[PATCH 0/8] current ACCESS_ONCE patch queue

[PATCH 0/8] current ACCESS_ONCE patch queue

[Christian Borntraeger]

Folks,

fyi, this is my current patch queue for the next merge window. It
does contain a patch that will disallow ACCESS_ONCE on non-scalar
types.

The tree is part of linux-next and can be found at
git://git.kernel.org/pub/scm/linux/kernel/git/borntraeger/linux.git linux-next


Christian Borntraeger (7):
  ppc/kvm: Replace ACCESS_ONCE with READ_ONCE
  ppc/hugetlbfs: Replace ACCESS_ONCE with READ_ONCE
  x86/xen/p2m: Replace ACCESS_ONCE with READ_ONCE
  x86/spinlock: Leftover conversion ACCESS_ONCE->READ_ONCE
  mm/gup: Replace ACCESS_ONCE with READ_ONCE
  kernel: tighten rules for ACCESS ONCE
  kernel: Fix sparse warning for ACCESS_ONCE

Guenter Roeck (1):
  next: sh: Fix compile error

 arch/powerpc/kvm/book3s_hv_rm_xics.c |  8 ++++----
 arch/powerpc/kvm/book3s_xics.c       | 16 ++++++++--------
 arch/powerpc/mm/hugetlbpage.c        |  4 ++--
 arch/sh/mm/gup.c                     |  2 +-
 arch/x86/include/asm/spinlock.h      |  2 +-
 arch/x86/xen/p2m.c                   |  2 +-
 include/linux/compiler.h             | 21 ++++++++++++++++-----
 mm/gup.c                             |  2 +-
 8 files changed, 34 insertions(+), 23 deletions(-)

-- 
1.9.3

[PATCH 1/8] ppc/kvm: Replace ACCESS_ONCE with READ_ONCE

[Christian Borntraeger]

ACCESS_ONCE does not work reliably on non-scalar types. For
example gcc 4.6 and 4.7 might remove the volatile tag for such
accesses during the SRA (scalar replacement of aggregates) step
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)

Change the ppc/kvm code to replace ACCESS_ONCE with READ_ONCE.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 arch/powerpc/kvm/book3s_hv_rm_xics.c |  8 ++++----
 arch/powerpc/kvm/book3s_xics.c       | 16 ++++++++--------
 2 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv_rm_xics.c b/arch/powerpc/kvm/book3s_hv_rm_xics.c
index 7b066f6..7c22997 100644
--- a/arch/powerpc/kvm/book3s_hv_rm_xics.c
+++ b/arch/powerpc/kvm/book3s_hv_rm_xics.c
@@ -152,7 +152,7 @@ static void icp_rm_down_cppr(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 	 * in virtual mode.
 	 */
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		/* Down_CPPR */
 		new_state.cppr = new_cppr;
@@ -211,7 +211,7 @@ unsigned long kvmppc_rm_h_xirr(struct kvm_vcpu *vcpu)
 	 * pending priority
 	 */
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		xirr = old_state.xisr | (((u32)old_state.cppr) << 24);
 		if (!old_state.xisr)
@@ -277,7 +277,7 @@ int kvmppc_rm_h_ipi(struct kvm_vcpu *vcpu, unsigned long server,
 	 * whenever the MFRR is made less favored.
 	 */
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		/* Set_MFRR */
 		new_state.mfrr = mfrr;
@@ -352,7 +352,7 @@ int kvmppc_rm_h_cppr(struct kvm_vcpu *vcpu, unsigned long cppr)
 	icp_rm_clr_vcpu_irq(icp->vcpu);
 
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		reject = 0;
 		new_state.cppr = cppr;
diff --git a/arch/powerpc/kvm/book3s_xics.c b/arch/powerpc/kvm/book3s_xics.c
index 807351f..a4a8d9f 100644
--- a/arch/powerpc/kvm/book3s_xics.c
+++ b/arch/powerpc/kvm/book3s_xics.c
@@ -327,7 +327,7 @@ static bool icp_try_to_deliver(struct kvmppc_icp *icp, u32 irq, u8 priority,
 		 icp->server_num);
 
