linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2 1/2] x86/pti: check the return value of pti_user_pagetable_walk_p4d
@ 2018-07-17  7:23 Jiang Biao
  2018-07-17  7:23 ` [PATCH v2 2/2] x86/pti: check the return value of pti_user_pagetable_walk_pmd Jiang Biao
  0 siblings, 1 reply; 3+ messages in thread
From: Jiang Biao @ 2018-07-17  7:23 UTC (permalink / raw)
  To: tglx, mingo
  Cc: dave.hansen, luto, hpa, x86, albcamus, linux-kernel,
	zhong.weidong, jiang.biao2

pti_user_pagetable_walk_p4d() may return NULL, we should check the
return value to avoid NULL pointer dereference. And add warning
for fail allocation where NULL returned.

Signed-off-by: Jiang Biao <jiang.biao2@zte.com.cn>
---
 arch/x86/mm/pti.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/arch/x86/mm/pti.c b/arch/x86/mm/pti.c
index 4d418e7..7c402e9 100644
--- a/arch/x86/mm/pti.c
+++ b/arch/x86/mm/pti.c
@@ -176,7 +176,7 @@ static p4d_t *pti_user_pagetable_walk_p4d(unsigned long address)
 
 	if (pgd_none(*pgd)) {
 		unsigned long new_p4d_page = __get_free_page(gfp);
-		if (!new_p4d_page)
+		if (WARN_ON(!new_p4d_page))
 			return NULL;
 
 		set_pgd(pgd, __pgd(_KERNPG_TABLE | __pa(new_p4d_page)));
@@ -195,8 +195,10 @@ static p4d_t *pti_user_pagetable_walk_p4d(unsigned long address)
 static pmd_t *pti_user_pagetable_walk_pmd(unsigned long address)
 {
 	gfp_t gfp = (GFP_KERNEL | __GFP_NOTRACK | __GFP_ZERO);
-	p4d_t *p4d = pti_user_pagetable_walk_p4d(address);
 	pud_t *pud;
+	p4d_t *p4d = pti_user_pagetable_walk_p4d(address);
+	if (!p4d)
+		return NULL;
 
 	BUILD_BUG_ON(p4d_large(*p4d) != 0);
 	if (p4d_none(*p4d)) {
@@ -354,6 +356,8 @@ static void __init pti_clone_p4d(unsigned long addr)
 	pgd_t *kernel_pgd;
 
 	user_p4d = pti_user_pagetable_walk_p4d(addr);
+	if (!user_p4d)
+		return;
 	kernel_pgd = pgd_offset_k(addr);
 	kernel_p4d = p4d_offset(kernel_pgd, addr);
 	*user_p4d = *kernel_p4d;
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* [PATCH v2 2/2] x86/pti: check the return value of pti_user_pagetable_walk_pmd
  2018-07-17  7:23 [PATCH v2 1/2] x86/pti: check the return value of pti_user_pagetable_walk_p4d Jiang Biao
@ 2018-07-17  7:23 ` Jiang Biao
  2018-07-17 18:03   ` Dave Hansen
  0 siblings, 1 reply; 3+ messages in thread
From: Jiang Biao @ 2018-07-17  7:23 UTC (permalink / raw)
  To: tglx, mingo
  Cc: dave.hansen, luto, hpa, x86, albcamus, linux-kernel,
	zhong.weidong, jiang.biao2

Check the return value of pti_user_pagetable_walk_pmd() to avoid
NULL pointer dereference. And add warning for fail allocation.

