[SOLVED] Marvell 88SE6111 IDE Problems

[pacman99]

Hi,

I'm having problems with my Marvell 88SE6111 IDE Controller found on my ASUS P5K SE board.
lspci says:

Code: Select all

03:00.0 IDE interface: Marvell Technology Group Ltd. Device 6121 (rev b2)

Using the drivers that came with my motherboard, I get:

Code: Select all

SERVERU Source code # make
make ARCH=x86_64 CC=cc LD=ld CROSS_COMPILE= V= -C /lib/modules/2.6.25-gentoo-r2/build M=`pwd` modules
make[1]: Entering directory `/usr/src/linux-2.6.25-gentoo-r2'
make[1]: *** No rule to make target `code'.  Stop.
make[1]: Leaving directory `/usr/src/linux-2.6.25-gentoo-r2'
make: *** [all] Error 2

Using drivers from Asus' site for the Marvel 6121 controller using the guidelines suggested in this post: http://forums.gentoo.org/viewtopic-t-55 ... e6111.html, I get:

Code: Select all

SERVERU thor-nonraid # make
make ARCH=x86_64 CC=cc LD=ld CROSS_COMPILE= V= -C /lib/modules/2.6.25-gentoo-r2/build M=`pwd` modules
make[1]: Entering directory `/usr/src/linux-2.6.25-gentoo-r2'
  CC [M]  /home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.o
/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.c: In function 'HBA_Translate_Req_Status_To_OS_Status':
/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.c:23: error: 'struct scsi_cmnd' has no member named 'use_sg'
/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.c:50: error: 'struct scsi_cmnd' has no member named 'use_sg'
/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.c:53: error: 'struct scsi_cmnd' has no member named 'request_buffer'
/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.c:54: error: 'struct scsi_cmnd' has no member named 'use_sg'
/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.c:60: error: 'struct scsi_cmnd' has no member named 'request_bufflen'
make[2]: *** [/home/Mar6121/1.0.0.9/thor-nonraid/linux/linux_sense.o] Error 1
make[1]: *** [_module_/home/Mar6121/1.0.0.9/thor-nonraid] Error 2
make[1]: Leaving directory `/usr/src/linux-2.6.25-gentoo-r2'
make: *** [all] Error 2

My kernel options have the following relevant options enabled:

Code: Select all

<*>     generic/default IDE chipset support
<*>   Marvell PATA support via legacy mode   
<*> SCSI disk support
<*> SCSI CDROM support

I'm not sure what to do at this point to get my IDE controller working. Fortunately gentoo is installed on a sata drive that's not affected by this controller.

Thanks

[pacman99]

figured it out. The guidelines posted at: http://forums.gentoo.org/viewtopic-t-55 ... e6111.html were correct but did not apply to the 2.6.25 kernel. I had to downgrade to 2.6.24-r7 then play around with the source as described in the last post of that thread to get it to work

[SVMartin80]

I might buy me an Asus P5Q, which also has a Marvell 88SE6111 chip. Does latest Gentoo version support it out of the box? Is performance ok? (incl. DMA) I plan to use it in a RAID-0 setup. Can you confirm that this also works?

[emerald]

I've modified the marvell driver found at the asus page to work with newer kernels and created a kernel patch, which can be found here:
http://www.home.liji.de/kernel/addons/p ... 61xx.patch
http://www.home.liji.de/kernel/addons/p ... 61xx.patch

Supported chips are Marvell 88SE6121 88SE6122 88SE6141 88SE8145.
The chip called 886111 is some manuals seems to be a 88SE6121.

Here http://www.home.liji.de/kernel/addons/patchsets/ you will also find a (probably working but untested) patch for 2.6.24 and soon for later kernels too till a working driver for this chip is included into mainline.
Btw, the driver only works in non-raid mode.


Don't panic if the links are not accessible, it's a pc at home, which might be turned off now and then, just try again later.

[slaterson]

I've modified the marvell driver found at the asus page to work with 2.6.25 and created a kernel patch, which can be found here:
http://www.home.liji.de/kernel/addons/p ... 61xx.patch

Here http://www.home.liji.de/kernel/addons/patchsets/ you will also find a (probably working but untested) patch for 2.6.24 and soon for later kernels too till a working driver for this chip is included into mainline.
Btw, the driver only works in non-raid mode.

