Is HDD Enclosure using Marvell 88SM9715 Supported?

[kamcm]

Hello,

I just purchased a 5-bay eSata enclosure(TowerRAID TR5M6GNC) made by Sansdigital.
It supported 6G SATA Speed and it's using Marvell 88SM9715 chipset.
I don't know if it is supported by hdsentinel ? I cannot found any info in the compatibility list.
Anyway, I will report back when I got the enclosure.

Thanks,

Kam

[kamcm]

I just confirmed HDSentinel 5.01 works perfectly with my TowerRAID TR5M6GNC.
It showed up everything correctly. You did a good job on it.

Btw, I had another TowerRAID TR5UT which using JMB394 chipset unable to showed up everything correctly with HDSentinel.
I had sent out a test report yesterday. Hopefully it will be fixed in the coming version.

Thanks!

[hdsentinel]

Thanks for the information and confirmation !

Generally yes, there should be absolutely no problems with Marvell 88SM9715 or other "true" SATA port multiplier chipsets: complete status of both drives should be detected and displayed in Hard Disk Sentinel. This is true for other SATA port multiplier chipsets from other manufacturers too.

Anyway, if you prefer, I'd be happy if you can use Report menu -> Send test report to developer option in Hard Disk Sentinel, as then I can examine the details, would be nice to check and verify - just to be 100% sure. It may also gives ideas, thoughts about possible further optimizations, modifications.


Thanks for the report about the JMB39x chipset. I can confirm that of course it is completely supported by Hard Disk Sentinel.
I examined the report and as I see, complete status information of all hard disks detected and displayed, exactly as should. JMB39x chipset is supported in Hard Disk Sentinel for very long time.

Yes, I see that the logical drives seem do not match with the different drives in this configuration.
Yes, I agree that this seems not too elegant - but it does not mean that the chip is not supported, but exactly the opposite: Hard Disk Sentinel correctly identifies the JMB39x chip and the drives managed by it.

The problem is that the chip is technically a RAID controller chip (even if used as port multiplier, without RAID function).
It means that it shows all drives like if they'd form a single RAID array.
As you can see, Hard Disk Sentinel can enumerate all physical hard disk drives, their complete status properly - just shows all physical disks like if they'd form an array (so same logical drive displayed for them).

If possible, please send different report with other connection - as these always help to investigate, examine the situation and gives further information about how to "correct" the drive relation provided by the chip and associate the proper logical drives to each physical hard disks.

Thanks for increasing attention, yes, by the reports, I hope things can be improved.

[kamcm]

Thanks for the information and confirmation !

Generally yes, there should be absolutely no problems with Marvell 88SM9715 or other "true" SATA port multiplier chipsets: complete status of both drives should be detected and displayed in Hard Disk Sentinel. This is true for other SATA port multiplier chipsets from other manufacturers too.

Anyway, if you prefer, I'd be happy if you can use Report menu -> Send test report to developer option in Hard Disk Sentinel, as then I can examine the details, would be nice to check and verify - just to be 100% sure. It may also gives ideas, thoughts about possible further optimizations, modifications.


Thanks for the report about the JMB39x chipset. I can confirm that of course it is completely supported by Hard Disk Sentinel.
I examined the report and as I see, complete status information of all hard disks detected and displayed, exactly as should. JMB39x chipset is supported in Hard Disk Sentinel for very long time.

Yes, I see that the logical drives seem do not match with the different drives in this configuration.
Yes, I agree that this seems not too elegant - but it does not mean that the chip is not supported, but exactly the opposite: Hard Disk Sentinel correctly identifies the JMB39x chip and the drives managed by it.

The problem is that the chip is technically a RAID controller chip (even if used as port multiplier, without RAID function).
It means that it shows all drives like if they'd form a single RAID array.
As you can see, Hard Disk Sentinel can enumerate all physical hard disk drives, their complete status properly - just shows all physical disks like if they'd form an array (so same logical drive displayed for them).

If possible, please send different report with other connection - as these always help to investigate, examine the situation and gives further information about how to "correct" the drive relation provided by the chip and associate the proper logical drives to each physical hard disks.

Thanks for increasing attention, yes, by the reports, I hope things can be improved.

I tested it again with my main desktop. The result is a little bit interesting.