Signed-off-by: Jiang Biao <jiang.biao2@zte.com.cn>
---
 arch/x86/mm/pti.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/arch/x86/mm/pti.c b/arch/x86/mm/pti.c
index 7c402e9..3dba6a7 100644
--- a/arch/x86/mm/pti.c
+++ b/arch/x86/mm/pti.c
@@ -203,7 +203,7 @@ static pmd_t *pti_user_pagetable_walk_pmd(unsigned long address)
 	BUILD_BUG_ON(p4d_large(*p4d) != 0);
 	if (p4d_none(*p4d)) {
 		unsigned long new_pud_page = __get_free_page(gfp);
-		if (!new_pud_page)
+		if (WARN_ON(!new_pud_page))
 			return NULL;
 
 		set_p4d(p4d, __p4d(_KERNPG_TABLE | __pa(new_pud_page)));
@@ -217,7 +217,7 @@ static pmd_t *pti_user_pagetable_walk_pmd(unsigned long address)
 	}
 	if (pud_none(*pud)) {
 		unsigned long new_pmd_page = __get_free_page(gfp);
-		if (!new_pmd_page)
+		if (WARN_ON(!new_pmd_page))
 			return NULL;
 
 		set_pud(pud, __pud(_KERNPG_TABLE | __pa(new_pmd_page)));
@@ -239,8 +239,10 @@ static pmd_t *pti_user_pagetable_walk_pmd(unsigned long address)
 static __init pte_t *pti_user_pagetable_walk_pte(unsigned long address)
 {
 	gfp_t gfp = (GFP_KERNEL | __GFP_NOTRACK | __GFP_ZERO);
-	pmd_t *pmd = pti_user_pagetable_walk_pmd(address);
 	pte_t *pte;
+	pmd_t *pmd = pti_user_pagetable_walk_pmd(address);
+	if (!pmd)
+		return NULL;
 
 	/* We can't do anything sensible if we hit a large mapping. */
 	if (pmd_large(*pmd)) {
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH v2 2/2] x86/pti: check the return value of pti_user_pagetable_walk_pmd
  2018-07-17  7:23 ` [PATCH v2 2/2] x86/pti: check the return value of pti_user_pagetable_walk_pmd Jiang Biao
@ 2018-07-17 18:03   ` Dave Hansen
  0 siblings, 0 replies; 3+ messages in thread
From: Dave Hansen @ 2018-07-17 18:03 UTC (permalink / raw)
  To: Jiang Biao, tglx, mingo
  Cc: luto, hpa, x86, albcamus, linux-kernel, zhong.weidong

On 07/17/2018 12:23 AM, Jiang Biao wrote:
> Check the return value of pti_user_pagetable_walk_pmd() to avoid
> NULL pointer dereference. And add warning for fail allocation.

For both of these:

Acked-by: Dave Hansen <dave.hansen@intel.com>

It's minor, but if you redo these, I'd appreciate a slightly different
form.  Instead of:

> @@ -239,8 +239,10 @@ static pmd_t *pti_user_pagetable_walk_pmd(unsigned long address)
>  static __init pte_t *pti_user_pagetable_walk_pte(unsigned long address)
>  {
>  	gfp_t gfp = (GFP_KERNEL | __GFP_NOTRACK | __GFP_ZERO);
> -	pmd_t *pmd = pti_user_pagetable_walk_pmd(address);
>  	pte_t *pte;
> +	pmd_t *pmd = pti_user_pagetable_walk_pmd(address);
> +	if (!pmd)
> +		return NULL;

I'd much rather see separation of code -- especially _important_ code
like an allocation -- from local variable definitions.  Like this:

	gfp_t gfp = (GFP_KERNEL | __GFP_NOTRACK | __GFP_ZERO);
	pmd_t *pmd;
  	pte_t *pte;

	pmd = pti_user_pagetable_walk_pmd(address);
	if (!pmd)
		return NULL;

That clearly separtes the variables from the _code_ and also nicely
pairs the action with the check for that action being successful.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2018-07-17 18:03 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-07-17  7:23 [PATCH v2 1/2] x86/pti: check the return value of pti_user_pagetable_walk_p4d Jiang Biao
2018-07-17  7:23 ` [PATCH v2 2/2] x86/pti: check the return value of pti_user_pagetable_walk_pmd Jiang Biao
2018-07-17 18:03   ` Dave Hansen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).