When trying to analyze I/O data for SCSI tape drives (with the patch
to enable tracing on sg devices), i found that blkparse does not
include the requests in the stats summary at the end.
The requests i am looking at are passed from scsi_execute_async() (in
drivers/scsi/scsi_lib.c) to the block layer and are marked as
REQ_TYPE_BLOCK_PC. In blkparse, these requests are only logged, but
not included in the overall accounting. I added some calls to log the
data in the draft patch below. But then i am wondering what is the
difference between dump_trace_pc() and dump_trace_fs(). Could these
functions be merged to get the same data for fs and pc requests?
Christof Schmitt
---
blkparse.c | 5 +++++
1 file changed, 5 insertions(+)
--- a/blkparse.c 2007-11-01 17:28:23.000000000 +0100
+++ b/blkparse.c 2007-12-14 09:04:55.000000000 +0100
@@ -1350,9 +1350,11 @@ static void log_pc(struct per_cpu_info *
static void dump_trace_pc(struct blk_io_trace *t, struct per_cpu_info *pci)
{
int act = t->action & 0xffff;
+ int w = (t->action & BLK_TC_ACT(BLK_TC_WRITE)) != 0;
switch (act) {
case __BLK_TA_QUEUE:
+ account_queue(t, pci, w);
log_generic(pci, t, "Q");
break;
case __BLK_TA_GETRQ:
@@ -1362,12 +1364,15 @@ static void dump_trace_pc(struct blk_io_
log_generic(pci, t, "S");
break;
case __BLK_TA_REQUEUE:
+ account_requeue(t, pci, w);
log_generic(pci, t, "R");
break;
case __BLK_TA_ISSUE:
+ account_issue(t, pci, w);
log_pc(pci, t, "D");
break;
case __BLK_TA_COMPLETE:
+ account_c(t, pci, w, t->bytes);
log_pc(pci, t, "C");
break;
case __BLK_TA_INSERT:
-
To unsubscribe from this list: send the line "unsubscribe linux-btrace" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
