Move generic non-atomic bitops from the asm-generic header which
gets included only when there are no architecture-specific
alternatives, to a separate independent file to make them always
available.
Signed-off-by: Alexander Lobakin <[email protected]>
---
.../asm-generic/bitops/generic-non-atomic.h | 124 ++++++++++++++++++
include/asm-generic/bitops/non-atomic.h | 109 ++-------------
2 files changed, 132 insertions(+), 101 deletions(-)
create mode 100644 include/asm-generic/bitops/generic-non-atomic.h
diff --git a/include/asm-generic/bitops/generic-non-atomic.h b/include/asm-generic/bitops/generic-non-atomic.h
new file mode 100644
index 000000000000..808bc4469886
--- /dev/null
+++ b/include/asm-generic/bitops/generic-non-atomic.h
@@ -0,0 +1,124 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef __ASM_GENERIC_BITOPS_GENERIC_NON_ATOMIC_H
+#define __ASM_GENERIC_BITOPS_GENERIC_NON_ATOMIC_H
+
+#include <linux/bits.h>
+
+#ifndef _LINUX_BITOPS_H
+#error only <linux/bitops.h> can be included directly
+#endif
+
+/*
+ * Generic definitions for bit operations, should not be used in regular code
+ * directly.
+ */
+
+/**
+ * generic___set_bit - Set a bit in memory
+ * @nr: the bit to set
+ * @addr: the address to start counting from
+ *
+ * Unlike set_bit(), this function is non-atomic and may be reordered.
+ * If it's called on the same region of memory simultaneously, the effect
+ * may be that only one operation succeeds.
+ */
+static __always_inline void
+generic___set_bit(unsigned int nr, volatile unsigned long *addr)
+{
+ unsigned long mask = BIT_MASK(nr);
+ unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
+
+ *p |= mask;
+}
+
+static __always_inline void
+generic___clear_bit(unsigned int nr, volatile unsigned long *addr)
+{
+ unsigned long mask = BIT_MASK(nr);
+ unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
+
+ *p &= ~mask;
+}
+
+/**
+ * generic___change_bit - Toggle a bit in memory
+ * @nr: the bit to change
+ * @addr: the address to start counting from
+ *
+ * Unlike change_bit(), this function is non-atomic and may be reordered.
+ * If it's called on the same region of memory simultaneously, the effect
+ * may be that only one operation succeeds.
+ */
+static __always_inline
+void generic___change_bit(unsigned int nr, volatile unsigned long *addr)
+{
+ unsigned long mask = BIT_MASK(nr);
+ unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
+
+ *p ^= mask;
+}
+
+/**
+ * generic___test_and_set_bit - Set a bit and return its old value
+ * @nr: Bit to set
+ * @addr: Address to count from
+ *
+ * This operation is non-atomic and can be reordered.
+ * If two examples of this operation race, one can appear to succeed
+ * but actually fail. You must protect multiple accesses with a lock.
+ */
+static __always_inline int
+generic___test_and_set_bit(unsigned int nr, volatile unsigned long *addr)
+{
+ unsigned long mask = BIT_MASK(nr);
+ unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
+ unsigned long old = *p;
+
+ *p = old | mask;
+ return (old & mask) != 0;
+}
+
+/**
+ * generic___test_and_clear_bit - Clear a bit and return its old value
+ * @nr: Bit to clear
+ * @addr: Address to count from
+ *
+ * This operation is non-atomic and can be reordered.
+ * If two examples of this operation race, one can appear to succeed
+ * but actually fail. You must protect multiple accesses with a lock.
