Skip to content

Add field-logging barrier and field unlog bit (JDK21) #359

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wks wants to merge 4 commits into
base: jdk-21
Choose a base branch
from
Draft

Add field-logging barrier and field unlog bit (JDK21) #359

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
1a1e403
Upstream API change w.r.t. field log bits
wks Jul 10, 2026
4e75de2
WIP: Make interpreter work with field-logging barrier.
wks Jul 17, 2026
73f73ee
Merge branch 'jdk-21' into feature/field-log-barrier
wks Jul 27, 2026
d198005
Rename barrier-related classes.
wks Jul 27, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 0 additions & 2 deletions mmtk/Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

4 changes: 2 additions & 2 deletions mmtk/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,9 +36,9 @@ probe = "0.5"
# - change branch
# - change repo name
# But other changes including adding/removing whitespaces in commented lines may break the CI.
mmtk = { git = "https://github.com/mmtk/mmtk-core.git", rev = "37d81218028d788f7a24158aac0588c3f6925d08" }
# mmtk = { git = "https://github.com/mmtk/mmtk-core.git", rev = "37d81218028d788f7a24158aac0588c3f6925d08" }
# Uncomment the following to build locally
# mmtk = { path = "../repos/mmtk-core" }
mmtk = { path = "../../mmtk-core" }

[build-dependencies]
built = { version = "0.7.7", features = ["git2"] }
Expand Down
3 changes: 2 additions & 1 deletion mmtk/src/object_model.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,8 @@ use mmtk::vm::*;
pub struct VMObjectModel<const COMPRESSED: bool> {}

impl<const COMPRESSED: bool> ObjectModel<OpenJDK<COMPRESSED>> for VMObjectModel<COMPRESSED> {
const GLOBAL_LOG_BIT_SPEC: VMGlobalLogBitSpec = vm_metadata::LOGGING_SIDE_METADATA_SPEC;
const GLOBAL_OBJECT_UNLOG_BIT_SPEC: VMGlobalObjectUnlogBitSpec = vm_metadata::OBJECT_LOGGING_SIDE_METADATA_SPEC;
const GLOBAL_FIELD_UNLOG_BIT_SPEC: VMGlobalFieldUnlogBitSpec = vm_metadata::FIELD_LOGGING_SIDE_METADATA_SPEC;

const LOCAL_FORWARDING_POINTER_SPEC: VMLocalForwardingPointerSpec =
vm_metadata::FORWARDING_POINTER_METADATA_SPEC;
Expand Down
5 changes: 5 additions & 0 deletions mmtk/src/slots.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -205,6 +205,11 @@ impl<const COMPRESSED: bool> Slot for OpenJDKSlot<COMPRESSED> {
unsafe { self.addr.store(object) }
}
}

fn to_address(&self) -> Address {
assert!(!COMPRESSED);
self.addr
Comment on lines +210 to +211

@k-sareen k-sareen Jul 23, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I just realized this issue in my branch. I think something like the below is better:

Suggested change
assert!(!COMPRESSED);
self.addr
if COMPRESSED {
self.untagged_address()
} else {
self.addr
}

@wks wks Jul 23, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will eventually do this. Currently the 1 bit per word field unlog bit side metadata doesn't work for compressed pointers, so I want it to fail loud. To support compressed pointers, I will introduce an alternative 1 bit per 32 bit field unlog bit side metadata and give the VM binding the option to select that.

@k-sareen k-sareen Jul 23, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah fair enough. Makes sense. I just found this bug in my branch so I thought it'd be better to make a comment here in case you also have issues with the same in the future.

}
}

/// A range of OpenJDKSlot, usually used for arrays.
Expand Down
6 changes: 5 additions & 1 deletion mmtk/src/vm_metadata/constants.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,11 @@ pub(crate) const FORWARDING_POINTER_OFFSET: isize = 0;

/// Global logging bit metadata spec
/// 1 bit per object
pub(crate) const LOGGING_SIDE_METADATA_SPEC: VMGlobalLogBitSpec = VMGlobalLogBitSpec::side_first();
pub(crate) const OBJECT_LOGGING_SIDE_METADATA_SPEC: VMGlobalObjectUnlogBitSpec =
VMGlobalObjectUnlogBitSpec::side_first();

pub(crate) const FIELD_LOGGING_SIDE_METADATA_SPEC: VMGlobalFieldUnlogBitSpec =
VMGlobalFieldUnlogBitSpec::side_after(OBJECT_LOGGING_SIDE_METADATA_SPEC.as_spec());

// Global MetadataSpecs - End

Expand Down
4 changes: 2 additions & 2 deletions openjdk/cpu/x86/mmtkUnlogBitBarrierSetAssembler_x86.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,8 +62,8 @@ void MMTkUnlogBitBarrierSetAssembler::object_reference_write_pre_or_post(MacroAs
// compressed oops, the `val` register will be holding a compressed pointer to the target object
// due to the way `BarrierSetAssembler::store_at` works. If the write barrier needs to know the
// target, we will need to decompress it before passing it to the barrier slow path.
__ xorptr(c_rarg1, c_rarg1);
__ xorptr(c_rarg2, c_rarg2);
__ lea(c_rarg1, dst);
__ movptr(c_rarg2, val);

address entry_point = mmtk_enable_barrier_fastpath ? FN_ADDR(MMTkBarrierSetRuntime::object_reference_write_slow_call)
: pre ? FN_ADDR(MMTkBarrierSetRuntime::object_reference_write_pre_call)
Expand Down
Loading