linux-kernel - Re: VMI broken on tip/master...

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: <49BECCB0.4010804@goop.org>
Date:	Mon, 16 Mar 2009 15:03:28 -0700
From:	Jeremy Fitzhardinge <jeremy@...p.org>
To:	Steven Rostedt <rostedt@...dmis.org>
CC:	Peter Zijlstra <a.p.zijlstra@...llo.nl>, akataria@...are.com,
	the arch/x86 maintainers <x86@...nel.org>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: VMI broken on tip/master...

Steven Rostedt wrote:
> On Sun, 15 Mar 2009, Peter Zijlstra wrote:
>   
>> Looking at arch/x86/include/asm/paravirt.h (god I wish paravirt would go
>> away, not only does it screw over ctags, it also hurts my brain), it
>> appears its playing icky games with primitives like
>> raw_local_irq_disable():
>>
>> static inline void raw_local_irq_disable(void)
>> {
>>         asm volatile(paravirt_alt(PARAVIRT_CALL)
>>                      :
>>                      : paravirt_type(pv_irq_ops.irq_disable),
>>                        paravirt_clobber(CLBR_EAX)
>>                      : "memory", "eax", "cc");
>> }
>>
>> So what was supposed to be a simple op, now gets expanded into god knows
>> what, and might lead to tracer recursion or something.
>>     
>
> It should only blow up if a guest is using tracing, and the code to call
> the hypervisor is also being traced.
>
>   
>> Maybe a simple notrace annotation somewhere in that paravirt code is all
>> it takes, who knows.
>>
>> Steve, you've been known to work on virt stuff too, happen to have a
>> bright idea? :-)
>>     
>
> I have noticed that some paravirt ops calls (like this 
> raw_local_irq_disable) does not get inlined. It sometimes gets made into a 
> function.  This would cause raw_local_irq_disable to actually be traced!
>
> One answer is to use "always_inline" or I can dig out a patch that makes 
> inline also include "notrace".

Yes, these should probably be always_inline and notrace (just in case 
gcc gets it into its head that it might be a good idea to put mcount 
calls into inlined functions).

    J
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/