Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by leosarev on Sunday July 21, 2024 @08:14AM
(#41024541)
There is sandboxing API in Windows. It's called running programs in userspace.
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by hello_moto on Sunday July 21, 2024 @10:44AM
(#41025462)
Run what a userspace?
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by layer8 on Sunday July 21, 2024 @12:26PM
(#41026163)
It’s a tough problem, because you also don’t want the system to start without the CrowdStrike protection. Or more generally, a kernel driver is supposedly installed for a reason, and presumably you don’t want to keep the system running if it doesn’t work. So the alternative would be to shut down the system upon detection of the faulty driver without rebooting, which wouldn’t be much of an improvement in the present case.
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by ratorx on Sunday July 21, 2024 @01:27PM
(#41026586)
I can imagine better defaults. Assuming the threat vector is malicious programs running in userspace (probably malicious programs in kernel space is game over anyway right?), then you could simply boot into safe mode or something instead of crashlooping.
One of the problems with this outage was that you couldn’t even boot into safe mode without having the bit locker recovery key.
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by layer8 on Sunday July 21, 2024 @01:31PM
(#41026621)
You don’t want to boot into safe mode with networking enabled if the software that is supposed to detect attacks from the network isn’t running. Safe mode doesn’t protect you from malicious code in userspace, it only “protects” you from faulty drivers. Safe mode is for troubleshooting system components, not for increasing security.
I don’t know the exact reasoning why safe mode requires the BitLocker recovery key, but presumably not doing so would open up an attack vector defeating the BitLocker protection.
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by Uvix on Sunday July 21, 2024 @03:44PM
(#41027579)
Normally BitLocker gets the key from the TPM, which will have its own driver that's likely disabled in Safe Mode.
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by discostrings on Sunday July 21, 2024 @04:43PM
(#41027930)
The BitLocker configurations I've seen over the last few days don't require the recovery key to enter safe mode.
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by sm_1024 on Sunday July 21, 2024 @06:23PM
(#41028793)
Doesn't microsoft support eBPF on Windows?
Re:Initial details about why CrowdStrike's CSAgent.sys crashed
by nullfrigid on Monday July 22, 2024 @10:18AM
(#41034691)
If you know the answer why are you asking the question?
6) some form of sandboxing/error handling/api changes to make it possible to write safer kernel modules (not sure if it already exists and was just not used). It seems like the design could be better if a bad kernel module can cause a boot loop in the OS…