Universal BIOS Backup ToolKit 2.0.exe

"Paul" schreef in bericht
...
Linea Recta wrote:

I assume that passwords are also in the CMOS backup?

Yes, and you never know, they might be stored in plaintext.

I never understood why the don't use some rechargeable battery connected
to the mains (even when the system is down). I suppose it all about $$$.

Someone brought this topic up a while back. There are
devices that use a LR2032 instead of a CR2032. The two
are not interchangeable. (You cannot charge a CR2032, and
the motherboards uses low leakage diodes in that area
of the circuit board.)

http://www.batteryjunction.com/lir2032----.html

The LR2032 is rechargeable lithium.
The amp-hour rating is pretty low. If the product is
unplugged for just a few days, the coin cell can run
down to zero. The LR2032 is practical (in say a laptop),
as long as the user doesn't leave the laptop with no
main battery pack in it.

If you have an electronics product that already
uses one of those, then you can buy a similar
type if it ever needs to be replaced.

It just doesn't seem all that practical, to me.



Deaming further on about this while driving my car today, I wondered
wouldn't it be better if they used a non-volatile memory, something like an
SD card, for keeping all the settings, and perhaps also the BIOS code
itself? No current of any sort needed to keep the information available?




--


|\ /|
| \/ |@rk
\../
\/os

Linea Recta wrote:


Deaming further on about this while driving my car today, I wondered
wouldn't it be better if they used a non-volatile memory, something like
an SD card, for keeping all the settings, and perhaps also the BIOS code
itself? No current of any sort needed to keep the information available?


Absolutely. The current practice is "barbaric" :-)

*******

But consider that, they want to run a real time clock, even
when the computer is not powered. So the computer always
has a local time_of_day reference for stamping file
system entries. The computer must "work properly", with
no network connection to use NTP protocol.

An RTC draws around 2 microamps, if implemented in the
Southbridge. You need the battery, to power that clock
time piece. The clock runs off a 32768 Hertz quartz crystal
(just like your digital watch), and will drift a bit in
terms of time keeping. (It's not "atomic clock" quality
in any case.)

Adding the 256 bytes of RAM, doesn't change things
all that much. The total power is 10 microamps,
with both RTC and CMOS RAM. Part of that is likely
to be "CMOS well" transmission gate leakage, through the
stuff that prevents the battery powered portion of
the Southbridge, leaking into the larger part of the chip.
If transmission gates were not used, for electrical isolation,
current from the CR2032 would "leak" into the rest of the
Southbridge, and the battery would drain in no time. The usage
of transmission gates, helps make this "chip emulation" inside
the Southbridge a practical possibility. Where the letter "T"
is like a moat, they "lift the drawbridge" so the little
island stays electrically isolated, when the computer
is powered off. When the chip is fully powered, the OS
reads the registers of the RTC, via some sort of
transmission gate protected path.

+--------------------------------------+
| Southbridge - SATA, PCI, LPC I/O etc |
| |
| | | | | The letter "T" is
| T T T | where transmission gates
| | | | | for I/O, are located.
| +----------+ |
| | RTC and |--------------------------- CR2032 3V power
| | CMOS RAM |--------------------------- 32768Hz crystal
| +----------+ |
| |
+--------------------------------------+

The end result, with a CR2032, is you can power that
time piece (and the 256 byte RAM sitting next to it),
for about three years with no AC power.

So if we replaced the 256 byte RAM with 256 bytes or
NOR flash, we still need a clock function, we still
need a battery, and the battery lasts a bit longer.

At one time, this function was a separate chip, and
the emulation that puts it into the Southbridge,
is intended to make PCs cheaper to build. You might
have had a Dallas chip in really old PCs. And the problem
with the old solution to this problem, is the entire
Dallas chip needed to be replaced, when the internal
battery in the chip went flat. At least the battery
is easy to replace now. Dallas used to pot the whole
thing in epoxy, so you couldn't repair it. It was
"Dremel time", for the people who used to retrofit
an external battery to their Dallas.

http://classic-computers.org.nz/blog...ttery-chip.htm

Paul

On Mon, 20 Oct 2014 17:46:19 +0200, Linea Recta wrote:

"Paul" schreef in bericht
...
Linea Recta wrote:

I assume that passwords are also in the CMOS backup?