Don't panic if the links are not accessible, it's a pc at home, which might be turned off now and then, just try again later.

i'd like to give this a try. how/where do i apply your patch? it looks like a kernel patch, not a patch for the downloaded driver/module.

also, is this better than sata_mv that is included in the kernel now?

thanks

[emerald]

i'd like to give this a try. how/where do i apply your patch? it looks like a kernel patch, not a patch for the downloaded driver/module.

also, is this better than sata_mv that is included in the kernel now?

thanks

It is a kernel patch, an external module is usually more hassle than internal (and as kernel patch it can also staically be compiled into the kernel).
The patch has to be applied like any other patch (assuming the /usr/src/linux symlink is correct):

Code: Select all

cd /usr/src/linux
cat /<path>/marvell_88se61xx.patch | patch -p1
# check for no rejects
make menuconfig
# select the driver in scsi submenu

The sata_mv driver supports other devices, not the Marvell 88SE6121/6122/6141/6145 chips.
The aforementioned 88SE6111 chip should in fact be the 88SE6121, probably check with lspci too.

[slaterson]

i installed the patch, built a kernel and booted. i got a kernel panic because the patched kernel with the marvell driver switched my hard drive order, sdaX became sdbX and sdbX became sdaX. i booted to the unpatched kernel, changed me grub config (changed root=/dev/sda3 to root=/dev/sdb3 for teh patched kernel line in grub.conf) and i also updated my fstab to use /dev/sdbX instead of /dev/sdaX. the system then booted up just fine, however i'd like to retain my 'normal' configuration, sda remains as sda and sdb remains as sdb.

my system consists of two hard drives and a dvd burner. there is one sata hard drive, one ata hard drive and the ata dvd burner. previously the sata hdd would be sda, now its sdb.

any way to fix that or is the driver simply ordering the drives different due to proper discovery (compared to the previous driver i was using)?

thanks

[emerald]

This driver is working as a SCSI driver and the SCSI devices are parsed first, this fact is not changeable.
So your option is to compile the 88SE61xx driver (mv) as module and let udev handle it automatically or auto-insert the driver on boot. If you are not using the AHCI driver it's no problem that way and it will work.
If you are using the AHCI driver you will somehow have to prevent the AHCI driver from claiming the marvell chip (possibly via kernel parameter), but i don't know how at the moment (except patching the AHCI driver). Since the AHCI driver will (without parameter) automatically claim control of the marvell chip on load it's also no use to compile both as module and insert the AHCI driver first, then the other one won't load and you can't use the ide port (same as without this mv driver).
In case you want both, the AHCI and the Marvell driver, as i do, you will have to live with the changed harddisk layout, at least till a better working marvell driver is incorporated in mainline.

You can try this patch (for 2.6.25), which should prevent the AHCI driver from claiming the marvell chips (and then use the marvell-module-method), but no test and guarantee from my part:
http://www.home.liji.de/kernel/tmp/ahci ... hips.patch

Another way is for grub to use the label-mount: root=LABEL=<hdd-label> which, depending on your grub configuration may/should work and for fstab too, or, as i do, at least for fstab, not to use the /dev/sdYX devices but the symlinks created by udev in /dev/disk/by-id/ and /dev/disk/by-path/ , where the id will always be the same for the same harddisk (harddisk-dependent id) and the path will be the same as long as the controller stay in the same slot (on the same mainboard), which means the PCI-bus-numbering will stay the same.

[slaterson]

thanks for the thorough reply.

i've got ahci enabled in my kernel along with the marvell driver from your patch, of course. i am not sure i need ahci, i need to better understand what it does. i just built a new computer, the last time i made a major upgrade was several years ago when sata didn't exist, so i am behind on the current technology.

one thing that i discovered today is that my cdrom is no longer showing up. i am not when it stopped being identified, it was before i applied your patch - i have been tweaking the kernel hear and there for a few weeks now...

[emerald]

The output of lspci will help a even more specific answer alot, dmesg output too.
Gernerally, if you enabled AHCI mode for any SATA controller in your mainboard BIOS you need the AHCI driver, otherwise (probably) not at all.

