external eSata enclosure drive recognition?

[Bill in Co.]

I just added an external eSata HD (in a Vantec enclosure) to my new Dell
system, but it is only "recognized" after rebooting with it turned on FIRST.

IOW, if the external eSata HD enclosure unit is OFF and I power up the
computer, and then I turn on the external eSata HD enclosure, it isn't
recognized (not properly recognized, but there at least is some disk
activity noted after turning it on).

The HD drive I put into the external enclosure was a new Western Digital
eSata drive, and it has no jumpers, but maybe that's normal for Sata drives
these days.

My question is: how do I tell if this weird problem is due to the hard
drive itself, or the external hard drive enclosure, or possibly even the
BIOS?

It's a real nuisance to have to first turn on the external unit BEFORE
turning on the computer to get it to be recognized. It should be able to
be recognized after the computer has booted up and THEN I turn it on. But
it isn't for some reason.

 

[Andrew E.]

By default SATA controller is off in most BIOS settings from the mfg,turn
it on & set to use with IDE hds.Also,if SATA is on in BIOS,then also by
default
any new hd installed to pc & recognized by BIOS,usually gets selected as
First
boot hd priority,set BIOS properly in BIOS before xp starts to load.

 

[Bill in Co.]

Well, the BIOS settings are on, so that's not the problem, as I checked it
again.

What is so weird is that the external eSata drive works fine (and is
properly recognized in windows) IF I initially turn it on, before booting up
the computer.

(Or if not, and I forgot to, I have to reboot after first turning it on).

And then if I turn it off while in windows, nothing appears to happen (it
still shows up in windows explorer, as if nothing happened). Which is
completely inaccurate, because when I turn it off, it IS gone.

Any other ideas? Maybe it has to do with some weirdness in the SATA II
spec.
This is a SATA II drive, but the Dell spec says its compatible with SATA I
and II.

 

[Bill in Co.]

Addendum.
There ought to be SOME way to narrow this pecularity down to being either 1)
an external eSata HD enclosure problem, or 2) the eSata hard drive itself,
or 3) a BIOS problem, or 4) a windows "communication" problem (in
continually checking the BIOS for connect/disconnect of the external drive).

(or maybe it's a combo of two of the above).

Or at least that's my understanding.

 

[Paul]

Addendum.
There ought to be SOME way to narrow this pecularity down to being either 1)
an external eSata HD enclosure problem, or 2) the eSata hard drive itself,
or 3) a BIOS problem, or 4) a windows "communication" problem (in
continually checking the BIOS for connect/disconnect of the external drive).

(or maybe it's a combo of two of the above).

Or at least that's my understanding.

You're missing "hotplug". It takes a combination of the right setting
for the port in the BIOS, and driver to go with it, to get support.
If it is working, then you'd see a "Safely Remove" icon at the bottom
of your Windows screen. AFAIK, a default IDE driver for a SATA drive,
doesn't have hotplug or hotswap capability, and it takes another mode
such as AHCI or RAID mode, to fix it.

If the ESATA port on the computer, is hosted by a separate chip, it
should be relatively painless to fix.

Hot-plugging is mentioned here, in this AHCI article.

http://en.wikipedia.org/wiki/Advanced_Host_Controller_Interface

Moving the boot drive, from one controller on the motherboard, to
another controller, or to a separate controller purchased and
installed in a PCI slot, may allow making changes to the driver
and BIOS setting. It all depends on where the ESATA port is hosted,
as to how messy the fix would be. That is why I mentioned a "separate
chip" issue as being easier to fix, as there is no interaction with
the boot drive and its driver, while you're fixing it.

Paul

 

[Bill in Co.]

You're missing "hotplug". It takes a combination of the right setting
for the port in the BIOS, and driver to go with it, to get support.
If it is working, then you'd see a "Safely Remove" icon at the bottom
of your Windows screen. AFAIK, a default IDE driver for a SATA drive,
doesn't have hotplug or hotswap capability, and it takes another mode
such as AHCI or RAID mode, to fix it.

Thanks Paul. I did some more reading on this last nite, and apparently
one needs AHCI to support this "hot plugging" capability. And since I only
have the older Intel ICH9 controller, it isn't supported, nor are drivers
available for AHCI from Intel that will update it (although there is an
unsupported patch for AHCI at one of the MSDN sites that *supposedly* will
allow the standard ICH9 driver to be patched to get AHCI, instead of just
IDE emulation of SATA.

If the ESATA port on the computer, is hosted by a separate chip, it
should be relatively painless to fix.

It seems it is not upgradeable with a driver from Intel (I mean for the
standard ICH9 controller; I read about that over at the Intel site, and its
requirements).

Hot-plugging is mentioned here, in this AHCI article.

http://en.wikipedia.org/wiki/Advanced_Host_Controller_Interface

Yup, that's one of the articles I read last night - thanks.
I think it also referenced that unsupported AHCI patch (for non RAID Intel
controller), too.

Moving the boot drive, from one controller on the motherboard, to
another controller, or to a separate controller purchased and
installed in a PCI slot, may allow making changes to the driver
and BIOS setting. It all depends on where the ESATA port is hosted,
as to how messy the fix would be.

That's what I'm getting, when reading about this some more. It's seems it
can be a bit problematic - and a bit risky.

That is why I mentioned a "separate
chip" issue as being easier to fix, as there is no interaction with
the boot drive and its driver, while you're fixing it.

Well, I'm just not sure about the "separate chip" thing here that your
mentioning.

