Re: [Qemu-devel] [PATCH] linux-user: elfload: Still use TARGET_PAGE_SIZE for i386 guest

[gchen gchen]

On 2015年08月12日 23:06, Richard Henderson wrote:<br>&gt; On 08/12/2015 12:59 AM, 
gchen gchen wrote:<br>&gt;&gt;&gt; Nack.  There's 99 problems with host page 
size&gt; guest page size.  This<br>&gt;&gt;&gt; solves none of them, and in the 
hackiest way possible.<br>&gt;&gt;&gt;<br>&gt;&gt;<br>&gt;&gt; Under alpha 
virtual machine, if set i386 guest page size 8KB, it will<br>&gt;&gt; cause 
failure directly (any dynamically linked binaries can not 
work).<br>&gt;<br>&gt; Yes, I know.  The same thing happens when running i386 
guests on other<br>&gt; (non-virtual) hosts.  E.g. Sparc64's 8kB page, 
PowerPC64's 64kB page.<br>&gt;<br><br>Yes. The reason why I am only focus on 
Alpha is "the machine which I am<br>working for is almost the same as Alpha. 
But this machine is very slow,<br>its performance maybe like 10 years ago's 
x86_64 laptop".<br><br>&gt;&gt; Do you have any other ideas for solving this 
issue?<br>&gt;<br>&gt; The only complete solution that I see is to use softmmu 
with linux-user, so<br>&gt; that we properly emulate the guest pages.  Yes, it 
will cause quite some<br>&gt; slow-down in emulation, but I believe it's the 
only reliable way.<br>&gt;<br><br>I have tried softmmu, for me, the performance 
is not acceptable, we can<br>not use this way.<br><br>Our main goal is "let 
most of Windows XP graphic free programs (but not<br>open source), can be used 
by user under Alpha like machine, also its<br>performance and stability are 
acceptable".<br><br><br>Thanks.<br>--<br>Chen Gang<br><br>Open, share, and 
attitude like air, water, and life which God blessed<br>