Date:   Mon, 13 Nov 2017 17:50:37 +0800
From:   Chao Fan <fanc.fnst@...fujitsu.com>
To:     Baoquan He <bhe@...hat.com>
CC:     <linux-kernel@...r.kernel.org>, <x86@...nel.org>, <hpa@...or.com>,
        <tglx@...utronix.de>, <mingo@...hat.com>, <keescook@...omium.org>,
        <yasu.isimatu@...il.com>, <indou.takao@...fujitsu.com>,
        <caoj.fnst@...fujitsu.com>, <douly.fnst@...fujitsu.com>
Subject: Re: [PATCH v2 2/4] kaslr: select the memory region in immovable node
 to process

On Mon, Nov 13, 2017 at 05:26:24PM +0800, Baoquan He wrote:
>On 11/13/17 at 05:18pm, Chao Fan wrote:
>> On Mon, Nov 13, 2017 at 04:31:31PM +0800, Baoquan He wrote:
>> >On 11/01/17 at 07:32pm, Chao Fan wrote:
>> >> Compare the region of memmap entry and immovable_mem, then choose the
>> >> intersection to process_mem_region.
>> >> 
>> >> Since the interrelationship between e820 or efi entries and memory
>> >> region in immovable_mem is different:
>> >> One memory region in one node may contain several entries of e820 or
>> >> efi sometimes, and one entry of e820 or efi may contain the memory in
>> >> different nodes sometimes.
>> >> It may split one node or one entry to several regions.
>> >> 
>> >> Signed-off-by: Chao Fan <fanc.fnst@...fujitsu.com>
>> >> ---
>> >>  arch/x86/boot/compressed/kaslr.c | 60 ++++++++++++++++++++++++++++++++++------
>> >>  1 file changed, 52 insertions(+), 8 deletions(-)
>> >> 
>> >> diff --git a/arch/x86/boot/compressed/kaslr.c b/arch/x86/boot/compressed/kaslr.c
>> >> index 0a591c0023f1..fcd640fdeaed 100644
>> >> --- a/arch/x86/boot/compressed/kaslr.c
>> >> +++ b/arch/x86/boot/compressed/kaslr.c
>> >> @@ -634,6 +634,54 @@ static void process_mem_region(struct mem_vector *entry,
>> >>  	}
>> >>  }
>> >>  
>> >> +static bool select_immovable_node(struct mem_vector region,
>> >> +				  unsigned long long minimum,
>> >> +				  unsigned long long image_size)
>> >> +{
>> >
>> >About this patch, I just want to notice two things:
>> >1) From the current code, 'movable_node' kernel parameter is exclusive.
>> >In find_zone_movable_pfns_for_nodes(), you can see that 'kernelcore' and
>> >'movablecore' will be ignored as long as 'movable_node' is specified.
>> >Please also consider this in your code here. If 'movable_node' has to be
>> >specified too, and need skip the kernel mirror handling.
>> >
>> 
>> Thanks for the notice, I will add the similar operation.
>
>No, I meant if movable_node is specified, we have to ignore
>'kernelcore=', then we may need skip the kernel mirror handling. Since
>kernel mirror is enabled by 'kernelcore=mirror'.
>

Yes, the two parameters are conflict.
Then if "movable_node" specified, we shoul not consider mirror region.
I think there can be one another patch to handle this, but not in this
serials. I will add the change.

Thanks,
Chao Fan

>
>> 
>> >2)process_mem_region() is a key function to process the available memory
>> >regions. Please don't make another process_mem_region() like below. You
>> >can write a small helper function to find the immovable_mem[] which is
>> >intersecting with the passed in memory region and use clamp() to get
>> >the real available region. You 'REALLY' don't need to split the region
>> >in your so called 'select_immovable_node()' function here.
>> >
>> 
>> OK, I will try to make a new method, which is smaller and better to
>> filter the regions.
>> 
>> Thanks,
>> Chao Fan
>> 
>> >PLEASE elaborate more on these details before post.
>> >
>> >Thanks
>> >Baoquan
>> >
>> >> +	int i;
>> >> +
>> >> +	/* If no immovable_mem stored, use region directly */
>> >> +	if (num_immovable_region == 0) {
>> >> +		process_mem_region(&region, minimum, image_size);
>> >> +
>> >> +		if (slot_area_index == MAX_SLOT_AREA) {
>> >> +			debug_putstr("Aborted memmap scan (slot_areas full)!\n");
>> >> +			return 1;
>> >> +		}
>> >> +	} else {
>> >> +		/*
>> >> +		 * Walk all immovable regions, and filter the intersection
>> >> +		 * to process_mem_region.
>> >> +		 */
>> >> +		for (i = 0; i < num_immovable_region; i++) {
>> >> +			struct mem_vector entry;
>> >> +			unsigned long long start, end, select_end, region_end;
>> >> +
>> >> +			region_end = region.start + region.size - 1;
>> >> +			start = immovable_mem[i].start;
>> >> +			end = start + immovable_mem[i].size - 1;
>> >> +
>> >> +			if (region_end < start || region.start > end)
>> >> +				continue;
>> >> +
>> >> +			/* May split one region to several entries. */
>> >> +			entry.start = start > region.start ?
>> >> +				      start : region.start;
>> >> +			select_end = end > region_end ? region_end : end;
>> >> +
>> >> +			entry.size = select_end - entry.start + 1;
>> >> +
>> >> +			process_mem_region(&entry, minimum, image_size);
>> >> +
>> >> +			if (slot_area_index == MAX_SLOT_AREA) {
>> >> +				debug_putstr("Aborted memmap scan (slot_areas full)!\n");
>> >> +				return 1;
>> >> +			}
>> >> +		}
>> >> +	}
>> >> +	return 0;
>> >> +}
>> >> +
>> >>  #ifdef CONFIG_EFI
>> >>  /*
>> >>   * Returns true if mirror region found (and must have been processed
>> >> @@ -699,11 +747,9 @@ process_efi_entries(unsigned long minimum, unsigned long image_size)
>> >>  
>> >>  		region.start = md->phys_addr;
>> >>  		region.size = md->num_pages << EFI_PAGE_SHIFT;
>> >> -		process_mem_region(&region, minimum, image_size);
>> >> -		if (slot_area_index == MAX_SLOT_AREA) {
>> >> -			debug_putstr("Aborted EFI scan (slot_areas full)!\n");
>> >> +
>> >> +		if (select_immovable_node(region, minimum, image_size))
>> >>  			break;
>> >> -		}
>> >>  	}
>> >>  	return true;
>> >>  }
>> >> @@ -730,11 +776,9 @@ static void process_e820_entries(unsigned long minimum,
>> >>  			continue;
>> >>  		region.start = entry->addr;
>> >>  		region.size = entry->size;
>> >> -		process_mem_region(&region, minimum, image_size);
>> >> -		if (slot_area_index == MAX_SLOT_AREA) {
>> >> -			debug_putstr("Aborted e820 scan (slot_areas full)!\n");
>> >> +
>> >> +		if (select_immovable_node(region, minimum, image_size))
>> >>  			break;
>> >> -		}
>> >>  	}
>> >>  }
>> >>  
>> >> -- 
>> >> 2.13.6
>> >> 
>> >> 
>> >> 
>> >
>> >
>> 
>> 
>
>

Powered by blists - more mailing lists