The only thing I can tell you is that I evidently have an Intel ICH9
controller on the MB, and NOT the newer Intel ICH9R (etc) series, that does
support AHCI and has hot plugging capability (since it is not just an IDE
emulation of SATA).

I can see that in BIOS too, since the only selection for SATA is the IDE
emulation one. In other words, no AHCI.

(I'm not sure if I can get around this by adding a PCI controller card (to
allow for AHCI). But perhaps - but I think it's a bit risky, given
what's already set up and configured on the MB (the ICH9).

 

[Paul]

Thanks Paul. I did some more reading on this last nite, and apparently
one needs AHCI to support this "hot plugging" capability. And since I only
have the older Intel ICH9 controller, it isn't supported, nor are drivers
available for AHCI from Intel that will update it (although there is an
unsupported patch for AHCI at one of the MSDN sites that *supposedly* will
allow the standard ICH9 driver to be patched to get AHCI, instead of just
IDE emulation of SATA.


It seems it is not upgradeable with a driver from Intel (I mean for the
standard ICH9 controller; I read about that over at the Intel site, and its
requirements).


Yup, that's one of the articles I read last night - thanks.
I think it also referenced that unsupported AHCI patch (for non RAID Intel
controller), too.


That's what I'm getting, when reading about this some more. It's seems it
can be a bit problematic - and a bit risky.


Well, I'm just not sure about the "separate chip" thing here that your
mentioning.

The only thing I can tell you is that I evidently have an Intel ICH9
controller on the MB, and NOT the newer Intel ICH9R (etc) series, that does
support AHCI and has hot plugging capability (since it is not just an IDE
emulation of SATA).

I can see that in BIOS too, since the only selection for SATA is the IDE
emulation one. In other words, no AHCI.

(I'm not sure if I can get around this by adding a PCI controller card (to
allow for AHCI). But perhaps - but I think it's a bit risky, given
what's already set up and configured on the MB (the ICH9).

You could add a card to the machine, if there is a slot empty. For example,
a SIL3132 or other SATA 3Gbit/sec chip, would possibly do the job. Some
are set up as ESATA cards, with one or two connectors on the faceplate.
The SIL3132 has a PCI Express x1 interface, and isn't a plain PCI chip.

You can add a card like that, and install a driver, without having to
do anything to the rest of your setup.

I didn't see mention of the Dell model number of your machine, and maybe
I could look at the motherboard picture, to see what other options are
available.

Paul

 

[Bill in Co.]

You could add a card to the machine, if there is a slot empty. For
example,
a SIL3132 or other SATA 3Gbit/sec chip, would possibly do the job. Some
are set up as ESATA cards, with one or two connectors on the faceplate.
The SIL3132 has a PCI Express x1 interface, and isn't a plain PCI chip.

You can add a card like that, and install a driver, without having to
do anything to the rest of your setup.

I do have a PCI and PCI Express slot available. But right now, I simply
put the eSata connector cable and mounting bracket in there, so that I can
plug in my external eSata HD enclosure into the computer in the back, using
the SATA cables.

I didn't see mention of the Dell model number of your machine, and maybe

I have a Dell Inspiron 530, and I am using Windows XP Home Edition.
I know it has the ICH9 controller, at least, and NOT the newer ICH9R, etc,
which has AHCI)

I could look at the motherboard picture, to see what other options are
available.

Thanks.
I also reread some more about that MSFN unofficial AHCI patch again, at this
web site (referenced in that wiki article you mentioned), but am slightly
wary of this. But, who knows??? And it is at one of the MSFN sites, so
that ought to count for something. Here is the link:

http://www.msfn.org/board/How-to-enable-AHCI-on-Intel-ICH9-under-X-t109450.html

If that patch worked, it sure would be a cheap way to get AHCI installed, as
it "just" involves updating some files (uses a few tweaks to get AHCI to be
"allowed").

But your idea of a separate controller card sounds perhaps better, and
safer, I guess, albeit more expensive. Although I'm not sure if it would
mess up any of my other SATA stuff (like the primary HD, and possibly the
DVD drive, although they are connected I presume through the ICH9 controller
on the MB, which hopefully not conflict in any way with another controller
being added to the system. I'm not sure how the Dell BIOS would handle or
even recognize this changeover, however. As it is now, the Dell BIOS only
shows the option for IDE emulation of SATA, since that goes with the Intel
MB and chipsets I have on this Dell 530, presumably.

 

[Paul]

I do have a PCI and PCI Express slot available. But right now, I simply
put the eSata connector cable and mounting bracket in there, so that I can
plug in my external eSata HD enclosure into the computer in the back, using
the SATA cables.


I have a Dell Inspiron 530, and I am using Windows XP Home Edition.
I know it has the ICH9 controller, at least, and NOT the newer ICH9R, etc,
which has AHCI)


Thanks.
I also reread some more about that MSFN unofficial AHCI patch again, at this
web site (referenced in that wiki article you mentioned), but am slightly
wary of this. But, who knows??? And it is at one of the MSFN sites, so
that ought to count for something. Here is the link:

http://www.msfn.org/board/How-to-enable-AHCI-on-Intel-ICH9-under-X-t109450.html

If that patch worked, it sure would be a cheap way to get AHCI installed, as
it "just" involves updating some files (uses a few tweaks to get AHCI to be
"allowed").