+ */
+static __always_inline int
+generic___test_and_clear_bit(unsigned int nr, volatile unsigned long *addr)
+{
+ unsigned long mask = BIT_MASK(nr);
+ unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
+ unsigned long old = *p;
+
+ *p = old & ~mask;
+ return (old & mask) != 0;
+}
+
+/* WARNING: non atomic and it can be reordered! */
+static __always_inline int
+generic___test_and_change_bit(unsigned int nr, volatile unsigned long *addr)
+{
+ unsigned long mask = BIT_MASK(nr);
+ unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
+ unsigned long old = *p;
+
+ *p = old ^ mask;
+ return (old & mask) != 0;
+}
+
+/**
+ * generic_test_bit - Determine whether a bit is set
+ * @nr: bit number to test
+ * @addr: Address to start counting from
+ */
+static __always_inline int
+generic_test_bit(unsigned int nr, const volatile unsigned long *addr)
+{
+ return 1UL & (addr[BIT_WORD(nr)] >> (nr & (BITS_PER_LONG-1)));
+}
+
+#endif /* __ASM_GENERIC_BITOPS_GENERIC_NON_ATOMIC_H */
diff --git a/include/asm-generic/bitops/non-atomic.h b/include/asm-generic/bitops/non-atomic.h
index 078cc68be2f1..a05bc090a6a3 100644
--- a/include/asm-generic/bitops/non-atomic.h
+++ b/include/asm-generic/bitops/non-atomic.h
@@ -2,121 +2,28 @@
#ifndef _ASM_GENERIC_BITOPS_NON_ATOMIC_H_
#define _ASM_GENERIC_BITOPS_NON_ATOMIC_H_
+#include <asm-generic/bitops/generic-non-atomic.h>
#include <asm/types.h>
-/**
- * arch___set_bit - Set a bit in memory
- * @nr: the bit to set
- * @addr: the address to start counting from
- *
- * Unlike set_bit(), this function is non-atomic and may be reordered.
- * If it's called on the same region of memory simultaneously, the effect
- * may be that only one operation succeeds.
- */
-static __always_inline void
-arch___set_bit(unsigned int nr, volatile unsigned long *addr)
-{
- unsigned long mask = BIT_MASK(nr);
- unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
-
- *p |= mask;
-}
+#define arch___set_bit generic___set_bit
#define __set_bit arch___set_bit
-static __always_inline void
-arch___clear_bit(unsigned int nr, volatile unsigned long *addr)
-{
- unsigned long mask = BIT_MASK(nr);
- unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
-
- *p &= ~mask;
-}
+#define arch___clear_bit generic___clear_bit
#define __clear_bit arch___clear_bit
-/**
- * arch___change_bit - Toggle a bit in memory
- * @nr: the bit to change
- * @addr: the address to start counting from
- *
- * Unlike change_bit(), this function is non-atomic and may be reordered.
- * If it's called on the same region of memory simultaneously, the effect
- * may be that only one operation succeeds.
- */
-static __always_inline
-void arch___change_bit(unsigned int nr, volatile unsigned long *addr)
-{
- unsigned long mask = BIT_MASK(nr);
- unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
-
- *p ^= mask;
-}
+#define arch___change_bit generic___change_bit
#define __change_bit arch___change_bit
-/**
- * arch___test_and_set_bit - Set a bit and return its old value
- * @nr: Bit to set
- * @addr: Address to count from
- *
- * This operation is non-atomic and can be reordered.
- * If two examples of this operation race, one can appear to succeed
- * but actually fail. You must protect multiple accesses with a lock.
- */
-static __always_inline int
-arch___test_and_set_bit(unsigned int nr, volatile unsigned long *addr)
-{
- unsigned long mask = BIT_MASK(nr);
- unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
- unsigned long old = *p;
-
- *p = old | mask;
- return (old & mask) != 0;
-}
+#define arch___test_and_set_bit generic___test_and_set_bit
#define __test_and_set_bit arch___test_and_set_bit
-/**
- * arch___test_and_clear_bit - Clear a bit and return its old value
- * @nr: Bit to clear
- * @addr: Address to count from
- *
- * This operation is non-atomic and can be reordered.
- * If two examples of this operation race, one can appear to succeed
- * but actually fail. You must protect multiple accesses with a lock.
- */
-static __always_inline int
-arch___test_and_clear_bit(unsigned int nr, volatile unsigned long *addr)
-{
- unsigned long mask = BIT_MASK(nr);
- unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
- unsigned long old = *p;
-
- *p = old & ~mask;
- return (old & mask) != 0;
-}
+#define arch___test_and_clear_bit generic___test_and_clear_bit
#define __test_and_clear_bit arch___test_and_clear_bit
-/* WARNING: non atomic and it can be reordered! */
-static __always_inline int
-arch___test_and_change_bit(unsigned int nr, volatile unsigned long *addr)
-{
- unsigned long mask = BIT_MASK(nr);
- unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
- unsigned long old = *p;
-
- *p = old ^ mask;
- return (old & mask) != 0;
-}
+#define arch___test_and_change_bit generic___test_and_change_bit
#define __test_and_change_bit arch___test_and_change_bit
-/**
- * arch_test_bit - Determine whether a bit is set
- * @nr: bit number to test
- * @addr: Address to start counting from
- */
-static __always_inline int
-arch_test_bit(unsigned int nr, const volatile unsigned long *addr)
-{
- return 1UL & (addr[BIT_WORD(nr)] >> (nr & (BITS_PER_LONG-1)));
-}
+#define arch_test_bit generic_test_bit
#define test_bit arch_test_bit
#endif /* _ASM_GENERIC_BITOPS_NON_ATOMIC_H_ */
--
2.36.1
On Fri, Jun 10, 2022 at 01:34:23PM +0200, Alexander Lobakin wrote:
> Move generic non-atomic bitops from the asm-generic header which
> gets included only when there are no architecture-specific
> alternatives, to a separate independent file to make them always
> available.
