eSata SheevaPlug

[tangleboss2]

That was the impression I got from posts by NewIT which means it's well out of date (in internet time terms ) and will not support port multipliers for which you need 2.6.30.something. Whether the rest of the SATA support is the same as more recent kernels I don't know, it almost certainly comes from the days when you needed to use the Orion GIT tree while all the required support is now available in the newer kernel.org sources. At least I understand that is the case, I haven't researched whether there is any newer work available outside kernel.org. That said I imagine the SATA IP in the processor is more or less a copy of other Marvel chips so their driver should have been fairly mature from the start.

My 2.6.32.7 kernel and modified plug seems to work fine, I've loaded a few hundred GB of data onto the system so far and copied the same amount between disks without any disk or other issues.

[fun]

hello

I made debian kernel packages (from trunk) and uImage/uInitrd with SATA patch.

 http://www.naobsd.org/sheeva/debian/

uInitrd-2.6.32-2-kirkwood.nfsroot is initrd for NFSROOT environment.

here is a part of dmesg:

  :
[    0.000000] Linux version 2.6.32-2-kirkwood (Debian 2.6.32-8) (waldi@debian.org) (gcc version 4.3.2 (Debian 4.3.2-1.1) ) #1 Mon Feb 8 18:45:09 JST 2010
  :
[   25.431170] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl F300)
[   25.451575] ata2.15: Port Multiplier 1.1, 0x1095:0x5744 r33, 3 ports, feat 0x1/0x9
[   25.459583] ata2.00: hard resetting link
[   25.971398] ata2.01: hard resetting link
[   26.481401] ata2.02: hard resetting link
[   26.871199] ata2.00: ATA-8: Hitachi HTS545025B9A300, PB2OC64G, max UDMA/133
[   26.878192] ata2.00: 488397168 sectors, multi 0: LBA48 NCQ (depth 31/32)
[   26.941260] ata2.00: configured for UDMA/133
[   26.981197] ata2.01: ATA-8: Hitachi HTS545025B9A300, PB2OC64G, max UDMA/133
[   26.988194] ata2.01: 488397168 sectors, multi 0: LBA48 NCQ (depth 31/32)
[   27.051246] ata2.01: configured for UDMA/133
  :

[portnawak]

Code:

make -j2 ARCH=arm CROSS_COMPILE=arm-none-linux-gnueabi- uImage

Then install using the instructions that are around. In my case installing to a SD card it's actually trivially easy to put the card in a SD reader on my linux box and copy the files - also easier to recover if the kernel doesn't work.

Good evening/afternoon,

Well, I managed to compile the kernel using the instructions provided, and get a uImage binary.

I'm also now using the SD card (re-installed with the Installer), so the copy of the uImage to the sdcard was also trivially easy (I only copied this binary, I hope this is correct).

Unfortunately, the Sheevaplug does not boot any longer, last messages during boot are :

console [ttyS0] enabled
brd: module loaded
loop: module loaded
sata_mv sata_mv.0: version 1.28

and then nothing.

Tried to boot with, and a second time without, the enclosure connected to the eSata port, that did not make any difference.

I think I followed all the instructions provided, may be I missed a step but I don't think so.

Kernel is 2.6.32-8 from kernel.org, with the 4 patches from sheeva.with-linux, the .config from sheeva.with-linux, and the modified sheevaplug-setup.c

Does someone have an idea of what could be the reason of this message during boot ?

Thanks in advance.

[portnawak]

Unfortunately, the Sheevaplug does not boot any longer, last messages during boot are :

console [ttyS0] enabled
brd: module loaded
loop: module loaded
sata_mv sata_mv.0: version 1.28

and then nothing.

UPDATE : have upgraded uboot to a modified version with eSata support (found in this forum), the IDE reset now works, my 4 drives are recognized by uboot and the Sheevaplug boots correctly.

I however does not see any drives after Debian boot is completed, the next step is to understand why Debian does not see my drives (if someone has an idea, do not hesitate )


UPDATE #2 : it indeed does work with a single 2.5 disk connected to eSata port, but not yet with the 4 bays enclosure: the sata link goes up in /var/log/messages, but no disk is detected when the enclosure is connected (but u-boot is able to detect the 4 disks). It seems like the PM option is not enabled in the kernel (but it is). Strange ...

[portnawak]

UPDATE #2 : it indeed does work with a single 2.5 disk connected to eSata port, but not yet with the 4 bays enclosure: the sata link goes up in /var/log/messages, but no disk is detected when the enclosure is connected (but u-boot is able to detect the 4 disks). It seems like the PM option is not enabled in the kernel (but it is). Strange ...

Made different tests, with unfortunately same results.

U-boot is able to detect the PM in the enclosure (supposed to use a SIL chipset, this is a Mediasonic HF2-SU2S2) and recognize the 4 drives.