Code: Select all

<*>   AHCI SATA support

If you enabled the AHCI driver, without this ahci-dont-claim-marvell-chips.patch patch the AHCI driver will support your harddisk connected to the SATA ports of the Marvell chip too. But by doing so you lose any chance to use the IDE port of the Marvell chip. So, to prevent the AHCI driver to take 'ownership' of the Marvell chip you should patch your kernel with this patch.
The Marvell 88SE61xx (mv) driver you should compile as module now

Code: Select all

<M>   Marvell Storage Controller 6121/6122/6141/6145

so udev will load the driver after the AHCI driver was already loaded.
That way your SATA harddisk should be sda, the IDE harddisk sdb and the IDE DVD drive stays sr0.
So, just to be sure, get clean sources, apply both patches, select the AHCI and mv driver as just explained and it should work your way.
(I care little about the naming of the harddisks so i don't use the ahci-dont-claim-marvell-chips.patch, so i can't tell about it working or not but it still should do the right thing.)

In case you want to try the older legacy mode for the Intel chipset you can probably avoid the AHCI driver but you will have to select

Code: Select all

<*>   Intel ESB, ICH, PIIX3, PIIX4 PATA/SATA support

and only in case there's a Intel IDE port, you need

Code: Select all

<*>   Intel PATA MPIIX support

too.

[slaterson]

ok, let me see if i understand fully. and a little more info from me...

i do have ahci enabled in my bios, i dual boot this computer with windows xp and want to keep the sata controller in ahci mode for performance reasons, at least in windows.

to get access to all my hard drives - ata dvd burner, ata hard disk and the sata drive - i need to patch my kernel with the 'dont-claim' patch you provided. also, i need to compile it as a module and compile the ahci driver into the kernel. this will give me access to the sata drive and the ata devices (dvd and hdd). my only concern is booting linux - the sata is the drive that gentoo is installed on, so my first thought is that i need the 6121 driver installed not as a module but statically in the kernel.

if i fully understand, a second option is to turn the ahci driver off, only use the 6121 driver (from your previous patch) compiled into the kernel (not as a module) and also turn on piix4 pata/sata support, if i have appropriate hardware. here is my lspci output:

Code: Select all

00:00.0 Host bridge: Intel Corporation DRAM Controller (rev 02)
00:01.0 PCI bridge: Intel Corporation PCI Express Root Port (rev 02)
00:1a.0 USB Controller: Intel Corporation USB UHCI Controller #4 (rev 02)
00:1a.1 USB Controller: Intel Corporation USB UHCI Controller #5 (rev 02)
00:1a.2 USB Controller: Intel Corporation USB UHCI Controller #6 (rev 02)
00:1a.7 USB Controller: Intel Corporation USB2 EHCI Controller #2 (rev 02)
00:1b.0 Audio device: Intel Corporation HD Audio Controller (rev 02)
00:1c.0 PCI bridge: Intel Corporation PCI Express Port 1 (rev 02)
00:1c.4 PCI bridge: Intel Corporation PCI Express Port 5 (rev 02)
00:1c.5 PCI bridge: Intel Corporation PCI Express Port 6 (rev 02)
00:1d.0 USB Controller: Intel Corporation USB UHCI Controller #1 (rev 02)
00:1d.1 USB Controller: Intel Corporation USB UHCI Controller #2 (rev 02)
00:1d.2 USB Controller: Intel Corporation USB UHCI Controller #3 (rev 02)
00:1d.7 USB Controller: Intel Corporation USB2 EHCI Controller #1 (rev 02)
00:1e.0 PCI bridge: Intel Corporation 82801 PCI Bridge (rev 92)
00:1f.0 ISA bridge: Intel Corporation LPC Interface Controller (rev 02)
00:1f.2 IDE interface: Intel Corporation 2 port SATA IDE Controller (rev 02)
00:1f.3 SMBus: Intel Corporation SMBus Controller (rev 02)
00:1f.5 IDE interface: Intel Corporation 2 port SATA IDE Controller (rev 02)
01:00.0 VGA compatible controller: nVidia Corporation Unknown device 0402 (rev a1)
02:00.0 Ethernet controller: Attansic Technology Corp. L1 Gigabit Ethernet Adapter (rev b0)
03:00.0 IDE interface: Marvell Technology Group Ltd. Unknown device 6121 (rev b1)

am i understanding correctly? i am hoping to try this out tomorrow, however the computer is currently processing about 10000 photos which may force me to wait a day or two.

thanks

[emerald]

i do have ahci enabled in my bios, i dual boot this computer with windows xp and want to keep the sata controller in ahci mode for performance reasons, at least in windows.

