Add missing definitions for QNX8 for several BPF-related structures and constants #4769

rafaeling · 2025-10-24T09:56:36Z

Description

These structures are required for working with BPF filters and network interfaces on the QNX8 platform. Their absence currently prevents building or using BPF-related functionality with libc on QNX8 targets.

Sources

Updated src/unix/nto/mod.rs accordingly

Checklist

[?] Relevant tests in libc-test/semver have been updated
-> No tests for QNX?
[*] No placeholder or unstable values like *LAST or *MAX are
included (see #3131)
[?] Tested locally (cd libc-test && cargo test --target mytarget); especially relevant for platforms that may not be checked in CI.
-> even main branch does not work.
Error: could not execute process /home/user/libc/target/x86_64-pc-nto-qnx800/debug/deps/libc-67e46525c5118905 (never executed)

tgross35 · 2025-10-27T21:31:07Z

This looks fine to me, cc target maintainers @flba-eb @gh-tr @jonathanpallant @japaric to verify things build

tgross35

Would you actually mind moving this to the src/new module? We're slowly starting to migrate things there since it's easier to match up with what a user can #include. I believe this would be src/new/nto/net/if.h and src/new/nto/net/bpf.h, unless nto has a different structure. You can either move the existing BPF items over or not, your call.

Then this update to src/new/nto/mod.rs:

pub(crate) mod net {
    pub(crate) mod route;
}

And at

libc/src/new/mod.rs

Lines 171 to 186 in 564e0ef

    
           // Per-OS headers we export 
        
           cfg_if! { 
        
               if #[cfg(target_os = "android")] { 
        
                   pub use sys::socket::*; 
        
               } else if #[cfg(target_os = "linux")] { 
        
                   pub use linux::can::bcm::*; 
        
                   pub use linux::can::j1939::*; 
        
                   pub use linux::can::raw::*; 
        
                   pub use linux::can::*; 
        
                   pub use linux::keyctl::*; 
        
                   #[cfg(target_env = "gnu")] 
        
                   pub use net::route::*; 
        
               } else if #[cfg(target_vendor = "apple")] { 
        
                   pub use signal::*; 
        
               } 
        
           }

, add a nto arm and re-export everything from the two modules.

After that and the __c_anonymous_ifc_ifcu removal I'll merge this given we haven't heard back from the maintainers but this is straightforward.

src/unix/nto/mod.rs

tgross35 · 2025-11-02T09:35:56Z

[?] Tested locally (cd libc-test && cargo test --target mytarget); especially relevant for platforms that may not be checked in CI.
-> even main branch does not work.
Error: could not execute process /home/user/libc/target/x86_64-pc-nto-qnx800/debug/deps/libc-67e46525c5118905 (never executed)

Can you usually cargo run --target ... for nto? The relevant test here woul be cargo test -p libc-test --test ctest --target ... if you want to narrow it down and try executing the binary directly.

rafaeling · 2025-11-13T16:23:41Z

Would you actually mind moving this to the src/new module? We're slowly starting to migrate things there since it's easier to match up with what a user can #include. I believe this would be src/new/nto/net/if.h and src/new/nto/net/bpf.h, unless nto has a different structure. You can either move the existing BPF items over or not, your call.

Then this update to src/new/nto/mod.rs:
pub(crate) mod net {
    pub(crate) mod route;
}
And at

libc/src/new/mod.rs

Lines 171 to 186 in 564e0ef

// Per-OS headers we export

cfg_if! {

if #[cfg(target_os = "android")] {

pub use sys::socket::*;

} else if #[cfg(target_os = "linux")] {

pub use linux::can::bcm::*;

pub use linux::can::j1939::*;

pub use linux::can::raw::*;

pub use linux::can::*;

pub use linux::keyctl::*;

#[cfg(target_env = "gnu")]

pub use net::route::*;

} else if #[cfg(target_vendor = "apple")] {

pub use signal::*;

}

}

, add a nto arm and re-export everything from the two modules.

After that and the __c_anonymous_ifc_ifcu removal I'll merge this given we haven't heard back from the maintainers but this is straightforward.

@tgross35 Moved structs to src/new, removed __c_anonymous_ifc_ifcu, and ran ./ci/style.sh.
However, the runners are failing, not sure why.

Sorry for the delay in fixing this! I was away for a bit.
I would be willing to support here gradually move things into src/new going forward.

tgross35

Few small requests and one question. Would you be able to rebase? I think the test errors should be fixed

tgross35 · 2025-11-19T12:18:51Z

src/unix/nto/mod.rs

 pub const SCM_TIMESTAMP: c_int = 0x02;
 cfg_if! {
-    if #[cfg(not(target_env = "nto71_iosock"))] {
+    if #[cfg(not(any(target_env = "nto71_iosock", target_env = "nto80")))] {


Are these not available on 8.0 either?

If this change is related to making tests pass on 8.0 rather than as part of the BPF, would you be able to put it in a separate commit?

No, these are not available on 8.0, so this is part of the BPF topic.

I still don't follow the connection - why are sysctl things like HW_IOSTATS related to BPF?

Sorry for the confusion.

I was just referring specifically to the BIOC* constants needed for BPF. I initially assumed that all of the constants in that section were BPF-related, but that turns out not to be the case. After checking the actual BIOC* values used for BPF on QNX 8, I added the new flag because the build was incorrectly picking up the constants from the else branch (intended for QNX 7.1/7.0 I think).