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		*reject = 0;
 
@@ -512,7 +512,7 @@ static void icp_down_cppr(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 	 * in virtual mode.
 	 */
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		/* Down_CPPR */
 		new_state.cppr = new_cppr;
@@ -567,7 +567,7 @@ static noinline unsigned long kvmppc_h_xirr(struct kvm_vcpu *vcpu)
 	 * pending priority
 	 */
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		xirr = old_state.xisr | (((u32)old_state.cppr) << 24);
 		if (!old_state.xisr)
@@ -634,7 +634,7 @@ static noinline int kvmppc_h_ipi(struct kvm_vcpu *vcpu, unsigned long server,
 	 * whenever the MFRR is made less favored.
 	 */
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		/* Set_MFRR */
 		new_state.mfrr = mfrr;
@@ -679,7 +679,7 @@ static int kvmppc_h_ipoll(struct kvm_vcpu *vcpu, unsigned long server)
 		if (!icp)
 			return H_PARAMETER;
 	}
-	state = ACCESS_ONCE(icp->state);
+	state = READ_ONCE(icp->state);
 	kvmppc_set_gpr(vcpu, 4, ((u32)state.cppr << 24) | state.xisr);
 	kvmppc_set_gpr(vcpu, 5, state.mfrr);
 	return H_SUCCESS;
@@ -721,7 +721,7 @@ static noinline void kvmppc_h_cppr(struct kvm_vcpu *vcpu, unsigned long cppr)
 				      BOOK3S_INTERRUPT_EXTERNAL_LEVEL);
 
 	do {
-		old_state = new_state = ACCESS_ONCE(icp->state);
+		old_state = new_state = READ_ONCE(icp->state);
 
 		reject = 0;
 		new_state.cppr = cppr;
@@ -885,7 +885,7 @@ static int xics_debug_show(struct seq_file *m, void *private)
 		if (!icp)
 			continue;
 
-		state.raw = ACCESS_ONCE(icp->state.raw);
+		state.raw = READ_ONCE(icp->state.raw);
 		seq_printf(m, "cpu server %#lx XIRR:%#x PPRI:%#x CPPR:%#x MFRR:%#x OUT:%d NR:%d\n",
 			   icp->server_num, state.xisr,
 			   state.pending_pri, state.cppr, state.mfrr,
@@ -1082,7 +1082,7 @@ int kvmppc_xics_set_icp(struct kvm_vcpu *vcpu, u64 icpval)
 	 * the ICS states before the ICP states.
 	 */
 	do {
-		old_state = ACCESS_ONCE(icp->state);
+		old_state = READ_ONCE(icp->state);
 
 		if (new_state.mfrr <= old_state.mfrr) {
 			resend = false;
-- 
1.9.3

Re: [PATCH 1/8] ppc/kvm: Replace ACCESS_ONCE with READ_ONCE

[Michael Ellerman]

On Thu, 2015-01-15 at 09:58 +0100, Christian Borntraeger wrote:
> ACCESS_ONCE does not work reliably on non-scalar types. For
> example gcc 4.6 and 4.7 might remove the volatile tag for such
> accesses during the SRA (scalar replacement of aggregates) step
> (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)
> 
> Change the ppc/kvm code to replace ACCESS_ONCE with READ_ONCE.
> 
> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
> ---
>  arch/powerpc/kvm/book3s_hv_rm_xics.c |  8 ++++----
>  arch/powerpc/kvm/book3s_xics.c       | 16 ++++++++--------
>  2 files changed, 12 insertions(+), 12 deletions(-)
> 
> diff --git a/arch/powerpc/kvm/book3s_hv_rm_xics.c b/arch/powerpc/kvm/book3s_hv_rm_xics.c
> index 7b066f6..7c22997 100644
> --- a/arch/powerpc/kvm/book3s_hv_rm_xics.c
> +++ b/arch/powerpc/kvm/book3s_hv_rm_xics.c
> @@ -152,7 +152,7 @@ static void icp_rm_down_cppr(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
>  	 * in virtual mode.
>  	 */
>  	do {
> -		old_state = new_state = ACCESS_ONCE(icp->state);
> +		old_state = new_state = READ_ONCE(icp->state);

These are all icp->state.

