Implement section attribute by rikkimax · Pull Request #22719 · dlang/dmd

rikkimax · 2026-03-11T05:29:27Z

Upstreaming this from ldc/gdc/opend.

However I did have to reimplement the flags. They were not correct according to Gemini, and from what I can tell, it is correct.

dlang-bot · 2026-03-11T05:29:31Z

Thanks for your pull request and interest in making D better, @rikkimax! We are looking forward to reviewing it, and you should be hearing from a maintainer soon.
Please verify that your PR follows this checklist:

My PR is fully covered with tests (you can see the coverage diff by visiting the details link of the codecov check)
My PR is as minimal as possible (smaller, focused PRs are easier to review than big ones)
I have provided a detailed rationale explaining my changes
New or modified functions have Ddoc comments (with Params: and Returns:)

Please see CONTRIBUTING.md for more information.

If you have addressed all reviews or aren't sure how to proceed, don't hesitate to ping us with a simple comment.

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub run digger -- build "master + dmd#22719"

rikkimax · 2026-03-11T06:52:20Z

I don't understand the c++ interop tests failing. As far as I know this shouldn't be messing with it.

Windows known.

I've added some debug messages to figure out why my test is failing.

rikkimax · 2026-03-11T07:15:52Z

ELF is passing.

~~Windows is asserting out in backend:~~

dmd/compiler/src/dmd/backend/mscoffobj.d

Line 1404 in a807811

assert(strlen(sectname) <= 8); // so it won't go into string_table

Disabled assert, works fine. Seems its more a in progress assert during some development spree and isn't an actual limitation of the code. Now passing requires the $M suffix.

~~Mach is not passing, I can't get the section start/end symbols to match or generate (I think).~~ Compiler kept adding an underscore, detect is specifically for section$ and only for when using pragma(mangle.

rikkimax · 2026-03-11T07:38:07Z

I will need to remove my complex 2^^power calculation, its changing alignment size that'll leave holes in the section and really mess up the slices. I will need to emit an error when a type's alignment is not a power of 2.

ibuclaw · 2026-03-11T16:24:21Z

compiler/src/dmd/declaration.d

    Expression edtor;               // if !=null, does the destruction of the variable
    IntRange* range;                // if !=null, the variable is known to be within the range

+    string userDefinedSection;      // the @section("name"), if present


Why is a field needed for this when gdc/ldc do just fine without?

(And the worst kind of field for C++ back-ends)

It does not.

This is how opend has it implemented, and this was based upon that (but I keep finding "fun things" wrong with it).

Processing during semantic is debatable as Udas are preserved for the codegen pass.

Indeed, I'm happy to move it. Just haven't done it quite yet. Hopefully it clears up the c++ interop issues, I still have no idea what is causing them.

Just duplicates information already kept elsewhere, so you're increasing the memory size of all variable/parameter/bitfields by 16 bytes for something 99.99(many 9's later)% of them won't use.

I've just realised that this is only putting global vars into custom sections. Is ignoring functions intentional?

Ignoring functions is more like I can't be bothered to come up with a way to verify it in CI.

Realistically people are not likely to use that for dmd.

It's pretty hard to use something that doesn't work, yes.

If only there was a tool that could do object dumps and that same tool can display information for a given section name. :-)

Yeah I was using objdump for testing this.

But no, I don't want to mess around with the CI and how to verify it.

WalterBright · 2026-03-13T15:20:13Z

What is the purpose of section attributes?

thewilsonator · 2026-03-13T15:23:09Z

druntime/src/core/attribute.d

    // GDC and LDC declare this attribute in their own modules.
 }

+/**


documentation starts here

Herringway · 2026-03-13T15:24:08Z

What is the purpose of section attributes?

Placing symbols in specific sections. Very useful for embedded applications. I personally use it to guarantee that structures appear at the correct location in ROM images (though the linker does most of the heavy lifting for this).

rikkimax · 2026-03-15T11:09:40Z

When incrementally linking with MSVC's link it'll add padding before merging.

As such, on Windows the type of the variable should be pointers: https://devblogs.microsoft.com/oldnewthing/20190114-00/?p=100695

If you use -g this turns on incremental linking.

