Critical Flaw in Windows 10 Could Corrupt Your Hard Drive

[R-T-B]

As long as you don't use the built-in Windows Firewall, that's not really a problem.

Yet. But someday it will be. Hopefully by the time that comes around though most users will have migrated.

Write-buffering actually completes the writes as some point.

True. At least, usually true. There are some exceptions but not generally worth mentioning. JFS on linux for example, has the amazing property of a write buffer that won't flush until it's full: that's right, no timeout.

No one really uses that anymore though.

 

[Aquinus]

I still wouldn't trust it. If I knew that a file was corrupt, I'd wouldn't trust any repair of it. Replace it from a known good source.

Now there's the rub. Without a checksum, how are you going to know that data become corrupted in the first place if you don't have a signature to check it against? You won't until you try to use it and if (and only if,) the application using it throws an error due to the corruption as opposed to still continuing to operate with the bad data.

Write-buffering actually completes the writes as some point.

So does CoW. Metadata gets copied, but why copy the data when it doesn't change? The data is already there and it's not going to change. If you as the consumer of data get your data when you ask for it, why do you care if under the hood there is structural sharing to save on space along with all the other benefits of going that route? You shouldn't. You should however care about things like snapshotting and disk space optimization because it enables you to have more control over your data without having to do all of this yourself and things like optimizing disk usage gets you more out of your drive. That's definitely not a bad thing.

 

[trparky]

has the amazing property of a write buffer that won't flush until it's full: that's right, no timeout.

I'm not sure I like that concept. We've already had discussions in other parts of these forums where people have complained about how SSDs (except for those expensive enterprise drives) don't have some form of onboard power device to complete the write transactions in case of a power failure. Or am I barking up the wrong tree here?

Now there's the rub. Without a checksum, how are you going to know that data become corrupted in the first place if you don't have a signature to check it against? You won't until you try to use it and if (and only if,) the application using it throws an error due to the corruption as opposed to still continuing to operate with the bad data.

For my most important data, I make my own checksums. Do I wish I didn't have to do that? Yes. But I only do that for about 1 to 2% of the data that I store, mostly ripped DVD ISO files (yes, I own them). As for the rest of the six or seven terabytes of data that I have, oh well... it's not a big deal.

If you guys must know, I have a TV series set that is really old and the DVD set is experiencing bit-rot. Some of the disks are physically delaminating. Needless to say, I have all the disks ripped to ISO files on my drives and store them in multiple locations with manually created SHA512 checksums.

 

[R-T-B]

I'm not sure I like that concept. We've already had discussions in other parts of these forums where people have complained about how SSDs (except for those expensive enterprise drives) don't have some form of onboard power device to complete the write transactions in case of a power failure. Or am I barking up the wrong tree here?

No, you are completely correct.

 

[trparky]

I thought so. Any write buffer that doesn't start spooling the data to the disk as fast as possible opens the door for data corruption the likes of which... yeah, I really don't want to think about it.

 

[lexluthermiester]

Yet. But someday it will be.

If you really think that, you need to brush up on how firewalls work. If a person using a third party firewall that is currently being maintained and continuing to support Win7, if properly configured, that user will not be exposed to anything that will put them at risk.

Hopefully by the time that comes around though most users will have migrated.

Time will tell.

No one really uses that anymore though.

Likely because it's not very useful.

I thought so. Any write buffer that doesn't start spooling the data to the disk as fast as possible opens the door for data corruption the likes of which... yeah, I really don't want to think about it.

Exactly.

 

[R-T-B]

If you really think that, you need to brush up on how firewalls work.

No, I don't. a firewall can't supersede the TCP/IP stack that accepts the incoming connection in the first place. Find a hole in that and you've got the keys to the kingdom.

It'll take time. It took A REAL LONG TIME for 2000 to get there. I don't think XP is even there yet. But it'll happen, because software isn't perfect. As I said, I doubt it'll be a real issue because by then, the install base will be gone. We hope anyways.

Likely because it's not very useful.

It's fast and has low cpu cycles, but it's largerly from IBM datacenter land and doesn't care if you lose power, screw you in that situation lol. You also should be using ECC according to them. Yeah. Old school IBM exec logic... lol.

NTFS as far as a filesystem has it's lineage in HPFS from OS/2. It's a very old design. That doesn't mean it's bad though. It's decently vetted and proven. But Microsoft is aware it is aging, and is working on a replacement already. That's why I mentioned ReFS. It's already available in Server, but it doesn't support extended attributes yet, so you can't install to it.

 

[lexluthermiester]

No, I don't. a firewall can't supersede the TCP/IP stack that accepts the incoming connection in the first place. Find a hole in that and you've got the keys to the kingdom.

Oh yes you do. Most good firewalls use kernel level implementations to regulate network traffic coming into and out of a system being firewalled. I'm aware that IT security is your line of work, however that doesn't mean you have nothing to learn.

But Microsoft is aware it is aging, and is working on a replacement already.

NTFS doesn't need replacement, it needs refinement. Nothing more. You don't fix something that isn't broken.

 

[R-T-B]

Oh yes you do. Most good firewalls use kernel level implementations to regulate network traffic coming into and out of a system being firewalled. I'm aware that IT security is your line of work, however that doesn't mean you have nothing to learn.

