FYI, this patch also clean applied to 4.9.y tree. I guess I don't need to send another one to 4.9 tree. Just applied it to 4.9 should be fine.
On Fri, Jan 12, 2018 at 1:48 PM, lepton ytht.net@gmail.com wrote:
The last version is "V4" and I added it in subject.
On Fri, Jan 12, 2018 at 1:46 PM, Guenter Roeck groeck@google.com wrote:
On Fri, Jan 12, 2018 at 1:41 PM, Lepton Wu ytht.net@gmail.com wrote:
This finally resolve crash if loaded under qemu + haxm. Haitao Shan pointed out that the reason of that crash is that NX bit get set for page tables. It seems we missed checking if _PAGE_NX is supported in kaiser_add_user_map
Link: https://www.spinics.net/lists/kernel/msg2689835.html
Signed-off-by: Lepton Wu ytht.net@gmail.com
Reviewed-by: Guenter Roeck groeck@chromium.org
Hint: You might want to version your patches; otherwise it is difficult for the maintainer(s) to find the latest version.
Note that the fix probably also applies to v4.9 as well as to older stable kernels.
Guenter
arch/x86/mm/kaiser.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/x86/mm/kaiser.c b/arch/x86/mm/kaiser.c index 6a7a77929a8c..8af98513d36c 100644 --- a/arch/x86/mm/kaiser.c +++ b/arch/x86/mm/kaiser.c @@ -198,6 +198,8 @@ static int kaiser_add_user_map(const void *__start_addr, unsigned long size, * requires that not to be #defined to 0): so mask it off here. */ flags &= ~_PAGE_GLOBAL;
if (!(__supported_pte_mask & _PAGE_NX))flags &= ~_PAGE_NX; for (; address < end_addr; address += PAGE_SIZE) { target_address = get_pa_from_mapping(address);-- 2.16.0.rc1.238.g530d649a79-goog