Re: [PATCH v3 1/4] kaslr: add immovable_mem=nn[KMG]@ss[KMG] to specify extracting memory

[Chao Fan]

On Thu, Dec 07, 2017 at 11:09:24AM +0800, Baoquan He wrote:
>On 12/07/17 at 10:53am, Chao Fan wrote:
>> On Wed, Dec 06, 2017 at 05:35:57PM +0800, Baoquan He wrote:
>> >Hi Chao,
>> >
>> >Yes, now the code looks much better than the last version.
>> >
>> >On 12/05/17 at 04:51pm, Chao Fan wrote:
>> >> In current code, kaslr may choose the memory region in movable
>> >> nodes to extract kernel, which will make the nodes can't be hot-removed.
>> >> To solve it, we can specify the memory region in immovable node.
>> >> Create immovable_mem to store the regions in immovable_mem, where should
>> >> be chosen by kaslr.
>> >> 
>> >> Multiple regions can be specified, comma delimited.
>> >> Considering the usage of memory, only support for 4 regions.
>> >> 4 regions contains 2 nodes at least, enough for kernel to extract.
>> >> 
>> >> Also change the "handle_mem_memmap" to "handle_mem_filter", since
>> >> it will not only handle memmap parameter now.
>> >
>> >One concern is whether it will fail to do KASLR if only specify
>> 
>> Sorry, I think I have not understood your point.
>> So if there is something wrong, please let me know.
>
>What I meant is whether we need check 'movable_node' and 
>'immovable_mem=' being specified together. If only specify 'movable_node',
>we may need to return and do not do kaslr or do not do physical kaslr
>since kernel could be located on movable mem region.

I think both are OK and have reasons, and I tend to not return.
Because if there is a parameter can solve the problem, but not specified.
It's a problem of user-level.
How do you think?