Which is a union, but it's only the size of unsigned long. So in practice there
shouldn't be a bug here right?

cheers

Re: [PATCH 1/8] ppc/kvm: Replace ACCESS_ONCE with READ_ONCE

[Christian Borntraeger]

Am 16.01.2015 um 00:09 schrieb Michael Ellerman:
> On Thu, 2015-01-15 at 09:58 +0100, Christian Borntraeger wrote:
>> ACCESS_ONCE does not work reliably on non-scalar types. For
>> example gcc 4.6 and 4.7 might remove the volatile tag for such
>> accesses during the SRA (scalar replacement of aggregates) step
>> (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)
>>
>> Change the ppc/kvm code to replace ACCESS_ONCE with READ_ONCE.
>>
>> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
>> ---
>>  arch/powerpc/kvm/book3s_hv_rm_xics.c |  8 ++++----
>>  arch/powerpc/kvm/book3s_xics.c       | 16 ++++++++--------
>>  2 files changed, 12 insertions(+), 12 deletions(-)
>>
>> diff --git a/arch/powerpc/kvm/book3s_hv_rm_xics.c b/arch/powerpc/kvm/book3s_hv_rm_xics.c
>> index 7b066f6..7c22997 100644
>> --- a/arch/powerpc/kvm/book3s_hv_rm_xics.c
>> +++ b/arch/powerpc/kvm/book3s_hv_rm_xics.c
>> @@ -152,7 +152,7 @@ static void icp_rm_down_cppr(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
>>  	 * in virtual mode.
>>  	 */
>>  	do {
>> -		old_state = new_state = ACCESS_ONCE(icp->state);
>> +		old_state = new_state = READ_ONCE(icp->state);
> 
> These are all icp->state.
> 
> Which is a union, but it's only the size of unsigned long. So in practice there
> shouldn't be a bug here right?

This bug was that gcc lost the volatile tag when propagating aggregates to scalar types.
So in theory a union could be affected. See the original problem
 ( http://marc.info/?i=54611D86.4040306%40de.ibm.com ) 
which happened on 

union ipte_control {
        unsigned long val;
        struct {
                unsigned long k  : 1;
                unsigned long kh : 31;
                unsigned long kg : 32;
        };
};

Christian

[PATCH 2/8] ppc/hugetlbfs: Replace ACCESS_ONCE with READ_ONCE

[Christian Borntraeger]

ACCESS_ONCE does not work reliably on non-scalar types. For
example gcc 4.6 and 4.7 might remove the volatile tag for such
accesses during the SRA (scalar replacement of aggregates) step
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)

Change the ppc/hugetlbfs code to replace ACCESS_ONCE with READ_ONCE.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 arch/powerpc/mm/hugetlbpage.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index 5ff4e07..620d0ec 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -978,7 +978,7 @@ pte_t *find_linux_pte_or_hugepte(pgd_t *pgdir, unsigned long ea, unsigned *shift
 		 */
 		pdshift = PUD_SHIFT;
 		pudp = pud_offset(&pgd, ea);
-		pud  = ACCESS_ONCE(*pudp);
+		pud  = READ_ONCE(*pudp);
 
 		if (pud_none(pud))
 			return NULL;
@@ -990,7 +990,7 @@ pte_t *find_linux_pte_or_hugepte(pgd_t *pgdir, unsigned long ea, unsigned *shift
 		else {
 			pdshift = PMD_SHIFT;
 			pmdp = pmd_offset(&pud, ea);
-			pmd  = ACCESS_ONCE(*pmdp);
+			pmd  = READ_ONCE(*pmdp);
 			/*
 			 * A hugepage collapse is captured by pmd_none, because
 			 * it mark the pmd none and do a hpte invalidate.
-- 
1.9.3

