/liballoc/boxed.rs:1017
#20 0x00007ffff43e9f2d in <alloc::boxed::Box<F> as core::ops::function::FnOnce<A>>::call_once ()
at /rustc/4fb7144ed159f94491249e86d5bbd033b5d60550/src/liballoc/boxed.rs:1017
#21 std::sys_common::thread::start_thread () at src/libstd/sys_common/thread.rs:13
#22 std::sys::unix::thread::Thread::new::thread_start () at src/libstd/sys/unix/thread.rs:80
#23 0x00007ffff4333432 in start_thread () from /lib64/libpthread.so.0
#24 0x00007ffff424f9d3 in clone () from /lib64/libc.so.6
(gdb)
|
|
for a successful build I have to do: ❯ while ! cargo build --release ; do :; done; |
|
There is a mismatch between #![no_std]
#[no_mangle]
pub extern "C" fn rust_eh_personality() {
}
#[no_mangle]
pub fn zzz(a: Option<usize>) -> usize {
a.unwrap()
}Using the same type signature seems to work fine, although it requires ZST to workaround requirement for at least one named argument. |
|
Oh, you are right.. I can totally remove it and it does not segfault anymore! |
|
so, I added |
camelid
added
I-prioritize
|
Hi folks, I think I encountered the same issue, I've commented here with a similar stacktrace. |
Code
Steps to reproduce are in this comment
and in this comment
Meta
rustc --version --verbose:and
Error output
gdb backtrace with 1.42.0
gdb backtrace with 1.43.0-nightly (c20d7ee 2020-03-11)
The text was updated successfully, but these errors were encountered: