diff mbox

mm: Add the ram_latent_entropy kernel parameter

Message ID 20160810222805.GA13733@www.outflux.net
State New, archived
Headers show

Commit Message

Kees Cook Aug. 10, 2016, 10:28 p.m. UTC
From: Emese Revfy <re.emese@gmail.com>

When "ram_latent_entropy" is passed on the kernel command line, entropy
will be extracted from up to the first 4GB of RAM while the runtime memory
allocator is being initialized. This entropy isn't cryptographically
secure, but does help provide additional unpredictability on otherwise
low-entropy systems.

Based on work created by the PaX Team.

Signed-off-by: Emese Revfy <re.emese@gmail.com>
[kees: renamed parameter, dropped relationship with plugin, updated log]
Signed-off-by: Kees Cook <keescook@chromium.org>
---
This patch has been extracted from the latent_entropy gcc plugin, as
suggested by Linus: https://lkml.org/lkml/2016/8/8/840
---
 Documentation/kernel-parameters.txt |  5 +++++
 mm/page_alloc.c                     | 21 +++++++++++++++++++++
 2 files changed, 26 insertions(+)

Comments

Dave Hansen Aug. 11, 2016, 3:39 p.m. UTC | #1
On 08/10/2016 03:28 PM, Kees Cook wrote:
> +	if (ram_latent_entropy && !PageHighMem(page) &&
> +		page_to_pfn(page) < 0x100000) {
> +		u64 hash = 0;
> +		size_t index, end = PAGE_SIZE * nr_pages / sizeof(hash);
> +		const u64 *data = lowmem_page_address(page);
> +
> +		for (index = 0; index < end; index++)
> +			hash ^= hash + data[index];
> +		add_device_randomness((const void *)&hash, sizeof(hash));
> +	}

When I was first reading this, I thought it was using the _addresses_ of
the freed memory for entropy.  But it's actually using the _contents_.
The description could probably use a wee bit of sprucing up.

It might also be nice to say in the patch description (and the
Documentation/) what you expect to be in this memory.  It will obviously
be zeros for the vast majority of the space, but I do wonder what else
ends up in there in practice.

Why is it limited to 4GB?  Just so it doesn't go and try to XOR the
contents of a multi-TB system if it got turned on there? :)
Balbir Singh Aug. 12, 2016, 3:52 a.m. UTC | #2
On 11/08/16 08:28, Kees Cook wrote:
> From: Emese Revfy <re.emese@gmail.com>
> 
> When "ram_latent_entropy" is passed on the kernel command line, entropy
> will be extracted from up to the first 4GB of RAM while the runtime memory
> allocator is being initialized. This entropy isn't cryptographically
> secure, but does help provide additional unpredictability on otherwise
> low-entropy systems.
> 
> Based on work created by the PaX Team.
> 
> Signed-off-by: Emese Revfy <re.emese@gmail.com>
> [kees: renamed parameter, dropped relationship with plugin, updated log]
> Signed-off-by: Kees Cook <keescook@chromium.org>
> ---
> This patch has been extracted from the latent_entropy gcc plugin, as
> suggested by Linus: https://lkml.org/lkml/2016/8/8/840
> ---
>  Documentation/kernel-parameters.txt |  5 +++++
>  mm/page_alloc.c                     | 21 +++++++++++++++++++++
>  2 files changed, 26 insertions(+)
> 
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index 46c030a49186..9d054984370f 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -3245,6 +3245,11 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
>  	raid=		[HW,RAID]
>  			See Documentation/md.txt.
>  
> +	ram_latent_entropy
> +			Enable a very simple form of latent entropy extraction
> +			from the first 4GB of memory as the bootmem allocator
> +			passes the memory pages to the buddy allocator.
> +
>  	ramdisk_size=	[RAM] Sizes of RAM disks in kilobytes
>  			See Documentation/blockdev/ramdisk.txt.
>  
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index fb975cec3518..1de94f0ff29d 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -64,6 +64,7 @@
>  #include <linux/page_owner.h>
>  #include <linux/kthread.h>
>  #include <linux/memcontrol.h>
> +#include <linux/random.h>
>  
>  #include <asm/sections.h>
>  #include <asm/tlbflush.h>
> @@ -1236,6 +1237,15 @@ static void __free_pages_ok(struct page *page, unsigned int order)
>  	local_irq_restore(flags);
>  }
>  
> +bool __meminitdata ram_latent_entropy;
> +
> +static int __init setup_ram_latent_entropy(char *str)
> +{
> +	ram_latent_entropy = true;
> +	return 0;
> +}
> +early_param("ram_latent_entropy", setup_ram_latent_entropy);
> +
>  static void __init __free_pages_boot_core(struct page *page, unsigned int order)
>  {
>  	unsigned int nr_pages = 1 << order;
> @@ -1251,6 +1261,17 @@ static void __init __free_pages_boot_core(struct page *page, unsigned int order)
>  	__ClearPageReserved(p);
>  	set_page_count(p, 0);
>  
> +	if (ram_latent_entropy && !PageHighMem(page) &&
> +		page_to_pfn(page) < 0x100000) {
> +		u64 hash = 0;
> +		size_t index, end = PAGE_SIZE * nr_pages / sizeof(hash);
> +		const u64 *data = lowmem_page_address(page);
> +
> +		for (index = 0; index < end; index++)
> +			hash ^= hash + data[index];