[PATCH 3/8] x86/xen/p2m: Replace ACCESS_ONCE with READ_ONCE

[Christian Borntraeger]

ACCESS_ONCE does not work reliably on non-scalar types. For
example gcc 4.6 and 4.7 might remove the volatile tag for such
accesses during the SRA (scalar replacement of aggregates) step
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)

Change the p2m code to replace ACCESS_ONCE with READ_ONCE.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 arch/x86/xen/p2m.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
index edbc7a6..cb71016 100644
--- a/arch/x86/xen/p2m.c
+++ b/arch/x86/xen/p2m.c
@@ -554,7 +554,7 @@ static bool alloc_p2m(unsigned long pfn)
 		mid_mfn = NULL;
 	}
 
-	p2m_pfn = pte_pfn(ACCESS_ONCE(*ptep));
+	p2m_pfn = pte_pfn(READ_ONCE(*ptep));
 	if (p2m_pfn == PFN_DOWN(__pa(p2m_identity)) ||
 	    p2m_pfn == PFN_DOWN(__pa(p2m_missing))) {
 		/* p2m leaf page is missing */
-- 
1.9.3

Re: [PATCH 3/8] x86/xen/p2m: Replace ACCESS_ONCE with READ_ONCE

[Jürgen Groß]

On 01/15/2015 09:58 AM, Christian Borntraeger wrote:
> ACCESS_ONCE does not work reliably on non-scalar types. For
> example gcc 4.6 and 4.7 might remove the volatile tag for such
> accesses during the SRA (scalar replacement of aggregates) step
> (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)
>
> Change the p2m code to replace ACCESS_ONCE with READ_ONCE.
>
> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>

Reviewed-by: Juergen Gross <jgross@suse.com>

> ---
>   arch/x86/xen/p2m.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/xen/p2m.c b/arch/x86/xen/p2m.c
> index edbc7a6..cb71016 100644
> --- a/arch/x86/xen/p2m.c
> +++ b/arch/x86/xen/p2m.c
> @@ -554,7 +554,7 @@ static bool alloc_p2m(unsigned long pfn)
>   		mid_mfn = NULL;
>   	}
>
> -	p2m_pfn = pte_pfn(ACCESS_ONCE(*ptep));
> +	p2m_pfn = pte_pfn(READ_ONCE(*ptep));
>   	if (p2m_pfn == PFN_DOWN(__pa(p2m_identity)) ||
>   	    p2m_pfn == PFN_DOWN(__pa(p2m_missing))) {
>   		/* p2m leaf page is missing */
>

Re: [Xen-devel] [PATCH 3/8] x86/xen/p2m: Replace ACCESS_ONCE with READ_ONCE

[David Vrabel]

On 15/01/15 08:58, Christian Borntraeger wrote:
> ACCESS_ONCE does not work reliably on non-scalar types. For
> example gcc 4.6 and 4.7 might remove the volatile tag for such
> accesses during the SRA (scalar replacement of aggregates) step
> (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)
> 
> Change the p2m code to replace ACCESS_ONCE with READ_ONCE.

Acked-by: David Vrabel <david.vrabel@citrix.com>

Let me know if you want me to merge this via the Xen tree.

David

Re: [Xen-devel] [PATCH 3/8] x86/xen/p2m: Replace ACCESS_ONCE with READ_ONCE

[Christian Borntraeger]

Am 15.01.2015 um 11:43 schrieb David Vrabel:
> On 15/01/15 08:58, Christian Borntraeger wrote:
>> ACCESS_ONCE does not work reliably on non-scalar types. For
>> example gcc 4.6 and 4.7 might remove the volatile tag for such
>> accesses during the SRA (scalar replacement of aggregates) step
>> (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)
>>
>> Change the p2m code to replace ACCESS_ONCE with READ_ONCE.
> 
> Acked-by: David Vrabel <david.vrabel@citrix.com>

Thanks

> Let me know if you want me to merge this via the Xen tree.


With your Acked-by, I can easily carry this in my tree. We can 
then ensure that this patch is merged before the ACCESS_ONCE is
made stricter - making bisecting easier.