Desktop: Gigabyte Z77X-UD3H motherboard esata (Marvell 88SE9172) -> connect to JMB39x enclosure box via eSata = SMART INFO missing
Laptop: HP elitebook laptop with a sil3132 eSata express card (Support PM) -> connect to JMB39x enclosure box via eSata = showed up SMART INFO

Desktop: Gigabyte Z77X-UD3H motherboard esata (Marvell 88SE9172) -> connect to JMB39x enclosure box via USB3.0 = showed up SMART INFO
Laptop: HP elitebook laptop with a USB3.0 express card -> connect to JMB39x enclosure box via USB3.0 = showed up SMART INFO

Btw, I am sure my desktop's eSata port working with all of the other enclosure without any problem, including the 5-bay enclosure box I received today (Marvell 88SM9715).
Desktop: Gigabyte Z77X-UD3H motherboard esata (Marvell 88SE9172) -> connect to Marvell 88SM9715 enclosure box via eSata = showed up SMART INFO

So is that means (Marvell 88SE9172) having compatibility issue with JMB39x enclosure box
I had sent out the test report right now.

Please let me know,

Thanks!

Kam

[hdsentinel]

So is that means (Marvell 88SE9172) having compatibility issue with JMB39x enclosure box
I had sent out the test report right now.

At least with the current driver of the Marvell 88SE9172 yes, there seems a compatibility issue: the JMicron (RAID) specific detection seems blocked and the status information could not be detected.

Other "true" port multiplier chips (where there is no RAID function possible) may work differently in this area, but as this JMicron chip has RAID function (even if not used) this requires special attention. I understand that this weird - but technically this chip works differently than non-RAID port multipliers.

As displayed on the Information page:

> Disk Controller . . . . . . . . . . . . . . . . : Marvell 91xx SATA 6G Controller (AHCI) [VEN: 1B4B, DEV: 9172] Version: 1.2.0.1020, 1-19-2012

Would be nice to examine with a different driver for the Marvell 91xx SATA 6G controller.
Personally I'd attempt to switch the current driver to Microsoft "standard AHCI controller" driver.
In some cases (for example with Marvell controllers) this may work better.

So I'd recommend to try
- open Windows Device Manager
- locate the Marvell 91xx controller (usually it is in the Disk controllers or RAID controllers section) and right click on that
- manually change the driver to a Windows "Standard AHCI Controller" driver or similar
("Standard AHCI 1.0 Compatible controller" or so).

After the update and restart, things may be better with the JMB39x chip too.

[kamcm]

So is that means (Marvell 88SE9172) having compatibility issue with JMB39x enclosure box
I had sent out the test report right now.

At least with the current driver of the Marvell 88SE9172 yes, there seems a compatibility issue: the JMicron (RAID) specific detection seems blocked and the status information could not be detected.

Other "true" port multiplier chips (where there is no RAID function possible) may work differently in this area, but as this JMicron chip has RAID function (even if not used) this requires special attention. I understand that this weird - but technically this chip works differently than non-RAID port multipliers.

As displayed on the Information page:

> Disk Controller . . . . . . . . . . . . . . . . : Marvell 91xx SATA 6G Controller (AHCI) [VEN: 1B4B, DEV: 9172] Version: 1.2.0.1020, 1-19-2012

Would be nice to examine with a different driver for the Marvell 91xx SATA 6G controller.
Personally I'd attempt to switch the current driver to Microsoft "standard AHCI controller" driver.
In some cases (for example with Marvell controllers) this may work better.

So I'd recommend to try
- open Windows Device Manager
- locate the Marvell 91xx controller (usually it is in the Disk controllers or RAID controllers section) and right click on that
- manually change the driver to a Windows "Standard AHCI Controller" driver or similar
("Standard AHCI 1.0 Compatible controller" or so).

After the update and restart, things may be better with the JMB39x chip too.

I still remember the reason why I had installed the Marvell 91xx controller driver.
I was using Windows "Standard AHCI Controller" driver for many years until I bought a KLEVV 960GB SSD last years.


I connect the 2.5' SSD to my computer via esata and draw the 5V power from the USB port via this kinds of adapter.

I had run benchmark software and everything working fine.
But when I try to keep copy large files from intel 730 SSD (which directly connect to motherboard Sata port) to my KLEVV SSD (Marvell 91xx port), the huge ISO (10GB+) files got corrupted. I knew it because I use fastcopy to copy my files with checksum verify enabled. I finally found out that install the Marvell 91xx controller driver can addressed this problem but at the same time I realized there is a performance drop in throughput (verify by benchmark software, Read / write speed is slower). Data corruption is a much more serious problem and that's why I kept the latest driver installed.


Anyway, I will try to downgrade it back to windows 7 default one ("Standard AHCI Controller") driver and see if it will works with JMB39x tonight.

[kamcm]

I just have time to do the test today.

