Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 6751ed65 authored by Tony Luck's avatar Tony Luck
Browse files

x86/mce: Fix siginfo_t->si_addr value for non-recoverable memory faults 



In commit dad1743e ("x86/mce: Only restart instruction after machine
check recovery if it is safe") we fixed mce_notify_process() to force a
signal to the current process if it was not restartable (RIPV bit not
set in MCG_STATUS). But doing it here means that the process doesn't
get told the virtual address of the fault via siginfo_t->si_addr. This
would prevent application level recovery from the fault.

Make a new MF_MUST_KILL flag bit for memory_failure() et al. to use so
that we will provide the right information with the signal.

Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
Acked-by: default avatarBorislav Petkov <borislav.petkov@amd.com>
Cc: stable@kernel.org    # 3.4+
parent 6887a413

arch/x86/kernel/cpu/mcheck/mce.c

+4 −2
Original line number Original line Diff line number Diff line
@@ -1186,6 +1186,7 @@ void mce_notify_process(void) 
{

{

	unsigned long pfn;

	unsigned long pfn;

	struct mce_info *mi = mce_find_info();

	struct mce_info *mi = mce_find_info();

	int flags = MF_ACTION_REQUIRED;





	if (!mi)

	if (!mi)

		mce_panic("Lost physical address for unconsumed uncorrectable error", NULL, NULL);

		mce_panic("Lost physical address for unconsumed uncorrectable error", NULL, NULL);
@@ -1200,8 +1201,9 @@ void mce_notify_process(void) 
	 * doomed. We still need to mark the page as poisoned and alert any

	 * doomed. We still need to mark the page as poisoned and alert any

	 * other users of the page.

	 * other users of the page.

	 */

	 */

	if (memory_failure(pfn, MCE_VECTOR, MF_ACTION_REQUIRED) < 0 ||

	if (!mi->restartable)

			   mi->restartable == 0) {

		flags |= MF_MUST_KILL;

	if (memory_failure(pfn, MCE_VECTOR, flags) < 0) {

		pr_err("Memory error not recovered");

		pr_err("Memory error not recovered");

		force_sig(SIGBUS, current);

		force_sig(SIGBUS, current);

	}

	}

include/linux/mm.h

+1 −0
Original line number Original line Diff line number Diff line
@@ -1591,6 +1591,7 @@ void vmemmap_populate_print_last(void); 
enum mf_flags {

enum mf_flags {

	MF_COUNT_INCREASED = 1 << 0,

	MF_COUNT_INCREASED = 1 << 0,

	MF_ACTION_REQUIRED = 1 << 1,

	MF_ACTION_REQUIRED = 1 << 1,

	MF_MUST_KILL = 1 << 2,

};

};

extern int memory_failure(unsigned long pfn, int trapno, int flags);

extern int memory_failure(unsigned long pfn, int trapno, int flags);

extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);

extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);

mm/memory-failure.c

+8 −6
Original line number Original line Diff line number Diff line
@@ -345,14 +345,14 @@ static void add_to_kill(struct task_struct *tsk, struct page *p, 
 * Also when FAIL is set do a force kill because something went

 * Also when FAIL is set do a force kill because something went

 * wrong earlier.

 * wrong earlier.

 */

 */

static void kill_procs(struct list_head *to_kill, int doit, int trapno,

static void kill_procs(struct list_head *to_kill, int forcekill, int trapno,

			  int fail, struct page *page, unsigned long pfn,

			  int fail, struct page *page, unsigned long pfn,

			  int flags)

			  int flags)

{

{

	struct to_kill *tk, *next;

	struct to_kill *tk, *next;





	list_for_each_entry_safe (tk, next, to_kill, nd) {

	list_for_each_entry_safe (tk, next, to_kill, nd) {

		if (doit) {

		if (forcekill) {

			/*

			/*

			 * In case something went wrong with munmapping

			 * In case something went wrong with munmapping

			 * make sure the process doesn't catch the

			 * make sure the process doesn't catch the
@@ -858,7 +858,7 @@ static int hwpoison_user_mappings(struct page *p, unsigned long pfn, 
	struct address_space *mapping;

	struct address_space *mapping;

	LIST_HEAD(tokill);

	LIST_HEAD(tokill);

	int ret;

	int ret;

	int kill = 1;

	int kill = 1, forcekill;

	struct page *hpage = compound_head(p);

	struct page *hpage = compound_head(p);

	struct page *ppage;

	struct page *ppage;
@@ -888,7 +888,7 @@ static int hwpoison_user_mappings(struct page *p, unsigned long pfn, 
	 * be called inside page lock (it's recommended but not enforced).

	 * be called inside page lock (it's recommended but not enforced).

	 */

	 */

	mapping = page_mapping(hpage);

	mapping = page_mapping(hpage);

	if (!PageDirty(hpage) && mapping &&

	if (!(flags & MF_MUST_KILL) && !PageDirty(hpage) && mapping &&

	    mapping_cap_writeback_dirty(mapping)) {

	    mapping_cap_writeback_dirty(mapping)) {

		if (page_mkclean(hpage)) {

		if (page_mkclean(hpage)) {

			SetPageDirty(hpage);

			SetPageDirty(hpage);
@@ -965,12 +965,14 @@ static int hwpoison_user_mappings(struct page *p, unsigned long pfn, 
	 * Now that the dirty bit has been propagated to the

	 * Now that the dirty bit has been propagated to the

	 * struct page and all unmaps done we can decide if

	 * struct page and all unmaps done we can decide if

	 * killing is needed or not.  Only kill when the page

	 * killing is needed or not.  Only kill when the page

	 * was dirty, otherwise the tokill list is merely

	 * was dirty or the process is not restartable,

	 * otherwise the tokill list is merely

	 * freed.  When there was a problem unmapping earlier

	 * freed.  When there was a problem unmapping earlier

	 * use a more force-full uncatchable kill to prevent

	 * use a more force-full uncatchable kill to prevent

	 * any accesses to the poisoned memory.

	 * any accesses to the poisoned memory.

	 */

	 */

	kill_procs(&tokill, !!PageDirty(ppage), trapno,

	forcekill = PageDirty(ppage) || (flags & MF_MUST_KILL);

	kill_procs(&tokill, forcekill, trapno,

		      ret != SWAP_SUCCESS, p, pfn, flags);

		      ret != SWAP_SUCCESS, p, pfn, flags);





	return ret;

	return ret;