gdb on aarch64-unknown-linux-gnu/windows-gnu cannot print unused by-value non-immediate arguments #128973

saethlin · 2024-08-11T14:55:46Z

This test does not work on aarch64: https://github.com/rust-lang/rust/blob/c9bd03cb724e13cca96ad320733046cbdb16fbbe/tests/debuginfo/by-value-non-immediate-argument.rs

Stripped down (a bit):

fn by_val_enum(x: (usize, usize, usize)) {
    zzz(); // #break
}

fn main() {
    by_val_enum((usize::MAX, usize::MAX, usize::MAX));
}

fn zzz() { }

Set a breakpoint on line 2 and try to print x. gdb says it is optimized out. This can be reproduced even with -Zmir-opt-level=0, and doesn't reproduce on x86_64. If the argument is Option<(usize, usize)>, gdb will try to read from address 0x0.

I'm not sure if this is a rustc bug or a gdb bug because if I add any use of the non-immediate argument inside the function, we gain some instructions that look like a function prelude, and gdb is able to print the argument. https://godbolt.org/z/GeGxE7ExE

The text was updated successfully, but these errors were encountered:

…=compiler-errors Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here.

…=<try> Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here. try-job: i686-gnu try-job: aarch64-gnu try-job: x86_64-msvc try-job: i686-mingw

…=compiler-errors Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here.

…=<try> Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here. try-job: x86_64-gnu-llvm-17 try-job: x86_64-gnu-tools try-job: aarch64-gnu try-job: arm-android try-job: armhf-gnu try-job: i686-gnu try-job: i686-gnu-nopt try-job: mingw-check try-job: test-various try-job: x86_64-fuchsia try-job: x86_64-rust-for-linux try-job: x86_64-gnu try-job: x86_64-gnu-stable try-job: x86_64-gnu-aux try-job: x86_64-gnu-debug try-job: x86_64-gnu-llvm-18 try-job: x86_64-gnu-nopt try-job: x86_64-apple-1 try-job: aarch64-apple try-job: x86_64-msvc try-job: i686-msvc try-job: i686-mingw try-job: x86_64-mingw

…=<try> Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here. try-job: aarch64-gnu try-job: arm-android try-job: armhf-gnu try-job: i686-gnu try-job: i686-gnu-nopt try-job: test-various try-job: x86_64-fuchsia try-job: x86_64-rust-for-linux try-job: x86_64-gnu try-job: x86_64-gnu-stable

…=<try> Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here. try-job: aarch64-apple try-job: aarch64-gnu try-job: dist-i586-gnu-i586-i686-musl try-job: i686-gnu try-job: i686-gnu-nopt try-job: i686-mingw try-job: i686-msvc try-job: test-various try-job: x86_64-apple-1

…=<try> Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here. try-job: x86_64-gnu try-job: x86_64-gnu-distcheck try-job: x86_64-gnu-llvm-17 try-job: x86_64-gnu-llvm-18 try-job: x86_64-gnu-nopt try-job: x86_64-gnu-stable try-job: x86_64-mingw try-job: x86_64-msvc

…=compiler-errors Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here.

…-errors Enable debuginfo tests that have been "temporarily disabled" for the past 6 years The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here. A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in rust-lang/rust#124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI. `tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: rust-lang/rust#128971 `tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: rust-lang/rust#128945 `tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: rust-lang/rust#128973 `tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: rust-lang/rust#128981 `tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue. I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here.

…ark-Simulacrum Make option-like-enum.rs UB-free and portable Fixes rust-lang#129662 (or, at least the parts of it that aren't rust-lang#128973)

Rollup merge of rust-lang#129672 - saethlin:enum-debuginfo-tests, r=Mark-Simulacrum Make option-like-enum.rs UB-free and portable Fixes rust-lang#129662 (or, at least the parts of it that aren't rust-lang#128973)

…acrum Make option-like-enum.rs UB-free and portable Fixes rust-lang/rust#129662 (or, at least the parts of it that aren't rust-lang/rust#128973)

saethlin added A-debuginfo Area: Debugging information in compiled programs (DWARF, PDB, etc.) T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. C-bug Category: This is a bug. O-AArch64 Armv8-A or later processors in AArch64 mode labels Aug 11, 2024

rustbot added the needs-triage This issue may need triage. Remove it if it has been sufficiently triaged. label Aug 11, 2024

saethlin mentioned this issue Aug 11, 2024

Enable debuginfo tests that have been "temporarily disabled" for the past 6 years #128913

Merged

saethlin added the WG-debugging Working group: Bad Rust debugging experiences label Aug 11, 2024

saethlin removed the needs-triage This issue may need triage. Remove it if it has been sufficiently triaged. label Aug 13, 2024

This was referenced Aug 27, 2024

Unexpected nulls in debuginfo tests on windows-gnu #129662

Closed

Make option-like-enum.rs UB-free and portable #129672

Merged

saethlin changed the title ~~gdb on aarch64-unknown-linux-gnu cannot print unused by-value non-immediate arguments~~ gdb on aarch64-unknown-linux-gnu/windows-gnu cannot print unused by-value non-immediate arguments Aug 31, 2024

saethlin added the A-LLVM Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues. label Aug 31, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

gdb on aarch64-unknown-linux-gnu/windows-gnu cannot print unused by-value non-immediate arguments #128973

gdb on aarch64-unknown-linux-gnu/windows-gnu cannot print unused by-value non-immediate arguments #128973

saethlin commented Aug 11, 2024

gdb on aarch64-unknown-linux-gnu/windows-gnu cannot print unused by-value non-immediate arguments #128973

gdb on aarch64-unknown-linux-gnu/windows-gnu cannot print unused by-value non-immediate arguments #128973

Comments

saethlin commented Aug 11, 2024