Re: sili(4) on Sparc64 (Blade 100)

Le 20/10/10 04:28, Nick Holland a icrit :
> On 07/04/10 13:56, E. L. Evans V wrote:
> 
> sorry, took me a while to get around to experimenting with this...
> 
>> Hi.
>>
>> I'm trying to expand internal storage in a Blade 100, most recently
>> running 4.6, will jump to 4.7 if that helps.
>>
>> The firmware for the internal native PATA controller is of course
>> limited to 120G.
> 
> something is, it isn't the firmware.  Firmware limits don't matter after
> boot.  And yes, there is a 128g issue on the blade100, it seems, not
> just the U5/U10.  Apparently, you knew that, I had to find out for
> myself. :)
> 
>> I am looking to add one or more additional large disks internally and I
>> do not need to boot from them.
>>
>> Some reading suggests that a Silicon Image PCI (or PCI-E ?) adapter to
>> use sili(4) may or should work.  Of course I'm limited to PCI on this box.
>>
>> I am using a SYBA SD-SATA2-2E2I PCI SATA II Controller Card PCI-X/PCI
>> SATA card sporting a Silicon Image 3124 device.  I have tried the card
>> in the upper and lower slots, and tried both internal drive connectors
>> on the card.  The Blade has OBP 4.0.45 currently.
> 
> hm. That sounds similar to a very bogus card I picked up recently.
> "6-Port SATA II 150", SiL 3114CTU chip.  I'm not even going to mention
> the RAID part, which I knew was bogus, but just about everything else
> was, uh, "creative".  The "II" after the word "SATA"?  Doesn't mean
> SATA-II.  Pure SATA1.  Thus, the 150 is more "right".  Six port?
> uh..sure.  Only four can be active (four internal, or four internal and
> two external), but there are six ports (somewhere on the box, it did say
> "Four channel").
> 
> However, it works.  pciide, though, not sili.
> 
>> Pre and post hardware installation inspection (manual and diff) of dmesg
>> and eeprom -p output doesn't show anything new or unidentifiable.
>>
>> I cannot at the moment test the card elsewhere, so I suppose I've a duff
>> card or I've misread and it simply isn't supported.
>>
>> Has anyone done something similar as I am attempting?
> 
> actually, just did.
> Had a 1.5TB SATA disk inside my U10.
> It did NOT work with the SiL chip card, just like yours, it just didn't
> show up in my dmesg.
> 
> HOWEVER, an older VIA-based card did show up and work just fine.
> 
> My guess (and I want to highlight the word "guess") is that the U10 is
> maybe an older PCI bus standard.  I've seen this before, an early wi(4)
> card I had would work in only one or two systems I had at the time with
> a newer PCI bus, on the older machines, it just didn't show up.  I've
> also seen this on other cards in the U10, some just don't show up.  I'm
> pretty sure Sun never planned on a lot of the things I put in them, so I
> don't think it is a firmware compatibility issue.
> 
> After having some curious problems with the U10, I've moved the
> controller and drive to my blade100, and sure enough, the card is
> recognized by OpenBSD, and the drive is working seemingly well...
> (and so far, the U10 is behaving itself with the SATA card out of it.
> Too early to say "cause!".)

I just did the same retrofit on one of my U10... first by adding a 256Mb
compact-flash for the boot & root and two sata disk managed by softraid
and two sata racks without disk-tray (open the door, put your bare disk
in there, lock the door).

the via SATA PCI controler is not so bad (2 sata port, 1 pata) and
really cheap but second port activation apparently needs some fiddling
in the configuration register within pciide sources when using 4.6 (and
i didn't had time for this). This chip is aimed to PCI 2.2 but
compatibilty with PCI 2.1 seems correct. I needed two controlers to
setup the configuration...

I replaced the two via controler by a sil3112 PCI controler with 2
internal SATA connectors ( a bare one with no commercial fancy on it but
with a visualy "clean" circuit board) : everything runs smoothly with
this. PCI compatibility is not an issue here in my case.

dmesg extract (you'll notice my lazyness, i still don't have moved the
swap partition to sd0b) :

pciide0 at pci1 dev 3 function 0 "CMD Technology PCI0646" rev 0x03: DMA,
channel
 0 configured to native-PCI, channel 1 configured to native-PCI
pciide0: using ivec 0x7e0 for native-PCI interruptwd0 at pciide0 channel
0 drive 0: <TOSHIBA THNCF256MMA>
wd0: 1-sector PIO, LBA, 244MB, 500736 sectors
wd0(pciide0:0:0): using PIO mode 2
pciide0: channel 1 disabled (no drives)
ppb1 at pci0 dev 1 function 0 "Sun Simba PCI-PCI" rev 0x13
pci2 at ppb1 bus 2
pciide1 at pci2 dev 3 function 0 "CMD Technology SiI3112 SATA" rev 0x02: DMA
pciide1: using ivec 0x7d8 for native-PCI interrupt
pciide1: port 0: device present, speed: 1.5Gb/s
wd1 at pciide1 channel 0 drive 0: <Hitachi HDS721050CLA362>
wd1: 16-sector PIO, LBA48, 476940MB, 976773168 sectors
wd1(pciide1:0:0): using BIOS timings, Ultra-DMA mode 2
pciide1: port 1: device present, speed: 1.5Gb/s
wd2 at pciide1 channel 1 drive 0: <Hitachi HDS721050CLA362>
wd2: 16-sector PIO, LBA48, 476940MB, 976773168 sectors
wd2(pciide1:1:0): using BIOS timings, Ultra-DMA mode 2
softraid0 at root
scsibus0 at softraid0: 1 targets
sd0 at scsibus0 targ 0 lun 0: <OPENBSD, SR RAID 0, 003> SCSI2 0/direct fixed
sd0: 953879MB, 512 bytes/sec, 1953545985 sec total
bootpath: /pci@1f,0/pci@1,1/ide@3,0/disk@0,0
root on wd0a swap on wd0b dump on wd0b