Your saying they use their own TCP stack? I really doubt that. Mind you stack level exploits are really rare. Like I said the last I'm aware of is in Windows 2000... or maybe it's even NT, I forget. They aren't common, is the point.
Regardless, I basically agree with you.

NTFS doesn't need replacement, it needs refinement. Nothing more. You don't fix something that isn't broken.

In the same sense that FAT32 doesn't need replacement, yes (I'm aware they are not in the same league but they both fill roles, so bear with me). See, it's good at what it does but having a newer flashier "heir to the throne" would not hurt the PR people. That's what MS is after, honestly. ReFS is slowly being retrofitted for that, I feel.

 

[trparky]

Your saying they use their own TCP stack? I really doubt that.

Exactly. The firewall still needs to insert itself into the networking stack. If at any stage of the networking stack there is a vulnerability below that of where the firewall is loaded, as @R-T-B has said before... you have the keys to the kingdom.

 

[R-T-B]

Exactly. The firewall still needs to insert itself into the networking stack. If at any stage of the networking stack there is a vulnerability, as @R-T-B has said before... you have the keys to the kingdom.

But again, these kind of exploits are rare, because the best coders are hired for that sort of stuff. Because they know that's entry point #1.

Not impossible, but pretty far out.

So we both have points here.

 

[trparky]

But again, these kind of exploits are rare, because the best coders are hired for that sort of stuff. Because they know that's entry point #1.

Ah but I still remember the days of using punters back in the old Windows 9x days. Oh those were the days. Losing an online game? Load the punter and sit back as your opponent dropped out of the game. I believe it used a malformed ICMP packet that crashed the old Windows 9x TCP/IP stack.

 

[R-T-B]

Ah but I still remember the days of using punters back in the old Windows 9x days. Oh those were the days. Losing an online game? Load the punter and sit back as your opponent dropped out of the game. I believe it used a malformed ICMP packet that crashed the old Windows 9x TCP/IP stack.

9x was kind of an abomination in that regard, lol.

I think it's the last one Bill Gates actually had lines of code in too. Makes me wonder how competent he really was...

 

[trparky]

9x was kind of an abomination in that regard, lol.

I think it's the last one Bill Gates actually had lines of code in too. Makes me wonder how competent he really was...

Oh yeah, it was only when Windows XP came out that the consumer world graduated to using a real OS. Everything before Windows XP was an absolute dumpster fire.

 

[windwhirl]

Everything before Windows XP was an absolute dumpster fire.

Hey, my Windows 2000 nostalgia is insulted by that comment!

I do agree, though. I would go even further and say XP before SP2 was garbage.

 

[trparky]

Hey, my Windows 2000 nostalgia is insulted by that comment!

Windows XP was essentially Windows 2000 with a candy shell.

I do agree, though. I would go even further and say XP before SP2 was garbage.

I agree.

 

[danwat1234]

Yep. I have another unwanted forced update last night.

And YES I have done all the registry policy and O&O shut ups to stop forced update. But somehow, MS, automagically does it anyway.

You can COMPLETELY disable Windows updates on Windows 10. Tested on 1903,

First go to Services and shut off Windows Update and disable it. Refresh Services to make sure it isn't running.

Now navigate to \windows\system32\ and find files wuaueng.dll and wuauclt.exe. If you can find the former, then look for wuauserv.dll.

For each, go to properties, security, advanced.
Click change owner and type in your user name, Click Check Names to select your user name and Click ok. Then Click Apply or Ok on the main window and close it and reopen it.
Now, you can change permissions for all users.
Delete/Remove permissions from all users and Click Ok.

If that doesn't work, then change owner to Administrator, close the window and try again.

That's it.
To re-enable, add "Read/Execute" permissions to System on wuaueng.dll or whichever dll you have. Doesn't need it on wuauclt.exe for some reason.

 

[lemonadesoda]

...we need an update script, to allow permissions, force manual update, then deny permissions.

 

[R-T-B]

...we need an update script, to allow permissions, force manual update, then deny permissions.

I mean you basically described WuMgr.

 

[lemonadesoda]

RTB, thanks for the tip. Installed. Looks nifty. But look at this screenshot:

Please tell me what i am doing wrong. Look, the screenshot shows that "Disable Automatic Update" was ALREADY set. But guess what, Windows autoupdates, auto reboots. Can you lot even begin to understand my frustration!

 

[R-T-B]

RTB, thanks for the tip. Installed. Looks nifty. But look at this screenshot:

View attachment 187436

Please tell me what i am doing wrong. Look, the screenshot shows that "Disable Automatic Update" was ALREADY set. But guess what, Windows autoupdates, auto reboots. Can you lot even begin to understand my frustration!

Try running as Administrator? Maybe it'll unlock other checkboxes there, I usually use "Disable Update Facilities."

 

[lexluthermiester]

Can you lot even begin to understand my frustration!

We really can.

Try running as Administrator?

This and....

"Disable Update Facilitators"

...this. Also check the "Disable Store Auto Update"

 

[InVasMani]

Is this specific to the C:/ drive only or any drive!?!? Curious if you can use this to corrupt then reformat a bitlocker encrypted drive.

 

[bsvols]

Hi,
1709 was the best performer benchmark wise.


I use update mini tool it hasn't failed me yet.

"I use update mini tool it hasn't failed me yet." ***** This!

 

[lexluthermiester]

"I use update mini tool it hasn't failed me yet." ***** This!

Why? Have you had problems with it?