linux-kernel - Re: [PATCH 1/2] objtool: Fix clang switch table edge case

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: <20200218195705.GA10545@ubuntu-m2-xlarge-x86>
Date:   Tue, 18 Feb 2020 12:57:05 -0700
From:   Nathan Chancellor <natechancellor@...il.com>
To:     Josh Poimboeuf <jpoimboe@...hat.com>
Cc:     x86@...nel.org, linux-kernel@...r.kernel.org,
        Peter Zijlstra <peterz@...radead.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>
Subject: Re: [PATCH 1/2] objtool: Fix clang switch table edge case

On Mon, Feb 17, 2020 at 09:41:53PM -0600, Josh Poimboeuf wrote:
> Clang has the ability to create a switch table which is not a jump
> table, but is rather a table of string pointers.  This confuses objtool,
> because it sees the relocations for the string pointers and assumes
> they're part of a jump table:
> 
>   drivers/ata/sata_dwc_460ex.o: warning: objtool: sata_dwc_bmdma_start_by_tag()+0x3a2: can't find switch jump table
>   net/ceph/messenger.o: warning: objtool: ceph_con_workfn()+0x47c: can't find switch jump table
> 
> Make objtool's find_jump_table() smart enough to distinguish between a
> switch jump table (which has relocations to text addresses in the same
> function as the original instruction) and other anonymous rodata (which
> may have relocations to elsewhere).
> 
> Link: https://github.com/ClangBuiltLinux/linux/issues/485
> Reported-by: Nick Desaulniers <ndesaulniers@...gle.com>
> Tested-by: Nick Desaulniers <ndesaulniers@...gle.com>
> Signed-off-by: Josh Poimboeuf <jpoimboe@...hat.com>

Tested with clang-11 and binutils 2.34, a combo that was previously
broken and I no longer see these warnings on either defconfig or
allyesconfig.

Tested-by: Nathan Chancellor <natechancellor@...il.com>