I am seeing a regression when building Rust on sparcv9-sun-solaris.
Environment
- host:
sparcv9-sun-solaris
- OS: Solaris 11.4 on
sun4v
- build type: normal
dist/nightly build
Bisection
git bisect points to this as the first bad commit: 3f34c31
Author: Edvin Bryntesson epost@edvinbryntesson.se
Date: Sun Feb 22 15:31:39 2026 +0100
add variant `ExpectedNameValueAsLastArgument` to `AttributeParseErrorReason`
compiler/rustc_attr_parsing/src/context.rs
compiler/rustc_attr_parsing/src/session_diagnostics.rs
Failure
The build fails while compiling rustc_attr_parsing:
thread 'lto cgu.04' (26) has overflowed its stack
fatal runtime error: stack overflow, aborting
rustc exited with signal: 6 (SIGABRT) (core dumped)
error: could not compile `rustc_attr_parsing` (lib)
Notes
- Changing ulimit -s did not help.
- The overflow is in a Rust-created worker thread, not the shell/main thread.
- From source inspection, rustc worker threads use the internal default stack size unless overridden via RUST_MIN_STACK.
Workaround
Setting:
RUST_MIN_STACK=16777216
makes the same build pass on this machine.
So the current picture is:
- default worker-thread stack (8 MiB) fails on this target/build
- 16 MiB is sufficient in my testing
I am seeing a regression when building Rust on
sparcv9-sun-solaris.Environment
sparcv9-sun-solarissun4vdist/nightly buildBisection
git bisectpoints to this as the first bad commit: 3f34c31Author: Edvin Bryntesson epost@edvinbryntesson.se
Date: Sun Feb 22 15:31:39 2026 +0100
Failure
The build fails while compiling
rustc_attr_parsing:Notes
Workaround
Setting:
RUST_MIN_STACK=16777216makes the same build pass on this machine.
So the current picture is: