[stgt] Why "Microsoft iscsi" cannot work well with stgt (but it worked well with ietd)?

[FUJITA Tomonori]

On Tue, 17 Feb 2009 14:27:00 +0800
Cheng Renquan <crquan at gmail.com> wrote:

> On Fri, Feb 13, 2009 at 12:29 PM, FUJITA Tomonori
> <fujita.tomonori at lab.ntt.co.jp> wrote:
> > Yes, I use tgt with Windows XP (initiator 2.08).
> >
> > When Windows finds an unknown device and ask you where you see for the
> > driver. Then you need to tell Windows that you want to do nothing
> > about the driver.
> 
> >
> >> I have tested within WindowsXP and Windows2003, after iSCSI Logon, the
> >> "IET Controller" (LUN 0) will be displayed in "Other unknown devices"
> >> in "Device Manager"; while the "IET Virtual Disk" (LUN 1) displayed as
> >> known device in "Disk Drives" category, which is the same as logon
> >> from IET;
> >> but at this time the "IET Virtual Disk" volume didn't appear in the
> >> "Disk Management", thus not usable,
> >
> > My Windows shows "IET Virtual Disk" volume in the "Disk Management":
> >
> > http://www.kernel.org/pub/linux/kernel/people/tomo/tgt/xp.jpg
> >
> > It's Japanese edition but I guess that you see how it can work for me.
> 
> Yes, your picture showed it did work,
> 
> Then I've tested with more Windows machines (XP & 2003), some worked
> while some not, it seems a random problem, or I have not perceived
> what problems happended to those windows,

There should be a reason, I think.


> I've also tested with SCST, with which all LUNs can be defined
> specificly, TYPE_RAID (0xc) can be used as the first LUN or not, it
> has scst_disk, scst_vdisk, scst_raid, all can be used alone, combined
> freely.
> I think that design maybe better?

Using a special LUN 0 for REPORT_LUNS support is correct from the
perspective of SCSI. I prefer not to complicate the code for wrong
things.

Something like the following trick to use a fake disk instead of raid
type (as some commercial boxes use) might work but I'm not sure.


diff --git a/usr/bs_null.c b/usr/bs_null.c
index 00137ff..8571e78 100644
--- a/usr/bs_null.c
+++ b/usr/bs_null.c
@@ -28,7 +28,7 @@
 #include "tgtd.h"
 #include "scsi.h"
 
-#define NULL_BS_DEV_SIZE        (1ULL << 40)
+#define NULL_BS_DEV_SIZE        0
 
 int bs_null_cmd_submit(struct scsi_cmd *cmd)
 {
diff --git a/usr/target.c b/usr/target.c
index dc30c87..54c4958 100644
--- a/usr/target.c
+++ b/usr/target.c
@@ -1644,6 +1644,7 @@ int tgt_target_create(int lld, int tid, char *args)
 	struct target *target, *pos;
 	char *p, *q, *targetname = NULL;
 	struct backingstore_template *bst;
+	char buf[128];
 
 	p = args;
 	while ((q = strsep(&p, ","))) {
@@ -1714,7 +1715,8 @@ int tgt_target_create(int lld, int tid, char *args)
 	INIT_LIST_HEAD(&target->acl_list);
 	INIT_LIST_HEAD(&target->it_nexus_list);
 
-	tgt_device_create(tid, TYPE_RAID, 0, NULL, 0);
+	strncpy(buf, "path=/dev/null,bstype=null", sizeof(buf));
+	tgt_device_create(tid, TYPE_DISK, 0, buf, 1);
 
 	if (tgt_drivers[lld]->target_create)
 		tgt_drivers[lld]->target_create(target);
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html