To keep AHCI mode is a wise decision, imho, not only in windows but Linux too
Since Windows is in AHCI mode switching to legacy mode would probably not even be possible.

to get access to all my hard drives - ata dvd burner, ata hard disk and the sata drive - i need to patch my kernel with the 'dont-claim' patch you provided.

To just gain access you only need the Marvell driver patch, in this case both, AHCI and Marvell driver compiled statically into the kernel. The caveat is, your IDE harddrive will become sda, your SATA harddrive besomes sdb, since the Marvell driver is enumerated first by the kernel.

Since gentoo is installed on the SATA drive you need the AHCI driver statically compiled in.
Now, to keep your SATA harddisk as sda, you need the AHCI driver statically, but patched with the don't claim patch, so the AHCI driver will leave the Marvell chip alone and the Marvell driver, compiled as module, should be inserted by udev later on so your IDE devices become accessible. Since you don't need them to boot it's no problem to have the driver as module.

if i fully understand, a second option is to turn the ahci driver off, only use the 6121 driver (from your previous patch) compiled into the kernel (not as a module) and also turn on piix4 pata/sata support, if i have appropriate hardware.

This - legacy mode - i'm not sure if it really works that way, but it should, use the Marvell driver as module, as before, and use the Intel drivers statically to gain access to your SATA harddrive. But since you might have to switch the harddisk driver in Windows it probably is not an option. (Whether the driver switch works in Windows or not i'm not sure.)

[slaterson]

To keep AHCI mode is a wise decision, imho, not only in windows but Linux too
Since Windows is in AHCI mode switching to legacy mode would probably not even be possible.

yes, i would definitely prefer to kept it enabled, i have noticed much better performance with it on.

To just gain access you only need the Marvell driver patch, in this case both, AHCI and Marvell driver compiled statically into the kernel. The caveat is, your IDE harddrive will become sda, your SATA harddrive besomes sdb, since the Marvell driver is enumerated first by the kernel.

Since gentoo is installed on the SATA drive you need the AHCI driver statically compiled in.
Now, to keep your SATA harddisk as sda, you need the AHCI driver statically, but patched with the don't claim patch, so the AHCI driver will leave the Marvell chip alone and the Marvell driver, compiled as module, should be inserted by udev later on so your IDE devices become accessible. Since you don't need them to boot it's no problem to have the driver as module.

ok, i think i have everything working without applying the dont-claim patch. i have ahci and marvell compiled statically into the kernel, my sata drive becomes sdb which i can live with if i know it is happening and this morning i did some coherent (meaning i was awake and thinking clearly) investigation and my dvd burner is recognised just fine. it to had its named changed somewhere along the line since i last used it.

am i missing anything?

thanks a lot for all the help! i've been trying to get the marvell driver to work since i got this new motherboard a month ago.

[emerald]

am i missing anything?

I don't know if you are missing any hardisks or such

I got the mainboard last wednesday (one week ago) and, lacking of IDE harddisks, only the DVD writer was unusable, it still bothered me so i prepared the driver and updated it do be compatible with 2.6.24 and 2.6.25.

I'm glad you have everything working now

[slaterson]

when i first got mine i had a major problem as no live cds would recognise my hard drives. i finally found the rescuecd that would work and then struggled to get the system to boot as i went through kernel config tweaking for a day or two before i found something that actually worked. now i have been tweaking and testing and trying get better performance, i think after the patch that you provided it is performing rather nicely.

how long before your patch is included in the kernel?

[emerald]

when i first got mine i had a major problem as no live cds would recognise my hard drives. i finally found the rescuecd that would work and then struggled to get the system to boot as i went through kernel config tweaking for a day or two before i found something that actually worked. now i have been tweaking and testing and trying get better performance, i think after the patch that you provided it is performing rather nicely.

I switched from an Athlon64 X2 to Intel Core2 Quad, recompiled the kernel on the old system to support the new hardware, and booted the kernel up in the new system, keeping everything as it was
As live-cd supporting much hardware i can recommend the SystemRescueCD www.sysresccd.org it allows booting a 32 and 64 bit kernel for chrooting too and can be installed to usb stick
I used this one to test the system before i booted up my working system, so i won't destroy anything important.

how long before your patch is included in the kernel?