On Debian, unfortunately, no disk is detected by the kernel when the enclosure is connected (it does work with a single 2.5 eSata hard disk, as well as with a eSata dock and a 3.5 HD). Tried with one or 4 drives in the enclosure, no change.

As U-Boot is able to detect all the disks in the enclosure, it seems to me that this is probably not a hardware issue, but something wrong with the linux driver ?

Does someone have an idea ?

Code:

Reset IDE:
Marvell Serial ATA Adapter
Integrated Sata device found
Port Multiplier found @ 0 1. Vendor: 13a8 ports: 4
[0 1 0]: Enable DMA mode (6)
  Device 1 @ 0 1 0:
Model: SAMSUNG HD154UI                          Firm: 1AG01118 Ser#: S1xxxxxxxxxxxx
            Type: Hard Disk
            Supports 48-bit addressing
            Capacity: 1430799.3 MB = 1397.2 GB (-1364690128 x 512)
[0 1 1]: Enable DMA mode (6)
  Device 2 @ 0 1 1:
Model: SAMSUNG HD154UI                          Firm: 1AG01118 Ser#: S1xxxxxxxxxxxx
            Type: Hard Disk
            Supports 48-bit addressing
            Capacity: 1430799.3 MB = 1397.2 GB (-1364690128 x 512)
[0 1 2]: Enable DMA mode (6)
  Device 3 @ 0 1 2:
Model: SAMSUNG HD154UI                          Firm: 1AG01118 Ser#: S1xxxxxxxxxxxx
            Type: Hard Disk
            Supports 48-bit addressing
            Capacity: 1430799.3 MB = 1397.2 GB (-1364690128 x 512)
[0 1 3]: Enable DMA mode (6)
  Device 4 @ 0 1 3:
Model: SAMSUNG HD154UI                          Firm: 1AG01118 Ser#: S1xxxxxxxxxxxx
            Type: Hard Disk
            Supports 48-bit addressing
            Capacity: 1430799.3 MB = 1397.2 GB (-1364690128 x 512)

Marvell>>

[pingtoo]

May be your kernel did not enable port multiplier support. check your .config file, you need to have CONFIG_SATA_PMP=y.

Good luck

[portnawak]

May be your kernel did not enable port multiplier support. check your .config file, you need to have CONFIG_SATA_PMP=y.

Thanks, but unfortunately, it did:

CONFIG_SATA_PMP=y

[pingtoo]

Then I don't know where else to begin. 

I ordered my Guru Plug server, so until I have a machine I can not run any test about esata. Everybody seems to have a esata enabled plug these day I wonder where did it came from since Globalscale is not shipping yet.

[UnaClocker]

Then I don't know where else to begin. 

I ordered my Guru Plug server, so until I have a machine I can not run any test about esata. Everybody seems to have a esata enabled plug these day I wonder where did it came from since Globalscale is not shipping yet.

I believe everyone who has one so far, has soldered the components on themselves.

[portnawak]

I believe everyone who has one so far, has soldered the components on themselves.

No, some of us bought the device from  NewIT in the UK (I live in Europe), they sell the black Sheevaplug with the eSata port, that's what I bought.

As the supplier of NewIT is Globalscale, it's really strange that this black eSata Sheeva is not listed on Globalscale site 

[NewIT]

I believe everyone who has one so far, has soldered the components on themselves.

No, some of us bought the device from  NewIT in the UK (I live in Europe), they sell the black Sheevaplug with the eSata port, that's what I bought.

As the supplier of NewIT is Globalscale, it's really strange that this black eSata Sheeva is not listed on Globalscale site 

We secured sole distribution rights for a period on this product, thas why its not on their site.

[boerner]

I believe everyone who has one so far, has soldered the components on themselves.

No, some of us bought the device from  NewIT in the UK (I live in Europe), they sell the black Sheevaplug with the eSata port, that's what I bought.

As the supplier of NewIT is Globalscale, it's really strange that this black eSata Sheeva is not listed on Globalscale site 

We secured sole distribution rights for a period on this product, thas why its not on their site.

So the mystery is solved. If it wasn't for the poor exchange rate to the U.S., I would have already ordered one from you :-)

[tangleboss2]

May be your kernel did not enable port multiplier support. check your .config file, you need to have CONFIG_SATA_PMP=y.

Thanks, but unfortunately, it did:

CONFIG_SATA_PMP=y

The other vital option to see more than one disk on a controler - whether SATA or USB or SCSI - is to enable 'scan all LUNS' in the SCSI options section and set a suitable maximum number of LUNs. But these options were already set in the sheeva.with-linux .config files I used.
If it works with a single SATA disk then it seems the hardware and mods to the stat-up file are good.