On Sat, 12 Feb 2022 at 03:28:21, Paul <nospam@needed.invalid> wrote (my
responses usually follow points raised):

[in a discussion of bitlocker and similar]

>Just as every time you boot the PC, there are writes to the
>BIOS chip. Eventually, the BIOS chip could wear out. The BIOS
>chip is unlikely to have wear leveling like an SSD.
>
> Paul

I won't claim to understand much of the discussion that I've snipped,
but that last paragraph worries me - is it true for all PCs, or only
ones that use encryption?

[Amusing: my spelling-checker suggests "bootlicker" for bitlocker.
Amusing because you _could_ see hardware encryption as something
hardware manufacturers are bowing down and incorporating because of
pressure from Microsoft etc., and intriguing for how "bootlicker" came
to be in my spellchecker's list!]
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

Veni Vidi Visa [I came, I saw, I did a little shopping] - Mik from S+AS Limited
(mik@saslimited.demon.co.uk), 1998

On 2/12/2022 8:05 AM, J. P. Gilliver (John) wrote:
> On Sat, 12 Feb 2022 at 03:28:21, Paul <nospam@needed.invalid> wrote (my responses usually follow points raised):
>
> [in a discussion of bitlocker and similar]
>
>> Just as every time you boot the PC, there are writes to the
>> BIOS chip. Eventually, the BIOS chip could wear out. The BIOS
>> chip is unlikely to have wear leveling like an SSD.
>>
>>   Paul
>
> I won't claim to understand much of the discussion that I've snipped, but that last paragraph worries me - is it true for all PCs, or only ones that use encryption?
>
> [Amusing: my spelling-checker suggests "bootlicker" for bitlocker. Amusing because you _could_ see hardware encryption as something hardware manufacturers are bowing down and incorporating because of pressure from Microsoft etc., and intriguing for how "bootlicker" came to be in my spellchecker's list!]

UEFI keeps boot paths in NVRAM.

NVRAM is a portion of the BIOS flash memory.

The CMOS RAM (256 bytes) is small and useless,
compared to the 4MB reserved in the BIOS chip
for NVRAM functions.

*******

The legacy BIOS does writes to DMI/ESCD when hardware
changes. Each time you add or remove a DIMM, a record is
kept. Intel DMI Explorer was named as an example of
an application that could dump the table. My work, once
attempted to inventory all the DIMMs using this function,
but the project sank without making a ripple :-)

The BIOS also has a 2KB storage space for a manual microcode
upgrade. Which is how I could run a Tualatin on a board that
did not support it. Modern processors use model-specific lengths
of microcode, so that exact design no longer works or is supported.

But the UEFI is more verbose, and byte for byte, you would
suspect (but not know for sure), that more is going on.
Early EFI *bricked* because the 4MB area was "full", and
that's the only observational datapoint available. The mature
UEFI designs are much better at metadata consolidation, so
that should not happen today.

Paul

On Sat, 12 Feb 2022 at 15:14:50, Paul <nospam@needed.invalid> wrote (my
responses usually follow points raised):
>On 2/12/2022 8:05 AM, J. P. Gilliver (John) wrote:
>> On Sat, 12 Feb 2022 at 03:28:21, Paul <nospam@needed.invalid> wrote
>>(my responses usually follow points raised):
>> [in a discussion of bitlocker and similar]
>>
>>> Just as every time you boot the PC, there are writes to the
>>> BIOS chip. Eventually, the BIOS chip could wear out. The BIOS
>>> chip is unlikely to have wear leveling like an SSD.
>>>
>>>   Paul
>> I won't claim to understand much of the discussion that I've
>>snipped, but that last paragraph worries me - is it true for all PCs,
>>or only ones that use encryption?
>> [Amusing: my spelling-checker suggests "bootlicker" for bitlocker.
>>Amusing because you _could_ see hardware encryption as something
>>hardware manufacturers are bowing down and incorporating because of
>>pressure from Microsoft etc., and intriguing for how "bootlicker" came
>>to be in my spellchecker's list!]
>
>UEFI keeps boot paths in NVRAM.
>
>NVRAM is a portion of the BIOS flash memory.
>
>The CMOS RAM (256 bytes) is small and useless,
>compared to the 4MB reserved in the BIOS chip
>for NVRAM functions.
>
>*******
>
>The legacy BIOS does writes to DMI/ESCD when hardware
>changes. Each time you add or remove a DIMM, a record is
>kept. Intel DMI Explorer was named as an example of
>an application that could dump the table. My work, once
>attempted to inventory all the DIMMs using this function,
>but the project sank without making a ripple :-)
>
>The BIOS also has a 2KB storage space for a manual microcode
>upgrade. Which is how I could run a Tualatin on a board that
>did not support it. Modern processors use model-specific lengths
>of microcode, so that exact design no longer works or is supported.
>
>But the UEFI is more verbose, and byte for byte, you would
>suspect (but not know for sure), that more is going on.
>Early EFI *bricked* because the 4MB area was "full", and
>that's the only observational datapoint available. The mature
>UEFI designs are much better at metadata consolidation, so
>that should not happen today.
>
> Paul

So something is written when you change hardware.

But other than that, are there "writes to the BIOS chip every time you
boot the PC", as you said earlier, even on PCs that _don't_ use
encryption?
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I'm the oldest woman on primetime not baking cakes.
- Anne Robinson, RT 2015/8/15-21

On 2/13/2022 5:16 PM, J. P. Gilliver (John) wrote:
> On Sat, 12 Feb 2022 at 15:14:50, Paul <nospam@needed.invalid> wrote (my responses usually follow points raised):
>> On 2/12/2022 8:05 AM, J. P. Gilliver (John) wrote:
>>> On Sat, 12 Feb 2022 at 03:28:21, Paul <nospam@needed.invalid> wrote (my responses usually follow points raised):
>>>  [in a discussion of bitlocker and similar]
>>>
>>>> Just as every time you boot the PC, there are writes to the
>>>> BIOS chip. Eventually, the BIOS chip could wear out. The BIOS
>>>> chip is unlikely to have wear leveling like an SSD.
>>>>
>>>>   Paul
>>>  I won't claim to understand much of the discussion that I've snipped, but that last paragraph worries me - is it true for all PCs, or only ones that use encryption?
>>>  [Amusing: my spelling-checker suggests "bootlicker" for bitlocker. Amusing because you _could_ see hardware encryption as something hardware manufacturers are bowing down and incorporating because of pressure from Microsoft etc., and intriguing for how "bootlicker" came to be in my spellchecker's list!]
>>
>> UEFI keeps boot paths in NVRAM.
>>
>> NVRAM is a portion of the BIOS flash memory.
>>
>> The CMOS RAM (256 bytes) is small and useless,
>> compared to the 4MB reserved in the BIOS chip
>> for NVRAM functions.
>>
>> *******
>>
>> The legacy BIOS does writes to DMI/ESCD when hardware
>> changes. Each time you add or remove a DIMM, a record is
>> kept. Intel DMI Explorer was named as an example of
>> an application that could dump the table. My work, once
>> attempted to inventory all the DIMMs using this function,
>> but the project sank without making a ripple :-)
>>
>> The BIOS also has a 2KB storage space for a manual microcode
>> upgrade. Which is how I could run a Tualatin on a board that
>> did not support it. Modern processors use model-specific lengths
>> of microcode, so that exact design no longer works or is supported.
>>
>> But the UEFI is more verbose, and byte for byte, you would
>> suspect (but not know for sure), that more is going on.
>> Early EFI *bricked* because the 4MB area was "full", and
>> that's the only observational datapoint available. The mature
>> UEFI designs are much better at metadata consolidation, so
>> that should not happen today.
>>
>>   Paul
>
> So something is written when you change hardware.
>
> But other than that, are there "writes to the BIOS chip every time you boot the PC", as you said earlier, even on PCs that _don't_ use encryption?

Think about how hard it would be, to overflow a
4MB area, by writing a couple boot strings each time
the machine comes up.

UEFI must be a busy beaver. There's got to be more to
it than that.

Paul

On Sun, 13 Feb 2022 at 18:01:18, Paul <nospam@needed.invalid> wrote (my
responses usually follow points raised):
>On 2/13/2022 5:16 PM, J. P. Gilliver (John) wrote:
[]
>>> On 2/12/2022 8:05 AM, J. P. Gilliver (John) wrote:
>>>> On Sat, 12 Feb 2022 at 03:28:21, Paul <nospam@needed.invalid> wrote
>>>>(my responses usually follow points raised):
>>>>  [in a discussion of bitlocker and similar]
>>>>
>>>>> Just as every time you boot the PC, there are writes to the
>>>>> BIOS chip. Eventually, the BIOS chip could wear out. The BIOS
>>>>> chip is unlikely to have wear leveling like an SSD.
>>>>>
>>>>>   Paul
>>>>  I won't claim to understand much of the discussion that I've
>>>>snipped, but that last paragraph worries me - is it true for all
>>>>PCs, or only ones that use encryption?
[snip]
>> So something is written when you change hardware.
>> But other than that, are there "writes to the BIOS chip every time
>>you boot the PC", as you said earlier, even on PCs that _don't_ use
>>encryption?
>
>Think about how hard it would be, to overflow a
>4MB area, by writing a couple boot strings each time
>the machine comes up.

But as you said above, its unlikely to have wear levelling, so isn't it
going to write those strings to the same part of the relevant memory -
if it does indeed do a write every boot? And thus wear it out?
>
>UEFI must be a busy beaver. There's got to be more to
>it than that.
>
> Paul
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

If you want to make people angry, lie to them. If you want to make them
absolutely livid, then tell 'em the truth.