Jump to content
Corsair Community

Corsair X series incompatible with nVidia MCP55 SATA controller?


niemayer

Recommended Posts

Hi,

 

I attached a Corsair CMFSSD-128D1 to a SATA channel on a SuperMicro H8DME-2 board (which employs an nVidia MCP55 pro chipset, including the 6-channel SATA controller). While it is possible to read sectors from the SSD, any attempt to write even a single sector results in a lock-up on the SATA channel (so even subsequent reads won't work), this lock-up will not resolve unless the system is unpowered (even pushing the reset button will not allow subsequent reads off the SSD).

 

Since even the nVidia RAID controller tries to write sectors, using the Corsair SSD locks up the system if entering the BIOS RAID setup utility - no operating system required to reproduce the problem.

 

Only if I disable the on-board RAID controller and avoid any writes to the SSD, I can boot into an operating system. Reading sectors, again, works fine.

Find below detailed error messages emitted from the "sata_nv" driver of linux-2.6.32.7 when trying to write a single sector (dd if=/dev/zero of=/dev/sdd bs=512 count=1):

 

Jan 29 19:59:26 kernel: ata6: EH in SWNCQ mode,QC:qc_active 0x1 sactive 0x1
Jan 29 19:59:26 kernel: ata6: SWNCQ:qc_active 0x1 defer_bits 0x0 last_issue_tag 0x0
Jan 29 19:59:26 kernel:   dhfis 0x1 dmafis 0x1 sdbfis 0x0
Jan 29 19:59:26 kernel: ata6: ATA_REG 0x41 ERR_REG 0x84
Jan 29 19:59:26 kernel: ata6: tag : dhfis dmafis sdbfis sacitve
Jan 29 19:59:26 kernel: ata6: tag 0x0: 1 1 0 1  
Jan 29 19:59:26 kernel: ata6.00: exception Emask 0x1 SAct 0x1 SErr 0x400000 action 0x6 frozen
Jan 29 19:59:26 kernel: ata6.00: Ata error. fis:0x21
Jan 29 19:59:26 kernel: ata6: SError: { Handshk }
Jan 29 19:59:26 kernel: ata6.00: failed command: WRITE FPDMA QUEUED
Jan 29 19:59:26 kernel: ata6.00: cmd 61/08:00:00:00:00/00:00:00:00:00/40 tag 0 ncq 4096 out
Jan 29 19:59:26 kernel: 	 res 41/84:00:00:00:00/84:00:00:00:00/40 Emask 0x10 (ATA bus error)
Jan 29 19:59:26 kernel: ata6.00: status: { DRDY ERR }
Jan 29 19:59:26 kernel: ata6.00: error: { ICRC ABRT }
Jan 29 19:59:26 kernel: ata6: hard resetting link
Jan 29 19:59:26 kernel: ata6: nv: skipping hardreset on occupied port
Jan 29 19:59:32 kernel: ata6: link is slow to respond, please be patient (ready=0)
Jan 29 19:59:36 kernel: ata6: SRST failed (errno=-16)
Jan 29 19:59:36 kernel: ata6: hard resetting link
Jan 29 19:59:36 kernel: ata6: nv: skipping hardreset on occupied port
Jan 29 19:59:42 kernel: ata6: link is slow to respond, please be patient (ready=0)
Jan 29 19:59:46 kernel: ata6: SRST failed (errno=-16)
Jan 29 19:59:46 kernel: ata6: hard resetting link
Jan 29 19:59:46 kernel: ata6: nv: skipping hardreset on occupied port
Jan 29 19:59:52 kernel: ata6: link is slow to respond, please be patient (ready=0)
Jan 29 20:00:21 kernel: ata6: SRST failed (errno=-16)
Jan 29 20:00:21 kernel: ata6: limiting SATA link speed to 1.5 Gbps
Jan 29 20:00:21 kernel: ata6: hard resetting link
Jan 29 20:00:21 kernel: ata6: nv: skipping hardreset on occupied port
Jan 29 20:00:26 kernel: ata6: SRST failed (errno=-16)
Jan 29 20:00:26 kernel: ata6: reset failed, giving up
Jan 29 20:00:26 kernel: ata6.00: disabled
Jan 29 20:00:26 kernel: ata6: EH complete
Jan 29 20:00:26 kernel: sd 6:0:0:0: [sdd] Unhandled error code
Jan 29 20:00:26 kernel: sd 6:0:0:0: [sdd] Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK
Jan 29 20:00:26 kernel: sd 6:0:0:0: [sdd] CDB: Write(10): 2a 00 00 00 00 00 00 00 08 00
Jan 29 20:00:26 kernel: end_request: I/O error, dev sdd, sector 0
Jan 29 20:00:26 kernel: Buffer I/O error on device sdd, logical block 0
Jan 29 20:00:26 kernel: lost page write due to I/O error on sdd

 

BTW: Disabling NCQ in the driver, often suggested with nVidia controllers, changes the error messages, but not the symptom - writing fails the same.

