Driver for LSI MegaRAID SAS family

[MrSTiMPy]

i will be testing SSD Cachecade on this setup and migrating both windows and mac to new disks soon.

If there's anything worthy reporting, i'll let you know.

 

 

Hey CycleBurns, did you have any luck with the CacheCade? Thanks!!

 

CycleBurns was meant to try cachecade on his setup: http://www.insanelymac.com/forum/topic/285197-driver-for-lsi-megaraid-sas-family/?p=1901976 so you may ask him

 

Just sent him a private msg, but seems he was last online on June 2013... so I maybe out of luck...

 

Also, I am about to test my 9260, will let you know my findings!

 

Thanks for everything man!

Another question sir...

 

Sorry I am a bit of a bugger...

 

Do you suggest upgrading to the latest firmware?

 

Thanks!

[MrSTiMPy]

OK, I am back to report my results... BIG SUCCESS!!

 

 

Tomorrow I will install the cache battery... Crossing my fingers to get a reply from any of the guys with CacheCade and Fastpath...

[dukzcry]

Do you suggest upgrading to the latest firmware?

 

I would say yes, unless the latest FW breaks things for you.

 

OK, I am back to report my results... BIG SUCCESS!!

Tomorrow I will install the cache battery... Crossing my fingers to get a reply from any of the guys with CacheCade and Fastpath...

 

Good!

[dogmaphobic]

Success with an Intel RT3WB080 Raid Controller.

 

I had one not being used and it didn't even occur to me to try on Mac OS. A quick google search (well, not so quick as I had to first find out to look for "MegaRaid") brought me here. Downloaded the kext, booted and it all came right up.

 

I have it with 4 2TB drives in Raid 5. When I first booted the disk was automatically mounted with its original NTFS partition and all files accessible. I reformatted and I am now in the process of restoring data into it.

 

A huge thanks for the effort!

 

The only question I have: Is there a way to have S.M.A.R.T. support?

 

Thanks!

 

 

[dukzcry]

 

Success with an Intel RT3WB080 Raid Controller.

 

An interesting controller you have, can find almost nothing about it

 

The only question I have: Is there a way to have S.M.A.R.T. support?

 

You can't see it directly in the OS, but you can monitor physical drives status as well as many other parameters using a console 'mfiutil' tool. Both utility and manpage are installed along with a kext.

 

Have fun!

[dogmaphobic]

I got it from newegg a few years ago. They still have it listed: http://www.newegg.com/Product/Product.aspx?Item=N82E16816117214

 

I had read something about mfiutil in passing but had not paid close attention to it. Thanks for pointing it out. I can't however, get the events (what I think it would be what I'm looking for). The config comes up OK:

 

>sudo mfiutil show config
mfi0 Configuration: 1 arrays, 1 volumes, 0 spares
    array 0 of 4 drives:
        drive  4 ( 1863G) ONLINE <WDC WD2002FAEX-0 1D05 serial=WD-WCAY000xxxx> SATA
        drive  5 ( 1863G) ONLINE <WDC WD2002FAEX-0 1D05 serial=WD-WCAY000xxxx> SATA
        drive  6 ( 1863G) ONLINE <WDC WD2002FAEX-0 1D05 serial=WD-WCAY000xxxx> SATA
        drive  7 ( 1863G) ONLINE <WDC WD2002FAEX-0 1D05 serial=WD-WCAY000xxxx> SATA
    volume mfid0 (5585G) RAID-5 256K OPTIMAL spans:
        array 0

 

The events cause an I/O error (edited with proper output):

 

> sudo mfiutil show events 1
<... snip ...>
12884 (Sat Jan 19 08:18:19 EST 2013/DRIVE/WARN) - Command timeout on PD 00(e0xfc/s3) Path 4433221100000000, CDB: 2f 00 a2 66 90 00 00 10 00 00
12885 (Sat Jan 19 08:18:19 EST 2013/DRIVE/WARN) - PD 00(e0xfc/s3) Path 4433221100000000  reset (Type 03)
mfiutil: Failed to fetch events: Input/output error

For what's worth, here is the log sequence:

 

>sudo mfiutil show logstate
mfi0 Event Log Sequence Numbers:
  Newest Seq #: 18575
  Oldest Seq #: 1
   Clear Seq #: 2
Shutdown Seq #: 18562
    Boot Seq #: 18563

 

Thanks!