Reviewed-by: Andy Shevchenko <[email protected]>
But see below.
> Signed-off-by: Alexander Lobakin <[email protected]>
> ---
> .../asm-generic/bitops/generic-non-atomic.h | 124 ++++++++++++++++++
> include/asm-generic/bitops/non-atomic.h | 109 ++-------------
> 2 files changed, 132 insertions(+), 101 deletions(-)
> create mode 100644 include/asm-generic/bitops/generic-non-atomic.h
>
> diff --git a/include/asm-generic/bitops/generic-non-atomic.h b/include/asm-generic/bitops/generic-non-atomic.h
> new file mode 100644
> index 000000000000..808bc4469886
> --- /dev/null
> +++ b/include/asm-generic/bitops/generic-non-atomic.h
> @@ -0,0 +1,124 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +#ifndef __ASM_GENERIC_BITOPS_GENERIC_NON_ATOMIC_H
> +#define __ASM_GENERIC_BITOPS_GENERIC_NON_ATOMIC_H
> +
> +#include <linux/bits.h>
> +
> +#ifndef _LINUX_BITOPS_H
> +#error only <linux/bitops.h> can be included directly
> +#endif
> +
> +/*
> + * Generic definitions for bit operations, should not be used in regular code
> + * directly.
> + */
> +
> +/**
> + * generic___set_bit - Set a bit in memory
> + * @nr: the bit to set
> + * @addr: the address to start counting from
> + *
> + * Unlike set_bit(), this function is non-atomic and may be reordered.
> + * If it's called on the same region of memory simultaneously, the effect
> + * may be that only one operation succeeds.
> + */
> +static __always_inline void
> +generic___set_bit(unsigned int nr, volatile unsigned long *addr)
> +{
> + unsigned long mask = BIT_MASK(nr);
> + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> +
> + *p |= mask;
> +}
> +
> +static __always_inline void
> +generic___clear_bit(unsigned int nr, volatile unsigned long *addr)
> +{
> + unsigned long mask = BIT_MASK(nr);
> + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> +
> + *p &= ~mask;
> +}
> +
> +/**
> + * generic___change_bit - Toggle a bit in memory
> + * @nr: the bit to change
> + * @addr: the address to start counting from
> + *
> + * Unlike change_bit(), this function is non-atomic and may be reordered.
> + * If it's called on the same region of memory simultaneously, the effect
> + * may be that only one operation succeeds.
> + */
> +static __always_inline
> +void generic___change_bit(unsigned int nr, volatile unsigned long *addr)
> +{
> + unsigned long mask = BIT_MASK(nr);
> + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> +
> + *p ^= mask;
> +}
> +
> +/**
> + * generic___test_and_set_bit - Set a bit and return its old value
> + * @nr: Bit to set
> + * @addr: Address to count from
> + *
> + * This operation is non-atomic and can be reordered.
> + * If two examples of this operation race, one can appear to succeed
> + * but actually fail. You must protect multiple accesses with a lock.
> + */
> +static __always_inline int
> +generic___test_and_set_bit(unsigned int nr, volatile unsigned long *addr)
> +{
> + unsigned long mask = BIT_MASK(nr);
> + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> + unsigned long old = *p;
> +
> + *p = old | mask;
> + return (old & mask) != 0;
> +}
> +
> +/**
> + * generic___test_and_clear_bit - Clear a bit and return its old value
> + * @nr: Bit to clear
> + * @addr: Address to count from
> + *
> + * This operation is non-atomic and can be reordered.
