Re: [PATCH v3] arm: remove !CPU_V6 and !GENERIC_ATOMIC64 builddependencies for XEN

From: Stefano Stabellini
Date: Mon Jan 20 2014 - 10:34:05 EST


On Thu, 16 Jan 2014, Will Deacon wrote:
> Hi Stefano,
>
> On Thu, Jan 16, 2014 at 04:27:55PM +0000, Stefano Stabellini wrote:
> > On Thu, 9 Jan 2014, Will Deacon wrote:
> > > Ok, thanks for the explanation. Looking at the patch, I wonder whether it's
> > > not cleaner just to implement xchg code separately for Xen? The Linux code
> > > isn't always sufficient (due to the GENERIC_ATOMIC64 stuff) and most of the
> > > churn coming out of this patch is an attempt to provide some small code
> > > reuse at the cost of code readability.
> > >
> > > What do others think?
> >
> > I am OK with that, in fact my first version of the patch did just that:
> >
> > http://marc.info/?l=linux-arm-kernel&m=138436406724990&w=2
> >
> > Is that what you had in mind?
>
> For the xchg part, yes, that looks a lot better. I don't like the #undef
> CONFIG_CPU_V6 though, can that be rewritten to use __LINUX_ARM_ARCH__?

The problem is that the 1 and 2 byte parameter size cases in __cmpxchg
are ifdef'ed CONFIG_CPU_V6 but drivers/xen/grant-table.c needs them.

So we can either undef CONFIG_CPU_V6 in grant-table.c or call a
different function.

If I switch from ifdef CONFIG_CPU_V6 to if __LINUX_ARM_ARCH__ > 6 in
__cmpxchg, we still have the problem that if __LINUX_ARM_ARCH__ == 6,
grant-table.c doesn't compile.

Maybe the approach taken by the other patch for cmpxchg is better, see
below.


diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index c1f1a7e..ae54ae0 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -1881,8 +1881,7 @@ config XEN_DOM0
config XEN
bool "Xen guest support on ARM (EXPERIMENTAL)"
depends on ARM && AEABI && OF
- depends on CPU_V7 && !CPU_V6
- depends on !GENERIC_ATOMIC64
+ depends on CPU_V7
select ARM_PSCI
select SWIOTLB_XEN
help
diff --git a/arch/arm/include/asm/cmpxchg.h b/arch/arm/include/asm/cmpxchg.h
index df2fbba..cc8a4a2 100644
--- a/arch/arm/include/asm/cmpxchg.h
+++ b/arch/arm/include/asm/cmpxchg.h
@@ -133,6 +133,44 @@ extern void __bad_cmpxchg(volatile void *ptr, int size);
* cmpxchg only support 32-bits operands on ARMv6.
*/

