From: Akio Takebe <takebe_akio@jp.fujitsu.com>
Subject: Re: [Patch][RFC] fix xc_ptrace_core for gdbserver-xen
	on?x86_64.
Date: Fri, 25 Aug 2006 18:03:36 +0900
Message-ID: <58C6C825592784takebe_akio@jp.fujitsu.com>
References: <20060825080039.17DF134047@koto.vergenet.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
In-Reply-To: <20060825080039.17DF134047@koto.vergenet.net>
Sender: xen-devel-bounces@lists.xensource.com
Errors-To: xen-devel-bounces@lists.xensource.com
To: Horms <horms@verge.net.au>
Cc: xen-devel@lists.xensource.com, Akio Takebe <takebe_akio@jp.fujitsu.com>

Hi, Horms

>
>It seems generaly consistent with the i386 code, and looks generaly
>fine, though I am not an expert on x86_64 either.
>
>There seem to be three distinct portions of your patch.
>
>1. Make nr_pages an unsigned long instead of a long
>
>   This seems like it should be a separate patch as its really
>   a trivial (but probably important) bugfix. Right?
>
Yes, This is trivial bugfix.

>2. Fill in map_domain_va_core() for x86_64
>
>3. Get rid of m2p_array.
>
>   I'm not sure what the relationship between 2 and 3 is.
>   Is m2p_array no longer needed because of the changes in 2?
>   Or is it a separate cleanup?
Yes, separate cleanup.

m2p_array for x86_64 is very big.
So I fix the code with the following way.

@@ -20,11 +19,18 @@ static unsigned long
 static unsigned long
 map_mtop_offset(unsigned long ma)
 {
-    return pages_offset + (m2p_array[ma >> PAGE_SHIFT] << PAGE_SHIFT);
-    return 0;
-}
-
-
+
+    unsigned long page_num;
+    for(page_num=0; page_num <nr_pages; page_num++){
+        if((ma >> PAGE_SHIFT) == p2m_array[page_num])
+		break;
+    }
+
+    return pages_offset + (page_num << PAGE_SHIFT);
+}
+
+

The way get page number of ma by searching the p2m_array without 
m2p_array.

Best Regards,

Akio Takebe