> + * If two examples of this operation race, one can appear to succeed
> + * but actually fail. You must protect multiple accesses with a lock.
> + */
> +static __always_inline int
> +generic___test_and_clear_bit(unsigned int nr, volatile unsigned long *addr)
> +{
> + unsigned long mask = BIT_MASK(nr);
> + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> + unsigned long old = *p;
> +
> + *p = old & ~mask;
> + return (old & mask) != 0;
> +}
> +
> +/* WARNING: non atomic and it can be reordered! */
> +static __always_inline int
> +generic___test_and_change_bit(unsigned int nr, volatile unsigned long *addr)
> +{
> + unsigned long mask = BIT_MASK(nr);
> + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> + unsigned long old = *p;
> +
> + *p = old ^ mask;
> + return (old & mask) != 0;
> +}
> +
> +/**
> + * generic_test_bit - Determine whether a bit is set
> + * @nr: bit number to test
> + * @addr: Address to start counting from
> + */
Shouldn't we add in this or in separate patch a big NOTE to explain that this
is actually atomic and must be kept as a such?
> +static __always_inline int
> +generic_test_bit(unsigned int nr, const volatile unsigned long *addr)
> +{
> + return 1UL & (addr[BIT_WORD(nr)] >> (nr & (BITS_PER_LONG-1)));
> +}
> +
> +#endif /* __ASM_GENERIC_BITOPS_GENERIC_NON_ATOMIC_H */
> diff --git a/include/asm-generic/bitops/non-atomic.h b/include/asm-generic/bitops/non-atomic.h
> index 078cc68be2f1..a05bc090a6a3 100644
> --- a/include/asm-generic/bitops/non-atomic.h
> +++ b/include/asm-generic/bitops/non-atomic.h
> @@ -2,121 +2,28 @@
> #ifndef _ASM_GENERIC_BITOPS_NON_ATOMIC_H_
> #define _ASM_GENERIC_BITOPS_NON_ATOMIC_H_
>
> +#include <asm-generic/bitops/generic-non-atomic.h>
> #include <asm/types.h>
>
> -/**
> - * arch___set_bit - Set a bit in memory
> - * @nr: the bit to set
> - * @addr: the address to start counting from
> - *
> - * Unlike set_bit(), this function is non-atomic and may be reordered.
> - * If it's called on the same region of memory simultaneously, the effect
> - * may be that only one operation succeeds.
> - */
> -static __always_inline void
> -arch___set_bit(unsigned int nr, volatile unsigned long *addr)
> -{
> - unsigned long mask = BIT_MASK(nr);
> - unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> -
> - *p |= mask;
> -}
> +#define arch___set_bit generic___set_bit
> #define __set_bit arch___set_bit
>
> -static __always_inline void
> -arch___clear_bit(unsigned int nr, volatile unsigned long *addr)
> -{
> - unsigned long mask = BIT_MASK(nr);
> - unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> -
> - *p &= ~mask;
> -}
> +#define arch___clear_bit generic___clear_bit
> #define __clear_bit arch___clear_bit
>
> -/**
> - * arch___change_bit - Toggle a bit in memory
> - * @nr: the bit to change
> - * @addr: the address to start counting from
> - *
> - * Unlike change_bit(), this function is non-atomic and may be reordered.
> - * If it's called on the same region of memory simultaneously, the effect
> - * may be that only one operation succeeds.
> - */
> -static __always_inline
> -void arch___change_bit(unsigned int nr, volatile unsigned long *addr)
> -{
> - unsigned long mask = BIT_MASK(nr);
> - unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> -
> - *p ^= mask;
> -}
> +#define arch___change_bit generic___change_bit
> #define __change_bit arch___change_bit
>
> -/**
> - * arch___test_and_set_bit - Set a bit and return its old value
> - * @nr: Bit to set
> - * @addr: Address to count from
> - *
> - * This operation is non-atomic and can be reordered.
> - * If two examples of this operation race, one can appear to succeed
> - * but actually fail. You must protect multiple accesses with a lock.
> - */
> -static __always_inline int
> -arch___test_and_set_bit(unsigned int nr, volatile unsigned long *addr)
> -{
> - unsigned long mask = BIT_MASK(nr);
> - unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> - unsigned long old = *p;
> -
> - *p = old | mask;
> - return (old & mask) != 0;
> -}
> +#define arch___test_and_set_bit generic___test_and_set_bit
> #define __test_and_set_bit arch___test_and_set_bit
>
> -/**
> - * arch___test_and_clear_bit - Clear a bit and return its old value
> - * @nr: Bit to clear
> - * @addr: Address to count from
> - *
> - * This operation is non-atomic and can be reordered.
