Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 15 Apr 2010 08:41:12 -0500
From:      Nathan Whitehorn <nwhitehorn@freebsd.org>
To:        John Baldwin <jhb@freebsd.org>
Cc:        pluknet <pluknet@gmail.com>, freebsd-current@freebsd.org
Subject:   Re: HEADS UP: COMPAT_IA32 renamed COMPAT_FREEBSD32
Message-ID:  <4BC71778.40505@freebsd.org>
In-Reply-To: <201004150913.50097.jhb@freebsd.org>
References:  <20100312171758.GB31089@dragon.NUXI.org> <201004071549.41672.jhb@freebsd.org> <i2ka31046fc1004150306l619e5b2btc158df99f06627f3@mail.gmail.com> <201004150913.50097.jhb@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 04/15/10 08:13, John Baldwin wrote:
> On Thursday 15 April 2010 6:06:24 am pluknet wrote:
>    
>> On 7 April 2010 23:49, John Baldwin<jhb@freebsd.org>  wrote:
>>      
>>> On Tuesday 06 April 2010 11:24:21 am Nathan Whitehorn wrote:
>>>        
>>>> pluknet wrote:
>>>>          
>>>>> Hi,
>>>>>
>>>>> the interesting part for me is how to properly assert now a value of
>>>>>            
> e.g.
>    
>>>>> KINFO_PROC_SIZE varying on err.. different COMPAT_FREEBSD32 arches
>>>>> (say, FreeBSD would have _kern_proc FreeBSD32 compat layer for
>>>>>            
> top/ps/).
>    
>>>>>
>>>>>            
>>>> Probably the cleanest thing would be to set KINFO_PROC_SIZE in
>>>> machine/proc.h instead of where it is now, and then also define a
>>>> KINFO_PROC32_SIZE or something in the same place. Also, that would be a
>>>> really nice feature.
>>>>          
>>> Yes, I think this sounds like the best approach.
>>>
>>>        
>> Something quick&  not clean (well, it passes universe) attached.
>> So, don't shoot me, please ;-).
>> It's unclear how to convert those mips o32/n32/o64/n64 though.
>> I had to make definitions out of _KERNEL visibility as far as
>> <sys/proc.h>  is included from<sys/user.h>  in !_KERNEL only too.
>>      
> Just one suggestion: don't make KINFO_PROC32 #define depenedent on
> COMPAT_FREEBSD32.  It should just be always defined.  I think that is the
> approach Nathan used for the 32-bit ELF machine type.
>    

I agree. There's no harm in making it a global definition. You also need 
a KINFO_PROC32 for ia64, which also implements i386 compatibility. Other 
than that, the patch looks good to me.
-Nathan



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BC71778.40505>