linux-kernel - Re: [PATCH] rust: print: Fix issue with rust_build

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: <20250922190127.GA2462108@joelbox2>
Date: Mon, 22 Sep 2025 15:01:27 -0400
From: Joel Fernandes <joelagnelf@...dia.com>
To: Miguel Ojeda <miguel.ojeda.sandonis@...il.com>
Cc: Benno Lossin <lossin@...nel.org>, linux-kernel@...r.kernel.org,
	Miguel Ojeda <ojeda@...nel.org>,
	Alex Gaynor <alex.gaynor@...il.com>,
	Boqun Feng <boqun.feng@...il.com>, Gary Guo <gary@...yguo.net>,
	Björn Roy Baron <bjorn3_gh@...tonmail.com>,
	Andreas Hindborg <a.hindborg@...nel.org>,
	Alice Ryhl <aliceryhl@...gle.com>, Trevor Gross <tmgross@...ch.edu>,
	Danilo Krummrich <dakr@...nel.org>, acourbot@...dia.com,
	Alistair Popple <apopple@...dia.com>, Timur Tabi <ttabi@...dia.com>,
	rust-for-linux@...r.kernel.org
Subject: Re: [PATCH] rust: print: Fix issue with rust_build_error

On Sun, Sep 21, 2025 at 11:13:11AM +0200, Miguel Ojeda wrote:
> On Sun, Sep 21, 2025 at 2:45 AM Joel Fernandes <joelagnelf@...dia.com> wrote:
> >
> > But even if the pointer is a C const pointer, LLVM seems to
> > always want to reload it.
> 
> What do you mean by this? I think I mentioned in the other thread that
> a C pointer to const still allows the callee to change the value.

Apologies, indeed a const pointer in C does not mean the pointee cannot be
modified.

I think I somewhat understand the issue but still not fully. MIR optimization
is supposed to optimize away the dead code in build_assert. This is what I
see for "good" cases when things work.

But the information that the data being printed is an immutable reference, is
lost somehow during MIR optimization phase when a printk is involved.  Per the
github issue [1], there is likely some provenance information in the immutable
reference to the data, that gets lost during "MIR inlining" optimization. In
other words, this is not an LLVM problem as I was pointing out, but an MIR
optimization problem.  Benno/Gary correct me anything I said is wrong.

I guess I still have a few more questions:

1. What is being inlined when we talk about MIR inlining? The print
statement? Constructor to the Argument object? Something else?

2. What does 'noalias' mean in the github report [1], and why would that effect
'MIR inlining'?

3. Why does LLVM inlining still succeed when MIR inlining is disabled? This
has something to do with a new llvm feature Niki referred to in the report.

thanks,

 - Joel
[1] https://github.com/rust-lang/rust/issues/146844