But your idea of a separate controller card sounds perhaps better, and
safer, I guess, albeit more expensive. Although I'm not sure if it would
mess up any of my other SATA stuff (like the primary HD, and possibly the
DVD drive, although they are connected I presume through the ICH9 controller
on the MB, which hopefully not conflict in any way with another controller
being added to the system. I'm not sure how the Dell BIOS would handle or
even recognize this changeover, however. As it is now, the Dell BIOS only
shows the option for IDE emulation of SATA, since that goes with the Intel
MB and chipsets I have on this Dell 530, presumably.

A separate card can be used two ways. It can be used for a "data only" disk,
one you don't care to boot from. In that case, it doesn't matter whether
the BIOS chip on the separate card, loads or not.

If you want to boot from a drive connected to a separate card, then the BIOS
chip on the card has to load. The BIOS does stuff like provide an extended
INT 0x13 routine, which provides the ability to read sectors from the disk
during bootup. Effectively, it is like a driver during boot time. Some
motherboard BIOS have an option called "Interrupt 19 capture". The difference
between the two statements I just made, is 0x13 is hexadecimal (1*16 + 3),
while the second statement uses a decimal number (19). Why the industry cannot
stick with one way of stating it, eludes me. Asus, in their BIOS, should really
refer to it as 0x13. I mean, when I first saw that setting, I didn't know
what it was for. Later, a little clue light bulb went on, that 0x13 and
19 were the same thing.

"Capture", means to load the BIOS on any separate cards. So my motherboard
has an option to disable all add-on ROMs, or allow them to be loaded.
If you allow them to be loaded, then the separate add-in cards can be
used for booting the computer.

On an add-in card that supports RAID, the BIOS chip on the card can even
include BIOS screens for declaring RAID arrays and the like. So the add-in
BIOS can add a user interface, for basic management functions before booting.

The Silicon Image cards I've seen, all have a BIOS chip on them. It can even
be reflashed, to change the functionality (RAID or non-RAID).

While I like the MSFN article you found, and don't doubt it would work,
I guess I look at the results, and ask how many situations will arise
at some future date, where I'd regret doing it. I keep very simple setups
here.

1) Ports set to vanilla IDE mode. This allows the default Microsoft driver
to load, so I don't need to press F6 during an install.
2) No dynamic disks.
3) No drive overlays.
4) No RAID. (If you're going to use RAID, then it pays to learn how to use
it. I recommend practicing repairing the array, installing a new disk and
the like, before real data goes on it. Nothing worse than that feeling
of panic, when the BIOS pops up a warning that a RAID array has "failed".
You don't want to learn how RAID really works, with live data on the disks
and no backup.)

Another way to look at it, is adding the ESATA card, means your computer
now supports up to six storage devices, rather than the four you had
previously.

About the only downside of add-in cards, is if there are too many add-in
BIOS ROMs in the system, you can actually run out of low memory to load
them all. Tyically 128KB, in the 0-640KB region, is reserved for add-in
BIOS ROMs to load, which is not a lot of space. The BIOS ROM on a video
card, can eat 64KB of that. Code loaded in that area, has a "shrinking"
function, where excess junk is jettisoned, so it is possible for
more stuff to fit in, if an add-in ROM is well written. (It means the
add-on ROM code is bigger initially, and after any initialization is
finished, the code can shrink to just the bits that are needed.)

As I understand it, the add-in ROMs are loaded in slot order, so the
card next to the processor is loaded first, then the next card, and so
on. That rule applies, as long as the motherboard designer set up the
address map properly, to make that happen. That is why there is a
recommendation, when there is trouble, to put the card you plan
to boot from, in slot #1. Now that computers have so many flavors of
slots, that isn't as practical a suggestion as it used to be.

I don't see any other storage options in the diagram here. As you note,
there is a PCI Express x1 connector, but you can only use it, if the
video card is thin enough. Some video cards are double slot width,
which would cover up the connector.

http://support.dell.com/support/edocs/systems/inspd530/EN/OM/parts.htm#wp1184332

Paul

 

[Bill in Co.]

A separate card can be used two ways. It can be used for a "data only"
disk,
one you don't care to boot from. In that case, it doesn't matter whether
the BIOS chip on the separate card, loads or not.

That would be my case, since I'm only using with an external HD enclosure
for image backup purposes of my system partition.

Still, when you boot up, I would think the system BIOS would need to
recognize this controller too, and show at least some other options in there
(in BIOS). Or maybe it does, but that's beside the point for what I'm
doing here (not booting with it).

If you want to boot from a drive connected to a separate card,

Not.

then the BIOS chip on the card has to load.

Seems like the BIOS chip on that PCI card would have to load anyways, just
to allow it and its controller to function, with any connected device. I
must not be understanding something then.

The BIOS does stuff like provide an extended
INT 0x13 routine, which provides the ability to read sectors from the disk
during bootup. Effectively, it is like a driver during boot time. Some
motherboard BIOS have an option called "Interrupt 19 capture". The
difference
between the two statements I just made, is 0x13 is hexadecimal (1*16 + 3),
while the second statement uses a decimal number (19). Why the industry
cannot
stick with one way of stating it, eludes me. Asus, in their BIOS, should
really refer to it as 0x13. I mean, when I first saw that setting, I
didn't
know what it was for. Later, a little clue light bulb went on, that 0x13
and
19 were the same thing.

Yeah. 13H = 16+3 = 19 decimal.
(Speaking of industry "standards", or rather the lack thereof, the gigabyte
ratings of most of the computer world (binary) VS that used by the hard
drive manufacturers, still bugs me.

"Capture", means to load the BIOS on any separate cards. So my motherboard
has an option to disable all add-on ROMs, or allow them to be loaded.

I'm not sure if I have that option, but if memory serves me right, I don't.

If you allow them to be loaded, then the separate add-in cards can be
used for booting the computer.

On an add-in card that supports RAID, the BIOS chip on the card can even
include BIOS screens for declaring RAID arrays and the like.

I don't need or want RAID, but then again, know little about it, anyways.
(I have noticed that this RAID capability often comes hand in hand with AHCI
controller stuff, however).

All I want is the hotplugable feature. That's the only reason I want
AHCI, so that I can turn on and off my external eSata drive at will, and
have it recognized that way.

So the add-in
BIOS can add a user interface, for basic management functions before
booting.

The Silicon Image cards I've seen, all have a BIOS chip on them.

I would expect they would have to to support their controller.

It can even
be reflashed, to change the functionality (RAID or non-RAID).

While I like the MSFN article you found, and don't doubt it would work,
I guess I look at the results, and ask how many situations will arise
at some future date, where I'd regret doing it.

Yeah, and that's a good point. And I think in rereading that article
(there are several pages) that one or two had reported such problems, along
with some other problems.

I keep very simple setups here.

1) Ports set to vanilla IDE mode. This allows the default Microsoft driver
to load, so I don't need to press F6 during an install.

OK, that sounds good. Although I don't know what this "needing to press
<F6> during install" thing is that they're talking about, but I saw that
mentioned before).

2) No dynamic disks.
3) No drive overlays.
4) No RAID. (If you're going to use RAID, then it pays to learn how to use
it. I recommend practicing repairing the array, installing a new disk
and
the like, before real data goes on it. Nothing worse than that feeling
of panic, when the BIOS pops up a warning that a RAID array has
"failed".
You don't want to learn how RAID really works, with live data on the
disks
and no backup.)

I agree - I don't want ANY of the above. I like to keep things simple,
too.

Another way to look at it, is adding the ESATA card, means your computer
now supports up to six storage devices, rather than the four you had
previously.

Which for me is a bit overkill, but whatever.

I believe I've got the ICH9 Intel 2920 and 2926 controllers, and the 2926 is
a two port controller (but not in use now, I guess), and the 2920 is a four
port controller (being used by my internal SATA drive and my DVD drive).

About the only downside of add-in cards, is if there are too many add-in
BIOS ROMs in the system, you can actually run out of low memory to load
them all. Tyically 128KB, in the 0-640KB region, is reserved for add-in
BIOS ROMs to load, which is not a lot of space. The BIOS ROM on a video
card, can eat 64KB of that.

Wow. But I expect that's for a higher end card, too, and that a more basic
card wouldn't use that full amount.

Code loaded in that area, has a "shrinking"
function, where excess junk is jettisoned, so it is possible for
more stuff to fit in, if an add-in ROM is well written. (It means the
add-on ROM code is bigger initially, and after any initialization is
finished, the code can shrink to just the bits that are needed.)

As I understand it, the add-in ROMs are loaded in slot order, so the
card next to the processor is loaded first, then the next card, and so
on. That rule applies, as long as the motherboard designer set up the
address map properly, to make that happen. That is why there is a
recommendation, when there is trouble, to put the card you plan
to boot from, in slot #1.

The card you plan to boot from? You mean, assuming you are adding a
controller card for the boot drive, I presume. (Is that really that
common? maybe, if people want higher performance, I guess)

Now that computers have so many flavors of
slots, that isn't as practical a suggestion as it used to be.

I don't see any other storage options in the diagram here.

??? I have both free PCI slots and PCI express slots at this point.

As you note,
there is a PCI Express x1 connector, but you can only use it, if the
video card is thin enough. Some video cards are double slot width,
which would cover up the connector.

I think it's clear. I'll take a look inside again. So at this point it
seems I could add a
PCI to external eSata II port card (seems most have two ports, but I only
need one).
Right?

And it could be EITHER a regular PCI card, OR a PCI express card.

Not sure which slot would be better to tie up with that card (leaving the
others free for something else later on). I assume it's a lot easier
finding std PCI cards, so maybe I should tie up the express slot, if I find
that card. I've been looking at amazon.com, and found some (not too
expensive, either).

http://support.dell.com/support/edocs/systems/ inspd530/EN/OM/parts.htm#wp1184332

Paul

Thanks again!

 

[Paul]

That would be my case, since I'm only using with an external HD enclosure
for image backup purposes of my system partition.

Still, when you boot up, I would think the system BIOS would need to
recognize this controller too, and show at least some other options in there
(in BIOS). Or maybe it does, but that's beside the point for what I'm
doing here (not booting with it).


Not.


Seems like the BIOS chip on that PCI card would have to load anyways, just
to allow it and its controller to function, with any connected device. I
must not be understanding something then.


Yeah. 13H = 16+3 = 19 decimal.
(Speaking of industry "standards", or rather the lack thereof, the gigabyte
ratings of most of the computer world (binary) VS that used by the hard
drive manufacturers, still bugs me.


I'm not sure if I have that option, but if memory serves me right, I don't.


I don't need or want RAID, but then again, know little about it, anyways.
(I have noticed that this RAID capability often comes hand in hand with AHCI
controller stuff, however).

All I want is the hotplugable feature. That's the only reason I want
AHCI, so that I can turn on and off my external eSata drive at will, and
have it recognized that way.


I would expect they would have to to support their controller.


Yeah, and that's a good point. And I think in rereading that article
(there are several pages) that one or two had reported such problems, along
with some other problems.



OK, that sounds good. Although I don't know what this "needing to press
<F6> during install" thing is that they're talking about, but I saw that
mentioned before).


I agree - I don't want ANY of the above. I like to keep things simple,
too.


Which for me is a bit overkill, but whatever.

I believe I've got the ICH9 Intel 2920 and 2926 controllers, and the 2926 is
a two port controller (but not in use now, I guess), and the 2920 is a four
port controller (being used by my internal SATA drive and my DVD drive).


Wow. But I expect that's for a higher end card, too, and that a more basic
card wouldn't use that full amount.


The card you plan to boot from? You mean, assuming you are adding a
controller card for the boot drive, I presume. (Is that really that
common? maybe, if people want higher performance, I guess)


??? I have both free PCI slots and PCI express slots at this point.


I think it's clear. I'll take a look inside again. So at this point it
seems I could add a
PCI to external eSata II port card (seems most have two ports, but I only
need one).
Right?

And it could be EITHER a regular PCI card, OR a PCI express card.

Not sure which slot would be better to tie up with that card (leaving the
others free for something else later on). I assume it's a lot easier
finding std PCI cards, so maybe I should tie up the express slot, if I find
that card. I've been looking at amazon.com, and found some (not too
expensive, either).


Thanks again!

OK. When the BIOS starts, it enumerates the add-on cards. It visits them
one at a time, and reads the configuration info from the main chip. No
BIOS chip is needed on the card to do that. You could unsolder the BIOS
chip from the card, and the only consequence, is you cannot boot from it.

The BIOS will effectively ignore a PCI card, after that point. It can
allocate a position in the address map, give the card an IRQ, and then
it is done with the card. In such a scenario, where you had a card with
a main chip on it, and no add-on BIOS ROM, the BIOS would ignore
the card, and could not boot from it.

If a BIOS ROM is detected, while the BIOS is doing its probes, then the
motherboard BIOS can attempt to load the add-in BIOS. That is, in a
sense, like loading a driver at the BIOS level, as it adds some simple
functions. Based on that added function, the BIOS then has more disks
in the potential boot order. The INT 0x13 code adds the bootability
function.

Now, when the OS starts up, it loads a bunch of its own drivers too. The
driver tells it how to access the functions on the card. Just like in the
BIOS scenario, if a card is present, but has no drivers, it is
effectively ignored and invisible. (Just an entry in Device Manager, and
an exclamation mark saying something isn't right.) From then on, you
could use the computer, and the add-in card without Windows driver won't
be doing anything. It is like so much dead weight.

When I referred to your motherboard, I was referring to the fact that
the only storage chip is the ICH9, with four ports. So there isn't an
additional chip, like a Jmicron chip on the motherboard. If there had
been such a chip present, it might have meant another option for
connecting disk drives. You do have expansion slots, and that is
how you can add storage capability.

With regard to PCI cards versus PCI Express, you may find that only
PCI Express has SATA ports at 3Gbit/sec. The older PCI cards have
SATA at 1.5Gbit/sec and no "hot plug". If I was solving the problem
you were working on, I'd be using a PCI Express x1 card, with SATA
ports rated at 3Gbit/sec. The difference is, the driver for the
3Gbit/sec chips, tends to support hot plugging, while the others
might not. There may be a hack for making the other ones do it,
but you know my answer about "keeping it simple"

This card is flashed with non-RAID BIOS, and has two ESATA connectors. $25.
Install it, install the driver, then you should see a "Safely Remove"
icon, when you connect the ESATA drive.

"Rosewill RC-219 Silicon Image PCI Express External eSATA II x2 NCQ non-RAID"
http://www.newegg.com/Product/Product.aspx?Item=N82E16816132020

Paul

 

[Bill in Co.]

OK. When the BIOS starts, it enumerates the add-on cards. It visits them
one at a time, and reads the configuration info from the main chip.

From the "main chip" on the add-on card, I presume that means. (Not sure
what the "main chip" is, but I guess that depends on the nature of the
card).

No BIOS chip is needed on the card to do that. You could unsolder the BIOS
chip from the card, and the only consequence, is you cannot boot from it.

Boot from it? What kind of "cards" could you "boot from"? Maybe you're
referring specifically to a controller card connected to a drive? And do
you mean you could use that type of card without invoking its own BIOS (for
that specific case) IF you didn't want to use it for a boot drive)?

The BIOS will effectively ignore a PCI card, after that point.

After it has been initially enumerated - I assume that's what you mean)

It can allocate a position in the address map, give the card an IRQ, and
then
it is done with the card. In such a scenario, where you had a card with
a main chip on it, and no add-on BIOS ROM, the BIOS would ignore
the card, and could not boot from it.

OK, except I still don't understand what you mean by "booting from a PCI
card", unless you were only referring to a controller card, with a drive
connected to it.

If a BIOS ROM is detected, while the BIOS is doing its probes, then the
motherboard BIOS can attempt to load the add-in BIOS.

But I would have thought that ithe system BIOS would *always* try to do
that.
(Or maybe you're saying it won't do that IF it is a controller card but is
not being used for the boot HD).

That is, in a
sense, like loading a driver at the BIOS level, as it adds some simple
functions. Based on that added function, the BIOS then has more disks
in the potential boot order. The INT 0x13 code adds the bootability
function.