[dukzcry]

I can't however, get the events (what I think it would be what I'm looking for). 

The events cause an I/O error (edited with proper output):

> sudo mfiutil show events 1
<... snip ...>
12884 (Sat Jan 19 08:18:19 EST 2013/DRIVE/WARN) - Command timeout on PD 00(e0xfc/s3) Path 4433221100000000, CDB: 2f 00 a2 66 90 00 00 10 00 00
12885 (Sat Jan 19 08:18:19 EST 2013/DRIVE/WARN) - PD 00(e0xfc/s3) Path 4433221100000000  reset (Type 03)
mfiutil: Failed to fetch events: Input/output error

 

By default, you're seeing only warning messages from the event log. If you need to see more, point to an appropriate class, like 'mfiutil show events -c info 1'. Classes are described at man page.

Just ignore I/O errors you'll see at the end of output, there's nothing to worry about. The reason is that mfiutil is written specially for FreeBSD's mfi driver, while this one is totally different from it.

 

For what's worth, here is the log sequence:

 

>sudo mfiutil show logstate
mfi0 Event Log Sequence Numbers:
  Newest Seq #: 18575
  Oldest Seq #: 1
   Clear Seq #: 2
Shutdown Seq #: 18562
    Boot Seq #: 18563

 

Instead of figuring out log entry numbers with 'logstate' cmd you may use aliases (see man for them). 'mfiutil show events -c info boot' for example.

[drowsyhaze]

Hey dukzcry, I have an LSI 9750-4i - I've read here that this driver isn't compatible because the firmware isn't, even though this family of cards uses the same processor. Is this true? Is there any workaround? I spent a lot of money on this card and would hate to have to buy another for 10.9 compatability.

 

I tried using the 9750-4i4e driver which is supposedly compatible through 10.8, having read it was broken in 10.8.3, but no dice. All drivers recognize an unformatted array but will not allow me to partition. Maybe I could inject the PCI-ID of the 4i4e into the DSDT and get results?

 

I may try gparted today to partition, but I understand that 10.9 still won't be able to mount the disk due to PCI-EFI not being available on the card's firmware.

 

I am using chameleon 2.2.1 if that's relevant.

 

Anyway, thanks so much for all of your hard work on this project, hope to hear from you soon!

[Shane 2]

I've been using your driver for a few days, and an issue has showed up.

 

Whenever I start Disk I/O on one of the partition/drive, I got sound assertion. This is my test routine:

 

a ) Clean boot.

b ) I start my music app and watch the console - Nothing

c ) I copy a file from one of the RAID drive to another drive

d ) sound assertion.

 

The console message about the assertion is this:

 

2014/1/17 上午3:45:18.000 kernel[0]: Sound assertion - 8086:8c20 FIFO error (Stream 7)

 

Is everyone experiencing similar problem or is it just me?

 

Is there other possibility for this to happen, how can I test it?

 

Beside using the Debug version of the kext which I will do after this weekend, what else can I do to help?

[dukzcry]

Hey dukzcry, I have an LSI 9750-4i - I've read here that this driver isn't compatible because the firmware isn't, even though this family of cards uses the same processor. Is this true? Is there any workaround? I spent a lot of money on this card and would hate to have to buy another for 10.9 compatability.

Maybe I could inject the PCI-ID of the 4i4e into the DSDT and get results?

 

drowsyhaze, it seems to be true  But isn't it possible to flash it with LSI 9261-4i firmware (or whatever card has the same hardware)?

As for DSDT i dunno. You may try.

 

 

I've been using your driver for a few days, and an issue has showed up.

 

Whenever I start Disk I/O on one of the partition/drive, I got sound assertion. This is my test routine:

 

a ) Clean boot.

b ) I start my music app and watch the console - Nothing

c ) I copy a file from one of the RAID drive to another drive

d ) sound assertion.

 

The console message about the assertion is this:

 

2014/1/17 上午3:45:18.000 kernel[0]: Sound assertion - 8086:8c20 FIFO error (Stream 7)

 

Is everyone experiencing similar problem or is it just me?

 

Is there other possibility for this to happen, how can I test it?

 

Beside using the Debug version of the kext which I will do after this weekend, what else can I do to help?

 

Not me at least. You may try to disable all third-party kexts you're not urgently need, and retest.

You may also try changing driver parameters in plist. For example disable MSI.

