Mailing List Archive

[PATCH v5 09/13] xen/spinlock: split recursive spinlocks from normal ones
Recursive and normal spinlocks are sharing the same data structure for
representation of the lock. This has two major disadvantages:

- it is not clear from the definition of a lock, whether it is intended
to be used recursive or not, while a mixture of both usage variants
needs to be

- in production builds (builds without CONFIG_DEBUG_LOCKS) the needed
data size of an ordinary spinlock is 8 bytes instead of 4, due to the
additional recursion data needed (associated with that the rwlock
data is using 12 instead of only 8 bytes)

Fix that by introducing a struct spinlock_recursive for recursive
spinlocks only, and switch recursive spinlock functions to require
pointers to this new struct.

This allows to check the correct usage at build time.

Signed-off-by: Juergen Gross <jgross@suse.com>
---
V2:
- use shorter names (Jan Beulich)
- don't embed spinlock_t in rspinlock_t (Jan Beulich)
V5:
- some style fixes (Jan Beulich)
- bool instead of int (Jan Beulich)
---
xen/common/spinlock.c | 50 ++++++++++++++++++++++++++
xen/include/xen/spinlock.h | 72 +++++++++++++++++++++++++++++---------
2 files changed, 105 insertions(+), 17 deletions(-)

diff --git a/xen/common/spinlock.c b/xen/common/spinlock.c
index c3f2f9b209..a88ad9b93c 100644
--- a/xen/common/spinlock.c
+++ b/xen/common/spinlock.c
@@ -541,6 +541,56 @@ void _rspin_unlock_irqrestore(rspinlock_t *lock, unsigned long flags)
local_irq_restore(flags);
}

+bool _nrspin_trylock(rspinlock_t *lock)
+{
+ check_lock(&lock->debug, true);
+
+ if ( unlikely(lock->recurse_cpu != SPINLOCK_NO_CPU) )
+ return false;
+
+ return spin_trylock_common(&lock->tickets, &lock->debug, LOCK_PROFILE_PAR);
+}
+
+void _nrspin_lock(rspinlock_t *lock)
+{
+ spin_lock_common(&lock->tickets, &lock->debug, LOCK_PROFILE_PAR, NULL,
+ NULL);
+}
+
+void _nrspin_unlock(rspinlock_t *lock)
+{
+ spin_unlock_common(&lock->tickets, &lock->debug, LOCK_PROFILE_PAR);
+}
+
+void _nrspin_lock_irq(rspinlock_t *lock)
+{
+ ASSERT(local_irq_is_enabled());
+ local_irq_disable();
+ _nrspin_lock(lock);
+}
+
+void _nrspin_unlock_irq(rspinlock_t *lock)
+{
+ _nrspin_unlock(lock);
+ local_irq_enable();
+}
+
+unsigned long _nrspin_lock_irqsave(rspinlock_t *lock)
+{
+ unsigned long flags;
+
+ local_irq_save(flags);
+ _nrspin_lock(lock);
+
+ return flags;
+}
+
+void _nrspin_unlock_irqrestore(rspinlock_t *lock, unsigned long flags)
+{
+ _nrspin_unlock(lock);
+ local_irq_restore(flags);
+}
+
#ifdef CONFIG_DEBUG_LOCK_PROFILE

struct lock_profile_anc {
diff --git a/xen/include/xen/spinlock.h b/xen/include/xen/spinlock.h
index 7dd11faab3..181e5c7d35 100644
--- a/xen/include/xen/spinlock.h
+++ b/xen/include/xen/spinlock.h
@@ -77,8 +77,6 @@ union lock_debug { };
*/

struct spinlock;
-/* Temporary hack until a dedicated struct rspinlock is existing. */
-#define rspinlock spinlock

struct lock_profile {
struct lock_profile *next; /* forward link */
@@ -110,6 +108,10 @@ struct lock_profile_qhead {
__used_section(".lockprofile.data") = \
&lock_profile_data__##name
#define SPIN_LOCK_UNLOCKED_(x) { \
+ .debug = LOCK_DEBUG_, \
+ .profile = x, \
+}
+#define RSPIN_LOCK_UNLOCKED_(x) { \
.recurse_cpu = SPINLOCK_NO_CPU, \
.debug = LOCK_DEBUG_, \
.profile = x, \
@@ -119,8 +121,9 @@ struct lock_profile_qhead {
spinlock_t l = SPIN_LOCK_UNLOCKED_(NULL); \
static struct lock_profile lock_profile_data__##l = LOCK_PROFILE_(l); \
LOCK_PROFILE_PTR_(l)
+#define RSPIN_LOCK_UNLOCKED RSPIN_LOCK_UNLOCKED_(NULL)
#define DEFINE_RSPINLOCK(l) \
- rspinlock_t l = SPIN_LOCK_UNLOCKED_(NULL); \
+ rspinlock_t l = RSPIN_LOCK_UNLOCKED_(NULL); \
static struct lock_profile lock_profile_data__##l = RLOCK_PROFILE_(l); \
LOCK_PROFILE_PTR_(l)

@@ -145,8 +148,11 @@ struct lock_profile_qhead {

#define spin_lock_init_prof(s, l) \
spin_lock_init_prof__(s, l, lock, spinlock_t, false)
-#define rspin_lock_init_prof(s, l) \
- spin_lock_init_prof__(s, l, rlock, rspinlock_t, true)
+#define rspin_lock_init_prof(s, l) do { \
+ spin_lock_init_prof__(s, l, rlock, rspinlock_t, true); \
+ (s)->l.recurse_cpu = SPINLOCK_NO_CPU; \
+ (s)->l.recurse_cnt = 0; \
+ } while (0)

void _lock_profile_register_struct(
int32_t type, struct lock_profile_qhead *qhead, int32_t idx);
@@ -168,11 +174,14 @@ struct lock_profile_qhead { };
struct lock_profile { };

