SiI 3112(A) controller cards get stuck on HDD initialization
November 6th, 2015, 18:26
Well, this has gone far enough. 
I hereby declare that, should I ever buy a controller card for the PCI port again, it will never be a Si3112/3112A again.
This is simply hard to comprehend, and I would really like some pros here to speak up WHAT the cause of this is (or might be).
Currently I have to say: in some way, it is the brand.
Most Seagate and (genuine) Maxtor drives (S/N ending in ...R0 or ...F0) work with the flakey chipset on these controller cards, except for 1+ TB Maxtor-branded ones (whilst 500 GB STM... one worked fine).
All the rest do not work. RAID or non-RAID does not matter. Nor does size/capacity. A 250 GB drive might get stuck the very same way if you're "lucky".
It's always the same thing. So a WDC Caviar Black drive would not be initialized the following way:
but...
and get stuck FOREVER, with the cursor placed where the underscore is.
What the HECK is this?!? And why does it always happen with that SiI chipset but never with VIA or Promise ones?
I dug in my H/W plastic box and found yet another Adaptec 2-port SATA controller card - same damn thing.
My eyes went wide open when I realized there was the same cr@ppy SiI chipset on there.
(well, it was because just did not expect that! An old Adaptec IDE expansion card I still have has got HighPoint (HPT) chipset, which was really the definition of quality)
I hereby declare that, should I ever buy a controller card for the PCI port again, it will never be a Si3112/3112A again.
This is simply hard to comprehend, and I would really like some pros here to speak up WHAT the cause of this is (or might be).
Currently I have to say: in some way, it is the brand.
Most Seagate and (genuine) Maxtor drives (S/N ending in ...R0 or ...F0) work with the flakey chipset on these controller cards, except for 1+ TB Maxtor-branded ones (whilst 500 GB STM... one worked fine).
All the rest do not work. RAID or non-RAID does not matter. Nor does size/capacity. A 250 GB drive might get stuck the very same way if you're "lucky".
It's always the same thing. So a WDC Caviar Black drive would not be initialized the following way:
- Code:
[1] WDC1001FALS ............................. 1024 GB
but...
- Code:
[1] WDC1001FALS ___
and get stuck FOREVER, with the cursor placed where the underscore is.
What the HECK is this?!? And why does it always happen with that SiI chipset but never with VIA or Promise ones?
I dug in my H/W plastic box and found yet another Adaptec 2-port SATA controller card - same damn thing.
My eyes went wide open when I realized there was the same cr@ppy SiI chipset on there.
(well, it was because just did not expect that! An old Adaptec IDE expansion card I still have has got HighPoint (HPT) chipset, which was really the definition of quality)
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 15th, 2016, 20:04
This is peculiar.
It seems with some old BIOSes, these SiI 3112A controllers have problems with HDDs of 1 TB and more in size?
(Nearly) any "smaller-size" drives (320, 400, 500, 750 GB) worked perfectly so far.
But once I connect a 1 TB one to it (e. g. a Toshiba DT01ACA100) the behavior is as described.
These 1 TB drives even DO work when hot-plugged to that card! However, it's a high risk as the system will very likely lock up fairly soon (as it did here on Linux).
Why, let me think: historically, there WAS a limitation of LBA that I clearly remember, but that was at 137 GB (drives that crossed this boundary with their size had to be treated specially by Windows 2000 and previous OS releases), however I can't remember that there was any other later one recently.
It seems with some old BIOSes, these SiI 3112A controllers have problems with HDDs of 1 TB and more in size?
(Nearly) any "smaller-size" drives (320, 400, 500, 750 GB) worked perfectly so far.
But once I connect a 1 TB one to it (e. g. a Toshiba DT01ACA100) the behavior is as described.
These 1 TB drives even DO work when hot-plugged to that card! However, it's a high risk as the system will very likely lock up fairly soon (as it did here on Linux).
Why, let me think: historically, there WAS a limitation of LBA that I clearly remember, but that was at 137 GB (drives that crossed this boundary with their size had to be treated specially by Windows 2000 and previous OS releases), however I can't remember that there was any other later one recently.
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 15th, 2016, 22:14
On the WD drives you can reduce the SATA link rate by jumpering OPT1. Have you tried that?
I remember that some NVIDIA drivers have a 31-bit LBA limit, so drives larger than 1TiB would have their capacity truncated, but I'm not aware of any such problem at 1TB. There is an old GigaByte BIOS bug that reduces the capacity of 1TB drives to 30MB or so, but that's a different thing.
I remember that some NVIDIA drivers have a 31-bit LBA limit, so drives larger than 1TiB would have their capacity truncated, but I'm not aware of any such problem at 1TB. There is an old GigaByte BIOS bug that reduces the capacity of 1TB drives to 30MB or so, but that's a different thing.
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 16th, 2016, 6:27
I don't know if it applies in your case, but it appears that there are bugs which affect certain Seagate drives attached to Sil3112 and Sil3114 controllers. These bugs manifest themselves under heavy usage. There is a corresponding blacklist in Linux for this combination.
