lkml.org 
[lkml]   [2012]   [Dec]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH 23/26] Generic dynamic per cpu refcounting
From
On Mon, Dec 3, 2012 at 9:58 PM, Kent Overstreet <koverstreet@google.com> wrote:
> +#define pr_fmt(fmt) "%s: " fmt "\n", __func__

Dynamic debug already allows to insert the function name. Please
consider leaving this line out entirely and move the "\n" to the end
of individual printed lines.

> +#define PCPU_REF_PTR 0
> +#define PCPU_REF_NONE 1
> +#define PCPU_REF_DYING 2
> +#define PCPU_REF_DEAD 3

Why has "define" been used here instead of an enum ? Using "define"
prevents the compiler to verify whether all possible values have been
handled in a switch statement.

> +#define REF_STATUS(count) ((unsigned long) count & PCPU_STATUS_MASK)

Please leave out the cast in the above expression. It prevents that
the compiler performs type checking on "count". Also, why is
REF_STATUS() a define and not an inline function ?

> +struct percpu_ref {
> + atomic64_t count;
> + unsigned __percpu *pcpu_count;
> +};

pcpu_count can either be a jiffies value (unsigned long) or a pointer
to unsigned __percpu. So please consider using a union here. Has this
code been checked with sparse ? I'm afraid that the above structure
definition will confuse sparse a lot.

> + if (now - last <= HZ << PCPU_STATUS_BITS) {

Please consider changing "HZ" into a symbolic name that makes it clear
that this value is the threshold in jiffies above which the switch is
made from a single atomic_t refcount to the per-cpu refcounting
scheme.

> +void __percpu_ref_get(struct percpu_ref *ref, bool alloc)
> +{
> + unsigned __percpu *pcpu_count;
> + uint64_t v;
> +
> + pcpu_count = rcu_dereference(ref->pcpu_count);
> +
> + if (REF_STATUS(pcpu_count) == PCPU_REF_PTR) {
> + __this_cpu_inc(*pcpu_count);
> + } else {
> + v = atomic64_add_return(1 + (1ULL << PCPU_COUNT_BITS),
> + &ref->count);
> +
> + if (!(v >> PCPU_COUNT_BITS) &&
> + REF_STATUS(pcpu_count) == PCPU_REF_NONE && alloc)
> + percpu_ref_alloc(ref, pcpu_count);
> + }
> +}

What will happen if another thread invokes percpu_ref_kill() after the
above rcu_dereference() call finished but before REF_STATUS() got
invoked ? Can that percpu_ref_kill() finish before
__this_cpu_inc(*pcpu_count) gets invoked ?
Documentation/RCU/checklist.txt says: "All RCU list-traversal
primitives, which include rcu_dereference(),
list_for_each_entry_rcu(), and list_for_each_safe_rcu(), must be
either within an RCU read-side critical section or must be protected
by appropriate update-side locks."

> +int percpu_ref_kill(struct percpu_ref *ref)

The meaning of the return value of this function has not been documented.

Thanks,

Fubo.


\
 
 \ /
  Last update: 2012-12-27 15:41    [W:0.176 / U:0.308 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site