linux-kernel - Re: [PATCH] overflow: Introduce struct

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: <20251127204342.05fc985e@robin>
Date: Thu, 27 Nov 2025 20:43:42 -0500
From: Steven Rostedt <rostedt@...dmis.org>
To: Kees Cook <kees@...nel.org>
Cc: LKML <linux-kernel@...r.kernel.org>, linux-hardening@...r.kernel.org,
 Linux Trace Kernel <linux-trace-kernel@...r.kernel.org>, Masami Hiramatsu
 <mhiramat@...nel.org>, Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
 Linus Torvalds <torvalds@...ux-foundation.org>, "Gustavo A. R. Silva"
 <gustavoars@...nel.org>
Subject: Re: [PATCH] overflow: Introduce struct_offset() to get offset of
 member

On Wed, 26 Nov 2025 23:58:01 -0800
Kees Cook <kees@...nel.org> wrote:

> > +/**
> > + * struct_offset() - Calculate the offset of a member within a struct
> > + * @p: Pointer to the struct
> > + * @member: Name of the member to get the offset of
> > + *
> > + * Calculates the offset of a particular @member of the structure pointed
> > + * to by @p.
> > + *
> > + * Return: number of bytes to the location of @member.
> > + */
> > +#define struct_offset(p, member) (offsetof(typeof(*(p)), member))  
> 
> I wonder if the kerndoc for this and offsetof() should reference each
> other? "For a type instead of a pointer, use offsetof()" etc...

I know I pushed this to my for-next branch already, but it's the top
patch. Looking at my code, I actually have a lot of places that use the
offsetof() for a structure variable and not a pointer to a structure.

Thus, I wonder if it is better to have this as:

#define struct_offset(s, member) (offsetof(typeof(s), member))

And then for pointers we have:

	size = struct_offset(*entry, id) + cnt;

If you forget the '*' it will error with a complaint that entry is not
a structure type. Then I could make changes like this:

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 1244d2c5c384..55f1bdab4ffa 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -587,19 +587,19 @@ int ring_buffer_print_page_header(struct trace_buffer *buffer, struct trace_seq

 	trace_seq_printf(s, "\tfield: local_t commit;\t"
 			 "offset:%u;\tsize:%u;\tsigned:%u;\n",
-			 (unsigned int)offsetof(typeof(field), commit),
+			 (unsigned int)struct_offset(field, commit),
 			 (unsigned int)sizeof(field.commit),
 			 (unsigned int)is_signed_type(long));

 	trace_seq_printf(s, "\tfield: int overwrite;\t"
 			 "offset:%u;\tsize:%u;\tsigned:%u;\n",
-			 (unsigned int)offsetof(typeof(field), commit),
+			 (unsigned int)struct_offset(field, commit),
 			 1,
 			 (unsigned int)is_signed_type(long));

 	trace_seq_printf(s, "\tfield: char data;\t"
 			 "offset:%u;\tsize:%u;\tsigned:%u;\n",
-			 (unsigned int)offsetof(typeof(field), data),
+			 (unsigned int)struct_offset(field, data),
 			 (unsigned int)buffer->subbuf_size,
 			 (unsigned int)is_signed_type(char));

I would have a lot more places I can make this update then if
struct_offset() took a pointer instead of the struct itself. As adding
a '*' isn't ugly I think I like this way better.

What are your thoughts?

-- Steve