Re: [RFC][PATCH] x86: introduce parse_early_param_alone() to parse param early (Was Re: linux-next: reservetop fix disables mem= )

["Américo Wang" <xiyou.wangcong@gmail.com>]

On Tue, Sep 01, 2009 at 03:55:14PM +0800, Xiao Guangrong wrote:
>Hi Ingo,
>
>I find the broken patch is still in -tip tree:
>
>Commit-ID:  8126dec32738421afa362114337331337b4be17f
>Gitweb:     http://git.kernel.org/tip/8126dec32738421afa362114337331337b4be17f
>Author:     Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
>AuthorDate: Thu, 20 Aug 2009 20:23:11 +0800
>Committer:  Ingo Molnar <mingo@elte.hu>
>CommitDate: Fri, 21 Aug 2009 16:40:30 +0200
>
>x86: Fix system crash when loading with "reservetop" parameter
>
>So, this patch is base on it.
>---------------------------------------------
>Subject: [RFC][PATCH] x86: introduce parse_early_param_alone() to parse param before parse_early_param()
>
>Sometimes, we should parse specific boot parameter alone before
>parse_early_param(), such as "reservetop" and "early_ioremap_debug",
>like below:
>
>> The main problem is that can't parse 'reservetop' and 'mem'/'memmap'
>> parameters at the same time, because we should parse 'mem'/'memmap'
>> after e820 map detected, and it might use early_memremap() during
>> e820 map detecting, so we should call early_ioremap_init() before it,
>> 'reservetop' parameter can change 'FIXADDR_TOP', so we parse
>> 'reservetop' first,
>> 
>> like below sequences:
>>   ......
>>   ___parse 'reservetop'___  ->
>>   early_ioremap_init()	  ->
>>   setup_memory_map(); /* might call early_memremap()*/  ->
>>   parse_setup_data(); /* might call early_memremap()*/  ->
>>   e820_reserve_setup_data();	->
>>   ___parse 'mem'/'memmap' and other parameters___
>>   ......
>> 
>> (the same as 'early_ioremap_debug' parameter)
>> 
>> So, I'm afraid that we need separate parse 'reservetop'/
>> 'early_ioremap_debug' parameters early
>> 
>
>This patch introduce a new api named parse_early_param_alone() to
>do this.
>
>Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
>---
> arch/x86/kernel/setup.c |   33 ++++++++++++++++++------------
> include/linux/init.h    |    1 +
> init/main.c             |   50 +++++++++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 71 insertions(+), 13 deletions(-)
>
>diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
>index 83ea092..f241c12 100644
>--- a/arch/x86/kernel/setup.c
>+++ b/arch/x86/kernel/setup.c
>@@ -696,10 +696,28 @@ void __init setup_arch(char **cmdline_p)
> 	printk(KERN_INFO "Command line: %s\n", boot_command_line);
> #endif
> 
>+#ifdef CONFIG_CMDLINE_BOOL
>+#ifdef CONFIG_CMDLINE_OVERRIDE
>+	strlcpy(boot_command_line, builtin_cmdline, COMMAND_LINE_SIZE);
>+#else
>+	if (builtin_cmdline[0]) {
>+		/* append boot loader cmdline to builtin */
>+		strlcat(builtin_cmdline, " ", COMMAND_LINE_SIZE);
>+		strlcat(builtin_cmdline, boot_command_line, COMMAND_LINE_SIZE);
>+		strlcpy(boot_command_line, builtin_cmdline, COMMAND_LINE_SIZE);
>+	}
>+#endif
>+#endif


This seems ugly.

CMDLINE_OVERRIDE depends on CMDLINE_BOOL, right? So the outer #ifdef
can be removed. :)


-- 
Live like a child, think like the god.