[PATCH v3 1/2] powerpc/fadump: reduce memory consumption for capture kernel

[Michael Ellerman]

Hari Bathini <hbathini at linux.vnet.ibm.com> writes:

> On Wednesday 26 April 2017 04:02 PM, Michael Ellerman wrote:
>> Hari Bathini <hbathini at linux.vnet.ibm.com> writes:
>>
>>> diff --git a/arch/powerpc/kernel/fadump.c b/arch/powerpc/kernel/fadump.c
>>> index 8ff0dd4..87edc7b 100644
>>> --- a/arch/powerpc/kernel/fadump.c
>>> +++ b/arch/powerpc/kernel/fadump.c
>>> @@ -338,6 +343,36 @@ unsigned long __init arch_reserved_kernel_pages(void)
>>>   	return memblock_reserved_size() / PAGE_SIZE;
>>>   }
>>>   
>>> +static void __init parse_fadump_append_params(const char *p)
>>> +{
>>> +	static char fadump_cmdline[COMMAND_LINE_SIZE] __initdata;
>>> +	char *token;
>>> +
>>> +	strlcpy(fadump_cmdline, p, COMMAND_LINE_SIZE);
>>> +	token = strchr(fadump_cmdline, ',');
>>> +	while (token) {
>>> +		*token = ' ';
>>> +		token = strchr(token, ',');
>>> +	}
>>> +
>>> +	pr_info("enforcing additional parameters (%s) passed through "
>>> +		"'fadump_append=' parameter\n", fadump_cmdline);
>>> +	parse_early_options(fadump_cmdline);
>> Using parse_early_options() means it only works for parameters defined
>> with early_param() or __setup() doesn't it?
>
> Yeah. Actually, is it better to change commandline from "$params 
> fadump_append=nr_cpus=1,numa=off"
> to "$params nr_cpus=1 numa=off" early in the boot process?

It's not better, it's more confusing for a user, because the command
line looks different to what they specified.

But it might be the only option because I don't know if there's an easy
way to trigger parsing for the non-early options.

cheers