Replies

OK, this isn’t a good thing to read. Might be time to update a BIOS (off line for Dell, CVE-2021-21571 and CVE-2021-21572 not being much mentioned..) if you can but basically we have to hope this one goes away.. Microsoft will no doubt adjust defender rules but will that be enough? Alternatively it might mitigate some added software which isn’t to everyone’s taste.

https://www.bleepingcomputer.com/news/security/microsoft-wpbt-flaw-lets-hackers-install-rootkits-on-windows-devices/

 

1 user thanked author for this post.

WCHS

PSU wise Corsair are usually good but that doesn’t mean every one they make is faultless.

If the BIOS won’t auto to the default speed try the values on the CPU data page.. and remember that information is fed through the SPD bus to and from the chipset when you read further down.

I Don’t think you’ll get a BIOS chip in the form you expect (It’s likely to be a tiny 8 pin affair soldered down under the heatsink bottom right looking at the pictures.) and a new motherboard would be the only way to fix the issue IF that is the problem as the BIOS chip also holds other information pertinent to the board.

Yes, the incomplete reset could indicate the BIOS setting tables could be corrupt, but it seems unlikely that would happen and leave the machine able to boot and run Windows. I’ve even had a memory module cause this sort of weirdness (and a non running CPU fan, in a laptop chassis..) and that was because the data it presented on the SPI (latterly SPD) bus was incorrect (that’s the bus connecting the tiny six or 8 pin module on the end of the memory – holds the memory information as read by CPU-Z and the like. Maybe you could check your report matches someone else with the same board in case something is being mis-read on that bus?

Perhaps your Windows installation’s drivers could be upsetting the firmware. If you unplug all the drives and boot the board only to the “failed to boot” condition, that will force the UEFI BIOS into a HSTI fail state (which should also return items like the boot menu and clear the UEFI state registered which might help it make sense of the situation as you can at least rule that out.

If that works and the power button acts normally, see if you can load defaults and if so try Windows again and see if it puts the fault back (or maybe it’s gone away..). If it still won’t load defaults try swapping in any other suitable graphics card, or memory, or each memory module in turn to see if any combo enables this as it might illuminate a component with problem SPD information. This might help explain where the last two steps may touch on the problem and BIOS update issues:

https://en.wikipedia.org/wiki/Serial_Peripheral_Interface#Intel_Enhanced_Serial_Peripheral_Interface_Bus

If the BIOS is beyond repair anyway perhaps try the BIOS update and firmware update again just in case as these processes work “block wise” on the flash and it might take.. the hard part is most updates check the existing version and may not reapply the update.

As to the original PSU, MTBF is 100000 hrs which is impressive – but it doesn’t mean every one lasts that long, in fact half of them do. I think it was just not up to the task required. https://www.corsair.com/uk/en/Categories/Products/Power-Supply-Units/cs-series-config/p/CP-9020086-UK#tab-tech-specs

Finally one old guy tip, not in the manual (and hasn’t been in any manual for a decade or two) for this one but might just still work so worth a go.. on really old machines if you pressed “insert” instead of “delete” during POST it loaded default values as opposed to entering setup. Any port in a storm..

1 user thanked author for this post.

wavy

“Oh no he’s back again..”

” Samsung’s enterprises SSDs use <b>1.25 watts of power in active mode and 0.3 watts in idle mode</b>. ” – assume it’s as long as its short and say six will take 10W (as yours are probably a bit adrift but the figure is tiny to what I expected) based on https://www.samsung.com/semiconductor/newsroom/news-events/new-samsung-enterprise-ssds-for-high-performance-applications-enable-low-power-consumption-and-high/#:~:text=Delivering%20the%20%22greenest%22%20storage%20solution,data%20center%20air%20conditioning%20system.

And the graphics card can top 200W (https://www.tomshardware.com/uk/reviews/amd-radeon-rx-580-review,5020-6.html) but is a 10th of that settled doing nothing.

So I make that a guestimate of strangely just over what your supply is capable of best case – nothing left to allow for old age or variability, let alone case fans and even decorative LED AURA RGB lights the motherboard can drive..

What to do? Run your rig at defaults speed and see if it is stable or not. if not uinplug everything you don’t need and see if it’s better. If it is try again with another PSU (even if you have to borrow one from a local gamer).

Run a memory test (MDSCHED.exe) and let it run overnight at minimum- any errors could indicate a memory problem as well as a stability issue. Memory issues tend to be fixed locations or bit positions, power errors seem more random.

That done, I have checked the downloads page – the BIOS (running on the CPU) works hand in hand with the Intel management engine firmware (which runs on the PCH and is instrumental in controlling the board). If you use Intel software for this the version is apparently on the main dialog, if you use another program, could it be this firmware is out of step with the BIOS version and that is the cause of the instability? Suggest you dig around on some forums and decide what to do there.

I suspect what you have is corruption of CMOS memory due to not following the manual.

https://dlcdnets.asus.com/pub/ASUS/mb/LGA1200/PRIME_Z590-A/E17776_PRIME_Z590-A_UM_WEB.pdf

To clear the CMOS memory. close the CLRTC the jumper as indicated in the manual.

This action results in the “logic high” of the battery good circuit not being removed, so the motherboard can detect the logic low forced at the memory supply as an instruction to default the settings and produce the BIOS settings screen at POST. Removing the battery causes both voltages to droop, but the memory one may actually droop more slowly as it’s not running the RTC clock, which produces an illogical input defined by an analogue variability and you wonder why the result isn’t predictable in short periods of removal?

The next thing I get to is the line “750w PSU decided to break down” – did you do a power budget at all? Lett’s make some assumptions, dangerous we all know, but we need a figure..

Motherboard, 750W (The manual says that is the minimum so that’s where to start!), memory , add 10W (again, no information)

CPU add 125W (https://ark.intel.com/content/www/us/en/ark/products/212321/intel-core-i911900kf-processor-16m-cache-up-to-5-30-ghz.html)

Cooler add 1.56W (yes there is a spec sheet for the cooler with information missing from details on other more expensive items..)

Sorry I have to do something. You get the idea. You need a PSU to sustain at least 30% more than you calculate if you want it to last as that initial gulp of power will be that much bigger.

My postulation is your PSU is too small and unstable (AKA “cheap”, “bargain”) to sustain the peaks when all that gear draws its first large breath of juice at power on, and that your patently adventurous overclocking is making it worse.

Overclocking should be done slowly – burning in the components (especially memory) by working them hard at default settings for a COUPLE WEEKS may improve the outcomes when you nudge the settings higher. also look at the relative timings of your kit – you can speed up one area, causing another to slow down as transactions miss their timing slot. again the steps are not analogue – it hits the time slot, or misses and takes the next slot in which case you lose.

1 user thanked author for this post.

wavy

I’m guessing you had one of those builds with a largish recovery partition 4 and a separate image partition 5 with a nice fat disk image under factory\dell or similar on that drive. (often there’s also an image of the boot partition as well though one poached from another machine seems to “work”) – They’re WIM images; have successfully rebuilt with these (with DISM) on a replacement drive having partitioned it in the usual layout. That’s if you want all the extra software, of course

If you ever do use the Dell software (download and update)  might be worth tracking down the fixes for CVE-2021-21571, CVE-2021-21572 and CVE-2021-21551. There seemed to be a lot of discussion about these and then absolutely nothing much about there being a solution. Of course if you don’t use Dell update you have one problem nailed, but never find out about the other!

The answer to both questions is “yes”.  The EFI holds the boot files, the NTFS your Windows installation, unless you mean the 800Mb extra to that. If that is the case I would have to say leave it alone – you can render systems unbootable, or recovery methods ineffective, by reordering or creating partitions. It’s not always obvious or possible to fix it either!

The layout you have only differs from the UEFI layout Microsoft indicate on the page below as the MSR isn’t a “normal” partition – more a sort of scratch space which is desirable, but not worth risking breakage of the disk structure to achieve (as Microsoft can’t tell if you have an OEM recovery solution, or how that might operate, and such a solution trying to operate on the wrong partition [if a partition were added] would potentially fail.)

https://docs.microsoft.com/en-us/windows-hardware/manufacture/desktop/configure-uefigpt-based-hard-drive-partitions

 

Microsoft want to run some of their security processes on a processor and software inaccessible to the operating system. It’s plain why they want to do that looking at the potential liabilities if they were found culpable for an exploit of their software – it’s risk management forced by the current malware environment. Also encryption incurs a small penalty in performance terms so if that’s dealt with by an assembly specialised in the operation rather than the CPU then the CPU can be doing better things and the encryption handshaking might be a shade quicker- a bit like the graphics workload has increasingly been passed from CPU to GPU.

Longer term Microsoft want to be selling software without incurring the wrath of the right to repair movement Susan Bradley refers to, so they need a business model where the old kit becomes less desirable, but without being rendered obsolete or unrepairable. Machines without TPM are generally older but not any any means past use so Microsoft is just making them less desirable..

On the other side the OEMs building the new systems need to be able to install the software without working through encryption (as that would require they handle the various parameters involved) so the current UEFI based method where the BIOS propagates settings that start encryption without user intervention at the end of the out of box process works fine for them. The TPM there is a part of the UEFI specification so without it though secure, the spec is incomplete. (I had a look having been taken to task recently – I believe a TPM this is what they mean by https://uefi.org/sites/default/files/resources/UEFI_Spec_2_8_final.pdf , section 2.6.2, item 26 among others.)

The question you would have to ask if what will happen in Windows 11 should the TPM fail after it’s been running a while – would the software then stop working? Someone’s going to hit that problem someday.

Here’s another question-

What should Microsoft build in Windows 11 to make it really desirable to the world in general.

If you can answer that for Microsoft.. it’s got to cost, we’ll be paying somehow.

There are places you can check the gossip with a bit more reputation than a random Lithuanian press release, and security news a bit ahead of the BBC (who’s click offering this week is more concerned with miscreant users than miscreant devices.. printmare? Petitpotam? they’re not news.. not a mention.)

https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=Huawei+

https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=Xiaomi+

https://www.bleepingcomputer.com/news/security/us-designates-chinas-huawei-and-zte-as-national-security-threats/

Both the above have “blog” pages good for knowing when you need to kick Chrome up the backside to get an update before you browse if nothing else.

https://nvd.nist.gov/general/nvd-dashboard

https://www.bleepingcomputer.com/news/security/

This does make the case for switching your phone off. (I haven’t got one which isn’t attached to a wall).

Smart phones seem to choose to install updates when you do that just like our favourite operating system. I have to deal with complaints from my mum about that (Yes, you read that right. She does more actual electronics based communication than I do now, I’m trying to keep up here! No she wasn’t a “teen mum”..)

Then again where I used to work we had a customer, a young lady of 86, who needed regular visits to update graphics drivers and the like as she thrashed those a quarter her age at Elder Scrolls, Everquest and the like!

 

 

 

1 user thanked author for this post.

Mr. Austin

Where did you download Windows 11 from? Oh, wait, – you know what your doing..  If that’s the standard of dialog MS can muster for the first thing you see from their front line product, I hope it’s not a reflection of the overall quality (as ultimately we’ll all end up having to use it or change to Linux etc). The stilted dialog reads a bit like one of those scam emails you just delete.

 

Guess what. you’re not alone.

Unfortunately employers in these parts (+044) are primarily interesting in keeping your workflow moving, not training for the future so I found myself redundant (the employer now gone in no small part due to their lack of a forward looking stance) past 50 years with qualifications most employers laugh at (or plain don’t remember) and 20 years experience in all manner of computer repair and OEM software deployment which nobody does within a sensible travel to work area (also somewhat nailed down by poor health..)- even my last (self funded) HND Electrical and electronics I used to get the job is now superseded by a FDSC and to be honest the reason I’m where I am is because the kit is now largely on a physical scale where economic repair by hand isn’t possible (as the faults now also tend to be PCB damage by bending or water damage as a factor behind the problems due to portability) so IT “experience” is all I have..

From Microsoft you only see the IT side of the cloud. Maybe look into things like cloud connected arduino / Pi applications and SCADA, as even a lot of the backwards companies are  looking to migrate old systems to the cloud to attain visibility of their processes as loss of say  half a day’s product, or a massive fine for environmental pollution, or both,  due to an undetected fault  is a pretty big incentive. PLC programming is also a thing – we did it with ladder logic on the course 20 years ago to find in the immediate “real world”  programming was done in C, and the programs in deployed units were locked in so you couldn’t even look at them (and reprogramming fees if you pulled that wrong cable and lost the program ran to £hundreds) and to be honest 90 percent we of repairs were one model and they all only had relay board faults. Thankfully they wrote a hardware test mode in the program.

Of course only those who have been around a few years are going to recognise the industrial implementations of the PC, let alone be able to work out how you might drag the implementation into the 21st century without shutting down the plant to analyse or replace the controller. If your lucky you get to work a few minutes at a time. If you’re unlucky they call you after it stops working or starts smoking (seriously!)

What did I do before Google existed? fixed welding machines (Mig, Tig, gas and plasma cutting), handled the supplies, and worked for a while in electronic assembly and even worked in oilfield drilling fluid laboratories.

Next bets, renewables .. recycling .. If I get lucky, retiring..

1 user thanked author for this post.

SueW

Never thought to try that. We can all learn things! Thanks.

So there isn’t a problem, but there is a difference.. but hey, it’s faster. Microsoft provided a tool for handling installation media so using it I feel to be suitable as set against maybe undoing the problem you cite by using IACLS to add S-1-0-0 with full control or something flaky.. (not going there with xcopy /k /n /o – that is so slow.)

I used to work for a small OEM so preparation speed wasn’t a concern – we applied the “no system files” recovery to the boot partition of UEFI media on SSD as indicated (as updating the PE at every version was a bit too much – we did little with answer files we couldn’t achieve elsewhere), we modified the boot.wim  file to incorporate our scripts and stationed the FFU images on what would usually be the Windows partition, so we could boot that SSD populated USB3 caddy and load the drive in a few minutes (needless to say we had quite a few caddies, replicated from a master). They did try PXE boot installations, but other users didn’t like the effect on network traffic when loading more than one machine (fairly gutless, some smart switches.. but the good stuff is at the top of the tree..) and patching applies to the boot media where the network is connected to the Internet and patching a PE is a bit of a grind (and it grows somewhat in size we found. probably did something wrong…)

The free space I gave is a bit higher as Windows needs space to work, you need near 4Gb for the archive, and last time I bothered installing Windows (1903) it inflated to 15GB (six products in there according to DISM) so that’s 19Gb for the work files and 11Gb for Windows to do its thing temporary file wise.

I believe we are discussing the difference between “can” and “should” so I’ll give you your glory .. my experience remains there never seems to be any DVD media to hand when you need it.. Oldguy is into avoiding viruses where he can so shopping for media is out..

Given the nice red text at the end of your post, I take it you realise you could achieve what we did but as back up and restore as you promote there, using just the tools in the Windows box – you would have to handle the partitions (creation and content) one by one (as FFUs have to be prepared with sysprep) but the joy of UEFI is you don’t have to fix the BCD – it just boots even if you started by partitioning a blank drive and load the content.

I was actually trying to offer our guest some background information supporting how and why B arrived at his recommendation. Then again I really should be elsewhere rather than typing just now..

 

The USB and ISO are two forms of the same software. Unless you specifically keep a USB key for the job, you probably need the ISO.

The recovery drive made from a PC (start > run > recoverydrive) is subtly different – if you copy the system files (default) it copies various windows files which were specifically needed for your Windows installation, so your PC comes out as it was in Windows installation terms (though obviously everything else isn’t put back unless you did a refresh instead of a reset), so basically the difference is not needing any updates which came out since the ISO you used, which could be months old..

If you have the ISO on your PC (… and find you can’t burn it as you cant find a double layer DVD … ) you can double click it in recent Windows 10 versions and mount the ISO (which explorer will open so you can see the files and drive letter ). You need to capture the file set from the mounted media and put it on a USB drive (the only speciality in the formatting is the drive is formatted FAT32 and labelled “RECOVERY”..)

Copying files won’t work – it messes with all manner of file attributes; you need to use DISM to replicate the file set, so you need probably 30GB free on your system drive to be sure..

Assuming your ISO is mounted as Drive e: and you have a suitably sized and correctly prepared USB key as drive f:, in an elevated CMD:

dism /capture-image /imagefile:c:\recovery.wim /capturedir:e: /name:name

dism /apply-image /imagefile:c:\recovery.wim /applydir:f: /index:1

Those fortunate enough to be upgrading to a SSD would be wise to keep the wim file. You can partition your new (blank) drive (UEFI hopefully, https://docs.microsoft.com/en-us/windows-hardware/manufacture/desktop/configure-uefigpt-based-hard-drive-partitions). That done, on the new “drive C” make a Windows10Upgrade folder and load the files` by changing applydir above. On the destination system (if you didn’t prepare the drive there) boot recovery to a command prompt, change drive (should be c: for UEFI) and CD to the Windows10Upgrade folder, REMOVE the recovery key and wait for about 10 seconds before running setup from there (or the boot will be placed on the USB which is pretty useless), and note how the install is that bit faster for not waiting for an optical drive to spin up and down and seek every few seconds..

If crystaldisk says the drive is fine and he interface is green then it likely is. it reads the SMART (Self Monitoring and Reporting Technology) information from the drive.

What you seem to have is a sector which is corrupt either as a result of a cell defect or logical damage to the controller level information in one location (a power loss during write could do it as flash devices flash one memory block at a time).

Unfortunately from there the method is to back everything up as the solution is to use manufacturer diagnostics to do a low level destructive write test, which should scrub out the problem or force the drive to reallocate the sector and thus return the drive to operation.

Best to use DISM from a recovery media. If your partition table has a “reserved” partition, that’s a MSR partition – there’s no volume, you just recreate it (the drive is UEFI – example https://docs.microsoft.com/en-us/windows-hardware/manufacture/desktop/configure-uefigpt-based-hard-drive-partitions). If you lack the reserved MSR partition, your drive is in MBR partition layout and you need to fix the boot sector after imaging – arrange to have the data at the final URL here to hand.

the form is (assuming we want to capture drive C:  on to an external drive f:, and e: is another partition on the local drive we can use for scratch space to stop a log jam with bytes going too and fro on the USB cable.. (recovery can be good, but if you cant skip the whole spec and it just takes longer..)

to capture:

dism /capture-image /imagefile:f:\mycdrive.wim /capturedir:c:\ /name:anyname /scratchdir:f:\

*****  Be sure to test extract any image if you are going to destroy the source – the image process can go awry, as with any backup. *****

to apply (assuming the same):

dism /apply-image /imagefile:f:\mycdrive.wim /applydir:c: /index:1

(you can also specify /name:anyname – DISM archives may contain multiple file sets with differing names numbered by the sequence of their addition, you can specify the one you want either way. I would not recommend compressing data for operations traversing different drives, and it’s also slow.)

In the unfortunate situation where the drive is not GPT partitioned you need to repair the boot sector and add the installation to the BCD store (as the new partition will have a different GUID) to get Windows to start, which is a bit fiddly (especially as sometimes you have to use diskpart to assign the boot partition a letter first..):

https://support.microsoft.com/en-us/topic/use-bootrec-exe-in-the-windows-re-to-troubleshoot-startup-issues-902ebb04-daa3-4f90-579f-0fbf51f7dd5d

 

Sounds to me like the CSM set to enabled has arrived of itself – The only thing I can think of is at some point you used the boot menu to boot something that needed CSM and it stayed enabled.

From what you’ve said, I’m with Alex5723 – as per the Microsoft article, turn CSM off, ie disabled. If it doesn’t boot there would have to be something pretty odd going on

You have the secure boot by virtue of not selecting an insecure boot, and supporting UEFI file system so it should all be fine. The thing you might notice is it gets very hard to access the boot or BIOS menus – you’ll need to use the Windows recovery options to access the BIOS settings to enable fast boot at which point you shouldn’t see a “seam” between the POST logo and Windows starting..

all the best..