Christian

[PATCH 4/8] x86/spinlock: Leftover conversion ACCESS_ONCE->READ_ONCE

[Christian Borntraeger]

commit 78bff1c8684f ("x86/ticketlock: Fix spin_unlock_wait() livelock")
introduced another ACCESS_ONCE case in x86 spinlock.h.

Change that as well.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Oleg Nesterov <oleg@redhat.com>
---
 arch/x86/include/asm/spinlock.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/spinlock.h b/arch/x86/include/asm/spinlock.h
index 625660f..9264f0f 100644
--- a/arch/x86/include/asm/spinlock.h
+++ b/arch/x86/include/asm/spinlock.h
@@ -186,7 +186,7 @@ static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
 	__ticket_t head = ACCESS_ONCE(lock->tickets.head);
 
 	for (;;) {
-		struct __raw_tickets tmp = ACCESS_ONCE(lock->tickets);
+		struct __raw_tickets tmp = READ_ONCE(lock->tickets);
 		/*
 		 * We need to check "unlocked" in a loop, tmp.head == head
 		 * can be false positive because of overflow.
-- 
1.9.3

Re: [PATCH 4/8] x86/spinlock: Leftover conversion ACCESS_ONCE->READ_ONCE

[Oleg Nesterov]

On 01/15, Christian Borntraeger wrote:
>
> --- a/arch/x86/include/asm/spinlock.h
> +++ b/arch/x86/include/asm/spinlock.h
> @@ -186,7 +186,7 @@ static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
>  	__ticket_t head = ACCESS_ONCE(lock->tickets.head);
>  
>  	for (;;) {
> -		struct __raw_tickets tmp = ACCESS_ONCE(lock->tickets);
> +		struct __raw_tickets tmp = READ_ONCE(lock->tickets);

Agreed, but what about another ACCESS_ONCE() above?

Oleg.

Re: [PATCH 4/8] x86/spinlock: Leftover conversion ACCESS_ONCE->READ_ONCE

[Christian Borntraeger]

Am 15.01.2015 um 20:38 schrieb Oleg Nesterov:
> On 01/15, Christian Borntraeger wrote:
>>
>> --- a/arch/x86/include/asm/spinlock.h
>> +++ b/arch/x86/include/asm/spinlock.h
>> @@ -186,7 +186,7 @@ static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
>>  	__ticket_t head = ACCESS_ONCE(lock->tickets.head);
>>  
>>  	for (;;) {
>> -		struct __raw_tickets tmp = ACCESS_ONCE(lock->tickets);
>> +		struct __raw_tickets tmp = READ_ONCE(lock->tickets);
> 
> Agreed, but what about another ACCESS_ONCE() above?
> 
> Oleg.
> 

tickets.head is a scalar type, so ACCESS_ONCE does work fine with gcc 4.6/4.7.
My goal was to convert all accesses on non-scalar types as until 
"kernel: tighten rules for ACCESS ONCE" is merged because anything else would be
a Whac-a-mole like adventure (I learned that during the last round in next: all
conversions in this series fix up changes made during this merge window)

We probably going to do a bigger bunch of bulk conversion later on when 
"kernel: tighten rules for ACCESS ONCE" prevents new problems.

Makes sense?

Christian

Re: [PATCH 4/8] x86/spinlock: Leftover conversion ACCESS_ONCE->READ_ONCE

[Oleg Nesterov]

On 01/15, Christian Borntraeger wrote:
>
> Am 15.01.2015 um 20:38 schrieb Oleg Nesterov:
> > On 01/15, Christian Borntraeger wrote:
> >>
> >> --- a/arch/x86/include/asm/spinlock.h
> >> +++ b/arch/x86/include/asm/spinlock.h
> >> @@ -186,7 +186,7 @@ static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
> >>  	__ticket_t head = ACCESS_ONCE(lock->tickets.head);
> >>
> >>  	for (;;) {
> >> -		struct __raw_tickets tmp = ACCESS_ONCE(lock->tickets);
> >> +		struct __raw_tickets tmp = READ_ONCE(lock->tickets);
> >
> > Agreed, but what about another ACCESS_ONCE() above?
> >
> > Oleg.
>
> tickets.head is a scalar type, so ACCESS_ONCE does work fine with gcc 4.6/4.7.
> My goal was to convert all accesses on non-scalar types

I understand, but READ_ONCE(lock->tickets.head) looks better anyway and
arch_spin_lock() already use READ_ONCE() for this.

