linux-kernel - Re: [GIT PULL] tracing: A couple more fixes to v6.18

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: <20251011193742.410cea9d@gandalf.local.home>
Date: Sat, 11 Oct 2025 19:37:42 -0400
From: Steven Rostedt <rostedt@...dmis.org>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Kees Cook <kees@...nel.org>, LKML <linux-kernel@...r.kernel.org>, Masami
 Hiramatsu <mhiramat@...nel.org>, Mathieu Desnoyers
 <mathieu.desnoyers@...icios.com>, Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [GIT PULL] tracing: A couple more fixes to v6.18

On Sat, 11 Oct 2025 16:09:18 -0700
Linus Torvalds <torvalds@...ux-foundation.org> wrote:

> On Sat, 11 Oct 2025 at 12:52, Steven Rostedt <rostedt@...dmis.org> wrote:
> >
> > +       /* cnt includes both the entry->id and the data behind it. */
> > +       size = struct_size(entry, buf, cnt - sizeof(entry->id));  
> 
> This seems very non-intuitive.
> 
> Why isn't it just saying
> 
>     size = cnt + offsetof(struct raw_data_entry, id);
> 
> which would seem to be much more straightforward than saying "go to
> the end, and then subtract out the size of this entry" - which then
> relies on the entry being the last thing in the struct.
> 
> And if somebody wants to have a helper like
> 
>    #define struct_offset(s,memb) (offsetof(typeof(*(s)), memb))
> 
> in order to get some kind of "typesafe offsetof", to go with 'struct
> size' then by all means.. That would make it be just
> 
>     size = cnt + struct_offset(entry, id);
> 
> which looks fairly legible and logical, and would go with the comment
> (and would pair fairly well with our other "struct_xyz()" helpers, I
> think).
> 
> I've pulled this, I just reacted to how odd that calculation looked.
> It makes very little sense  to me, since the calculation really has
> _nothing_ do with the size of the struct, and you explicitly have to
> play tricks to get the offset that way.

I could make this change for the next merge window.

I originally was going to change it to:

	size = sizeof(*entry) + cnt - sizeof(entry->id);

But figured using a macro would be better.

Honestly, I didn't like either solution. But having a:

	size = struct_offset(entry, id) + cnt

Would probably look better.

But again, it can probably wait till 6.19.

-- Steve