Yes, and you never know, they might be stored in plaintext.

I never understood why the don't use some rechargeable battery connected
to the mains (even when the system is down). I suppose it all about $$$.

Someone brought this topic up a while back. There are
devices that use a LR2032 instead of a CR2032. The two
are not interchangeable. (You cannot charge a CR2032, and
the motherboards uses low leakage diodes in that area
of the circuit board.)

http://www.batteryjunction.com/lir2032----.html

The LR2032 is rechargeable lithium.
The amp-hour rating is pretty low. If the product is
unplugged for just a few days, the coin cell can run
down to zero. The LR2032 is practical (in say a laptop),
as long as the user doesn't leave the laptop with no
main battery pack in it.

If you have an electronics product that already
uses one of those, then you can buy a similar
type if it ever needs to be replaced.

It just doesn't seem all that practical, to me.


Deaming further on about this while driving my car today, I wondered
wouldn't it be better if they used a non-volatile memory, something like an
SD card, for keeping all the settings, and perhaps also the BIOS code
itself? No current of any sort needed to keep the information available?

The BIOS code is in non-volatile memory. It *has* to be...

--
Gene E. Bloch (Stumbling Bloch)

Gene E. Bloch wrote:
On Mon, 20 Oct 2014 17:46:19 +0200, Linea Recta wrote:

"Paul" schreef in bericht
...
Linea Recta wrote:

I assume that passwords are also in the CMOS backup?
Yes, and you never know, they might be stored in plaintext.

I never understood why the don't use some rechargeable battery connected
to the mains (even when the system is down). I suppose it all about $$$.
Someone brought this topic up a while back. There are
devices that use a LR2032 instead of a CR2032. The two
are not interchangeable. (You cannot charge a CR2032, and
the motherboards uses low leakage diodes in that area
of the circuit board.)

http://www.batteryjunction.com/lir2032----.html

The LR2032 is rechargeable lithium.
The amp-hour rating is pretty low. If the product is
unplugged for just a few days, the coin cell can run
down to zero. The LR2032 is practical (in say a laptop),
as long as the user doesn't leave the laptop with no
main battery pack in it.

If you have an electronics product that already
uses one of those, then you can buy a similar
type if it ever needs to be replaced.

It just doesn't seem all that practical, to me.

Deaming further on about this while driving my car today, I wondered
wouldn't it be better if they used a non-volatile memory, something like an
SD card, for keeping all the settings, and perhaps also the BIOS code
itself? No current of any sort needed to keep the information available?

The BIOS code is in non-volatile memory. It *has* to be...


True in the most general sense. But there are details
as to how the BIOS chip is used. It's not "read-only" in the
way you might think. The DMI/ESCD area can be updated
during POST.

Paul

On Mon, 20 Oct 2014 14:57:43 -0400, Paul wrote:

Gene E. Bloch wrote:
On Mon, 20 Oct 2014 17:46:19 +0200, Linea Recta wrote:

"Paul" schreef in bericht
...
Linea Recta wrote:

I assume that passwords are also in the CMOS backup?
Yes, and you never know, they might be stored in plaintext.

I never understood why the don't use some rechargeable battery connected
to the mains (even when the system is down). I suppose it all about $$$.
Someone brought this topic up a while back. There are
devices that use a LR2032 instead of a CR2032. The two
are not interchangeable. (You cannot charge a CR2032, and
the motherboards uses low leakage diodes in that area
of the circuit board.)

http://www.batteryjunction.com/lir2032----.html

The LR2032 is rechargeable lithium.
The amp-hour rating is pretty low. If the product is
unplugged for just a few days, the coin cell can run
down to zero. The LR2032 is practical (in say a laptop),
as long as the user doesn't leave the laptop with no
main battery pack in it.

If you have an electronics product that already
uses one of those, then you can buy a similar
type if it ever needs to be replaced.

It just doesn't seem all that practical, to me.

Deaming further on about this while driving my car today, I wondered
wouldn't it be better if they used a non-volatile memory, something like an
SD card, for keeping all the settings, and perhaps also the BIOS code
itself? No current of any sort needed to keep the information available?

