linux-kernel - Re: [PATCH 17/31] gpu: nova-core: Hopper/Blackwell: calculate reserved FB heap size

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: <ac0338f2-6b7d-4b34-90a4-7be041682d2b@nvidia.com>
Date: Wed, 3 Dec 2025 23:34:45 -0800
From: John Hubbard <jhubbard@...dia.com>
To: Timur Tabi <ttabi@...dia.com>, "dakr@...nel.org" <dakr@...nel.org>
Cc: Alexandre Courbot <acourbot@...dia.com>,
 "lossin@...nel.org" <lossin@...nel.org>,
 "a.hindborg@...nel.org" <a.hindborg@...nel.org>,
 "boqun.feng@...il.com" <boqun.feng@...il.com>,
 "aliceryhl@...gle.com" <aliceryhl@...gle.com>, Zhi Wang <zhiw@...dia.com>,
 "simona@...ll.ch" <simona@...ll.ch>,
 "alex.gaynor@...il.com" <alex.gaynor@...il.com>,
 "ojeda@...nel.org" <ojeda@...nel.org>, "tmgross@...ch.edu"
 <tmgross@...ch.edu>,
 "nouveau@...ts.freedesktop.org" <nouveau@...ts.freedesktop.org>,
 "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
 "rust-for-linux@...r.kernel.org" <rust-for-linux@...r.kernel.org>,
 "bjorn3_gh@...tonmail.com" <bjorn3_gh@...tonmail.com>,
 Edwin Peer <epeer@...dia.com>, "airlied@...il.com" <airlied@...il.com>,
 Joel Fernandes <joelagnelf@...dia.com>,
 "bhelgaas@...gle.com" <bhelgaas@...gle.com>,
 "gary@...yguo.net" <gary@...yguo.net>, Alistair Popple <apopple@...dia.com>
Subject: Re: [PATCH 17/31] gpu: nova-core: Hopper/Blackwell: calculate
 reserved FB heap size

On 12/3/25 12:48 PM, Timur Tabi wrote:
> On Tue, 2025-12-02 at 21:59 -0800, John Hubbard wrote:
>> +#[inline(always)]
>> +pub(crate) const fn const_align_up<const ALIGN: usize>(value: usize) -> usize {
>> +    build_assert!(ALIGN.is_power_of_two());
>> +    (value + (ALIGN - 1)) & !(ALIGN - 1)
>> +}
> 
> So this is just like the issue I'm having with .next_multiple_of() in my patch #10.
> 
> Shouldn't you have a check to make sure that value + (ALIGN - 1) doesn't overflow?  Since I need to
> align up to the nearest const power of two, I could use this function instead of align_up() and
> avoid testing for an error condition.
> 

Const items are special: they are evaluated at compile time, rather
than at runtime. And so this will fail to compile, for values that
would overflow.

For example:

error[E0080]: evaluation of constant value failed
    --> drivers/gpu/nova-core/num.rs:226:5
     |
226 |     (value + (ALIGN - 1)) & !(ALIGN - 1)
     |     ^^^^^^^^^^^^^^^^^^^^^ attempt to compute `usize::MAX + 15_usize`, which would overflow
     |
note: inside `const_align_up::<16>`
    --> drivers/gpu/nova-core/num.rs:226:5
     |
226 |     (value + (ALIGN - 1)) & !(ALIGN - 1)
     |     ^^^^^^^^^^^^^^^^^^^^^
note: inside `_OVERFLOW_TEST`
    --> drivers/gpu/nova-core/num.rs:231:31
     |
231 | const _OVERFLOW_TEST: usize = const_align_up::<16>(usize::MAX);
     |                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^


thanks,
-- 
John Hubbard