Thanks,
Chao Fan

>
>Otherwise it will do physical kaslr anyway, memory hotplug will be
>impacted later.
>
>> 
>> I don't think if only specify "movable_node" will fail KASLR.
>> Since in this patchset(3/4), only disable kernel mirror. KASLR in
>> current upstream code didn't parse "movable_node".
>> 
>> >"movable_node". Surely in this case it won't fail system, just hotplug
>> >memory might be impacted if kernel is located on that, will FJ mind
>> 
>> Yes, it's the reason why I make this patchset.
>> In my personal understanding, "movable_node" is a beginning why I make
>> this patchset, but not the whole reason.
>> Only "movable_node" specified might cause hotplug memory can't be
>> removed if kernel is located on that, so we need the help of
>> "immovable_mem=". "movable_node" help hotplug memory can be removed, and
>> "immovable_mem=" works for the same target, but just in kaslr.
>> So up to now, there is not a very tight coupling between "movable_node"
>> and "immovable_mem=". The independence of "immovable_mem=" is that,
>> help kaslr selects the right regions, avoid the memory in hotpluggable
>> NUMA nodes, which causes the memory can't removed. It's a independent
>> reason why we need a parameter like "immovable_mem=".
>> So I think we should also handle it if only specify "immovable_mem="
>> without "movable_node".
>> 
>> Thanks,
>> Chao Fan
>> 
>> >this? And what if only specify 'immovable_mem=' but without 'movable_node'?
>> >
>> >Thanks
>> >Baoquan
>> >
>> >> 
>> >> Signed-off-by: Chao Fan <fanc.fnst@xxxxxxxxxxxxxx>
>> >> ---
>> >>  arch/x86/boot/compressed/kaslr.c | 80 ++++++++++++++++++++++++++++++++++++++--
>> >>  1 file changed, 77 insertions(+), 3 deletions(-)
>> >> 
>> >> diff --git a/arch/x86/boot/compressed/kaslr.c b/arch/x86/boot/compressed/kaslr.c
>> >> index a63fbc25ce84..0bbbaf5f6370 100644
>> >> --- a/arch/x86/boot/compressed/kaslr.c
>> >> +++ b/arch/x86/boot/compressed/kaslr.c
>> >> @@ -108,6 +108,15 @@ enum mem_avoid_index {
>> >>  
>> >>  static struct mem_vector mem_avoid[MEM_AVOID_MAX];
>> >>  
>> >> +/* Only supporting at most 4 immovable memory regions with kaslr */
>> >> +#define MAX_IMMOVABLE_MEM	4
>> >> +
>> >> +/* Store the memory regions in immovable node */
>> >> +static struct mem_vector immovable_mem[MAX_IMMOVABLE_MEM];
>> >> +
>> >> +/* The immovable regions user specify, not more than 4 */
>> >> +static int num_immovable_region;
>> >> +
>> >>  static bool mem_overlaps(struct mem_vector *one, struct mem_vector *two)
>> >>  {
>> >>  	/* Item one is entirely before item two. */
>> >> @@ -168,6 +177,38 @@ parse_memmap(char *p, unsigned long long *start, unsigned long long *size)
>> >>  	return -EINVAL;
>> >>  }
>> >>  
>> >> +static int parse_immovable_mem(char *p,
>> >> +			       unsigned long long *start,
>> >> +			       unsigned long long *size)
>> >> +{
>> >> +	char *oldp;
>> >> +
>> >> +	if (!p)
>> >> +		return -EINVAL;
>> >> +
>> >> +	oldp = p;
>> >> +	*size = memparse(p, &p);
>> >> +	if (p == oldp)
>> >> +		return -EINVAL;
>> >> +
>> >> +	/* We support nn[KMG]@ss[KMG] and nn[KMG]. */
>> >> +	switch (*p) {
>> >> +	case '@':
>> >> +		*start = memparse(p + 1, &p);
>> >> +		return 0;
>> >> +	default:
>> >> +		/*
>> >> +		 * If w/o offset, only size specified, immovable_mem=nn[KMG]
>> >> +		 * has the same behaviour as immovable_mem=nn[KMG]@0. It means
>> >> +		 * the region starts from 0.
>> >> +		 */
>> >> +		*start = 0;
>> >> +		return 0;
>> >> +	}
>> >> +
>> >> +	return -EINVAL;
>> >> +}
>> >> +
>> >>  static void mem_avoid_memmap(char *str)
>> >>  {
>> >>  	static int i;
>> >> @@ -207,7 +248,37 @@ static void mem_avoid_memmap(char *str)
>> >>  		memmap_too_large = true;
>> >>  }
>> >>  
>> >> -static int handle_mem_memmap(void)
>> >> +#ifdef CONFIG_MEMORY_HOTPLUG
>> >> +static void parse_immovable_mem_regions(char *str)
>> >> +{
>> >> +	static int i;
>> >> +
>> >> +	while (str && (i < MAX_IMMOVABLE_MEM)) {
>> >> +		int rc;
>> >> +		unsigned long long start, size;
>> >> +		char *k = strchr(str, ',');
>> >> +
>> >> +		if (k)
>> >> +			*k++ = 0;
>> >> +
>> >> +		rc = parse_immovable_mem(str, &start, &size);
>> >> +		if (rc < 0)
>> >> +			break;
>> >> +		str = k;
>> >> +
>> >> +		immovable_mem[i].start = start;
>> >> +		immovable_mem[i].size = size;
>> >> +		i++;
>> >> +	}
>> >> +	num_immovable_region = i;
>> >> +}
>> >> +#else
>> >> +static inline void parse_immovable_mem_regions(char *str)
>> >> +{
>> >> +}
>> >> +#endif
>> >> +
>> >> +static int handle_mem_filter(void)
>> >>  {
>> >>  	char *args = (char *)get_cmd_line_ptr();
>> >>  	size_t len = strlen((char *)args);
>> >> @@ -215,7 +286,8 @@ static int handle_mem_memmap(void)
>> >>  	char *param, *val;
>> >>  	u64 mem_size;
>> >>  
>> >> -	if (!strstr(args, "memmap=") && !strstr(args, "mem="))
>> >> +	if (!strstr(args, "memmap=") && !strstr(args, "mem=") &&
>> >> +	    !strstr(args, "immovable_mem="))
>> >>  		return 0;
>> >>  
>> >>  	tmp_cmdline = malloc(len + 1);
>> >> @@ -240,6 +312,8 @@ static int handle_mem_memmap(void)
>> >>  
>> >>  		if (!strcmp(param, "memmap")) {
>> >>  			mem_avoid_memmap(val);
>> >> +		} else if (!strcmp(param, "immovable_mem")) {
>> >> +			parse_immovable_mem_regions(val);
>> >>  		} else if (!strcmp(param, "mem")) {
>> >>  			char *p = val;
>> >>  
>> >> @@ -379,7 +453,7 @@ static void mem_avoid_init(unsigned long input, unsigned long input_size,
>> >>  	/* We don't need to set a mapping for setup_data. */
>> >>  
>> >>  	/* Mark the memmap regions we need to avoid */
>> >> -	handle_mem_memmap();
>> >> +	handle_mem_filter();
>> >>  
>> >>  #ifdef CONFIG_X86_VERBOSE_BOOTUP
>> >>  	/* Make sure video RAM can be used. */
>> >> -- 
>> >> 2.14.3
>> >> 
>> >> 
>> >> 
>> >
>> >
>> 
>> 
>
>