linux-kernel - Re: [PATCH 1/1] pahole/Rust: Check that we're adding DW_TAG

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] [day] [month] [year] [list]

Message-ID: <Y+v7kNNoRC2ql/a0@kernel.org>
Date:   Tue, 14 Feb 2023 18:22:24 -0300
From:   Arnaldo Carvalho de Melo <acme@...nel.org>
To:     Miguel Ojeda <miguel.ojeda.sandonis@...il.com>
Cc:     Daniel Borkmann <daniel@...earbox.net>,
        Martin Rodriguez Reboredo <yakoyoku@...il.com>,
        linux-kernel@...r.kernel.org, Neal Gompa <neal@...pa.dev>,
        Eric Curtin <ecurtin@...hat.com>, bpf@...r.kernel.org,
        rust-for-linux@...r.kernel.org,
        Alexei Starovoitov <ast@...nel.org>,
        Andrii Nakryiko <andrii@...nel.org>,
        Jiri Olsa <jolsa@...nel.org>, Yonghong Song <yhs@...com>
Subject: Re: [PATCH 1/1] pahole/Rust: Check that we're adding DW_TAG_member
 sorted by byte offset

Em Fri, Feb 10, 2023 at 05:48:36PM -0300, Arnaldo Carvalho de Melo escreveu:
> Hi Miguel, after a long winter, I'm trying to get Rust properly
> supported on pahole, please check that this specific use case is working
> for you as well.
> 
> I'll go thru the others to see if they are easy (or at least restricted
> to Rust CUs) as this one.

I needed to add this one on top:

>From 1231b6b9b4d88e0084bef4254eb1a05eb9935c99 Mon Sep 17 00:00:00 2001
From: Arnaldo Carvalho de Melo <acme@...hat.com>
Date: Tue, 14 Feb 2023 18:13:59 -0300
Subject: [PATCH 1/1] dwarf_loader: Fix sorting of Rust structs

We may have structs with fields on the same offset, don't move anything
in that case, otherwise we get into an eternal loop, doh.

Tested with the Linux kernel built with CONFIG_RUST + all the code
examples, first Rust struct where this happened was:

  (gdb) p type->namespace.name
  $2 = 0x7fffda938497 "((), char)"
  (gdb) p type->nr_members
  $3 = 2
  (gdb) p current_member->name
  $4 = 0x7fffda918f36 "__1"
  (gdb) p next_member->name
  $5 = 0x7fffda91765c "__0"
  (gdb) p current_member->byte_offset
  $6 = 0
  (gdb) p next_member->byte_offset
  $7 = 0

But this shows that --lang_exclude should be better positioned as we're
now needlessly loading all the tags for Rust DWARF to then just trow it
away at the cu__filter() in pahole :-\

Too late in the 1.25 release cycle for that, optimize it in 1.26.

Signed-off-by: Arnaldo Carvalho de Melo <acme@...hat.com>
---
 dwarf_loader.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dwarf_loader.c b/dwarf_loader.c
index a77598dc3affca88..acdb68d5dd33f148 100644
--- a/dwarf_loader.c
+++ b/dwarf_loader.c
@@ -2857,7 +2857,7 @@ restart:
 
 		struct class_member *next_member = list_entry(current_member->tag.node.next, typeof(*current_member), tag.node);
 
-		if (current_member->byte_offset < next_member->byte_offset)
+		if (current_member->byte_offset <= next_member->byte_offset)
 			continue;
 
 		list_del(&current_member->tag.node);
-- 
2.39.1