linux-kernel - Re: linux-next: build failure after merge of the rust tree

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <2025032429-curliness-unblock-7240@gregkh>
Date: Mon, 24 Mar 2025 06:52:38 -0700
From: Greg KH <greg@...ah.com>
To: Danilo Krummrich <dakr@...nel.org>
Cc: Stephen Rothwell <sfr@...b.auug.org.au>,
	Miguel Ojeda <ojeda@...nel.org>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Linux Next Mailing List <linux-next@...r.kernel.org>
Subject: Re: linux-next: build failure after merge of the rust tree

On Mon, Mar 24, 2025 at 02:43:27PM +0100, Danilo Krummrich wrote:
> On Mon, Mar 24, 2025 at 06:29:30AM -0700, Greg KH wrote:
> > On Mon, Mar 24, 2025 at 12:59:27PM +0100, Danilo Krummrich wrote:
> > > Hi Stephen,
> > > 
> > > On Mon, Mar 24, 2025 at 10:06:29PM +1100, Stephen Rothwell wrote:
> > > > Hi all,
> > > > 
> > > > On Mon, 24 Mar 2025 21:57:02 +1100 Stephen Rothwell <sfr@...b.auug.org.au> wrote:
> > > > >
> > > > > After merging the rust tree, today's linux-next build (x86_64
> > > > > allmodconfig) failed like this:
> > > > > 
> > > > > error[E0277]: `*mut MyStruct` cannot be sent between threads safely
> > > > >   --> samples/rust/rust_dma.rs:47:22  
> > > > >    |
> > > > > 47 | impl pci::Driver for DmaSampleDriver {
> > > > >    |                      ^^^^^^^^^^^^^^^ `*mut MyStruct` cannot be sent between threads safely
> > > > >    |
> > > > >    = help: within `DmaSampleDriver`, the trait `Send` is not implemented for `*mut MyStruct`, which is required by `DmaSampleDriver: Send`
> > > > > note: required because it appears within the type `CoherentAllocation<MyStruct>`
> > > > >   --> rust/kernel/dma.rs:132:12  
> > > > > note: required because it appears within the type `DmaSampleDriver`
> > > > >   --> samples/rust/rust_dma.rs:9:8  
> > > > >    |
> > > > > 9  | struct DmaSampleDriver {
> > > > >    |        ^^^^^^^^^^^^^^^
> > > > > note: required by a bound in `kernel::pci::Driver`
> > > > >   --> rust/kernel/pci.rs:225:1  
> > > > > 
> > > > > error: aborting due to 1 previous error
> > > > > 
> > > > > For more information about this error, try `rustc --explain E0277`.
> > > > > 
> > > > > I have no idea what caused this - it built in next-20250321, but that
> > > > > no longer builds, so I have reset to the version of the rust tree in
> > > > > next-20250320 (commit 4a47eec07be6).
> > > > 
> > > > Actually, the driver-core tree gained these commits over the weekend:
> > > > 
> > > >   51d0de7596a4 ("rust: platform: require Send for Driver trait implementers")
> > > >   935e1d90bf6f ("rust: pci: require Send for Driver trait implementers")
> > > >   455943aa187f ("rust: platform: impl Send + Sync for platform::Device")
> > > >   e2942bb4e629 ("rust: pci: impl Send + Sync for pci::Device")
> > > > 
> > > > A heads up would have been nice ... and maybe even a test merge and
> > > > build against -next (given how late we are in the cycle).
> > > 
> > > Commit 935e1d90bf6f ("rust: pci: require Send for Driver trait implementers")
> > > from the driver-core tree fixes a missing concurrency requirement, which commit
> > > 9901addae63b ("samples: rust: add Rust dma test sample driver") from the Rust
> > > tree did not yet consider.
> > > 
> > > Technically, it did what it is supposed to do -- catch a concurrency issue at
> > > compile time. However, since I was involved into both sides, I could have
