linux-kernel - Re: [PATCH] ARC: ARCv2: jump label: implement jump label patching

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: <BN6PR1201MB0035E304FF25609CBCACBD09B6C90@BN6PR1201MB0035.namprd12.prod.outlook.com>
Date:   Wed, 17 Jul 2019 17:45:52 +0000
From:   Vineet Gupta <Vineet.Gupta1@...opsys.com>
To:     Eugeniy Paltsev <Eugeniy.Paltsev@...opsys.com>,
        "linux-snps-arc@...ts.infradead.org" 
        <linux-snps-arc@...ts.infradead.org>
CC:     "peterz@...radead.org" <peterz@...radead.org>,
        Alexey Brodkin <Alexey.Brodkin@...opsys.com>,
        "jbaron@...hat.com" <jbaron@...hat.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "pbonzini@...hat.com" <pbonzini@...hat.com>,
        "ard.biesheuvel@...aro.org" <ard.biesheuvel@...aro.org>,
        "linux-arch@...r.kernel.org" <linux-arch@...r.kernel.org>
Subject: Re: [PATCH] ARC: ARCv2: jump label: implement jump label patching

On 7/17/19 8:09 AM, Eugeniy Paltsev wrote:
>>> +/* Halt system on fatal error to make debug easier */
>>> +#define arc_jl_fatal(format...)						\
>>> +({									\
>>> +	pr_err(JUMPLABEL_ERR format);					\
>>> +	BUG();								\
>> Does it make sense to bring down the whole system vs. failing and returning.
>> I see there is no error propagation to core code, but still.
> I totally agree with Peter, and I prefer to stop the system on this errors. Here is few arguments:
> All this checks can't be toggle in system operating normally and only may be caused by bad code generation (or some code/data corruption):
> 1) We got our instruction to patch unaligned by 4 bytes (despite the fact that we used '.balign 4' to align it)
> 2) We got branch offset unaligned (which means that we calculate it wrong at build time or corrupt it in run time)
> 3) We got branch offset which not fits into s25. As this is offset inside one function (inside one 'if' statement actually) we newer get such huge
> offset in real life if code generation is ok.

I understand that. But AFAIKR in implementation arc_jl_fatal() gets called before
we have done the actual code patching and/or flushing the caches to that effect.
So harm has been done just yet. We just need to make sure that any book-keeping of
true/false is not yet done or unrolled.

> If we only warn to log and return we will face with compromised kernel flow later.
> I.E. it could be 'if' statement in kernel text which is switched to wrong state: the condition is true but we take the false branch.
> And It will be the issue which is much more difficult to debug.
>
> Does it sound reasonably?

I'm still not convinced that by hitting the _fatal() we are in some inconsistent
state already. But if u feel strongly lets keep it.

>
> If you really don't want to have BUG here, I can make it conditionally enabled

Not a good idea. It is unconditionally present or not. Not something in between.

> in depend on CONFIG_ARC_STATIC_KEYS_DEBUG as I want to have it enabled at least on ARC development platforms.