There are many other constants defined there that are unrelated to BPF. My assumption is that nto71_iosock and nto80 share the same constant layout, but I dont work for QNX, so I had to assume a few things.

That makes more sense to me, but I'd like the maintainers to double check since it's a bit surprising.

@flba-eb @gh-tr @jonathanpallant @japaric could you review and/or test the changes to these constants on QNX 8?

This looks okay, but I suggest to inverse the logic:

Suggested change

if #[cfg(not(any(target_env = "nto71_iosock", target_env = "nto80")))] {

if #[cfg(any(target_env = "nto70", target_env = "nto71"))] {

Positive logic is easier to understand

It is more likely that the list of QNX versions using the older network stack will not need to be modified in the future, but the versions with the new stack will probably need to be adjusted with each new QNX version.

Are these constants all only for the old network stack then? A comment would be nice then if you could add one @rafaeling.

src/new/nto/net/bpf.rs

rustbot · 2025-11-19T12:23:14Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

rustbot · 2025-12-03T09:10:56Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

tgross35 · 2025-12-03T09:42:21Z

src/new/nto/net/if_.rs

+    pub struct ifreq {
+        /// if name, e.g. "en0"
+        pub ifr_name: [c_char; crate::IFNAMSIZ],
+        pub ifr_ifru: __c_anonymous_ifr_ifru,
+    }


This probably has to go in s_no_extra_traits since the union doesn't have them (you can check that things work with --feature extra_traits)

I moved the structs into s_no_extra_traits (as linux does) but building with --features extra_traitsstill fails because the union would need Eq, Hash, and PartialEq.

On the BSD targets (Apple/FreeBSD/OpenBSD) these unions do implement Eq, Hash, and PartialEq and struct ifreq is into s! macro, should I mirror that behavior here as well? (I tried it and also fails with same error, not sure how to proceed)

What exactly is failing? If __c_anonymous_ifr_ifru and ifreq are both in s_no_extra_traits, that macro shouldn't try to derive them.

Don't handwrite trait impls, we do have them in some places but we're trying to move away from that because of the easy accidental unsoundness with unions.

Dont know why, but retried again and works now, maybe did something wrong before.
Check last commit

LGTM! I'm going to need to wait for somebody to look at #4769 (comment) though.

This will need a squash before merge.

rustbot added O-unix S-waiting-on-review labels Oct 24, 2025

rafaeling force-pushed the nto-bpf-ifreq-support branch from f1b2a57 to 7d27d60 Compare October 24, 2025 10:07

tgross35 requested changes Nov 2, 2025

View reviewed changes

src/unix/nto/mod.rs Outdated Show resolved Hide resolved

rustbot added S-waiting-on-author and removed S-waiting-on-review labels Nov 2, 2025

rafaeling force-pushed the nto-bpf-ifreq-support branch from 7d27d60 to f62f2d4 Compare November 13, 2025 16:12

This comment has been minimized.

Sign in to view

rafaeling force-pushed the nto-bpf-ifreq-support branch from 879b73f to 1687b82 Compare November 13, 2025 16:53

tgross35 requested changes Nov 19, 2025

View reviewed changes

rafaeling and others added 3 commits December 3, 2025 09:51

qnx: add missing BPF and ifreq structures

3bf1c6a

Relocate nto net headers to src/new

18de902

Add extern C f! macro and final newline

400b741

rafaeling force-pushed the nto-bpf-ifreq-support branch from 1687b82 to 400b741 Compare December 3, 2025 09:10

tgross35 reviewed Dec 3, 2025

View reviewed changes

Move structs into no_extra_traits macro

c244d16

	// Per-OS headers we export
	cfg_if! {
	if #[cfg(target_os = "android")] {
	pub use sys::socket::*;
	} else if #[cfg(target_os = "linux")] {
	pub use linux::can::bcm::*;
	pub use linux::can::j1939::*;
	pub use linux::can::raw::*;
	pub use linux::can::*;
	pub use linux::keyctl::*;
	#[cfg(target_env = "gnu")]
	pub use net::route::*;
	} else if #[cfg(target_vendor = "apple")] {
	pub use signal::*;
	}
	}

	if #[cfg(not(any(target_env = "nto71_iosock", target_env = "nto80")))] {
	if #[cfg(any(target_env = "nto70", target_env = "nto71"))] {

Add missing definitions for QNX8 for several BPF-related structures and constants #4769

Are you sure you want to change the base?

Add missing definitions for QNX8 for several BPF-related structures and constants #4769

Conversation

rafaeling commented Oct 24, 2025

Description

Sources

Checklist

Uh oh!

tgross35 commented Oct 27, 2025

Uh oh!

tgross35 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

tgross35 commented Nov 2, 2025

Uh oh!

This comment has been minimized.

rafaeling commented Nov 13, 2025

Uh oh!

tgross35 left a comment

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!

flba-eb Dec 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

rustbot commented Nov 19, 2025

Uh oh!

rustbot commented Dec 3, 2025

Uh oh!

tgross35 Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

rafaeling Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

flba-eb Dec 6, 2025 •

edited

Loading

tgross35 Dec 3, 2025 •

edited

Loading

rafaeling Dec 3, 2025 •

edited

Loading