public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed
* ioblock.stp major/minor number patch
@ 2006-10-23 15:37 Mike Mason
  2006-10-23 15:56 ` Frank Ch. Eigler
  0 siblings, 1 reply; 2+ messages in thread
From: Mike Mason @ 2006-10-23 15:37 UTC (permalink / raw)
  To: SystemTAP

The following patch adds major/minor number functions to the ioblock tapset.  I found them useful in a block IO script I'm writing.  It also fixes a few misleading comments.

- Mike Mason


--- src/tapset/ioblock.stp	2006-08-15 18:54:46.000000000 -0700
+++ src.save2/tapset/ioblock.stp	2006-10-23 08:30:08.000000000 -0700
@@ -10,9 +10,10 @@
 %{
 #include <linux/bio.h>
 #include <linux/genhd.h>
+#include <linux/kdev_t.h>
 %}
 
-/* get i-node number of mapped file */
+/* get i-node number of device file */
 function __bio_ino:long(bio:long)
 %{
         struct bio  *bio;
@@ -113,6 +114,52 @@
         }
 %}
 
+/* returns block device major number */
+function bio_major:long(bio:long)
+%{
+	struct bio *bio;
+	struct block_device *bi_bdev;
+
+	bio = (struct bio *)(long)THIS->bio;
+	bi_bdev = (struct block_device *)deref(sizeof(bio->bi_bdev),
+			&(bio->bi_bdev));
+	if (bi_bdev == NULL) {
+		THIS->__retvalue = -1;
+		goto end;
+	}
+
+	THIS->__retvalue = MAJOR(bi_bdev->bd_dev);
+
+	if (0) {
+deref_fault:
+		CONTEXT->last_error = "pointer dereference fault";
+	}
+end: ;
+%}
+
+/* returns block device minor number */
+function bio_minor:long(bio:long)
+%{
+	struct bio *bio;
+	struct block_device *bi_bdev;
+
+	bio = (struct bio *)(long)THIS->bio;
+	bi_bdev = (struct block_device *)deref(sizeof(bio->bi_bdev),
+			&(bio->bi_bdev));
+	if (bi_bdev == NULL) {
+		THIS->__retvalue = -1;
+		goto end;
+	}
+
+	THIS->__retvalue = MINOR(bi_bdev->bd_dev);
+
+	if (0) {
+deref_fault:
+		CONTEXT->last_error = "pointer dereference fault";
+	}
+end: ;
+%}
+
 global BIO_READ, BIO_WRITE
 probe begin
 {
@@ -129,7 +176,7 @@
  *
  * Variables:
  *  devname   - block device name
- *  ino       - i-node number of the mapped file 
+ *  ino       - i-node number of the device file 
  *  sector    - beginning sector for the entire bio 
  *  flags     - see below 
  *  	BIO_UPTODATE    0       ok after I/O completion
@@ -184,8 +231,8 @@
  *
  * Variables:
  *  devname   - block device name
- *  ino       - i-node number of the mapped file
- *  byte_done - number of bytes transferred 
+ *  ino       - i-node number of the device file
+ *  bytes_done - number of bytes transferred 
  *  sector    - beginning sector for the entire bio
  *  flags     - see below
  *      BIO_UPTODATE    0       ok after I/O completion

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: ioblock.stp major/minor number patch
  2006-10-23 15:37 ioblock.stp major/minor number patch Mike Mason
@ 2006-10-23 15:56 ` Frank Ch. Eigler
  0 siblings, 0 replies; 2+ messages in thread
From: Frank Ch. Eigler @ 2006-10-23 15:56 UTC (permalink / raw)
  To: Mike Mason; +Cc: SystemTAP


Mike Mason <mmlnx@us.ibm.com> writes:

> The following patch adds major/minor number functions to the ioblock
> tapset.  I found them useful in a block IO script I'm writing.

Good stuff, it's especially nice to see the deref() protection.
To complete the work, the new functions will need documentation
in one of the stapprobes* man pages, and test case or two.
(Something triggering those deref_faults too please.)

> It also fixes a few misleading comments.

It sounds like you should get yourself CVS access so that you can
check in small & obvious things like that.  (See the web site for
details.)

- FChE

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2006-10-23 15:56 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2006-10-23 15:37 ioblock.stp major/minor number patch Mike Mason
2006-10-23 15:56 ` Frank Ch. Eigler

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).