It may be worth disabling NCQ. I've seen NCQ bugs in various HDD brands, some of which were addressed in blacklists.
http://www.google.com/search?q=%22SiI+3 ... 2A+seagate
https://lists.debian.org/debian-user/20 ... 01226.html
http://web.archive.org/web/200703272306 ... 21928.html
http://osdir.com/ml/linux.ide/2005-05/msg00080.html
https://www.redhat.com/archives/rhl-lis ... 04835.html
It may be worth disabling NCQ. I've seen NCQ bugs in various HDD brands, some of which were addressed in blacklists.
http://www.google.com/search?q=%22SiI+3 ... 2A+seagate
https://lists.debian.org/debian-user/20 ... 01226.html
http://web.archive.org/web/200703272306 ... 21928.html
http://osdir.com/ml/linux.ide/2005-05/msg00080.html
https://www.redhat.com/archives/rhl-lis ... 04835.html
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 16th, 2016, 10:10
Great post fzabkar, unfortunately, there is no way to disable NCQ at that stage. Remember it hangs at boot time, when the BIOS initializes my SiI card and tries to locate the drive(s) connected there.
This would have to be done via hardware mod (e. g. a jumper), and I don't remember any drive that allows you to disable NCQ via jumper. Because at the stage where it happens first, I can't do anything at all about it.
This would have to be done via hardware mod (e. g. a jumper), and I don't remember any drive that allows you to disable NCQ via jumper. Because at the stage where it happens first, I can't do anything at all about it.
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 16th, 2016, 22:54
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 18th, 2016, 13:42
Not to a great extent IMHO, but many thanks for the link - it was a most interesting read.
Looks like these SiI3112'ers are actually bug hell, with a couple other issues likely to occur.
However, it's hard to look on e-bay for those that are NOT 3112/3114 - they're way too widespread. And, for instance, with the VIA ones, I've found out that my mainboard is allergic to two of them at a time...it's as if they're trying to compete in who's the more important one when putting both of them in.
In other words, it's wise that you do mix various vendors when you use 2 or more at a time, avoiding the use of identical ones at all costs - unless you're fond of some funny surprises, that is.
Looks like these SiI3112'ers are actually bug hell, with a couple other issues likely to occur.
However, it's hard to look on e-bay for those that are NOT 3112/3114 - they're way too widespread. And, for instance, with the VIA ones, I've found out that my mainboard is allergic to two of them at a time...it's as if they're trying to compete in who's the more important one when putting both of them in.
In other words, it's wise that you do mix various vendors when you use 2 or more at a time, avoiding the use of identical ones at all costs - unless you're fond of some funny surprises, that is.
Re: SiI 3112(A) controller cards get stuck on HDD initializa
January 19th, 2016, 5:58
SUCCESS!
Franc, a long nightmare is finally OVER!
"BIOS" was the right keyword; somewhere deep inside the Tom's Hardware thread you had pointed me to, I got the idea of what to do, and it's a success story!
OH MY GOSH.
There was really an ANCIENT bios flashed onto one of those cards -- a SiI 3114 4-port one by the way --, to be exact: version 5.0.xx (!!) from the very beginnings of the 2000's, and Silicon Image are offering the 5.4.0.3 one for add-in cards with that chipset (choose b5403.bin for non-RAID, for instance).
And lo' and behold, no more hangs!!
So thanks again for your help, even though you've helped me indirectly by directing me to the right thing to do.
Well, I still do consider myself an experienced user - but where did my thinking fail?
- Why, it's easy to answer. I'm a sort of old-schooler, and since VGA BIOS updates were rare as hen's teeth for decades (merely workarounds for card BIOS shortcomings were available by subsequent driver updates), I was almost sure that mass storage add-in cards would have the same "fate" (i. e. you'd be doomed to use the BIOS flashed onto that card by the manufacturer/vendor, and buying a used one off ebay would mean that you'd play a kind of lottery trying to get hold of one with a recent BIOS).
Franc, a long nightmare is finally OVER!
"BIOS" was the right keyword; somewhere deep inside the Tom's Hardware thread you had pointed me to, I got the idea of what to do, and it's a success story!
OH MY GOSH.
There was really an ANCIENT bios flashed onto one of those cards -- a SiI 3114 4-port one by the way --, to be exact: version 5.0.xx (!!) from the very beginnings of the 2000's, and Silicon Image are offering the 5.4.0.3 one for add-in cards with that chipset (choose b5403.bin for non-RAID, for instance).
And lo' and behold, no more hangs!!
So thanks again for your help, even though you've helped me indirectly by directing me to the right thing to do.
Well, I still do consider myself an experienced user - but where did my thinking fail?
- Why, it's easy to answer. I'm a sort of old-schooler, and since VGA BIOS updates were rare as hen's teeth for decades (merely workarounds for card BIOS shortcomings were available by subsequent driver updates), I was almost sure that mass storage add-in cards would have the same "fate" (i. e. you'd be doomed to use the BIOS flashed onto that card by the manufacturer/vendor, and buying a used one off ebay would mean that you'd play a kind of lottery trying to get hold of one with a recent BIOS).
Powered by phpBB © phpBB Group.