Now, when the OS starts up, it loads a bunch of its own drivers too. The
driver tells it how to access the functions on the card. Just like in the
BIOS scenario, if a card is present, but has no drivers, it is
effectively ignored and invisible. (Just an entry in Device Manager, and
an exclamation mark saying something isn't right.) From then on, you
could use the computer, and the add-in card without Windows driver won't
be doing anything. It is like so much dead weight.

(And even with a windows driver, as in the case over here, it won't be
recognized via hot plugging, but only if it was first initialized at bootup
(due to the lack of AHCI).

When I referred to your motherboard, I was referring to the fact that
the only storage chip is the ICH9, with four ports. So there isn't an
additional chip, like a Jmicron chip on the motherboard. If there had
been such a chip present, it might have meant another option for
connecting disk drives. You do have expansion slots, and that is
how you can add storage capability.

OK. OR, just as I have been doing (w/o a card), by just running an eSata
cable over into a free SATA connector (header?) on my MB. (The main
disadvantage to me being, no hotplugging, of course).

There are four connectors (headers?) here on the MB; one is being used for
the main HD; one for the DVD drive; and now one for my external eSATA
enclosure drive (unless and until I get an eSata PCI card)

With regard to PCI cards versus PCI Express, you may find that only
PCI Express has SATA ports at 3Gbit/sec. The older PCI cards have
SATA at 1.5Gbit/sec and no "hot plug".

Oh, ok. Good to know, and that seems to be what I'm finding, too.

If I was solving the problem
you were working on, I'd be using a PCI Express x1 card, with SATA
ports rated at 3Gbit/sec. The difference is, the driver for the
3Gbit/sec chips, tends to support hot plugging, while the others
might not. There may be a hack for making the other ones do it,
but you know my answer about "keeping it simple"

Right.

This card is flashed with non-RAID BIOS, and has two ESATA connectors.
$25.
Install it, install the driver, then you should see a "Safely Remove"
icon, when you connect the ESATA drive.

"Rosewill RC-219 Silicon Image PCI Express External eSATA II x2 NCQ
non-RAID"
http://www.newegg.com/Product/Product.aspx?Item=N82E16816132020

Paul

Thanks. I looked at that one, and (what appears to be) a slightly simpler
SATA II card, using that SIL3132 chip that you mentioned. Here is the link
for that card over at amazon.com, which seems pretty similar to the one you
mentioned.
Its also a PCI Express to 2 External e-SATA II Ports Controller Card
(SIL3132):

http://www.amazon.com/External-e-SA...7?ie=UTF8&s=electronics&qid=1217387126&sr=1-7

Thanks again, Paul

 

[Dragomir Kollaric]

A separate card can be used two ways. It can be used for a
"data only" disk, one you don't care to boot from. In that
case, it doesn't matter whether the BIOS chip on the
separate card, loads or not.

If you want to boot from a drive connected to a separate
card, then the BIOS chip on the card has to load. The BIOS
does stuff like provide an extended INT 0x13 routine,
which provides the ability to read sectors from the disk
during bootup. Effectively, it is like a driver during
boot time. Some motherboard BIOS have an option called
"Interrupt 19 capture". The difference between the two
statements I just made, is 0x13 is hexadecimal (1*16 + 3),
while the second statement uses a decimal number (19). Why
the industry cannot stick with one way of stating it,
eludes me. Asus, in their BIOS, should really refer to it
as 0x13. I mean, when I first saw that setting, I didn't
know what it was for. Later, a little clue light bulb went
on, that 0x13 and 19 were the same thing.

"Capture", means to load the BIOS on any separate cards.
So my motherboard has an option to disable all add-on
ROMs, or allow them to be loaded. If you allow them to be
loaded, then the separate add-in cards can be used for
booting the computer.

I'll just add this.

If one uses more HDD's in the PC and they are powered by the
PSU it could turn out that the PSU is too weak. I had one
gone bad on me, when I had 4 HDD and 2 CD/CD-RW installed.
the 4 HDD were set up on a raid-card, albeit not used as
raid. Still the PSU was too weak and died on me. In the Shop
they put in another unit, as I had removed the home-build
array and stuck in a single HDD, they couldn't figure out
why the PSU failed. How much more power do the newer eSata
HDD need, that the point I make could be of concern?



Dragomir Kollaric

 

[Dragomir Kollaric]

On 2008-07-30, Bill in Co. hit the keyboard and wrote:

Still, when you boot up, I would think the system BIOS
would need to recognize this controller too, and show at
least some other options in there (in BIOS). Or maybe it
does, but that's beside the point for what I'm doing here
(not booting with it).

card,

Not.

Well I'd like to boot from mine again.

And yes when I start up the PC I see messages related to the
card, it shows me among other things how many HDD (Names)
are connected. If a cable f.e fails (as it happened to me
once) one can easily see that something is gone bad.

In my older PC (Athlon 1800XP) I was able to boot from it,
but not from this motherboard. In any case the BIOS would
give one the option to boot from such cards. I suspect that
my NIC is interfering with this option, but this is another
story, for another NG.

Thanks again!

Dragomir Kollaric

 

[Bill in Co.]

Addendum.
I stand corrected, Paul Your newegg controller card might be a bit more
basic (compared to the one I mentioned below at amazon.com), but I'm still
not sure (in addition to some other comments I wasn't sure about below).