Unfortunately this means that linker lists are going to need some more thinking.

rikkimax · 2026-03-17T01:15:42Z

I have updated the spec, to include that pragma(mangle does not add the leading underscore on macOS if it begins with section$.

rikkimax · 2026-03-17T10:41:26Z

I am happy with current state.

compiler/src/dmd/backend/melf.d

thewilsonator

Presumably Clang/GCC have the ability to do this too, can you add a C++ test that checks we match what they do?

compiler/src/dmd/glue/toobj.d

compiler/test/runnable/sectiondefs.d

rikkimax · 2026-03-18T22:20:07Z

I've gone ahead and removed the class detection for mutability. Class references (which are what a global variable stores) can be in read-only memory. Its the class instance that can't be, which this doesn't affect.

rikkimax requested a review from ibuclaw as a code owner March 11, 2026 05:29

rikkimax force-pushed the upstream-section branch 7 times, most recently from 62be678 to b9deb45 Compare March 11, 2026 06:50

rikkimax force-pushed the upstream-section branch from b9deb45 to b5f0610 Compare March 11, 2026 07:15

rikkimax force-pushed the upstream-section branch from b5f0610 to 762982c Compare March 11, 2026 07:29

rikkimax force-pushed the upstream-section branch from 762982c to 165f324 Compare March 11, 2026 07:59

ibuclaw reviewed Mar 11, 2026

View reviewed changes

rikkimax force-pushed the upstream-section branch 9 times, most recently from a6d8c85 to 94cb8f1 Compare March 11, 2026 22:47

thewilsonator reviewed Mar 13, 2026

View reviewed changes

rikkimax force-pushed the upstream-section branch 2 times, most recently from 87507e0 to f1ea646 Compare March 15, 2026 07:37

rikkimax force-pushed the upstream-section branch from f1ea646 to 8bd9ad1 Compare March 15, 2026 11:07

rikkimax force-pushed the upstream-section branch 2 times, most recently from 00829e4 to ae3f223 Compare March 15, 2026 11:45

rikkimax marked this pull request as draft March 15, 2026 12:17

rikkimax force-pushed the upstream-section branch 2 times, most recently from 8e46945 to 2b0c683 Compare March 17, 2026 01:14

rikkimax marked this pull request as ready for review March 17, 2026 01:14

thewilsonator reviewed Mar 17, 2026

View reviewed changes

compiler/src/dmd/backend/melf.d Outdated Show resolved Hide resolved

rikkimax force-pushed the upstream-section branch from 2b0c683 to 86435de Compare March 18, 2026 00:22

thewilsonator reviewed Mar 18, 2026

View reviewed changes

compiler/src/dmd/glue/toobj.d Show resolved Hide resolved

compiler/test/runnable/sectiondefs.d Outdated Show resolved Hide resolved

rikkimax force-pushed the upstream-section branch from 86435de to 2912f11 Compare March 18, 2026 01:31

Implement section attribute

d3175b7

rikkimax force-pushed the upstream-section branch from 2912f11 to d3175b7 Compare March 18, 2026 22:20

thewilsonator approved these changes Mar 20, 2026

View reviewed changes

thewilsonator merged commit 8c44fed into dlang:master Mar 20, 2026
42 checks passed

Uh oh!

Conversation

rikkimax commented Mar 11, 2026

Uh oh!

dlang-bot commented Mar 11, 2026

Bugzilla references

Testing this PR locally

Uh oh!

rikkimax commented Mar 11, 2026

Uh oh!

rikkimax commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rikkimax commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

WalterBright commented Mar 13, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Herringway commented Mar 13, 2026

Uh oh!

rikkimax commented Mar 15, 2026

Uh oh!

rikkimax commented Mar 17, 2026

Uh oh!

rikkimax commented Mar 17, 2026

Uh oh!

Uh oh!

thewilsonator left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

rikkimax commented Mar 18, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

rikkimax commented Mar 11, 2026 •

edited

Loading

rikkimax commented Mar 11, 2026 •

edited

Loading