From mboxrd@z Thu Jan  1 00:00:00 1970
From: pbonzini@redhat.com (Paolo Bonzini)
Date: Wed, 16 Jun 2010 14:23:18 +0200
Subject: [ltt-dev] [PATCH RFC] Force build failure on unknown
	architectures
In-Reply-To: <20100616001228.GM2304@linux.vnet.ibm.com>
References: <20100615231423.GA16064@linux.vnet.ibm.com>
	<4C1812E1.5080103@redhat.com>
	<20100616001228.GM2304@linux.vnet.ibm.com>
Message-ID: <4C18C236.3030100@redhat.com>

On 06/16/2010 02:12 AM, Paul E. McKenney wrote:
> On Wed, Jun 16, 2010 at 01:55:13AM +0200, Paolo Bonzini wrote:
>> On 06/16/2010 01:14 AM, Paul E. McKenney wrote:
>>> Create urcu/arch_unknown.h and urcu/uatomic_arch_unknown.h, which
>>> contain only #error statements and explanatory comments.  This forces
>>> build failures on unrecognized architectures in preference to trying
>>> to guess at what operations might be safe on such architectures.
>>>
>>> One other semi-feasible alternative is to use hashed arrays of locks
>>> that are acquired with signals disabled.  However, this seems a bit
>>> too ornate, especially for architectures for which the gcc __sync_
>>> primitives work correctly.
>>
>> ia64 is one.
>
> Does ia64 work with the current code?  (My guess is "no" -- I believe
> that you would get build errors.  But I have been surprised before!)

I don't know. :)

> If my guess is correct, my thought would be to create urcu/arch_gcc.h
> and urcu/uatomic_arch_gcc.h files with the appropriate definitions
> based on __sync_ primitives.  If you can test on IA64, I would also
> be happy to add support for it based on these new files.

Sure, thanks!

Paolo