 

[Paul]

From the "main chip" on the add-on card, I presume that means. (Not sure
what the "main chip" is, but I guess that depends on the nature of the
card).

The chip connected to the bus present in that expansion slot. The SIL3132
chip has a PCI Express x1 interface on it, and it connects to the slot.
Probes by the BIOS, to the configuration space of the SIL3132, will
uncover it is a storage controller card. If a BIOS chip is detected,
then the next step would be to load code from the BIOS_chip.

SIL3132 ------ BIOS_chip
|
| (Enumerates via bus)
|
PCI_Express_x1_slot

If the above BIOS chip, has INT 0x13 code present, then that storage card
could contribute a drive to the boot list.

If the card is designed like this, then a disk drive connected to the SIL3132
cannot be used to boot the computer. But once you get into Windows,
and install a SIL3132 driver, it is still possible for the OS to read/write
a data hard drive.

SIL3132 ---/
|
| (Enumerates via bus, no BIOS to load)
|
PCI_Express_x1_slot

Boot from it? What kind of "cards" could you "boot from"? Maybe you're
referring specifically to a controller card connected to a drive? And do
you mean you could use that type of card without invoking its own BIOS (for
that specific case) IF you didn't want to use it for a boot drive)?

You can boot from network cards, as another example of bootable devices.
A network card may have a BIOS chip, containing a PXE code module. If so, and
there is some kind of BOOTP server on the network, then it is possible
to boot a computer via an OS stored on a server. (I'm not up on the
details of this, and have just heard of the basic mechanism.)

After it has been initially enumerated - I assume that's what you mean)

Yes. Once resources are set up for a card, there is nothing else
to do.

But I would have thought that ithe system BIOS would *always* try to do
that.
(Or maybe you're saying it won't do that IF it is a controller card but is
not being used for the boot HD).

It won't do that, if the "Capture" option in the BIOS screen, has disabled
it. It also won't do it, if there is something wrong with the add-in BIOS
code. Or there isn't room to store the code in the 128KB area reserved for such
things, in low memory. If other cards have used up the 128KB area, and
no memory is left, then the BIOS will not attempt to load the BIOS contents
from the next add-in card. Typically, that might happen with a PC having
three SCSI cards in it. The BIOS for the third card might not load.

Thanks. I looked at that one, and (what appears to be) a slightly simpler
SATA II card, using that SIL3132 chip that you mentioned. Here is the link
for that card over at amazon.com, which seems pretty similar to the one you
mentioned.
Its also a PCI Express to 2 External e-SATA II Ports Controller Card
(SIL3132):

http://www.amazon.com/External-e-SA...7?ie=UTF8&s=electronics&qid=1217387126&sr=1-7

Thanks again, Paul

While I have no experience with that card on Amazon, I can say I don't approve
of "two-headed" card designs. This is what the Amazon card is doing. It
offers four connectors, but the chip only has two electrical interfaces.
And that means, they are hanging two connectors from the same bus. The
design looks like this.

SIL3132 -------------internal_connector--------external_connector
-------------internal_connector--------external_connector

You can only use one connector on each branch, at any one time. And electrically
speaking, connecting a cable and drive to the internal_connector, leaves
a stub off the high speed bus.

The reason I picked the Rosewill card (even though I'm not a fan of Rosewill),
is because it is set up like this.

SIL3132 -------------external_connector
-------------external_connector

That should give better electrical characteristics.

You can find other card designs, such as a card with one ESATA and
one internal SATA, if you want some flexibility. That would be like
this.

SIL3132 -------------external_connector
-------------internal_connector

Paul

 

[Bill in Co.]

The chip connected to the bus present in that expansion slot. The SIL3132
chip has a PCI Express x1 interface on it, and it connects to the slot.
Probes by the BIOS, to the configuration space of the SIL3132, will
uncover it is a storage controller card. If a BIOS chip is detected,
then the next step would be to load code from the BIOS_chip.

SIL3132 ------ BIOS_chip
|
| (Enumerates via bus)
|
PCI_Express_x1_slot

If the above BIOS chip, has INT 0x13 code present, then that storage card
could contribute a drive to the boot list.

I wonder if it does (or how many actually do). Interesting.

If the card is designed like this, then a disk drive connected to the
SIL3132
cannot be used to boot the computer.

Cannot? And yet it would appear in the boot list. I don't understand
why if the system reads the BIOS on the card it's not able to boot the
computer (not that I want that, anyways). It seems you are saying above
that if the BIOS on the card is read, that it can't be used for booting?
OR that this is somehow tied into the INT 0 x 13 thing you mentioned, and
THAT is what creates that limitation?

So I'm trying to figure out exactly what *would* allow the card (and the
drive connected to it) to be bootable - just curious.

But once you get into Windows,
and install a SIL3132 driver, it is still possible for the OS to
read/write
a data hard drive.

SIL3132 ---/
|
| (Enumerates via bus, no BIOS to load)
|
PCI_Express_x1_slot

OK, now here you are saying that the card BIOS *is* needed for boot
capability.

You can boot from network cards, as another example of bootable devices.
A network card may have a BIOS chip, containing a PXE code module. If so,
and
there is some kind of BOOTP server on the network, then it is possible
to boot a computer via an OS stored on a server. (I'm not up on the
details of this, and have just heard of the basic mechanism.)


Yes. Once resources are set up for a card, there is nothing else to do.


It won't do that, if the "Capture" option in the BIOS screen, has disabled
it.