 

P.S.: Do you use an AppleHDA? You may try VodooHDA instead if it supports your sound.

[Shane 2]

I am using AppleHDA for now. I will try to disable unnecessary kexts and try. Thanks!

[bichobola]

Hi. First thanks for the MegaRaid SAS kext. Now I can use my fujitsu d2616 (SAS2108) in OSX Mavericks 10.9.1!!

I have only a little problem. Now OSX can't sleep, the monitor goes off, but the computer remains enabled.

¿Anyone know a workaround?

 

Thanks.

[dukzcry]

Hi. First thanks for the MegaRaid SAS kext. Now I can use my fujitsu d2616 (SAS2108) in OSX Mavericks 10.9.1!!

I have only a little problem. Now OSX can't sleep, the monitor goes off, but the computer remains enabled.

¿Anyone know a workaround?

 

Thanks.

 

I'm sorry, but i was forced to add a hack to this driver which prevents the system from going to sleep. The reason is that i'm getting data corruption of boot volume and/or kernel panic on either wakeup or next boot after sleep->wakeup was made, when i'm booting OS from the drive attached to a controller (if i'm booting from a drive not handled by this driver, everything is ok). I've tried to fix this, but nothing helped so far.

I'll send you kexts with this hack disabled via PM, if you mind to check, maybe you got another experience... But, as a disclaimer, you'll need to disconnect all volumes with valuable data, and move your OS X boot drive to an internal controller.

 

P.S.: Note that round a minute should pass after the wakeup before you'll can access your drives.

[bichobola]

Thanks a lot dukzcry. My boot hd is in another controller, and the minute for controller's wakeup is a normal behavior (under windows).

I'm going to try it.

Thanks, thanks and thanks again.

[CycleBurns]

Hey dukzcry,

 

i just can't thank you enough!

i am actually happy to see more and more people enjoying your work  

 

I've been using your kext for a while, and i sort of found a small issue.

 

 

It has to do with the visibility of the virtual disks, which is ' sort of' associated with the internal enumeration of the virtual disks.

Let me explain; In windows you always see all virtual disks. In OSX this depends. It depends on the availability of the disks PREVIOUS to the actual disk number on which your install is located.

I encountered this a while ago, when i just started, but disregarded it (after finding out what it was) to the firmware.

In the latest firmware update an issue was solved that implied the same behaviour. But now i encounter it again.

The situation is always the same; when i migrate to new disks. I found out because AFTER migrating an install, it would work.... untill i removed old disks (and associated virtual disks weren't available.

 

when i suspected to understand this behavior, i made new virtual disks (in WebBios), and since a new VD0 was available, the boot worked.

It appears that the enumeration of disks must be continous, or else it will not show the rest of the disks... IMPORTANT; Remember that if you remove disks (not deleting the virtual disks on the physical disks) the controller will remember them, causing 'gaps' in your VD enumeration.

 

 

See the image attached; for some reason ID 7 doesnt have the 'MFID' string prefixed, and the partition on this VD are not visible.

I suspect because there are no disks ID's 5 and 6 available. Similar situation can/will happen when you migrate your install from existing disks (hosted by the controller) to other disks on the controller.

I haven fully tested all possible scenario's, and once aware you can ' circumvent' this more or less, but i thought you (and others) might like to know.. 

 

Off course i might be mistaken, but i think this is worth checking out....

 

 

 

[dukzcry]

Thanks for the reporting, CycleBurns!

 

Please test whether kext in attach works for you.

[fisherwei]

can this driver support Dell perc H310(be flashed as IT mode) HBA card in 10.9?

I tried to install(download from osx86.com), but, it seems not work.

 

Kindly, would you advise what cmds/steps can be used for troubleshooting?

[dukzcry]

can this driver support Dell perc H310(be flashed as IT mode) HBA card in 10.9?

I tried to install(download from osx86.com), but, it seems not work.

 

Kindly, would you advise what cmds/steps can be used for troubleshooting?

The card is supported, but only with IR firmware.

[CycleBurns]

Thanks for the reporting, CycleBurns!

 

Please test whether kext in attach works for you.

 

YOU! Sir, are awesome!

Seems to work like a charm.

 

BTW; Why is it so much smaller?

And... will you be supporting sleep (eventually)?

[dukzcry]

YOU! Sir, are awesome!

Seems to work like a charm.

