From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jan Kiszka <jan.kiszka@siemens.com>
Subject: Re: [PATCH] kvm-userspace: x86: Support for breakpoints in ROM code
Date: Tue, 17 Mar 2009 11:43:03 +0100
Message-ID: <49BF7EB7.4060603@siemens.com>
References: <49BE9EF7.7010005@siemens.com> <49BEA02A.1070909@siemens.com> <49BF6E75.1070802@redhat.com> <49BF7109.5000207@siemens.com> <49BF72B6.4040907@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-15
Content-Transfer-Encoding: 7bit
Cc: kvm-devel <kvm@vger.kernel.org>
To: Avi Kivity <avi@redhat.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from gecko.sbs.de ([194.138.37.40]:18995 "EHLO gecko.sbs.de"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1755361AbZCQKnW (ORCPT <rfc822;kvm@vger.kernel.org>);
	Tue, 17 Mar 2009 06:43:22 -0400
In-Reply-To: <49BF72B6.4040907@redhat.com>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

Avi Kivity wrote:
> Jan Kiszka wrote:
>> This is not a hack (it shouldn't have been merged upstream otherwise):
>> cpu_physical_memory_write_rom() takes system-wide physical addresses
>> while kvm_patch_opcode_byte() works with per-CPU linear addresses.
>>   
> 
> From exec.c:
> 
>> /* virtual memory access for debug */
>> int cpu_memory_rw_debug(CPUState *env, target_ulong addr,
>>                         uint8_t *buf, int len, int is_write)
>> {
>>     int l;
>>     target_phys_addr_t phys_addr;
>>     target_ulong page;
>>
>>     while (len > 0) {
>>         page = addr & TARGET_PAGE_MASK;
>>         phys_addr = cpu_get_phys_page_debug(env, page);
>>         /* if no physical page mapped, return an error */
>>         if (phys_addr == -1)
>>             return -1;
>>         l = (page + TARGET_PAGE_SIZE) - addr;
>>         if (l > len)
>>             l = len;
>>         cpu_physical_memory_rw(phys_addr + (addr & ~TARGET_PAGE_MASK),
>>                                buf, l, is_write);
>>         len -= l;
>>         buf += l;
>>         addr += l;
>>     }
>>     return 0;
>> }
> 
> I'm talking about replacing cpu_physical_memory_rw() with
> cpu_physical_memory_write_rom() (for the write case).

Ah, miss-read your point.

>  Is there a case
> where the debugger should be prevented from writing into ROM?  If so,
> maybe cpu_memory_rw_debug_rom() for breakpoints?

Good and valid question. So far only the 'M' gdb packet uses
cpu_physical_memory_rw in write mode, and I don't know if there is any
reason to deny modifying ROM code that way. Hmm, will check with
upstream and change there first. So forget about this patch for now.

Jan

-- 
Siemens AG, Corporate Technology, CT SE 2
Corporate Competence Center Embedded Linux