Two bug fixes:
1) missing entries in the l1d_param array; this can cause a host crash
if an access attempts to reach the missing entry. Future-proof the get
function against any overflows as well. However, the two entries
VMENTER_L1D_FLUSH_EPT_DISABLED and VMENTER_L1D_FLUSH_NOT_REQUIRED must
not be accepted by the parse function, so disable them there.
2) invalid values must be rejected even if the CPU does not have the
bug, so test for them before checking boot_cpu_has(X86_BUG_L1TF)
... and a small refactoring, since the .cmd field is redundant with
the index in the array.
Reported-by: Bandan Das <[email protected]>
Cc: [email protected]
Fixes: a7b9020b06ec6d7c3f3b0d4ef1a9eba12654f4f7
Signed-off-by: Paolo Bonzini <[email protected]>
---
arch/x86/kvm/vmx.c | 26 ++++++++++++++++----------
1 file changed, 16 insertions(+), 10 deletions(-)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index c76ca8c4befa..8dae47e7267a 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -198,12 +198,14 @@
static const struct {
const char *option;
- enum vmx_l1d_flush_state cmd;
+ bool for_parse;
} vmentry_l1d_param[] = {
- {"auto", VMENTER_L1D_FLUSH_AUTO},
- {"never", VMENTER_L1D_FLUSH_NEVER},
- {"cond", VMENTER_L1D_FLUSH_COND},
- {"always", VMENTER_L1D_FLUSH_ALWAYS},
+ [VMENTER_L1D_FLUSH_AUTO] = {"auto", true},
+ [VMENTER_L1D_FLUSH_NEVER] = {"never", true},
+ [VMENTER_L1D_FLUSH_COND] = {"cond", true},
+ [VMENTER_L1D_FLUSH_ALWAYS] = {"always", true},
+ [VMENTER_L1D_FLUSH_EPT_DISABLED] = {"EPT disabled", false},
+ [VMENTER_L1D_FLUSH_NOT_REQUIRED] = {"not required", false},
};
#define L1D_CACHE_ORDER 4
@@ -287,8 +289,9 @@ static int vmentry_l1d_flush_parse(const char *s)
if (s) {
for (i = 0; i < ARRAY_SIZE(vmentry_l1d_param); i++) {
- if (sysfs_streq(s, vmentry_l1d_param[i].option))
- return vmentry_l1d_param[i].cmd;
+ if (vmentry_l1d_param[i].for_parse &&
+ sysfs_streq(s, vmentry_l1d_param[i].option))
+ return i;
}
}
return -EINVAL;
@@ -298,13 +301,13 @@ static int vmentry_l1d_flush_set(const char *s, const struct kernel_param *kp)
{
int l1tf, ret;
- if (!boot_cpu_has(X86_BUG_L1TF))
- return 0;
-
l1tf = vmentry_l1d_flush_parse(s);
if (l1tf < 0)
return l1tf;
+ if (!boot_cpu_has(X86_BUG_L1TF))
+ return 0;
+
/*
* Has vmx_init() run already? If not then this is the pre init
* parameter parsing. In that case just store the value and let
@@ -324,6 +327,9 @@ static int vmentry_l1d_flush_set(const char *s, const struct kernel_param *kp)
static int vmentry_l1d_flush_get(char *s, const struct kernel_param *kp)
{
+ if (WARN_ON_ONCE(l1tf_vmx_mitigation >= ARRAY_SIZE(vmentry_l1d_param)))
+ return sprintf(s, "???\n");
+
return sprintf(s, "%s\n", vmentry_l1d_param[l1tf_vmx_mitigation].option);
}
--
1.8.3.1
On Wed, Aug 22, 2018 at 04:53:00PM +0200, Paolo Bonzini wrote:
> Two bug fixes:
>
> 1) missing entries in the l1d_param array; this can cause a host crash
> if an access attempts to reach the missing entry. Future-proof the get
> function against any overflows as well. However, the two entries
> VMENTER_L1D_FLUSH_EPT_DISABLED and VMENTER_L1D_FLUSH_NOT_REQUIRED must
> not be accepted by the parse function, so disable them there.