> - * If two examples of this operation race, one can appear to succeed
> - * but actually fail. You must protect multiple accesses with a lock.
> - */
> -static __always_inline int
> -arch___test_and_clear_bit(unsigned int nr, volatile unsigned long *addr)
> -{
> - unsigned long mask = BIT_MASK(nr);
> - unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> - unsigned long old = *p;
> -
> - *p = old & ~mask;
> - return (old & mask) != 0;
> -}
> +#define arch___test_and_clear_bit generic___test_and_clear_bit
> #define __test_and_clear_bit arch___test_and_clear_bit
>
> -/* WARNING: non atomic and it can be reordered! */
> -static __always_inline int
> -arch___test_and_change_bit(unsigned int nr, volatile unsigned long *addr)
> -{
> - unsigned long mask = BIT_MASK(nr);
> - unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr);
> - unsigned long old = *p;
> -
> - *p = old ^ mask;
> - return (old & mask) != 0;
> -}
> +#define arch___test_and_change_bit generic___test_and_change_bit
> #define __test_and_change_bit arch___test_and_change_bit
>
> -/**
> - * arch_test_bit - Determine whether a bit is set
> - * @nr: bit number to test
> - * @addr: Address to start counting from
> - */
> -static __always_inline int
> -arch_test_bit(unsigned int nr, const volatile unsigned long *addr)
> -{
> - return 1UL & (addr[BIT_WORD(nr)] >> (nr & (BITS_PER_LONG-1)));
> -}
> +#define arch_test_bit generic_test_bit
> #define test_bit arch_test_bit
>
> #endif /* _ASM_GENERIC_BITOPS_NON_ATOMIC_H_ */
> --
> 2.36.1
>
--
With Best Regards,
Andy Shevchenko
On Fri, 10 Jun 2022 at 18:02, Luck, Tony <[email protected]> wrote:
>
> > > +/**
> > > + * generic_test_bit - Determine whether a bit is set
> > > + * @nr: bit number to test
> > > + * @addr: Address to start counting from
> > > + */
> >
> > Shouldn't we add in this or in separate patch a big NOTE to explain that this
> > is actually atomic and must be kept as a such?
>
> "atomic" isn't really the right word. The volatile access makes sure that the
> compiler does the test at the point that the source code asked, and doesn't
> move it before/after other operations.
It's listed in Documentation/atomic_bitops.txt.
It is as "atomic" as READ_ONCE() or atomic_read() is. Though you are
right that the "atomicity" of reading one bit is almost a given,
because we can't really read half a bit.
The main thing is that the compiler keeps it "atomic" and e.g. doesn't
fuse the load with another or elide it completely, and then transforms
the code in concurrency-unfriendly ways.
Like READ_ONCE() and friends, test_bit(), unlike non-atomic bitops,
may also be used to dependency-order some subsequent marked (viz.
atomic) operations.
> But there is no such thing as an atomic test_bit() operation:
>
> if (test_bit(5, addr)) {
> /* some other CPU nukes bit 5 */
>
> /* I know it was set when I looked, but now, could be anything */
The operation itself is atomic, because reading half a bit is
impossible. Whether or not that bit is modified concurrently is a
different problem.
Thanks,
-- Marco
> > +/**
> > + * generic_test_bit - Determine whether a bit is set
> > + * @nr: bit number to test
> > + * @addr: Address to start counting from
> > + */
>
> Shouldn't we add in this or in separate patch a big NOTE to explain that this
> is actually atomic and must be kept as a such?
"atomic" isn't really the right word. The volatile access makes sure that the
compiler does the test at the point that the source code asked, and doesn't
move it before/after other operations.
But there is no such thing as an atomic test_bit() operation:
if (test_bit(5, addr)) {
/* some other CPU nukes bit 5 */
/* I know it was set when I looked, but now, could be anything */
...
}
-Tony
From: Marco Elver <[email protected]>
Date: Fri, 10 Jun 2022 18:32:36 +0200
> On Fri, 10 Jun 2022 at 18:02, Luck, Tony <[email protected]> wrote:
> >
> > > > +/**
> > > > + * generic_test_bit - Determine whether a bit is set
> > > > + * @nr: bit number to test
> > > > + * @addr: Address to start counting from
> > > > + */
> > >
> > > Shouldn't we add in this or in separate patch a big NOTE to explain that this
> > > is actually atomic and must be kept as a such?