> > > thought of this, but unfortunately in this case it was too subtle for me to
> > > spot -- sorry.
> > > 
> > > There are two options, 1. simply drop the commit [1] that introduces the
> > > affected sample DMA code, or 2. apply the fix below to [2]. My preference would
> > > be (2).
> > > 
> > > --
> > > 
> > > diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
> > > index 9d00f9c49f47..18de693c4924 100644
> > > --- a/rust/kernel/dma.rs
> > > +++ b/rust/kernel/dma.rs
> > > @@ -301,6 +301,10 @@ fn drop(&mut self) {
> > >      }
> > >  }
> > > 
> > > +// SAFETY: It is safe to send a `CoherentAllocation` to another thread if `T`
> > > +// can be send to another thread.
> > > +unsafe impl<T: AsBytes + FromBytes + Send> Send for CoherentAllocation<T> {}
> > > +
> > >  /// Reads a field of an item from an allocated region of structs.
> > >  ///
> > >  /// # Examples
> > > 
> > 
> > I can't "drop" anything here as that would be a mess.
> 
> It's the DMA commit that has a bug, that was revealed by the fix in the
> driver-core tree. So, the patch to drop is in the rust tree (not sure if Miguel
> changes history at this point though).
> 
> Anyways, I think the fix is simple enough.
> 
> > Maybe we just
> > consider a merge of the driver core and rust trees at this point in time
> > and fix things up and do a combined pull request to Linus so he doesn't
> > have to deal with the fixups?
> 
> I think it's not that bad, the full diff for the conflicts between driver-core
> and rust is:
> 
> diff --git a/rust/kernel/dma.rs b/rust/kernel/dma.rs
> index 9d00f9c49f47..18de693c4924 100644
> --- a/rust/kernel/dma.rs
> +++ b/rust/kernel/dma.rs
> @@ -301,6 +301,10 @@ fn drop(&mut self) {
>      }
>  }
> 
> +// SAFETY: It is safe to send a `CoherentAllocation` to another thread if `T`
> +// can be send to another thread.
> +unsafe impl<T: AsBytes + FromBytes + Send> Send for CoherentAllocation<T> {}
> +
>  /// Reads a field of an item from an allocated region of structs.
>  ///
>  /// # Examples
> diff --git a/samples/rust/rust_dma.rs b/samples/rust/rust_dma.rs
> index 908acd34b8db..874c2c964afa 100644
> --- a/samples/rust/rust_dma.rs
> +++ b/samples/rust/rust_dma.rs
> @@ -4,10 +4,10 @@
>  //!
>  //! To make this driver probe, QEMU must be run with `-device pci-testdev`.
> 
> -use kernel::{bindings, dma::CoherentAllocation, pci, prelude::*};
> +use kernel::{bindings, device::Core, dma::CoherentAllocation, pci, prelude::*, types::ARef};
> 
>  struct DmaSampleDriver {
> -    pdev: pci::Device,
> +    pdev: ARef<pci::Device>,
>      ca: CoherentAllocation<MyStruct>,
>  }
> 
> @@ -48,7 +48,7 @@ impl pci::Driver for DmaSampleDriver {
>      type IdInfo = ();
>      const ID_TABLE: pci::IdTable<Self::IdInfo> = &PCI_TABLE;
> 
> -    fn probe(pdev: &mut pci::Device, _info: &Self::IdInfo) -> Result<Pin<KBox<Self>>> {
> +    fn probe(pdev: &pci::Device<Core>, _info: &Self::IdInfo) -> Result<Pin<KBox<Self>>> {
>          dev_info!(pdev.as_ref(), "Probe DMA test driver.\n");
> 
>          let ca: CoherentAllocation<MyStruct> =
> @@ -64,7 +64,7 @@ fn probe(pdev: &mut pci::Device, _info: &Self::IdInfo) -> Result<Pin<KBox<Self>>
> 
>          let drvdata = KBox::new(
>              Self {
> -                pdev: pdev.clone(),
> +                pdev: pdev.into(),
>                  ca,
>              },
>              GFP_KERNEL,

Ok, let's just leave it as-is for now...