Re: [PATCH] net/9p: Convert net/9p protocol dumps to tracepoints

[Steven Rostedt <srostedt@redhat.com>]

On Tue, 2011-08-09 at 18:12 +0530, Aneesh Kumar K.V wrote:

> +#define P9_PROTO_DUMP_SZ (16 * 3 + 2 + 16 + 1)

Wow, you're making your event have 67 byte arrays, two of them.

> +TRACE_EVENT(9p_protocol_dump,
> +	    TP_PROTO(struct p9_client *clnt, struct p9_fcall *pdu),
> +
> +	    TP_ARGS(clnt, pdu),
> +
> +	    TP_STRUCT__entry(
> +		    __field(	__u64,		clnt				)
> +		    __field(	__u8,		type				)
> +		    __field(	__u16,		tag				)
> +		    __array(	unsigned char,	line1,	P9_PROTO_DUMP_SZ	)
> +		    __array(	unsigned char,	line2,	P9_PROTO_DUMP_SZ	)
> +		    ),

Wouldn't it be better to put this as two unsigned char arrays of 16
bytes?

> +
> +	    TP_fast_assign(
> +		    const u8 *ptr   =  pdu->sdata;
> +		    __entry->clnt   =  (__u64)clnt;
> +		    __entry->type   =  pdu->id;
> +		    __entry->tag    =  pdu->tag;
> +		    hex_dump_to_buffer(ptr, 16, 16,
> +				       1, __entry->line1, P9_PROTO_DUMP_SZ, true);
> +		    hex_dump_to_buffer(ptr + 16, 16, 16,
> +				       1, __entry->line2, P9_PROTO_DUMP_SZ, true);
> +		    ),
> +
> +	    TP_printk("clnt %lu %s(tag = %d)\n%.8x: %s\n%.8x: %s\n",
> +		      (long)__entry->clnt, show_9p_op(__entry->type),
> +		      __entry->tag, 0, __entry->line1, 16 , __entry->line2)

Yeah, you would need to make the above ugly to print out the array, but
it's not that hard. And you will be saving 102 bytes per event in the
ring buffer, which is very expensive real-estate. Not to mention the
time it takes to copy all that.

-- Steve