> >
> > "atomic" isn't really the right word. The volatile access makes sure that the
> > compiler does the test at the point that the source code asked, and doesn't
> > move it before/after other operations.
>
> It's listed in Documentation/atomic_bitops.txt.
Oh, so my memory was actually correct that I saw it in the docs
somewhere.
WDYT, should I mention this here in the code (block comment) as well
that it's atomic and must not lose `volatile` as Andy suggested or
it's sufficient to have it in the docs (+ it's not underscored)?
>
> It is as "atomic" as READ_ONCE() or atomic_read() is. Though you are
> right that the "atomicity" of reading one bit is almost a given,
> because we can't really read half a bit.
> The main thing is that the compiler keeps it "atomic" and e.g. doesn't
> fuse the load with another or elide it completely, and then transforms
> the code in concurrency-unfriendly ways.
>
> Like READ_ONCE() and friends, test_bit(), unlike non-atomic bitops,
> may also be used to dependency-order some subsequent marked (viz.
> atomic) operations.
>
> > But there is no such thing as an atomic test_bit() operation:
> >
> > if (test_bit(5, addr)) {
> > /* some other CPU nukes bit 5 */
> >
> > /* I know it was set when I looked, but now, could be anything */
>
> The operation itself is atomic, because reading half a bit is
> impossible. Whether or not that bit is modified concurrently is a
> different problem.
>
> Thanks,
> -- Marco
Thanks,
Olek
On Mon, 13 Jun 2022 at 16:21, Alexander Lobakin
<[email protected]> wrote:
>
> From: Marco Elver <[email protected]>
> Date: Fri, 10 Jun 2022 18:32:36 +0200
>
> > On Fri, 10 Jun 2022 at 18:02, Luck, Tony <[email protected]> wrote:
> > >
> > > > > +/**
> > > > > + * generic_test_bit - Determine whether a bit is set
> > > > > + * @nr: bit number to test
> > > > > + * @addr: Address to start counting from
> > > > > + */
> > > >
> > > > Shouldn't we add in this or in separate patch a big NOTE to explain that this
> > > > is actually atomic and must be kept as a such?
> > >
> > > "atomic" isn't really the right word. The volatile access makes sure that the
> > > compiler does the test at the point that the source code asked, and doesn't
> > > move it before/after other operations.
> >
> > It's listed in Documentation/atomic_bitops.txt.
>
> Oh, so my memory was actually correct that I saw it in the docs
> somewhere.
> WDYT, should I mention this here in the code (block comment) as well
> that it's atomic and must not lose `volatile` as Andy suggested or
> it's sufficient to have it in the docs (+ it's not underscored)?
Perhaps a quick comment in the code (not kerneldoc above) will be
sufficient, with reference to Documentation/atomic_bitops.txt.
>> It's listed in Documentation/atomic_bitops.txt.
>
> Oh, so my memory was actually correct that I saw it in the docs
> somewhere.
> WDYT, should I mention this here in the code (block comment) as well
> that it's atomic and must not lose `volatile` as Andy suggested or
> it's sufficient to have it in the docs (+ it's not underscored)?
I think a comment that the "volatile" is required to prevent re-ordering
is enough.
But maybe others are sufficiently clear on the meaning? I once wasted
time looking for the non-atomic __test_bit() version (to use in some code
that was already protected by a spin lock, so didn't need the overhead
of an "atomic" version) before realizing there wasn't a non-atomic one.
-Tony
From: Luck, Tony
> Sent: 13 June 2022 17:27
>
> >> It's listed in Documentation/atomic_bitops.txt.
> >
> > Oh, so my memory was actually correct that I saw it in the docs
> > somewhere.
> > WDYT, should I mention this here in the code (block comment) as well
> > that it's atomic and must not lose `volatile` as Andy suggested or
> > it's sufficient to have it in the docs (+ it's not underscored)?
>
> I think a comment that the "volatile" is required to prevent re-ordering
> is enough.
>
> But maybe others are sufficiently clear on the meaning? I once wasted
> time looking for the non-atomic __test_bit() version (to use in some code
> that was already protected by a spin lock, so didn't need the overhead
> of an "atomic" version) before realizing there wasn't a non-atomic one.
