eSATA eject - camcontrol devlist still sees disk

J

johnedstone

I am using the the HighPoint RocketRaid 2302 PCI-e card
http://www.highpoint-tech.com/usa/rr2302.htm

I have used their cards successfully for several years.
This is the first eSATA card I've used.

Hotswapping has the following problem. Has anyone seen this problem, and found a solution?

Problem: When the command
Code: 
camcontrol eject da2
is issued, and the eSATA is physically disconnected, and after
Code: 
camcontrol rescan all
is run, the command
Code: 
camcontrol devlist
still shows the disk

Here is an example of an eSATA (da2) being successfully recognized, when it is attached:
Code: 
tail -3 /var/log/messages
Jun 11 11:40:57 potter34 kernel: hptrr: start channel [0,0]
Jun 11 11:40:57 potter34 kernel: hptrr: [0 0] Start channel soft reset.
Jun 11 11:40:57 potter34 kernel: hptrr: channel [0,0] started successfully

camcontrol devlist
<HPT DISK 0_0 4.00>                at scbus0 target 0 lun 0 (pass0,da0)
<HPT DISK 0_1 4.00>                at scbus0 target 1 lun 0 (pass1,da1)
<PIONEER DVD-RW  DVR-218L 1.02>    at scbus1 target 0 lun 0 (pass2,cd0)

camcontrol rescan all
Re-scan of bus 0 was successful
Re-scan of bus 1 was successful
Re-scan of bus 2 was successful
Re-scan of bus 3 was successful
Re-scan of bus 4 was successful
Re-scan of bus 5 was successful

camcontrol devlist
<HPT DISK 0_0 4.00>                at scbus0 target 0 lun 0 (pass0,da0)
<HPT DISK 0_1 4.00>                at scbus0 target 1 lun 0 (pass1,da1)
<HPT DISK 0_2 4.00>                at scbus0 target 2 lun 0 (da2,pass3)
<PIONEER DVD-RW  DVR-218L 1.02>    at scbus1 target 0 lun 0 (pass2,cd0)

Hardware details listed here, note at boot time the eSATA is not turned on
Code: 
uname -a
FreeBSD potter34.johnedstone.net 8.0-RELEASE-p2 FreeBSD 8.0-RELEASE-p2 #0: Tue Mar 30 07:56:28 EDT 2010     root@ejstone34.johnedstone.net:/usr/obj/usr/src/sys/GENERIC  amd64

egrep hptrr /var/run/dmesg.boot 
hptrr0: <sx508x> port 0x2000-0x20ff mem 0xd3200000-0xd32fffff irq 17 at device 0.0 on pci5
hptrr: adapter at PCI 5:0:0, IRQ 17
hptrr: start channel [0,0]
hptrr: start channel [0,1]
hptrr: start channel [0,2]
hptrr: start channel [0,3]
hptrr: [0 2] Start channel soft reset.
hptrr: [0 3] Start channel soft reset.
hptrr: channel [0,2] started successfully
hptrr: channel [0,3] started successfully
hptrr: [0 0] Failed to perform channel hard reset.
hptrr: [0 1] Failed to perform channel hard reset.
hptrr0: [GIANT-LOCKED]
hptrr0: [ITHREAD]
da0 at hptrr0 bus 0 target 0 lun 0
da1 at hptrr0 bus 0 target 1 lun 0
 
Additional information for current environment

Code: 
cat /boot/loader.conf
## snd_driver_load="YES"
snd_hda_load="YES"
atapicam_load="YES"
hw.ata.atapi_dma="1"
geom_mirror_load="YES"
ext2fs_load="YES"
vboxdrv_load="YES"
nvidia_load="YES"

kldstat
Id Refs Address            Size     Name
 1   29 0xffffffff80100000 d188c0   kernel
 2    1 0xffffffff80e19000 20ab0    geom_mirror.ko
 3    1 0xffffffff80e3a000 23ed0    snd_hda.ko
 4    2 0xffffffff80e5e000 75708    sound.ko
 5    1 0xffffffff80ed4000 52d0     atapicam.ko
 6    1 0xffffffff80eda000 15310    ext2fs.ko
 7    3 0xffffffff80ef0000 45e18    vboxdrv.ko
 8    1 0xffffffff80f36000 d440e0   nvidia.ko
 9    2 0xffffffff81e22000 28ce     vboxnetflt.ko
10    2 0xffffffff81e25000 8d44     netgraph.ko
11    1 0xffffffff81e2e000 13be     ng_ether.ko
12    1 0xffffffff81e30000 d2c      vboxnetadp.ko
 