 

BTW; Why is it so much smaller?

And... will you be supporting sleep (eventually)?

 

Because it's just a kext, not installer

I'll add sleep enabling option into 'unsafe settings' plist, as i have no workaround for the case when the boot volume is on MegaRAID controller.

 

UPDATE: zefix for you 'll not go upstream, because in case with supporting of uncontiguous configurations, boot time aries significally (OS probes even non-existing drives). So you'd better to find the firmware that does proper enumeration or always use that kext from attach with slow boot.

[schafdog]

I am looking into buying a RAID controller. Have read this thread, but is not quite sure what the full requirements are to use this driver: 

 

chipset based on LSI 2008 (some other numbers are mentioned, but they are perhaps rebranded?)

Firmware should be IR (raid mode?) 

Some "other firmware" requirements? 

 

So looking at the market in Denmark I seem to be able to get

 

- PERC 5/6i (out of stock, but may come back)

- IBM ServeRAID M1015 & M5014

- Dell H310

- LSI 9211-8i 

 

The 9211-8i is a little more expensive but supports SATA-600. Is that supported? 

(UPDATE: Seems to be the same as M1015, just rebranded ?)

 

If not which is is the best choice? 

 

cheers, 

:-D

[CycleBurns]

Because it's just a kext, not installer

I'll add sleep enabling option into 'unsafe settings' plist, as i have no workaround for the case when the boot volume is on MegaRAID controller.

 

UPDDATE: zefix for you 'll not go upstream, because in case with supporting of uncontiguous configurations, boot time aries significally (OS probes even non-existing drives). So you'd better to find the firmware that does proper enumeration or always use that kext from attach with slow boot.

Still... The kext is almost 50% smaller :-)

 

- IBM ServeRAID M1015 & M5014

- Dell H310

 

 

Are all SATA III (600)

 

 

Be very(!) careful with the Dell H310.

The chance your consumer motherboard supports it, without any backdraws, is not that good as with the other cards.

Off course, if you plan to use it in a Dell workstation, you'll be golden.

[schafdog]

Be very(!) careful with the Dell H310.

The chance your consumer motherboard supports it, without any backdraws, is not that good as with the other cards.

Off course, if you plan to use it in a Dell workstation, you'll be golden.

Thanks! I'll stay away from H310. Don't like Dell anyway. 

[dukzcry]

Have read this thread, but is not quite sure what the full requirements are to use this driver: 

 

chipset based on LSI 2008 (some other numbers are mentioned, but they are perhaps rebranded?)

 

Few chips are supported. Not only the SAS2008. Here they are: 1064R, 1068, 1078/DE, 2108. Not all products based on these chips will pass thought, as an appropriate IOP should be used by vendor in pair.

 

 

Firmware should be IR (raid mode?) 

Some "other firmware" requirements? 

 

Yup.

Dunno, but if someone here has a success using the same controller while you haven't, you may ask his/her FW revision and try after flashing that one.

 

 

The 9211-8i is a little more expensive but supports SATA-600. Is that supported? 

(UPDATE: Seems to be the same as M1015, just rebranded ?)

 

If not which is is the best choice? 

 

Can't say 'bout the best choice, but a good choice is to pick a card someone already successfully tested under this kext. Just look through the thread and make your decision

Note that one person in this topic had failed with a M1015 (was using a wrong firmware all time?), while others got a success with it. 9211 is unchecked, but if you found it being the same as M1015, then, in theory, it should go.

[CycleBurns]

Thanks for the reporting, CycleBurns!

 

Please test whether kext in attach works for you.

 

 

An update....

 

 

My IBM m5014 broke down.

 

Fortunately i had the chance to aquire an original LSI 9260-8i with BBU and 512Mb. (i have some for sale also)

I also replaced my slow and unsupported WD30ERZRX with Hitachi 4TB drives

 

All should work well, but at this point (i haven't tested and analyzed all yet) it seems that if the read/write speeds get too high, OS X will fail.

It will become unresponsive... as in... no drive activity/awareness anymore, rendering it useless.

 

So far;

• I de-activated disk cache, but it still happens.
• I recreated the array/changed properties. This helps when doing a migration from a flashed IBM to a original LSI. Apparently there are some small differences. Just change some attributes and change them back later to get it going correctly....

 

How could i best determine whether this is driver related? OS related? HDD related? (Which it absolutely shouldn't be!)