W dniu 27.09.2015 o 04:07, Nicholas A. Bellinger pisze:
Hi Andrzej,
(Adding Sebastian CC')
On Thu, 2015-09-24 at 12:25 +0200, Andrzej Pietrasiewicz wrote:
Hello All,
I am trying to set up a tcm_usb_gadget on a board with DWC3 UDC.
I've downloaded targetcli from
https://github.com/Datera/targetcli
built a deb package and installed it.
I downloaded it instead of installing from distribution package
to be sure I get the latest version.
Prior to that I have built and properly deployed tcm_usb_gadget.ko,
that is copied it to system location /lib/modules/`uname -r` and
calculated module dependencies with depmod -a.
When targetcli deb package was installed, a bunch of messages
appeared, including this:
[....] Fabric module tcm_usb_gadget is not loaded ... (warning).
Any idea why is that?
Mmmm. With the same userspace builds on a Debian based v4.3-rc1 setup,
v4.3-rc3, still not loaded.
Only tcm_usb_gadget is a module, the rest is compiled-in.
Can this be an issue?
tcm_usb_gadget is getting loaded automatically:
root@scsi-mq:/usr/src/target-pending.git# /etc/init.d/target start
[ ok ] The Linux SCSI Target is already stopped.
[info] The Linux SCSI Target looks properly installed.
[info] The configfs filesystem is already mounted.
[info] Loaded core module target_core_mod.
[info] Loaded core module target_core_pscsi.
[info] Loaded core module target_core_iblock.
[info] Loaded core module target_core_file.
[info] Loaded fabric module vhost_scsi.
[info] Loaded fabric module tcm_fc.
[info] Loaded fabric module ib_srpt.
[info] Loaded fabric module tcm_qla2xxx.
[info] Loaded fabric module tcm_loop.
[info] Loaded fabric module tcm_usb_gadget.
[info] Loaded fabric module iscsi_target_mod.
[info] Loaded /etc/target/scsi_target.lio.
[ ok ] Started The Linux SCSI Target.
Module Size Used by
iscsi_target_mod 251714 7
tcm_usb_gadget 21050 4
tcm_loop 15500 1
tcm_qla2xxx 23666 1
ib_srpt 36234 1
tcm_fc 20357 1
vhost_scsi 22522 1
target_core_file 10281 2
target_core_iblock 9322 0
target_core_pscsi 13248 0
target_core_mod 300866 35 target_core_iblock,tcm_usb_gadget,tcm_qla2xxx,target_core_pscsi,iscsi_target_mod,tcm_fc,ib_srpt,target_core_file,vhost_scsi,tcm_loop
Then I manually load tcm_usb_gadget: modprobe tcm_usb_gadget.
After that when targetcli is run usb_gadget fabric directory
appears.
Then I create the following hierarchy:
/> ls
o- / ..................................................................... [...]
o- backstores .......................................................... [...]
| o- fileio ............................................... [1 Storage Object]
| | o- abc ...................................... [10.0M, /root/f.img, in use]
| o- iblock ............................................... [0 Storage Object]
| o- pscsi ................................................ [0 Storage Object]
| o- rd_mcp ............................................... [0 Storage Object]
o- iscsi ......................................................... [0 Targets]
o- loopback ...................................................... [0 Targets]
o- usb_gadget ..................................................... [1 Target]
o- naa.60014051babd31e4 ....................... [nexus WWN 600140515a36d7db]
After patches from Nicholas:
o- naa.60014055d35bc331 ........................ [nexus WWN naa.600140534dac0d37]
o- luns .......................................................... [1 LUN]
o- lun0 ..................................... [fileio/abc (/root/f.img)]
and exit targetcli:
/> exit
Comparing startup and running configs...
Invalid fabric identifier 'usb_gadget': expected type 'loopback', 'iscsi', 'ib_srpt', 'vhost', 'tcm_fc', 'qla2xxx'
Why is 'usb_gadget' fabric unknown?
This problem is now gone...
For rtslib/targetcli v3.0 code, usb-gadget is missing the following
policy file:
root@scsi-mq:/# cat /var/target/policy/fabric_usb_gadget.lio
fabric usb_gadget {
target %naa {
nexus_wwn %naa
lun %int backend %backend
}
}
... with this added.
Also, it looks like usb_gadget is stripping off the 'naa.' from the
nexus input, that is making the rtslib v3.0 sanity checks for nexus WWPN
formatting unhappy.