And, for completeness sake, the following command
Code: 
camcontrol reset 0:2:0
or
Code: 
camcontrol reset 0:2:0
camcontrol rescan all
does not fix the problem, that is
Code: 
camcontrol devlist
still 'sees' the disk that is not there.
 
I had thought of that, but this disk is seen as a SCSI disk, not as an ATA disk, so it turns out that
Code: 
atacontrol list
doesn't even see the disk. This is confirmed by the man page.
Thanks though, for the suggestion.
 
johnedstone said:
And, for completeness sake, the following command
Code: 
camcontrol reset 0:2:0
or
Code: 
camcontrol reset 0:2:0
camcontrol rescan all
does not fix the problem
Click to expand...
Can you try:

Code: 
camcontrol reset 0
camcontrol rescan 0

?
 
Code: 
camcontrol reset 0

never returns to the prompt. And, in the process table you see this
Code: 
root        1449  0.0  0.0  6052  1120  v0  D+   10:06AM   0:00.00 camcontrol reset 0
Which means, as you know
Code: 
D       Marks a process in disk (or other short term, uninter‐ruptible) wait

hmmm ... in uninterruptible wait.

I can not simply, from another xterm
Code: 
kill -9 1449

But, I can open another terminal and do
Code: 
halt -p

I should also note that the following sequence where 'camcontrol reset 0:2:0' returns an error
And, in a moment I'm going to place a note about 'halt -p' off my first thread, about the alarm on the card going off, and it not powering off (later this evening)

Look at this sequence and the error
Code: 
# camcontrol eject da2
Unit stopped successfully, Media ejected

# camcontrol reset 0:2:0
Reset of 0:2:0 returned error 0x6

# camcontrol rescan all
Re-scan of bus 0 was successful
Re-scan of bus 1 was successful
Re-scan of bus 2 was successful
Re-scan of bus 3 was successful
Re-scan of bus 4 was successful
Re-scan of bus 5 was successful

# camcontrol reset 0:2:0
Reset of 0:2:0 returned error 0x6

# camcontrol devlist
<HPT DISK 0_0 4.00>                at scbus0 target 0 lun 0 (pass0,da0)
<HPT DISK 0_1 4.00>                at scbus0 target 1 lun 0 (pass1,da1)
<HPT DISK 0_2 4.00>                at scbus0 target 2 lun 0 (da2,pass3)
<PIONEER DVD-RW  DVR-218L 1.02>    at scbus1 target 0 lun 0 (pass2,cd0)

Argg, it's still there.
 
Also hptrr hangs at halt -p

Also, a relevent observation is that upon issuing
Code: 
halt -p
all of the disks get sync'd properly, and then the following message appears on the console

Code: 
hptrr: [0 1] Failed to perform channel hard reset
hptrr: [0,1,0] device is disconnected on channel

and then either the RocketRaid card starts, and continues to beep, or the system board beeps, I'm not sure which one. And then, I have to manually power it off. It reboots w/o error since the disks have been sync'd

This leads me to believe that when the command
Code: 
camcontrol eject da2
somehow the hptrr driver is not 'letting go' of the device.
Or, perhaps the RocketRaid card can not 'let go', and so this never gets back to the hptrr driver.

I will ask the contact in the man page for hptrr to look at this, and maybe HighPoint to take a look at this.
 
Solved - HighPoint supplied 2 packages

This has been fixed.
I ask http://www.highpoint-tech.com to take a look at this thread.

