On 3/2/20 6:24 PM, Bartlomiej Zolnierkiewicz wrote:
>
> On 2/13/20 10:53 AM, Hannes Reinecke wrote:
>> Add tracepoints for bus-master DMA and taskfile related functions.
>> That allows us to drop the relevant DPRINTK() calls.
>
> ..DPRINTK() & VPRINTK() calls.
>
> Also please see comments below (from v1 review).
>
>> Signed-off-by: Hannes Reinecke <hare@xxxxxxx>
>> ---
>> drivers/ata/libata-core.c | 5 +++
>> drivers/ata/libata-sff.c | 23 +++++++---
>> include/trace/events/libata.h | 97 +++++++++++++++++++++++++++++++++++++++++++
>> 3 files changed, 120 insertions(+), 5 deletions(-)
>>
[ .. ]
>> + TP_STRUCT__entry(
>> + __field( unsigned int, ata_port )
>> + __field( unsigned char, cmd )
>> + __field( unsigned char, dev )
>> + __field( unsigned char, lbal )
>> + __field( unsigned char, lbam )
>> + __field( unsigned char, lbah )
>> + __field( unsigned char, nsect )
>> + __field( unsigned char, feature )
>> + __field( unsigned char, hob_lbal )
>> + __field( unsigned char, hob_lbam )
>> + __field( unsigned char, hob_lbah )
>> + __field( unsigned char, hob_nsect )
>> + __field( unsigned char, hob_feature )
>> + __field( unsigned char, proto )
>> + __field( unsigned long, flags )
>> + ),
>
> 'flags' field doesn't seem to be used?
>
Indeed.
[ .. ]
>> + TP_printk("ata_port=%u cmd=%s%s tag=%d",
>> + __entry->ata_port,
>> + show_protocol_name(__entry->proto),
>> + show_opcode_name(__entry->cmd),
>> + __entry->tag)
>
> Please keep both new events consistent regarding 'proto' and
> 'cmd' fields printing (add "proto=%s" in this event or rework
> printing in ata_tf_load one).
> right, will be doing so in the next round.
Cheers,
Hannes
--
Dr. Hannes Reinecke Kernel Storage Architect
hare@xxxxxxx +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