What "capture option"? (I mean, I know when you go to the BIOS screen,
you can see if it has been recognized there or not, and that this was due to
the earlier enumeration, but that's all I know about that).

It also won't do it, if there is something wrong with the add-in BIOS
code. Or there isn't room to store the code in the 128KB area reserved for
such things, in low memory. If other cards have used up the 128KB area,
and
no memory is left, then the BIOS will not attempt to load the BIOS
contents
from the next add-in card. Typically, that might happen with a PC having
three SCSI cards in it. The BIOS for the third card might not load.
OK.


While I have no experience with that card on Amazon, I can say I don't
approve
of "two-headed" card designs. This is what the Amazon card is doing. It
offers four connectors,

Where are you seeing that? I'm seeing only TWO external eSata
connectors, and those are only for TWO (max) eSata drives - NOT four. Nor
are there any internal SATA connectors, from what I can tell.

but the chip only has two electrical interfaces.
And that means, they are hanging two connectors from the same bus. The
design looks like this.

SIL3132 -------------internal_connector--------external_connector
-------------internal_connector--------external_connector

You can only use one connector on each branch, at any one time. And
electrically speaking, connecting a cable and drive to the
internal_connector, leaves a stub off the high speed bus.

But again, I must be missing something, as I don't see it for that card.
(The PCI Express to 2 External e-SATA II Ports Controller Card at amazon)

The reason I picked the Rosewill card (even though I'm not a fan of
Rosewill),
is because it is set up like this.

SIL3132 -------------external_connector
-------------external_connector

That should give better electrical characteristics.

You can find other card designs, such as a card with one ESATA and
one internal SATA, if you want some flexibility. That would be like
this.

SIL3132 -------------external_connector
-------------internal_connector

Right, I thought about that, but I've already some internal eSata headers
free, so I probably don't need it, as far as I can tell. I've got four
SATA headers on the MB, and three are in use now (but one being used for the
external eSata drive right now), but maybe it's worth thinking about.

 

[Paul]

But again, I must be missing something, as I don't see it for that card.
(The PCI Express to 2 External e-SATA II Ports Controller Card at amazon)

OK. I downloaded the image of the card and zoomed in. What I was seeing,
is solder pads for internal connectors. There are no internal connectors
installed in this case. But I still don't like to see that style of
wiring. The idea here, is they could make two versions of card, but
if they decided to make a card with the two internal ones installed,
and the two external ones missing, then there would be stubs off the
bus. So this particular card is *OK*. But if they ship a version where
the internal connectors are installed, but the external ones are not,
that is still a stub off the bus. I prefer to see one connector per port,
for peace of mind. I have seen cards designed the same way this one is,
but with all four connectors soldered in place. And I would not buy
one designed like that.

This is the Amazon card.

http://ecx.images-amazon.com/images/I/51LHEMcClEL._SS400_.jpg

*OK* design. Since the stub of the solder pads is minimal.

SIL3132 ------------- (none) --------external_connector
------------- (none) --------external_connector

This, on the other hand, I would not buy.

SIL3132 -------------internal_connector-------- (none)
-------------internal_connector-------- (none)

I prefer to inspect and see designs that look like this. The absence
of a position for a second pair of connectors, means no wiring issues.

SIL3132 -------------internal_connector
-------------internal_connector

Paul

 

[Paul]

OK. I downloaded the image of the card and zoomed in. What I was seeing,
is solder pads for internal connectors. There are no internal connectors
installed in this case. But I still don't like to see that style of
wiring. The idea here, is they could make two versions of card, but
if they decided to make a card with the two internal ones installed,
and the two external ones missing, then there would be stubs off the
bus. So this particular card is *OK*. But if they ship a version where
the internal connectors are installed, but the external ones are not,
that is still a stub off the bus. I prefer to see one connector per port,
for peace of mind. I have seen cards designed the same way this one is,
but with all four connectors soldered in place. And I would not buy
one designed like that.

This is the Amazon card.

http://ecx.images-amazon.com/images/I/51LHEMcClEL._SS400_.jpg

*OK* design. Since the stub of the solder pads is minimal.

SIL3132 ------------- (none) --------external_connector
------------- (none) --------external_connector

This, on the other hand, I would not buy.

SIL3132 -------------internal_connector-------- (none)
-------------internal_connector-------- (none)

I prefer to inspect and see designs that look like this. The absence
of a position for a second pair of connectors, means no wiring issues.

SIL3132 -------------internal_connector
-------------internal_connector

Paul

I notice the other card on Amazon, with the SIL3132 on it, doesn't reuse
the PCB with the four connector locations on it. This card offers two internal
connectors, and there are no stubs to be seen. I wouldn't mind buying this.

(Two port internal SIL3132)
http://ecx.images-amazon.com/images/I/51VuFyHaUNL._SS400_.jpg

Paul

 

[Bill Blanton]

I wonder if it does (or how many actually do). Interesting.


Cannot? And yet it would appear in the boot list. I don't understand why if the system reads the BIOS on the card it's not
able to boot the computer (not that I want that, anyways). It seems you are saying above that if the BIOS on the card is read,
that it can't be used for booting? OR that this is somehow tied into the INT 0 x 13 thing you mentioned, and THAT is what creates
that limitation?

AIUI, theorectically you could boot the device as long as the BIOS can load and
jump into the first sector. Practically, "basic disk" Windows MBR boot code requires
the device to be recognised by the 0x13 interrupts. IOW, Windows MBR code
uses int 0x13 to load the volume boot sector of the boot device.