I had switch back to the original AHCI Driver provided by Microsoft. HDSentinel able to display all 5 drives's SMART info immediately.
But the disk management only recognized one of them, the default driver didn't support port multiplier really good.

So I switch back to Marvell driver and reboot my computer. And after the reboot , the smart info now showed up on HDSentinel correctly with marvell driver installed





I don't know how it can be happened. may be the re-installation of driver change some settings
At the time it unable to showed up SMART. HDSentinel recognized the correct drive letter and drive capacity.
But when HDSentinel able to showed up SMART, it no longer able to recognized the correct drive letter and drive capacity

I think It would be great if HDSentinel have an new function in the coming version -> Add a "optimized for JBOD enclosure mode for non-raid user" checkbox in the setting and when this checkbox is ticked, HDSentinel will verify the disk details from windows and make sure it can populate the correct disk info.

Thanks!

Kam

[kamcm]

I was curious on it and I finally found a way to replicate the problem.

In this test I used 6 different HDD enclosure having different chipset from 2 bay to 5 bay to make it happened. When I had connected my marvell eSata port to many different HDD enclosure one by one without reinstall the marvell sata driver, the marvell sata will finally no longer able to communicate with my JMB39x enclosure box until I reinstall the marvell Sata driver):lol:

Kam

[hdsentinel]

> I had switch back to the original AHCI Driver provided by Microsoft. HDSentinel able to display all 5 drives's SMART info immediately.
> But the disk management only recognized one of them, the default driver didn't support port multiplier really good.

Oh, it is good to know. Sorry, as I remember, the default AHCI driver should also support port multiplier (at least we'd expect this).
Maybe depends on controller / OS version also....


> So I switch back to Marvell driver and reboot my computer. And after the reboot , the
> smart info now showed up on HDSentinel correctly with marvell driver installed

Thanks for the images, however, they do not give any information about the internals, how the information could be detected (differently), what is the response of the bridge chip and so.
In such situations, a developer report (Report menu -> Send test report to developer option) is the best: as then it is possible to examine the raw response of the chip, examine the behaviour of the driver - and this gives information ans tips about possible changes (I mean workarounds for bugs/limitations of the chip, disk controller, driver and their combination).

> I don't know how it can be happened. may be the re-installation of driver change some settings
> At the time it unable to showed up SMART. HDSentinel recognized the correct drive letter and drive capacity.

Yes. Then the chip did not provide any status information at all, did not respond to the specific detection code, designed exactly for this chip, which enumerates the drives and their status.
This is not rare when driver changed and the disk controller restarts - usually a power cycle of the external enclosure required to make things work correctly.


> But when HDSentinel able to showed up SMART, it no longer able to recognized the correct drive letter and drive capacity

Yes. As I tried to explain, when the chip properly responds to detection, it shows all drives like if they'd form a single RAID array.
So then it is not really possible to determine which one physical hard disk belongs to which logical drive.


> I think It would be great if HDSentinel have an new function in
> the coming version -> Add a "optimized for JBOD enclosure mode for non-raid user" checkbox

Yes I agree. But if that would be possible - then that option would be already available - or even better: there would be no need for this option as then Hard Disk Sentinel could automatically determine the correct relation between the physical drives and the logical drives.

Hard Disk Sentinel can detect this scenario: when the RAID chip is used in simple, non-RAID (port multiplier) mode.
But the problem is that (even if this sounds weird) the order of physical drives may still not match the logical drives - so (because of the chip) we may still have not really perfect association.

I'm still researching and trying the best, but this requires developer reports. I mean I need to see the other status, when the logical drives seem correctly reported.

To make a such report, you may try the following:

1) completely close Hard Disk Sentinel by File -> Exit
2) open the folder of the software and move DETJM.DLL and DETJM2.DLL files to a different (temporary) folder.
3) launch Hard Disk Sentinel.
This way the JMicron specific detection will not run at all - and I suspect all drives with correct logical drive letters displayed (just without the status information, without self-monitoring details).
Please use Report menu -> Send test report to developer option NOW.
4) completely close Hard Disk Sentinel by File -> Exit
5) move back the DETJM.DLL and DETJM2.DLL files
6) launch Hard Disk Sentinel so then it will detect status again - just due to limitation of the chip, the logical drive letters may be missing.
Please use Report menu -> Send test report to developer option NOW again.

By comparing the two different situations, things may be seen better and these give ideas about possible improvements.

Now also I recommend to use the comment feature: on the Overview page, you may see the "Click to add comment ..." field.
There it is possible to specify any comment for the actual physical hard disk. You may specify (for example) the correct logical drive here for all drives, so then this will be displayed in the software and the reports - and helps to properly identify logical drives.