>
> 2) invalid values must be rejected even if the CPU does not have the
> bug, so test for them before checking boot_cpu_has(X86_BUG_L1TF)
>
> ... and a small refactoring, since the .cmd field is redundant with
> the index in the array.
>
> Reported-by: Bandan Das <[email protected]>
> Cc: [email protected]
> Fixes: a7b9020b06ec6d7c3f3b0d4ef1a9eba12654f4f7
> Signed-off-by: Paolo Bonzini <[email protected]>
Reviewed-by: Konrad Rzeszutek Wilk <[email protected]>
with one little nitpick
> ---
> arch/x86/kvm/vmx.c | 26 ++++++++++++++++----------
> 1 file changed, 16 insertions(+), 10 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index c76ca8c4befa..8dae47e7267a 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -198,12 +198,14 @@
>
> static const struct {
> const char *option;
> - enum vmx_l1d_flush_state cmd;
> + bool for_parse;
> } vmentry_l1d_param[] = {
> - {"auto", VMENTER_L1D_FLUSH_AUTO},
> - {"never", VMENTER_L1D_FLUSH_NEVER},
> - {"cond", VMENTER_L1D_FLUSH_COND},
> - {"always", VMENTER_L1D_FLUSH_ALWAYS},
> + [VMENTER_L1D_FLUSH_AUTO] = {"auto", true},
> + [VMENTER_L1D_FLUSH_NEVER] = {"never", true},
> + [VMENTER_L1D_FLUSH_COND] = {"cond", true},
> + [VMENTER_L1D_FLUSH_ALWAYS] = {"always", true},
> + [VMENTER_L1D_FLUSH_EPT_DISABLED] = {"EPT disabled", false},
> + [VMENTER_L1D_FLUSH_NOT_REQUIRED] = {"not required", false},
Ingo likes these to have tabs. Any chance you could do:
[VMENTER_L1D_FLUSH_NOT_REQUIRED] = {"not required", false},
or so?
> };
>
> #define L1D_CACHE_ORDER 4
> @@ -287,8 +289,9 @@ static int vmentry_l1d_flush_parse(const char *s)
>
> if (s) {
> for (i = 0; i < ARRAY_SIZE(vmentry_l1d_param); i++) {
> - if (sysfs_streq(s, vmentry_l1d_param[i].option))
> - return vmentry_l1d_param[i].cmd;
> + if (vmentry_l1d_param[i].for_parse &&
> + sysfs_streq(s, vmentry_l1d_param[i].option))
> + return i;
> }
> }
> return -EINVAL;
> @@ -298,13 +301,13 @@ static int vmentry_l1d_flush_set(const char *s, const struct kernel_param *kp)
> {
> int l1tf, ret;
>
> - if (!boot_cpu_has(X86_BUG_L1TF))
> - return 0;
> -
> l1tf = vmentry_l1d_flush_parse(s);
> if (l1tf < 0)
> return l1tf;
>
> + if (!boot_cpu_has(X86_BUG_L1TF))
> + return 0;
> +
> /*
> * Has vmx_init() run already? If not then this is the pre init
> * parameter parsing. In that case just store the value and let
> @@ -324,6 +327,9 @@ static int vmentry_l1d_flush_set(const char *s, const struct kernel_param *kp)
>
> static int vmentry_l1d_flush_get(char *s, const struct kernel_param *kp)
> {
> + if (WARN_ON_ONCE(l1tf_vmx_mitigation >= ARRAY_SIZE(vmentry_l1d_param)))
> + return sprintf(s, "???\n");
> +
> return sprintf(s, "%s\n", vmentry_l1d_param[l1tf_vmx_mitigation].option);
> }
>
> --
> 1.8.3.1
>
Paolo Bonzini <[email protected]> 于2018年8月22日周三 下午4:56写道：
>
> Two bug fixes:
>
> 1) missing entries in the l1d_param array; this can cause a host crash
> if an access attempts to reach the missing entry. Future-proof the get
> function against any overflows as well. However, the two entries
> VMENTER_L1D_FLUSH_EPT_DISABLED and VMENTER_L1D_FLUSH_NOT_REQUIRED must
> not be accepted by the parse function, so disable them there.
>
> 2) invalid values must be rejected even if the CPU does not have the
> bug, so test for them before checking boot_cpu_has(X86_BUG_L1TF)
>
> ... and a small refactoring, since the .cmd field is redundant with
> the index in the array.