Won't the hash be the same across boots? Is this entropy addition for
KASLR, since it is so early in boot?q

> +		add_device_randomness((const void *)&hash, sizeof(hash));
> +	}
> +
>  	page_zone(page)->managed_pages += nr_pages;
>  	set_page_refcounted(page);
>  	__free_pages(page, order);
> 


Balbir Singh
diff mbox

Patch

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 46c030a49186..9d054984370f 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -3245,6 +3245,11 @@  bytes respectively. Such letter suffixes can also be entirely omitted.
 	raid=		[HW,RAID]
 			See Documentation/md.txt.
 
+	ram_latent_entropy
+			Enable a very simple form of latent entropy extraction
+			from the first 4GB of memory as the bootmem allocator
+			passes the memory pages to the buddy allocator.
+
 	ramdisk_size=	[RAM] Sizes of RAM disks in kilobytes
 			See Documentation/blockdev/ramdisk.txt.
 
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index fb975cec3518..1de94f0ff29d 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -64,6 +64,7 @@ 
 #include <linux/page_owner.h>
 #include <linux/kthread.h>
 #include <linux/memcontrol.h>
+#include <linux/random.h>
 
 #include <asm/sections.h>
 #include <asm/tlbflush.h>
@@ -1236,6 +1237,15 @@  static void __free_pages_ok(struct page *page, unsigned int order)
 	local_irq_restore(flags);
 }
 
+bool __meminitdata ram_latent_entropy;
+
+static int __init setup_ram_latent_entropy(char *str)
+{
+	ram_latent_entropy = true;
+	return 0;
+}
+early_param("ram_latent_entropy", setup_ram_latent_entropy);
+
 static void __init __free_pages_boot_core(struct page *page, unsigned int order)
 {
 	unsigned int nr_pages = 1 << order;
@@ -1251,6 +1261,17 @@  static void __init __free_pages_boot_core(struct page *page, unsigned int order)
 	__ClearPageReserved(p);
 	set_page_count(p, 0);
 
+	if (ram_latent_entropy && !PageHighMem(page) &&
+		page_to_pfn(page) < 0x100000) {
+		u64 hash = 0;
+		size_t index, end = PAGE_SIZE * nr_pages / sizeof(hash);
+		const u64 *data = lowmem_page_address(page);
+
+		for (index = 0; index < end; index++)
+			hash ^= hash + data[index];
+		add_device_randomness((const void *)&hash, sizeof(hash));
+	}
+
 	page_zone(page)->managed_pages += nr_pages;
 	set_page_refcounted(page);
 	__free_pages(page, order);