Link to comment
Share on other sites

Stupid question to begin with - have you got the latest drivers for your chipset? If not head off to nvidia.com and see if there's anything new.

 

Failing that, get in touch with nvidia and ask if they have any working solutions to the problem, or if they are working on one. It's well know that SSD's (no matter who's they are) have mega problems with nVidia chipsets.

Link to comment
Share on other sites

My Asus motherboard has a 780i chipset and I have had zero issues as far as compatability with the SSD. I have never heard that nVidia had mega problems with SSDs. I thought that nVida was always rated as top notch. Oh well, sorry that you are having issues. I would suggest checking for updated chipset drivers. NVidia usually is on top of things as far as drivers go.
Link to comment
Share on other sites

Stupid question to begin with - have you got the latest drivers for your chipset?
As I wrote, the error occurs even before any operating system is booted - just entering the BIOS RAID configuration is enough to reproduce the problem.

But yes, linux kernel 2.6.32.7 is the most recent stable kernel, and the nvidia web pages officially point to the kernel driver sata_nv as being "the driver".

 

Failing that, get in touch with nvidia and ask if they have any working solutions to the problem, or if they are working on one. It's well know that SSD's (no matter who's they are) have mega problems with nVidia chipsets.
As this would probably be censored (according to forum rules), I'll rather not enter into a discussion on other companies behavior in this thread...
Link to comment
Share on other sites

NVidia are not in direct competition with Corsair. They don't produce RAM, PSU's, Cases or SSD'd. Without feedback from NVidia people here can only guess as to your problem. Posting feedback from NVidia should not be breaking any rules.

 

True. Just no flaming or other violations ;):

Link to comment
Share on other sites

I have a machine with a NVidia 590 SLI - with 6 SATA ports. I found I had to connect the X32 SSD to a low numbered port to get it to work. Connecting to port 5 or 6, like I tried first, caused 'system freeze' when attempting quick format. Actually, if you leave the machine long enough, it will probably 'unfreeze', reporting time-out. But I gave up and moved to the second port and it is working perfectly now.
Link to comment
Share on other sites

I have a machine with a NVidia 590 SLI - with 6 SATA ports. I found I had to connect the X32 SSD to a low numbered port to get it to work. Connecting to port 5 or 6, like I tried first, caused 'system freeze' when attempting quick format.
Interesting, in fact I connected the X128 to port number 6 (of 6).

 

The nVidia 590 is said to have 8 SATA ports - 6 driven by one controller chip, 2 driven by another, 1 of the 2 being an eSATA port. On which of the chips did your SSD fail?

 

I will try connecting the SSD to port number 0 later. But I would be very surprised if that made a difference, since on the MCP55, all ports are of the same type, driven by the same chip.

Link to comment
Share on other sites

It may be the case that some ports are configured slightly differently in the controller FW than others. For example on the ICH10R, there are 6 SATA ports but only 4 have the ability to act in fully "Native" mode, ie the OS assigns the IRQ's to them.

 

I'm not bashing NVidia, I have a HTPC built around one of their boards but it is known that their SB and SATA implementation is probably the poorest of all the manufacturers (Intel, AM|D, NVidia, VIA etc). One can see noticably poorer results on NVidia HW, especially if one uses their own SATA drivers. MS have made a better job with their NV SATA driver.

 

True. Just no flaming or other violations ;):

 

Absolutely. I like to think that the forum is moderated tightly but with understanding. Sometimes one can't help but mention a competitor if that component may be called into question re the problem experienced. Any promotion or the opposite, ridicule of a competing brand should never be tolerated.

 

However... send me a free Obsidian and a H50 and I'll promote |Corsair on any forum you like. ;):

Link to comment
Share on other sites

The nVidia 590 is said to have 8 SATA ports - 6 driven by one controller chip, 2 driven by another, 1 of the 2 being an eSATA port. On which of the chips did your SSD fail?

 

I think the 2 extra ports are running off a Silicon Image controller. Anyway, the failing connection in my case was to the last port on the main block of 6 ports - dunno how better to describe it. Just for fun I enabled the SI controller in the bios, but win7 nagged me about providing a driver disk, so I gave up. It's only a test machine, and always in bits.

Link to comment
Share on other sites

  • Corsair Employees

I seem to remember seeing something about that on a Linux forum I am sorry but Linux is open source and you are more or less on your own, check with the software vendor and see what they say about the issue.

Have you tested the drive using Windows any version and do you see any problems?

Link to comment
Share on other sites

Interesting, in fact I connected the X128 to port number 6 (of 6). .. I will try connecting the SSD to port number 0 later. But I would be very surprised if that made a difference, since on the MCP55, all ports are of the same type, driven by the same chip.
Unbelievable - the SSD works perfectly on Port number 0 (of 6)! It even works with the often blamed NCQ enabled.

 

So there actually is a subtle difference between the SATA ports provided by the MCP55 chipset.

 

And btw., I did test that port 6 works well with ordinary magnetic disks.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...