+static inline unsigned long __cmpxchg8(volatile void *ptr, unsigned long old,
+ unsigned long new)
+{
+ unsigned long oldval, res;
+
+ do {
+ asm volatile("@ __cmpxchg1\n"
+ " ldrexb %1, [%2]\n"
+ " mov %0, #0\n"
+ " teq %1, %3\n"
+ " strexbeq %0, %4, [%2]\n"
+ : "=&r" (res), "=&r" (oldval)
+ : "r" (ptr), "Ir" (old), "r" (new)
+ : "memory", "cc");
+ } while (res);
+
+ return oldval;
+}
+
+static inline unsigned long __cmpxchg16(volatile void *ptr, unsigned long old,
+ unsigned long new)
+{
+ unsigned long oldval, res;
+
+ do {
+ asm volatile("@ __cmpxchg1\n"
+ " ldrexh %1, [%2]\n"
+ " mov %0, #0\n"
+ " teq %1, %3\n"
+ " strexheq %0, %4, [%2]\n"
+ : "=&r" (res), "=&r" (oldval)
+ : "r" (ptr), "Ir" (old), "r" (new)
+ : "memory", "cc");
+ } while (res);
+
+ return oldval;
+}
+
static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old,
unsigned long new, int size)
{
@@ -141,28 +179,10 @@ static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old,
switch (size) {
#ifndef CONFIG_CPU_V6 /* min ARCH >= ARMv6K */
case 1:
- do {
- asm volatile("@ __cmpxchg1\n"
- " ldrexb %1, [%2]\n"
- " mov %0, #0\n"
- " teq %1, %3\n"
- " strexbeq %0, %4, [%2]\n"
- : "=&r" (res), "=&r" (oldval)
- : "r" (ptr), "Ir" (old), "r" (new)
- : "memory", "cc");
- } while (res);
+ oldval = __cmpxchg8(ptr, old, new);
break;
case 2:
- do {
- asm volatile("@ __cmpxchg1\n"
- " ldrexh %1, [%2]\n"
- " mov %0, #0\n"
- " teq %1, %3\n"
- " strexheq %0, %4, [%2]\n"
- : "=&r" (res), "=&r" (oldval)
- : "r" (ptr), "Ir" (old), "r" (new)
- : "memory", "cc");
- } while (res);
+ oldval = __cmpxchg16(ptr, old, new);
break;
#endif
case 4:
diff --git a/arch/arm/include/asm/sync_bitops.h b/arch/arm/include/asm/sync_bitops.h
index 63479ee..942659a 100644
--- a/arch/arm/include/asm/sync_bitops.h
+++ b/arch/arm/include/asm/sync_bitops.h
@@ -21,7 +21,29 @@
#define sync_test_and_clear_bit(nr, p) _test_and_clear_bit(nr, p)
#define sync_test_and_change_bit(nr, p) _test_and_change_bit(nr, p)
#define sync_test_bit(nr, addr) test_bit(nr, addr)
-#define sync_cmpxchg cmpxchg

+static inline unsigned long sync_cmpxchg(volatile void *ptr,
+ unsigned long old,
+ unsigned long new)
+{
+ unsigned long oldval;
+ int size = sizeof(*(ptr));
+
+ smp_mb();
+ switch (size) {
+ case 1:
+ oldval = __cmpxchg8(ptr, old, new);
+ break;
+ case 2:
+ oldval = __cmpxchg16(ptr, old, new);
+ break;
+ default:
+ oldval = __cmpxchg(ptr, old, new, size);
+ break;
+ }
+ smp_mb();
+
+ return oldval;
+}

#endif
diff --git a/arch/arm/include/asm/xen/events.h b/arch/arm/include/asm/xen/events.h
index 8b1f37b..2032ee6 100644
--- a/arch/arm/include/asm/xen/events.h
+++ b/arch/arm/include/asm/xen/events.h
@@ -16,7 +16,37 @@ static inline int xen_irqs_disabled(struct pt_regs *regs)
return raw_irqs_disabled_flags(regs->ARM_cpsr);
}

-#define xchg_xen_ulong(ptr, val) atomic64_xchg(container_of((ptr), \
+#ifdef CONFIG_GENERIC_ATOMIC64
+/* if CONFIG_GENERIC_ATOMIC64 is defined we cannot use the generic
+ * atomic64_xchg function because it is implemented using spin locks.
+ * Here we need proper atomic instructions to read and write memory
+ * shared with the hypervisor.
+ */
+static inline u64 xen_atomic64_xchg(atomic64_t *ptr, u64 new)
+{
+ u64 result;
+ unsigned long tmp;
+
+ smp_mb();
+
+ __asm__ __volatile__("@ xen_atomic64_xchg\n"
+"1: ldrexd %0, %H0, [%3]\n"
+" strexd %1, %4, %H4, [%3]\n"
+" teq %1, #0\n"
+" bne 1b"
+ : "=&r" (result), "=&r" (tmp), "+Qo" (ptr->counter)
+ : "r" (&ptr->counter), "r" (new)
+ : "cc");
+
+ smp_mb();
+
+ return result;
+}
+#else
+#define xen_atomic64_xchg atomic64_xchg
+#endif
+
+#define xchg_xen_ulong(ptr, val) xen_atomic64_xchg(container_of((ptr), \
atomic64_t, \
counter), (val))

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/