#define SPIN_LOCK_UNLOCKED { \
+ .debug = LOCK_DEBUG_, \
+}
+#define RSPIN_LOCK_UNLOCKED { \
.recurse_cpu = SPINLOCK_NO_CPU, \
.debug = LOCK_DEBUG_, \
}
#define DEFINE_SPINLOCK(l) spinlock_t l = SPIN_LOCK_UNLOCKED
-#define DEFINE_RSPINLOCK(l) rspinlock_t l = SPIN_LOCK_UNLOCKED
+#define DEFINE_RSPINLOCK(l) rspinlock_t l = RSPIN_LOCK_UNLOCKED

#define spin_lock_init_prof(s, l) spin_lock_init(&((s)->l))
#define rspin_lock_init_prof(s, l) rspin_lock_init(&((s)->l))
@@ -193,6 +202,14 @@ typedef union {
#define SPINLOCK_TICKET_INC { .head_tail = 0x10000, }

typedef struct spinlock {
+ spinlock_tickets_t tickets;
+ union lock_debug debug;
+#ifdef CONFIG_DEBUG_LOCK_PROFILE
+ struct lock_profile *profile;
+#endif
+} spinlock_t;
+
+typedef struct rspinlock {
spinlock_tickets_t tickets;
uint16_t recurse_cpu:SPINLOCK_CPU_BITS;
#define SPINLOCK_NO_CPU ((1u << SPINLOCK_CPU_BITS) - 1)
@@ -203,12 +220,10 @@ typedef struct spinlock {
#ifdef CONFIG_DEBUG_LOCK_PROFILE
struct lock_profile *profile;
#endif
-} spinlock_t;
-
-typedef spinlock_t rspinlock_t;
+} rspinlock_t;

#define spin_lock_init(l) (*(l) = (spinlock_t)SPIN_LOCK_UNLOCKED)
-#define rspin_lock_init(l) (*(l) = (rspinlock_t)SPIN_LOCK_UNLOCKED)
+#define rspin_lock_init(l) (*(l) = (rspinlock_t)RSPIN_LOCK_UNLOCKED)

void _spin_lock(spinlock_t *lock);
void _spin_lock_cb(spinlock_t *lock, void (*cb)(void *data), void *data);
@@ -312,12 +327,35 @@ static always_inline void rspin_lock(rspinlock_t *lock)
#define rspin_barrier(l) _rspin_barrier(l)
#define rspin_is_locked(l) _rspin_is_locked(l)

-#define nrspin_trylock(l) spin_trylock(l)
-#define nrspin_lock(l) spin_lock(l)
-#define nrspin_unlock(l) spin_unlock(l)
-#define nrspin_lock_irq(l) spin_lock_irq(l)
-#define nrspin_unlock_irq(l) spin_unlock_irq(l)
-#define nrspin_lock_irqsave(l, f) spin_lock_irqsave(l, f)
-#define nrspin_unlock_irqrestore(l, f) spin_unlock_irqrestore(l, f)
+bool _nrspin_trylock(rspinlock_t *lock);
+void _nrspin_lock(rspinlock_t *lock);
+#define nrspin_lock_irqsave(l, f) \
+ ({ \
+ BUILD_BUG_ON(sizeof(f) != sizeof(unsigned long)); \
+ (f) = _nrspin_lock_irqsave(l); \
+ block_lock_speculation(); \
+ })
+unsigned long _nrspin_lock_irqsave(rspinlock_t *lock);
+void _nrspin_unlock(rspinlock_t *lock);
+void _nrspin_lock_irq(rspinlock_t *lock);
+void _nrspin_unlock_irq(rspinlock_t *lock);
+void _nrspin_unlock_irqrestore(rspinlock_t *lock, unsigned long flags);
+
+static always_inline void nrspin_lock(rspinlock_t *lock)
+{
+ _nrspin_lock(lock);
+ block_lock_speculation();
+}
+
+static always_inline void nrspin_lock_irq(rspinlock_t *l)
+{
+ _nrspin_lock_irq(l);
+ block_lock_speculation();
+}
+
+#define nrspin_trylock(l) lock_evaluate_nospec(_nrspin_trylock(l))
+#define nrspin_unlock(l) _nrspin_unlock(l)
+#define nrspin_unlock_irqrestore(l, f) _nrspin_unlock_irqrestore(l, f)
+#define nrspin_unlock_irq(l) _nrspin_unlock_irq(l)

#endif /* __SPINLOCK_H__ */
--
2.35.3
Re: [PATCH v5 09/13] xen/spinlock: split recursive spinlocks from normal ones [ In reply to ]
On 14.03.2024 08:20, Juergen Gross wrote:
> Recursive and normal spinlocks are sharing the same data structure for
> representation of the lock. This has two major disadvantages:
>
> - it is not clear from the definition of a lock, whether it is intended
> to be used recursive or not, while a mixture of both usage variants
> needs to be
>
> - in production builds (builds without CONFIG_DEBUG_LOCKS) the needed
> data size of an ordinary spinlock is 8 bytes instead of 4, due to the
> additional recursion data needed (associated with that the rwlock
> data is using 12 instead of only 8 bytes)
>
> Fix that by introducing a struct spinlock_recursive for recursive
> spinlocks only, and switch recursive spinlock functions to require
> pointers to this new struct.
>
> This allows to check the correct usage at build time.
>
> Signed-off-by: Juergen Gross <jgross@suse.com>

Reviewed-by: Jan Beulich <jbeulich@suse.com>