Skip to content

Documentation changes for 3.14

Paul E. McKenney lkml 2 versions – In review

Get series with b4
b4 am [email protected]

Versions newest first · linked by author and title

  1. v2 Latest4 patches 1 replyIn review
  2. v114 patches 2 repliesSuperseded

Patches in v2 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/4rcu: Add comment on evaluate-once properties of rcu_assign_pointer().include/linux/rcupdate.h+8 -00 replies
2/4bonding: Use RCU_INIT_POINTER() for better overhead and for sparsedrivers/net/bonding/bond_main.c+1 -10 replies
3/4rcu: Make rcu_assign_pointer's assignment volatile and type-safeinclude/linux/rcupdate.h+1 -10 replies
4/4rcu: Add an RCU_INITIALIZER for global RCU-protected pointersinclude/linux/rcupdate.h+42 -380 replies

Reviews trailers per patch

None of the 4 patches has a Reviewed-by or Acked-by of their own; the series has one for all of them.

PatchJosh Triplett
SeriesR-b
1/4
2/4
3/4
4/4

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v2 thread

Cover letter [PATCH v2 tip/core/rcu 0/4] Documentation changes for 3.14

Hello!

This series once again attempts to improve rcu_assign_pointer()'s relationship with sparse.

1. Add a comment indicating that despite appearances,
rcu_assign_pointer() really only evaluates its arguments once, as a cpp macro should.

2. Replace rcu_assign_pointer() of NULL with RCU_INIT_POINTER() to
silence a sparse warning.

3. Apply ACCESS_ONCE() to rcu_assign_pointer()'s target to prevent
comiler mischief. Also require that the source pointer be from the kernel address space. Sometimes it can be from the RCU address space, which necessitates the remaining patches in this series. Which, it must be admitted, apply to a very small fraction of the rcu_assign_pointer() invocations in the kernel. This commit courtesy of Josh Triplett.

4. Add an RCU_INITIALIZER() for compile-time initialization of
global RCU-protected pointers.

Changes from v1 (candidate for v3.14):

o Remove the requirement that values passed to rcu_assign_pointer()
needed to be in sparse's __kernel address space. This found
no bugs and produced a number of ugly false positives. (Hey,
it seemed like a good idea at the time...)

o Remove lots of rcu_assign_pointer()-to-ACCESS_ONCE() changes
that are no longer needed given the above change.

Changes from v3 (candidate for 3.13):

o Remove the replacements of rcu_assign_pointer() with ACCESS_ONCE()
where new data really was being exposed to readers.

Changes from v2:

o Switch from rcu_assign_pointer() to ACCESS_ONCE() given that
the pointers are all --rcu and already visible to readers, as suggested by Eric Dumazet and Josh Triplett.

o Place the commit adding the rcu_assign_pointer()'s ACCESS_ONCE()
at the end to allow better bisectability, as suggested by Josh Triplett.

o Add a comment to rcu_assign_pointer() noting that it only evaluates
its arguments once, as suggested by Josh Triplett.

Changes from v1:

o Fix grammar nit in commit logs.

Thanx, Paul

------------------------------------------------------------------------

 b/drivers/net/bonding/bond_main.c |    2 
 b/include/linux/rcupdate.h        |   90 +++++++++++++++++++++-----------------
 2 files changed, 52 insertions(+), 40 deletions(-)