The BIOS code is in non-volatile memory. It *has* to be...


True in the most general sense. But there are details
as to how the BIOS chip is used. It's not "read-only" in the
way you might think. The DMI/ESCD area can be updated
during POST.

Paul

I don't recall saying read-only...nor thinking it.

In fact, in my post I intentionally didn't talk about BIOS updates,
which, IIRC :-), involve *writing* to the BIOS's non-volatile memory,
since I saw it as irrelevant to the remark I made.

Many[1] BIOSes do have read-only portion(s) which contain, among other
things, the firmware needed to update the BIOS firmware itself.

[1] It's now 2014. That probably should read *all*, not *many*,
BIOSes...

--
Gene E. Bloch (Stumbling Bloch)

Gene E. Bloch wrote:
On Mon, 20 Oct 2014 14:57:43 -0400, Paul wrote:

Gene E. Bloch wrote:
On Mon, 20 Oct 2014 17:46:19 +0200, Linea Recta wrote:

"Paul" schreef in bericht
...
Linea Recta wrote:

I assume that passwords are also in the CMOS backup?
Yes, and you never know, they might be stored in plaintext.

I never understood why the don't use some rechargeable battery connected
to the mains (even when the system is down). I suppose it all about $$$.
Someone brought this topic up a while back. There are
devices that use a LR2032 instead of a CR2032. The two
are not interchangeable. (You cannot charge a CR2032, and
the motherboards uses low leakage diodes in that area
of the circuit board.)

http://www.batteryjunction.com/lir2032----.html

The LR2032 is rechargeable lithium.
The amp-hour rating is pretty low. If the product is
unplugged for just a few days, the coin cell can run
down to zero. The LR2032 is practical (in say a laptop),
as long as the user doesn't leave the laptop with no
main battery pack in it.

If you have an electronics product that already
uses one of those, then you can buy a similar
type if it ever needs to be replaced.

It just doesn't seem all that practical, to me.

Deaming further on about this while driving my car today, I wondered
wouldn't it be better if they used a non-volatile memory, something like an
SD card, for keeping all the settings, and perhaps also the BIOS code
itself? No current of any sort needed to keep the information available?
The BIOS code is in non-volatile memory. It *has* to be...

True in the most general sense. But there are details
as to how the BIOS chip is used. It's not "read-only" in the
way you might think. The DMI/ESCD area can be updated
during POST.

Paul

I don't recall saying read-only...nor thinking it.

In fact, in my post I intentionally didn't talk about BIOS updates,
which, IIRC :-), involve *writing* to the BIOS's non-volatile memory,
since I saw it as irrelevant to the remark I made.

Many[1] BIOSes do have read-only portion(s) which contain, among other
things, the firmware needed to update the BIOS firmware itself.

[1] It's now 2014. That probably should read *all*, not *many*,
BIOSes...


You could draw it like this. For a legacy BIOS.

+-------------------------------------+
| Main BIOS code
| Many separate code modules
+--
| (Read Only, except on a BIOS update)
+--
| Checksum protected
+--------------------------------------+
| Microcode Cache
| (Read Only, except when updated
| by the insertion of a new processor)
+--------------------------------------+
| DMI/ESCD
| (Read/write, in the presence of
| POST detected hardware changes,
| or user updates with Intel utility)
+--------------------------------------+
| Boot Block
| (Should really be read-only, but
| frequently gets updated on a BIOS
| upgrade.)
+--------------------------------------+

The BIOS main code, is mirrored in RAM. This
allows updates to the Flash chip, without
concern for whether the code is being used
right now.

The chip has erasable pages, so the whole device
doesn't have to be erased, to update a small area.
The size of the pages, helps define the boundaries
of the smaller areas. Maybe the boot block is 8KB,
the Microcode Cache 2KB, and so on.

The bytes in the main code area are compressed. If
you intended to disassemble the code, you need to
decompress it first. It's a lot easier to read
that way, and has give-away text strings in it.