So why we should keep the last ACCESS_ONCE() in spinlock.h ? Just to make
another cosmetic cleanup which touches the same function later?

Oleg.

Re: [PATCH 4/8] x86/spinlock: Leftover conversion ACCESS_ONCE->READ_ONCE

[Christian Borntraeger]

Am 15.01.2015 um 21:01 schrieb Oleg Nesterov:
> On 01/15, Christian Borntraeger wrote:
>>
>> Am 15.01.2015 um 20:38 schrieb Oleg Nesterov:
>>> On 01/15, Christian Borntraeger wrote:
>>>>
>>>> --- a/arch/x86/include/asm/spinlock.h
>>>> +++ b/arch/x86/include/asm/spinlock.h
>>>> @@ -186,7 +186,7 @@ static inline void arch_spin_unlock_wait(arch_spinlock_t *lock)
>>>>  	__ticket_t head = ACCESS_ONCE(lock->tickets.head);
>>>>
>>>>  	for (;;) {
>>>> -		struct __raw_tickets tmp = ACCESS_ONCE(lock->tickets);
>>>> +		struct __raw_tickets tmp = READ_ONCE(lock->tickets);
>>>
>>> Agreed, but what about another ACCESS_ONCE() above?
>>>
>>> Oleg.
>>
>> tickets.head is a scalar type, so ACCESS_ONCE does work fine with gcc 4.6/4.7.
>> My goal was to convert all accesses on non-scalar types
> 
> I understand, but READ_ONCE(lock->tickets.head) looks better anyway and
> arch_spin_lock() already use READ_ONCE() for this.
> 
> So why we should keep the last ACCESS_ONCE() in spinlock.h ? Just to make
> another cosmetic cleanup which touches the same function later?

OK, I will change that one as well.

[PATCH 5/8] mm/gup: Replace ACCESS_ONCE with READ_ONCE

[Christian Borntraeger]

ACCESS_ONCE does not work reliably on non-scalar types. For
example gcc 4.6 and 4.7 might remove the volatile tag for such
accesses during the SRA (scalar replacement of aggregates) step
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145)

Fixup gup_pmd_range.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 mm/gup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/gup.c b/mm/gup.c
index a900759..bed30efa 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -926,7 +926,7 @@ static int gup_pmd_range(pud_t pud, unsigned long addr, unsigned long end,
 
 	pmdp = pmd_offset(&pud, addr);
 	do {
-		pmd_t pmd = ACCESS_ONCE(*pmdp);
+		pmd_t pmd = READ_ONCE(*pmdp);
 
 		next = pmd_addr_end(addr, end);
 		if (pmd_none(pmd) || pmd_trans_splitting(pmd))
-- 
1.9.3

[PATCH 6/8] kernel: tighten rules for ACCESS ONCE

[Christian Borntraeger]