.. i'd say never since this work is based on the work of a marvell programmer, just not kept up-to-date.

[slaterson]

ok, one more question from something i noticed...

this is a sata ii controller however dmesg shows this:

Code: Select all

ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)

how do i get it to come up at 3.0 Gbps?

thanks!

[emerald]

how do i get it to come up at 3.0 Gbps?

probably using a SATA 2 harddisk:

Code: Select all

ata2: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
ata2.00: ATA-6: WDC WD3200JD-22KLB0, 08.05J08, max UDMA/100
ata2.00: 625142448 sectors, multi 0: LBA48
ata2.00: configured for UDMA/100
ata3: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
ata3.00: ATA-7: SAMSUNG HD753LJ, 1AA01109, max UDMA7
ata3.00: 1465149168 sectors, multi 0: LBA48 NCQ (depth 31/32)
ata3.00: configured for UDMA/133

[Monkeh]

Actually, more likely by removing the tiny jumper on the drive which is limiting the speed.

[slaterson]

i have a sata ii disk. its a seagate ST3500630AS

here is some info:

Code: Select all

#dmesg | grep ata
ata_piix 0000:00:1f.2: version 2.12
ata_piix 0000:00:1f.2: MAP [ P0 -- P1 -- ]
scsi1 : ata_piix
scsi2 : ata_piix
ata1: SATA max UDMA/133 cmd 0xb000 ctl 0xac00 bmdma 0xa480 irq 22
ata2: SATA max UDMA/133 cmd 0xa880 ctl 0xa800 bmdma 0xa488 irq 22
ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
ata1.00: ATA-7: ST3500630AS, 3.AAK, max UDMA/133
ata1.00: 976773168 sectors, multi 16: LBA48 NCQ (depth 0/32)
ata1.00: configured for UDMA/133
ata2: SATA link down (SStatus 0 SControl 300)
ata_piix 0000:00:1f.5: MAP [ P0 -- P1 -- ]
scsi3 : ata_piix
scsi4 : ata_piix
ata3: SATA max UDMA/133 cmd 0xc000 ctl 0xbc00 bmdma 0xb480 irq 22
ata4: SATA max UDMA/133 cmd 0xb880 ctl 0xb800 bmdma 0xb488 irq 22
ata3: SATA link down (SStatus 0 SControl 300)
ata4: SATA link down (SStatus 0 SControl 300)

is the intel piix driver taking over here? disable it in the kernel?

[emerald]

hmm, since the ata_piix takes the controller i wonder if it's running in ahci mode?
deactivate the ata_piix in kernel too so this one won't claim the harddrive controller.

Code: Select all

ata1.00: ATA-7: ST3500630AS, 3.AAK, max UDMA/133 

ATA-7 seems to mean it's a sata 2 drive, so it's probably the legacy driver

[slaterson]

i'll make the kernel update as soon as openoffice is done emerging...

in a previous comment you mentioned upgrading to a quad core... i went from a p4 3.0 ghz (hyperthreading) with 2 gb of ram to a q6600 with 8 gb of ram. the speed difference is jaw dropping. i can do things in 20 minutes with my photography that used to take me a week. literally.

[slaterson]

Actually, more likely by removing the tiny jumper on the drive which is limiting the speed.

i missed this message. i'll have a look at the drive also to see if it has the jumper in place.

thanks!

[emerald]

i'll make the kernel update as soon as openoffice is done emerging...

in a previous comment you mentioned upgrading to a quad core... i went from a p4 3.0 ghz (hyperthreading) with 2 gb of ram to a q6600 with 8 gb of ram. the speed difference is jaw dropping. i can do things in 20 minutes with my photography that used to take me a week. literally.

I went from a Athlon64 X2 4400+ with 4G DDR Ram to a C2Q9450 with 8G DDR3 Ram, the effect was similar for me. Most obvious was Openoffice installation, it took like 8+ hours on the Athlon and on the new system not much more than 1 hour (without ccache).

[Monkeh]

Actually, more likely by removing the tiny jumper on the drive which is limiting the speed.

i missed this message. i'll have a look at the drive also to see if it has the jumper in place.

thanks!

It will have. Seagate drives ship with them in place. Though really, there's little point in removing them.