The boot block on the other hand, is less likely
to be compressed. It's probably got a decompresser,
or knows where to find one. I've never attempted
to do anything with a boot block, so haven't a clue
how it is structured. I've taken apart a number
of main BIOS code blocks, for a quick look.

The UEFI BIOS is much more complicated than this,
and one article I was reading hinted at as many
as 200 files being present inside. And the UEFI BIOS
also offers some sort of access to the file system,
when the OS is running. I haven't really found any
good tutorial articles, on just how exposed to exploits
a UEFI BIOS might be. It sounds complicated enough, to
be a disaster area waiting to happen.

Paul

On Mon, 20 Oct 2014 17:28:37 -0400, Paul wrote:


The BIOS main code, is mirrored in RAM. This
allows updates to the Flash chip, without
concern for whether the code is being used
right now.


Interesting, thanks! I never realized that before.

Ken Blake wrote:
On Mon, 20 Oct 2014 17:28:37 -0400, Paul wrote:


The BIOS main code, is mirrored in RAM. This
allows updates to the Flash chip, without
concern for whether the code is being used
right now.


Interesting, thanks! I never realized that before.


https://web.archive.org/web/20000831...adowing-c.html

"One problem with ROMs such as those used for the
system BIOS and video BIOS, is that it is relatively slow."

"I'm sure you can see where this is heading. Since there
is RAM hiding underneath the ROMs anyway, most systems have
the ability to "mirror" the ROM code into this RAM to
improve performance. This is called ROM Shadowing, and is
controlled using a set of BIOS parameters. There is normally
a separate parameter to control the shadowing of the system BIOS,
the video BIOS and adapter ROM areas."

The details have likely changed with time. Partially, because
the Flash chip is larger than it used to be. There is certainly
lots of RAM to go around, but the archaic usage rules (and
backward compatibility), kinda handcuff things. Maybe my first
motherboard, is guaranteed to have that feature.

And the comment about "relatively slow", still applies. The
current generation of 8 pin serial flash chips, are no
speed demons. Moving from a byte wide interface to a
serial interface, did not help matters.

One of the slowest flash ever put on a motherboard, was
the Asus "bitching betty" vocal error reporter setup. It
used a serial chip to hold canned vocal error messages.
It would take somewhere around 20 minutes to half an hour
to reprogram one of those. The BIOS Flash chip is faster
than that, by quite a bit.

Paul

In message , Paul
writes:
[]
But consider that, they want to run a real time clock, even
when the computer is not powered. So the computer always
has a local time_of_day reference for stamping file
system entries. The computer must "work properly", with
no network connection to use NTP protocol.

An RTC draws around 2 microamps, if implemented in the
Southbridge. You need the battery, to power that clock
time piece. The clock runs off a 32768 Hertz quartz crystal
(just like your digital watch), and will drift a bit in
terms of time keeping. (It's not "atomic clock" quality
in any case.)
[]
It certainly isn't; it is highly likely to use a 32768 hertz* crystal
that was indeed designed for the digital watch market; such crystals are
designed to run at a fairly constant temperature (strapped to your
wrist!), so drift quite a bit in a cold PC case. Only relatively - I
doubt more than a few seconds a day in practice - but more than when
used in a wris****ch, and certainly more than the PC when it's turned on
and getting reasonably regular timechecks over the 'net.

* ISTR reading somewhere that the SI units, when written out in full
(hertz, joule, newton, watt etc.) _don't_ take a capital letter;
obviously if you're talking about the person they're named after, they
do. Also, the SI _abbreviation_ - Hz, J, N, W - may be capitalised. (I'm
trying to think of one where the abbreviation/symbol _isn't_, but the
only one I can think of is gram[me] [symbol g], and I don't think that
one's named after anyone.)
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I'd be a middle-class hero if I had the time, but I've got to go to Waitrose
first. - Tim Vine, RT 2014/2/15-21

On Wed, 22 Oct 2014 21:51:42 +0100, "J. P. Gilliver (John)"
wrote:

I'm
trying to think of one where the abbreviation/symbol _isn't_, but the
only one I can think of is gram[me] [symbol g], and I don't think that
one's named after anyone.)


If you pronounce her name the way my wife does, perhaps it's named
after Martha Graham.