And after some discussion w/ them, they sent two packages,

hptraidconf-3.4-1.tbz
hptsvr-3.13-6.tbz

and a Readme and a pdf.

These packages provided a cli tool (I believe they also have a gui, and a web version) that allows me to 'unplug' the eSATA.

There is a lot of other functionality, e.g. creating raid arrays, with these packages, but I am only using them to 'unplug' the eSATA.

Together with their command line interface, and camcontrol I was able to cleanly unplug the eSATA. It is also useful, of course to watch /var/log/messages at every step of the way.

Here's an example of plugging in and unplugging an eSATA
I've intermixed their command line tool with the OS commands, in the order I ran them.

Watch da2 (WDC WD10EALS-00Z8A0) apppear, and then disappear. I will not post /var/log/messages, but it showed no errors.

The eSATA was first powered on (AC), then plugged into the eSATA connection. Then the following commands were run.

Code: 
HighPoint CLI>query devices
ID      Capacity    MaxFree     Flag    Status    ModelNumber
-------------------------------------------------------------------------------
1/3     1000.20     0           SINGLE  LEGACY    ST31000528AS
1/4     1000.20     0           SINGLE  LEGACY    ST31000528AS
-------------------------------------------------------------------------------
HighPoint CLI>rescan devices
HighPoint CLI>query devices
ID      Capacity    MaxFree     Flag    Status    ModelNumber
-------------------------------------------------------------------------------
1/2     1000.20     0           SINGLE  LEGACY    WDC WD10EALS-00Z8A0
1/3     1000.20     0           SINGLE  LEGACY    ST31000528AS
1/4     1000.20     0           SINGLE  LEGACY    ST31000528AS

# camcontrol devlist
<HPT DISK 0_0 4.00>                at scbus0 target 0 lun 0 (pass0,da0)
<HPT DISK 0_1 4.00>                at scbus0 target 1 lun 0 (pass1,da1)
<PIONEER DVD-RW  DVR-218L 1.02>    at scbus1 target 0 lun 0 (pass2,cd0)

# camcontrol rescan all
Re-scan of bus 0 was successful
Re-scan of bus 1 was successful
Re-scan of bus 2 was successful
Re-scan of bus 3 was successful
Re-scan of bus 4 was successful
Re-scan of bus 5 was successful

# camcontrol devlist
<HPT DISK 0_0 4.00>                at scbus0 target 0 lun 0 (pass0,da0)
<HPT DISK 0_1 4.00>                at scbus0 target 1 lun 0 (pass1,da1)
<HPT DISK 0_2 4.00>                at scbus0 target 2 lun 0 (da2,pass3)
<PIONEER DVD-RW  DVR-218L 1.02>    at scbus1 target 0 lun 0 (pass2,cd0)

# echo starting the unplug sequence
starting the unplug sequence

# camcontrol eject da2
Unit stopped successfully, Media ejected

HighPoint CLI>unplug 1/2
HighPoint CLI>query devices
ID      Capacity    MaxFree     Flag    Status    ModelNumber
-------------------------------------------------------------------------------
1/3     1000.20     0           SINGLE  LEGACY    ST31000528AS
1/4     1000.20     0           SINGLE  LEGACY    ST31000528AS
-------------------------------------------------------------------------------

# camcontrol rescan all
Re-scan of bus 0 was successful
Re-scan of bus 1 was successful
Re-scan of bus 2 was successful
Re-scan of bus 3 was successful
Re-scan of bus 4 was successful
Re-scan of bus 5 was successful

# camcontrol devlist
<HPT DISK 0_0 4.00>                at scbus0 target 0 lun 0 (pass0,da0)
<HPT DISK 0_1 4.00>                at scbus0 target 1 lun 0 (pass1,da1)
<PIONEER DVD-RW  DVR-218L 1.02>    at scbus1 target 0 lun 0 (pass2,cd0)

After this sequence of commands, the eSATA was unplugged from the eSATA connection, and then powered down, in that order.

Thanks everyone for help thinking through this
-- johnedstone
 
You must log in or register to reply here.
Back
Top