Now that all non-scalar users of ACCESS_ONCE have been converted
to READ_ONCE or ASSIGN once, lets tighten ACCESS_ONCE to only
work on scalar types.
This variant was proposed by Alexei Starovoitov.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 include/linux/compiler.h | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index a1c81f8..5e186bf 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -447,12 +447,23 @@ static __always_inline void __assign_once_size(volatile void *p, void *res, int
  * to make the compiler aware of ordering is to put the two invocations of
  * ACCESS_ONCE() in different C statements.
  *
- * This macro does absolutely -nothing- to prevent the CPU from reordering,
- * merging, or refetching absolutely anything at any time.  Its main intended
- * use is to mediate communication between process-level code and irq/NMI
- * handlers, all running on the same CPU.
+ * ACCESS_ONCE will only work on scalar types. For union types, ACCESS_ONCE
+ * on a union member will work as long as the size of the member matches the
+ * size of the union and the size is smaller than word size.
+ *
+ * The major use cases of ACCESS_ONCE used to be (1) Mediating communication
+ * between process-level code and irq/NMI handlers, all running on the same CPU,
+ * and (2) Ensuring that the compiler does not  fold, spindle, or otherwise
+ * mutilate accesses that either do not require ordering or that interact
+ * with an explicit memory barrier or atomic instruction that provides the
+ * required ordering.
+ *
+ * If possible use READ_ONCE/ASSIGN_ONCE instead.
  */
-#define ACCESS_ONCE(x) (*(volatile typeof(x) *)&(x))
+#define __ACCESS_ONCE(x) ({ \
+	 __maybe_unused typeof(x) __var = 0; \
+	(volatile typeof(x) *)&(x); })
+#define ACCESS_ONCE(x) (*__ACCESS_ONCE(x))
 
 /* Ignore/forbid kprobes attach on very low level functions marked by this attribute: */
 #ifdef CONFIG_KPROBES
-- 
1.9.3

[PATCH 7/8] next: sh: Fix compile error

[Christian Borntraeger]

From: Guenter Roeck <linux@roeck-us.net>

Commit a91ed664749c ("kernel: tighten rules for ACCESS ONCE") results in a
compile failure for sh builds with CONFIG_X2TLB enabled.

arch/sh/mm/gup.c: In function 'gup_get_pte':
arch/sh/mm/gup.c:20:2: error: invalid initializer
make[1]: *** [arch/sh/mm/gup.o] Error 1

Replace ACCESS_ONCE with READ_ONCE to fix the problem.

Fixes: a91ed664749c ("kernel: tighten rules for ACCESS ONCE")
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 arch/sh/mm/gup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/sh/mm/gup.c b/arch/sh/mm/gup.c
index 37458f3..e113bb4 100644
--- a/arch/sh/mm/gup.c
+++ b/arch/sh/mm/gup.c
@@ -17,7 +17,7 @@
 static inline pte_t gup_get_pte(pte_t *ptep)
 {
 #ifndef CONFIG_X2TLB
-	return ACCESS_ONCE(*ptep);
+	return READ_ONCE(*ptep);
 #else
 	/*
 	 * With get_user_pages_fast, we walk down the pagetables without
-- 
1.9.3

[PATCH 8/8] kernel: Fix sparse warning for ACCESS_ONCE

[Christian Borntraeger]

Commit a91ed664749c ("kernel: tighten rules for ACCESS ONCE") results in
sparse warnings like "Using plain integer as NULL pointer" - Let's add a
type cast to the dummy assignment.
To avoid warnings lik "sparse: warning: cast to restricted __hc32" we also
use __force on that cast.

Fixes: a91ed664749c ("kernel: tighten rules for ACCESS ONCE")
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 include/linux/compiler.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 5e186bf..7bebf05 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -461,7 +461,7 @@ static __always_inline void __assign_once_size(volatile void *p, void *res, int
  * If possible use READ_ONCE/ASSIGN_ONCE instead.
  */
 #define __ACCESS_ONCE(x) ({ \
-	 __maybe_unused typeof(x) __var = 0; \
+	 __maybe_unused typeof(x) __var = (__force typeof(x)) 0; \
 	(volatile typeof(x) *)&(x); })
 #define ACCESS_ONCE(x) (*__ACCESS_ONCE(x))
 
-- 
1.9.3

Re: [PATCH 0/8] current ACCESS_ONCE patch queue

[Alexander Graf]

On 15.01.15 09:58, Christian Borntraeger wrote:
> Folks,
> 
> fyi, this is my current patch queue for the next merge window. It
> does contain a patch that will disallow ACCESS_ONCE on non-scalar
> types.
> 
> The tree is part of linux-next and can be found at
> git://git.kernel.org/pub/scm/linux/kernel/git/borntraeger/linux.git linux-next

KVM PPC bits are:

 Acked-by: Alexander Graf <agraf@suse.de>



Alex