Here's a quick patch to fix that:
diff --git a/drivers/usb/gadget/legacy/tcm_usb_gadget.c b/drivers/usb/gadget/legacy/tcm_usb_g
index c3c4808..b3631aa 100644
--- a/drivers/usb/gadget/legacy/tcm_usb_gadget.c
+++ b/drivers/usb/gadget/legacy/tcm_usb_gadget.c
@@ -1665,7 +1669,7 @@ static ssize_t tcm_usbg_tpg_store_nexus(
if (i_port[strlen(i_port) - 1] == '\n')
i_port[strlen(i_port) - 1] = '\0';
- ret = tcm_usbg_make_nexus(tpg, &i_port[4]);
+ ret = tcm_usbg_make_nexus(tpg, &i_port[0]);
if (ret < 0)
return ret;
return count;
The gadget goes alive, I can list it on my Linux host, but I
repeatedly get:
[ 177.628876] g_target gadget: high-speed config #1: Linux Target
[ 177.633411] Using the BOT protocol
[ 198.459512] ERR bot_status_complete(64)
[ 198.621077] g_target gadget: high-speed config #1: Linux Target
[ 198.625607] Using the BOT protocol
[ 219.451560] ERR bot_status_complete(64)
The same problem remains. The device keeps resetting.
USB analyzer shows that the device sends many (hundreds of thousands)
of IN transaction NAKs.
...
Any ideas, why?
With the above patch in place, v4.3-rc1 code appears to be working
now using dummy_hcd.
/> cd /usb_gadget
/usb_gadget> create
Created target naa.60014051d601ad20.
/usb_gadget> cd naa.60014051d601ad20/
/usb_gadget/n...14051d601ad20> ./luns create lun=0 storage_object=/backstores/fileio/tmpfile
Created LUN 0.
/usb_gadget/n...14051d601ad20> disable
The TPG has been disabled.
/usb_gadget/n...14051d601ad20> enable
The TPG has been enabled.
/usb_gadget/n...14051d601ad20> ls /usb_gadget/
o- usb_gadget .................................................................... [1 Target]
o- naa.60014051d601ad20 .................................. [nexus WWN naa.6001405c8f91fa92]
o- luns ......................................................................... [1 LUN]
o- lun0 ............................................... [fileio/tmpfile (/root/fileio)]
/usb_gadget/n...14051d601ad20>
[ 63.698312] dummy_hcd dummy_hcd.0: USB Host+Gadget Emulator, driver 02 May 2005
[ 63.709515] dummy_hcd dummy_hcd.0: Dummy host controller
[ 63.722241] dummy_hcd dummy_hcd.0: new USB bus registered, assigned bus number 1
[ 63.758657] hub 1-0:1.0: USB hub found
[ 63.763240] hub 1-0:1.0: 1 port detected
[ 72.839346] g_target gadget: g_target ready
[ 72.846854] udc_bind_to_driver: 0
[ 73.060250] usb 1-1: new high-speed USB device number 2 using dummy_hcd
[ 73.244219] g_target gadget: high-speed config #1: Linux Target
[ 73.247685] Using the BOT protocol
[ 73.317603] usb-storage 1-1:1.0: USB Mass Storage device detected
[ 73.325750] usb-storage 1-1:1.0: Quirks match for vid 0525 pid a4a5: 10000
[ 73.331335] scsi host0: usb-storage 1-1:1.0
[ 73.351713] usbcore: registered new interface driver usb-storage
[ 74.352193] No LUNs configured?
[ 84.618554] usb 1-1: USB disconnect, device number 2
[ 85.597708] g_target gadget: g_target ready
[ 85.600641] udc_bind_to_driver: 0
[ 85.816185] usb 1-1: new high-speed USB device number 3 using dummy_hcd
[ 85.996223] g_target gadget: high-speed config #1: Linux Target
[ 85.999791] Using the BOT protocol
[ 86.013201] usb-storage 1-1:1.0: USB Mass Storage device detected
[ 86.018596] usb-storage 1-1:1.0: Quirks match for vid 0525 pid a4a5: 10000
[ 86.022647] scsi host1: usb-storage 1-1:1.0
[ 87.052737] scsi 1:0:0:0: Direct-Access LIO-ORG FILEIO 4.0 PQ: 0 ANSI: 5
[ 87.088839] sd 1:0:0:0: [sda] 4194304 512-byte logical blocks: (2.14 GB/2.00 GiB)
[ 87.108419] sd 1:0:0:0: [sda] Write Protect is off
[ 87.113042] sd 1:0:0:0: [sda] Mode Sense: 43 00 00 08
[ 87.120664] sd 1:0:0:0: Attached scsi generic sg0 type 0
[ 87.124484] sd 1:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
[ 87.264468] sd 1:0:0:0: [sda] Attached SCSI disk
[ 87.288299] target: Unknown data direction for SCSI Opcode 0xa1
Note there still appears to be a nexus memory leak on shutdown and/or
usb_composite_probe() failure, but the above should at least get you up
and running.
Will send out a separate patch to address the reference leak.
Thanks for reporting!
--nab
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