>
> Reported-by: Bandan Das <[email protected]>
> Cc: [email protected]
> Fixes: a7b9020b06ec6d7c3f3b0d4ef1a9eba12654f4f7
> Signed-off-by: Paolo Bonzini <[email protected]>
> ---
> arch/x86/kvm/vmx.c | 26 ++++++++++++++++----------
> 1 file changed, 16 insertions(+), 10 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index c76ca8c4befa..8dae47e7267a 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -198,12 +198,14 @@
>
> static const struct {
> const char *option;
> - enum vmx_l1d_flush_state cmd;
> + bool for_parse;
> } vmentry_l1d_param[] = {
> - {"auto", VMENTER_L1D_FLUSH_AUTO},
> - {"never", VMENTER_L1D_FLUSH_NEVER},
> - {"cond", VMENTER_L1D_FLUSH_COND},
> - {"always", VMENTER_L1D_FLUSH_ALWAYS},
> + [VMENTER_L1D_FLUSH_AUTO] = {"auto", true},
> + [VMENTER_L1D_FLUSH_NEVER] = {"never", true},
> + [VMENTER_L1D_FLUSH_COND] = {"cond", true},
> + [VMENTER_L1D_FLUSH_ALWAYS] = {"always", true},
> + [VMENTER_L1D_FLUSH_EPT_DISABLED] = {"EPT disabled", false},
> + [VMENTER_L1D_FLUSH_NOT_REQUIRED] = {"not required", false},
> };
>
> #define L1D_CACHE_ORDER 4
> @@ -287,8 +289,9 @@ static int vmentry_l1d_flush_parse(const char *s)
>
> if (s) {
> for (i = 0; i < ARRAY_SIZE(vmentry_l1d_param); i++) {
> - if (sysfs_streq(s, vmentry_l1d_param[i].option))
> - return vmentry_l1d_param[i].cmd;
> + if (vmentry_l1d_param[i].for_parse &&
> + sysfs_streq(s, vmentry_l1d_param[i].option))
> + return i;
> }
> }
> return -EINVAL;
> @@ -298,13 +301,13 @@ static int vmentry_l1d_flush_set(const char *s, const struct kernel_param *kp)
> {
> int l1tf, ret;
>
> - if (!boot_cpu_has(X86_BUG_L1TF))
> - return 0;
> -
> l1tf = vmentry_l1d_flush_parse(s);
> if (l1tf < 0)
> return l1tf;
>
> + if (!boot_cpu_has(X86_BUG_L1TF))
> + return 0;
> +
> /*
> * Has vmx_init() run already? If not then this is the pre init
> * parameter parsing. In that case just store the value and let
> @@ -324,6 +327,9 @@ static int vmentry_l1d_flush_set(const char *s, const struct kernel_param *kp)
>
> static int vmentry_l1d_flush_get(char *s, const struct kernel_param *kp)
> {
> + if (WARN_ON_ONCE(l1tf_vmx_mitigation >= ARRAY_SIZE(vmentry_l1d_param)))
> + return sprintf(s, "???\n");
> +
> return sprintf(s, "%s\n", vmentry_l1d_param[l1tf_vmx_mitigation].option);
> }
>
> --
> 1.8.3.1
>
Tested-by: Jack Wang <[email protected]>
Thanks,
Jack
On 22/08/2018 17:29, Konrad Rzeszutek Wilk wrote:
>> + [VMENTER_L1D_FLUSH_AUTO] = {"auto", true},
>> + [VMENTER_L1D_FLUSH_NEVER] = {"never", true},
>> + [VMENTER_L1D_FLUSH_COND] = {"cond", true},
>> + [VMENTER_L1D_FLUSH_ALWAYS] = {"always", true},
>> + [VMENTER_L1D_FLUSH_EPT_DISABLED] = {"EPT disabled", false},
>> + [VMENTER_L1D_FLUSH_NOT_REQUIRED] = {"not required", false},
> Ingo likes these to have tabs. Any chance you could do:
>
> [VMENTER_L1D_FLUSH_NOT_REQUIRED] = {"not required", false},
>
> or so?
>
Hmm, too late... Sorry about that.
Paolo