linux-kernel - Re: [PATCH v1 04/19] x86/insn-eval: Handle return values from the decoder

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20201228190615.GC20321@zn.tnic>
Date:   Mon, 28 Dec 2020 20:06:15 +0100
From:   Borislav Petkov <bp@...en8.de>
To:     Sean Christopherson <seanjc@...gle.com>
Cc:     Andy Lutomirski <luto@...capital.net>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        X86 ML <x86@...nel.org>, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v1 04/19] x86/insn-eval: Handle return values from the
 decoder

On Mon, Dec 28, 2020 at 10:51:15AM -0800, Sean Christopherson wrote:
> This patch is incomplete/inconsistent, and arguably wrong.
> 
>   - get_eff_addr_reg() and get_eff_addr_modrm() still ignore the return of
>     insn_get_modrm() after this patch.

Ah, will fix, thx.

>   - Calling insn_get_modrm() from get_eff_addr_sib() is unnecessary (unless the
>     caller passed uninitialized garbage in @insn) as get_eff_addr_sib() is
>     called if and only if sib.nbytes!=0, and sib.nbytes can be non-zero if and
>     only if the modrm and sib have been got.
> 
>   - get_addr_ref_16() does insn_get_displacement, i.e. guarantees the modrm is
>     parsed, while the 32/64 variants do not.
> 
> What about adding a prereq patch (or three) to call insn_get_displacement() in
> insn_get_addr_ref() prior to switching on insn->addr_bytes?  Then the various
> internal helpers could be changed to either omit the sanity checks entirely or
> WARN on invalid calls?  Or better yet, add an INSN_WARN_ON() macro that compiles
> out the checks by default?  E.g. something like:

So the idea is one construction site at a time (that's a German saying :)).

This set deals with whacking the insn decoder into returning
proper error/success values. The next set should do
simplifications/fixes/cleanups/whatever but not all at the same time for
obvious reasons.

So yeah, I'm all for omitting useless code but let's do that ontop.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette