Hi Andrey
Yes, you are right, & I got this one wrong.
Before my previous post, I had only just
a quick glance at the code & I thought I
remembered what this function did.
But my memory was wrong.
Maybe I was getting confused with the target code
for a different OS!
I should have checked, before posting.
Apologies for the confusion.

I've been trying out one of Brendan Gregg's DTrace scripts
to look at the code flow.
And yes "iscsi_full_feature" is called many times!
http://www.brendangregg.com/DTrace/dapptrace
(The multiple threads do confuse the flow in this script.)
Here is a example:

[EMAIL PROTECTED]:~# ps -ef | grep iscsi
    root  7417     1   0 16:29:21 ?           1:18 /usr/sbin/iscsitgtd
[EMAIL PROTECTED]:~# ./dapptrace  -F -p 659
PID/LWP   CALL(args)             = return
7417/22:  -> iscsi_full_feature(0x80CF8C8, 0x0, 0xD1613A00)
7417/22:    -> iscsi_cmd_remove(0x80CF8C8, 0x8, 0x3)
7417/22:      -> sna_lt(0x7, 0x8, 0x3)
7417/22:      <- sna_lt = 67
7417/22:    <- iscsi_cmd_remove = 294
7417/22:    -> handle_scsi_cmd(0x80CF8C8, 0xD1014EC0, 0x0)
7417/22:      -> iscsi_cmd_alloc(0x80CF8C8, 0x1, 0x0)
7417/22:      <- iscsi_cmd_alloc = 264
7417/22:      -> spc_decode_lu_addr(0xD1014EC8, 0x8, 0x80DAE40)
7417/22:      <- spc_decode_lu_addr = 196
7417/22:      -> queue_message_set(0x80C5AF0, 0x0, 0x0)
7417/22:      <- queue_message_set = 181
7417/22:    <- handle_scsi_cmd = 686
7417/22:  <- iscsi_full_feature = 871

Andrey, are you going to raise a bug for that memory leak
you spotted in the code for the AHS related error handling?
I think you should.
Best Regards
Nigel Smith
 
 
This message posted from opensolaris.org
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to