[RFC PATCH 4/6] relay: add relay_reserve_cpu()

[RFC PATCH 4/6] relay: add relay_reserve_cpu()

[Tom Zanussi]

This patch adds the ability to explicitly specify the per-cpu buffer
to reserve space in.  Needed for early DTI tracing.

Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
Signed-off-by: David Wilder <dwilder@us.ibm.com>
---
 relay.h |   33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/include/linux/relay.h b/include/linux/relay.h
index 6caedef..37a7306 100644
--- a/include/linux/relay.h
+++ b/include/linux/relay.h
@@ -269,6 +269,39 @@ static inline void *relay_reserve(struct rchan *chan, size_t length)
 }
 
 /**
+ *	relay_reserve_cpu - reserve slot in given cpu's channel buffer
+ *	@chan: relay channel
+ *	@length: number of bytes to reserve
+ *	@cpu: cpu to log to
+ *
+ *	Returns pointer to reserved slot, NULL if full.
+ *
+ *	Reserves a slot in the given cpu's channel buffer.
+ *	Does not protect the buffer at all - caller must provide
+ *	appropriate synchronization.
+ *
+ *	NOTE: this is almost certainly not the function you want -
+ *	use relay_reserve() instead for normal logging.  This version
+ *	is specialized for things like early tracing.
+ */
+static inline void *relay_reserve_cpu(struct rchan *chan, size_t length,
+				      unsigned int cpu)
+{
+	void *reserved;
+	struct rchan_buf *buf = chan->buf[cpu];
+
+	if (unlikely(buf->offset + length > buf->chan->subbuf_size)) {
+		length = relay_switch_subbuf(buf, length);
+		if (!length)
+			return NULL;
+	}
+	reserved = buf->data + buf->offset;
+	buf->offset += length;
+
+	return reserved;
+}
+
+/**
  *	subbuf_start_reserve - reserve bytes at the start of a sub-buffer
  *	@buf: relay channel buffer
  *	@length: number of bytes to reserve

Re: [RFC PATCH 4/6] relay: add relay_reserve_cpu()

[Mathieu Desnoyers]

* Tom Zanussi (zanussi@us.ibm.com) wrote:
> This patch adds the ability to explicitly specify the per-cpu buffer
> to reserve space in.  Needed for early DTI tracing.
> 

This should come with the ability to tell relay where the already
allocated static buffers are. Can you tell me a little more on why the
standard relay_reserve would not do the job? Is smp_processor_id() not
ready to output a sensible CPU id yet ?

> Signed-off-by: Tom Zanussi <zanussi@us.ibm.com>
> Signed-off-by: David Wilder <dwilder@us.ibm.com>
> ---
>  relay.h |   33 +++++++++++++++++++++++++++++++++
>  1 file changed, 33 insertions(+)
> 
> diff --git a/include/linux/relay.h b/include/linux/relay.h
> index 6caedef..37a7306 100644
> --- a/include/linux/relay.h
> +++ b/include/linux/relay.h
> @@ -269,6 +269,39 @@ static inline void *relay_reserve(struct rchan *chan, size_t length)
>  }
>  
>  /**
> + *	relay_reserve_cpu - reserve slot in given cpu's channel buffer
> + *	@chan: relay channel
> + *	@length: number of bytes to reserve
> + *	@cpu: cpu to log to
> + *
> + *	Returns pointer to reserved slot, NULL if full.
> + *
> + *	Reserves a slot in the given cpu's channel buffer.
> + *	Does not protect the buffer at all - caller must provide
> + *	appropriate synchronization.
> + *
> + *	NOTE: this is almost certainly not the function you want -
> + *	use relay_reserve() instead for normal logging.  This version
> + *	is specialized for things like early tracing.
> + */
> +static inline void *relay_reserve_cpu(struct rchan *chan, size_t length,
> +				      unsigned int cpu)
> +{
> +	void *reserved;
> +	struct rchan_buf *buf = chan->buf[cpu];
> +
> +	if (unlikely(buf->offset + length > buf->chan->subbuf_size)) {
> +		length = relay_switch_subbuf(buf, length);
> +		if (!length)
> +			return NULL;
> +	}
> +	reserved = buf->data + buf->offset;
> +	buf->offset += length;
> +
> +	return reserved;
> +}
> +
> +/**
>   *	subbuf_start_reserve - reserve bytes at the start of a sub-buffer
>   *	@buf: relay channel buffer
>   *	@length: number of bytes to reserve
> 
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

-- 
Mathieu Desnoyers
Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

Re: [RFC PATCH 4/6] relay: add relay_reserve_cpu()

[Tom Zanussi]

On Tue, 2007-07-03 at 23:17 -0400, Mathieu Desnoyers wrote:
> * Tom Zanussi (zanussi@us.ibm.com) wrote:
> > This patch adds the ability to explicitly specify the per-cpu buffer
> > to reserve space in.  Needed for early DTI tracing.
> > 
> 
> This should come with the ability to tell relay where the already
> allocated static buffers are. Can you tell me a little more on why the
> standard relay_reserve would not do the job? Is smp_processor_id() not
> ready to output a sensible CPU id yet ?
> 

In the case of early tracing, the DTI code uses a global buffer and
saves the cpuid along with each event.  Later, when the relay channel is
available, the relogging code wants to put each event in the appropriate
per-cpu relay buffer.  The standard relay_reserve() always uses the
current cpuid, rather than the cpuid attached to the event, which is not
what we want.  relay_reserve_cpu() simply allows a cpuid other than the
current one to be used.

Tom