Does it make any sense for 'test bit' to be atomic?
I'm not even sure is needs any ordering constraints either.
The result is always stale - the value can be changed by
another cpu at any time.
The set/clear atomic bit-ops require a RMW bus cycle - which has
to be locked (or similar) to avoid corruption.
The atomic 'test and set' (etc) are RMW and return a valid state.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
On Mon, Jun 13, 2022 at 04:33:17PM +0200, Marco Elver wrote:
> On Mon, 13 Jun 2022 at 16:21, Alexander Lobakin
> <[email protected]> wrote:
> >
> > From: Marco Elver <[email protected]>
> > Date: Fri, 10 Jun 2022 18:32:36 +0200
> >
> > > On Fri, 10 Jun 2022 at 18:02, Luck, Tony <[email protected]> wrote:
> > > >
> > > > > > +/**
> > > > > > + * generic_test_bit - Determine whether a bit is set
> > > > > > + * @nr: bit number to test
> > > > > > + * @addr: Address to start counting from
> > > > > > + */
> > > > >
> > > > > Shouldn't we add in this or in separate patch a big NOTE to explain that this
> > > > > is actually atomic and must be kept as a such?
> > > >
> > > > "atomic" isn't really the right word. The volatile access makes sure that the
> > > > compiler does the test at the point that the source code asked, and doesn't
> > > > move it before/after other operations.
> > >
> > > It's listed in Documentation/atomic_bitops.txt.
> >
> > Oh, so my memory was actually correct that I saw it in the docs
> > somewhere.
> > WDYT, should I mention this here in the code (block comment) as well
> > that it's atomic and must not lose `volatile` as Andy suggested or
> > it's sufficient to have it in the docs (+ it's not underscored)?
>
> Perhaps a quick comment in the code (not kerneldoc above) will be
> sufficient, with reference to Documentation/atomic_bitops.txt.
If it may help, we can do:
/*
* Bit testing is a naturally atomic operation because bit is
* a minimal quantum of information.
*/
#define __test_bit test_bit
On Wed, 15 Jun 2022 at 04:47, Yury Norov <[email protected]> wrote:
>
> On Mon, Jun 13, 2022 at 04:33:17PM +0200, Marco Elver wrote:
> > On Mon, 13 Jun 2022 at 16:21, Alexander Lobakin
> > <[email protected]> wrote:
> > >
> > > From: Marco Elver <[email protected]>
> > > Date: Fri, 10 Jun 2022 18:32:36 +0200
> > >
> > > > On Fri, 10 Jun 2022 at 18:02, Luck, Tony <[email protected]> wrote:
> > > > >
> > > > > > > +/**
> > > > > > > + * generic_test_bit - Determine whether a bit is set
> > > > > > > + * @nr: bit number to test
> > > > > > > + * @addr: Address to start counting from
> > > > > > > + */
> > > > > >
> > > > > > Shouldn't we add in this or in separate patch a big NOTE to explain that this
> > > > > > is actually atomic and must be kept as a such?
> > > > >
> > > > > "atomic" isn't really the right word. The volatile access makes sure that the
> > > > > compiler does the test at the point that the source code asked, and doesn't
> > > > > move it before/after other operations.
> > > >
> > > > It's listed in Documentation/atomic_bitops.txt.
> > >
> > > Oh, so my memory was actually correct that I saw it in the docs
> > > somewhere.
> > > WDYT, should I mention this here in the code (block comment) as well
> > > that it's atomic and must not lose `volatile` as Andy suggested or
> > > it's sufficient to have it in the docs (+ it's not underscored)?
> >
> > Perhaps a quick comment in the code (not kerneldoc above) will be
> > sufficient, with reference to Documentation/atomic_bitops.txt.
>
> If it may help, we can do:
>
> /*
> * Bit testing is a naturally atomic operation because bit is
> * a minimal quantum of information.
> */
> #define __test_bit test_bit
That's redundant and we'll end up with a random mix of both.
What'd be more interesting is having a __test_bit without the volatile
that allows compilers to optimize things more. But I think that also
becomes mostly redundant with the optimizations that this series seeks
out to do.
The distinction is ever so subtle, and clever compilers *will* break
concurrent code in ways that are rather hard to imagine:
https://lwn.net/Articles/793253/