[PATCH v2] timekeeping: don't use seqcount loop in ktime_mono_to_any on 64-bit systems

[PATCH v2] timekeeping: don't use seqcount loop in ktime_mono_to_any on 64-bit systems

[Jeff Layton]

ktime_mono_to_any only fetches the offset inside the loop. This is a
single word on 64-bit hosts, and seqcount_read_begin implies a full SMP
barrier.

When BITS_PER_LONG == 64, just do a simple ktime_add instead as there
is no possibility of getting a torn offset value.

Cc: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
Thanks for the review so far, Thomas! Hopefully this looks better.
Disregard my earlier question about making this a static inline too.
That would require making offsets[] a global symbol, which I don't think
we want to do.
---
Changes in v2:
- drop the READ_ONCE
- clean up changelog
- Link to v1: https://lore.kernel.org/r/20240910-mgtime-v1-1-35fb64bd0af5@kernel.org
---
 kernel/time/timekeeping.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 3f524d43d685..a3872f087fbc 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -946,6 +946,14 @@ EXPORT_SYMBOL_GPL(ktime_get_coarse_with_offset_and_floor);
  * @tmono:	time to convert.
  * @offs:	which offset to use
  */
+#if BITS_PER_LONG == 64
+ktime_t ktime_mono_to_any(ktime_t tmono, enum tk_offsets offs)
+{
+	ktime_t *offset = offsets[offs];
+
+	return ktime_add(tmono, *offset);
+}
+#else /* BITS_PER_LONG == 64 */
 ktime_t ktime_mono_to_any(ktime_t tmono, enum tk_offsets offs)
 {
 	ktime_t *offset = offsets[offs];
@@ -959,6 +967,7 @@ ktime_t ktime_mono_to_any(ktime_t tmono, enum tk_offsets offs)
 
 	return tconv;
 }
+#endif /* BITS_PER_LONG == 64 */
 EXPORT_SYMBOL_GPL(ktime_mono_to_any);
 
 /**

---
base-commit: dc8f9bbf6ceb160c7fc5e1a03c5a9318cb61bbd7
change-id: 20240910-mgtime-731eace7cca5

Best regards,
-- 
Jeff Layton <jlayton@kernel.org>

Re: [PATCH v2] timekeeping: don't use seqcount loop in ktime_mono_to_any on 64-bit systems

[Thomas Gleixner]

On Tue, Sep 10 2024 at 08:55, Jeff Layton wrote:
> ktime_mono_to_any only fetches the offset inside the loop. This is a
> single word on 64-bit hosts, and seqcount_read_begin implies a full SMP
> barrier.
>
> When BITS_PER_LONG == 64, just do a simple ktime_add instead as there
> is no possibility of getting a torn offset value.
>
> Cc: Vadim Fedorenko <vadim.fedorenko@linux.dev>
> Signed-off-by: Jeff Layton <jlayton@kernel.org>
> ---
> Thanks for the review so far, Thomas! Hopefully this looks better.
> Disregard my earlier question about making this a static inline too.
> That would require making offsets[] a global symbol, which I don't think
> we want to do.

This still fails to address these:

> Please describe functions with foo() and not foo.

>  Which will make KCSAN complain ...
>
>  So yes, READ_ONCE() is the correct thing todo, but then we want to have
>  the counterpart at the write sides.

KCSAN requires this to be annotated and it's also a good documentation
that this is a intentional unprotected access.

Thanks,

        tglx