On 03/28/2016 04:54 PM, Konrad Rzeszutek Wilk wrote:
On Tue, Mar 29, 2016 at 01:55:07AM +0800, fu.wei@linaro.org wrote:
From: Fu Wei fu.wei@linaro.org
This patch add a check_xsm_signature static function for detecting XSM from the second unknown module.
If xen can't get the kind of module from compatible, we guess the kind of these first two unknown respectively: (1) The first unknown must be kernel; (2) The second unknown is ramdisk, only if we have ramdisk; (3) Start from the 2nd unknown, detect the XSM binary signature; (4) If we got XSM in the 2nd unknown, that means we don't load initrd.
Pls make the 'xen' be 'Xen'.
Signed-off-by: Fu Wei fu.wei@linaro.org
Cc-ing also Daniel (XSM maintainer).
And Julien (linaro.org != arm.com) and Stefano.
v2: Using XEN_MAGIC macro instead of 0xf97cff8c : uint32_t selinux_magic = 0xf97cff8c; --> uint32_t xen_magic = XEN_MAGIC; Comment out the code(return 0 directly), if CONFIG_FLASK is not set.
v1: http://lists.xen.org/archives/html/xen-devel/2016-03/msg02430.html The first upstream patch to xen-devel mailing lists.
xen/arch/arm/bootfdt.c | 57 +++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 56 insertions(+), 1 deletion(-)
diff --git a/xen/arch/arm/bootfdt.c b/xen/arch/arm/bootfdt.c index 8a14015..322f17f 100644 --- a/xen/arch/arm/bootfdt.c +++ b/xen/arch/arm/bootfdt.c @@ -163,6 +163,52 @@ static void __init process_memory_node(const void *fdt, int node, } }
+/**
- check_xsm_signature - Check XSM Magic and Signature of the module header
- A XSM module has a special header
- uint magic | uint target_len | uchar target[8] |
- 0xf97cff8c | 8 | "XenFlask" |
- 0xf97cff8c is policy magic number.
- So we only read the first 16 Bytes of the module, then check these three
s/Bytes/bytes/
- parts.
Is it possible for the hypervisor to chnage the policy magic number? Perhaps you should have :
BUILD_BUG_ON(0xf97cff8c != XSM_MAGIC);
to guard against changes?
The value of XSM_MAGIC will always be that constant if FLASK is the enabled security module; the value was different when the (now-removed) ACM module was selected.
[...]
- if (strncmp(buff, (char *) &xen_magic, sizeof(u32)) ||
strncmp(buff + sizeof(u32), (char *) &target_len, sizeof(u32)) ||strncmp(buff + sizeof(u32) * 2, "XenFlask", target_len))return 0;
memcmp() is more correct than strncmp() here, especially since target_len will have embedded NULLs. It also assumes little endian byte order; is that worth commenting on?