3 ms·

> that can reliably be updated without the boot partition getting full I have the suspicion that this is caused by users who add extra kernels and don't use th

> that can reliably be updated without the boot partition getting full I have the suspicion that this is caused by users who add extra kernels and don't use the system provided ones. Ubuntu retains only a certain number of (distro-provided) kernels. Ubuntu desktop version doesn't even use a separate partition for `/boot`, so there's no risk. I remember this being a problem for Ubuntu Server, but it was long ago (I think 16.04, or maybe earlier). > sleep states being faulty (or the machine not working properly after waking from sleep) This problem can depend on two things. S3 (suspend to RAM) became a bloody mess in recent times (possibly, because of Microsoft's push for the Connected standby). The firmware of some laptops (and even desktops!) doesn't advertise the state. One can verify this with `sudo dmesg | grep "supports S"`: if there is no S3, it's the producers fault. In such cases, I actually don't know how Windows works (for example, if they patch the ACPI tables). If S3 is advertised, but it doesn't work well, then the producer didn't release proper drivers. Best thing is to open a bug in the kernel tracker; some producers do actually react. Kernel v6.1 released a lot of fixes for laptops, so it's worth giving a shot. I'm definitely not justifying the status quo, but it's a chicken and egg situation (small user base -> underdeveloped drivers -> small user base). > random things like thunderbolt docks not working I suspect this is a similar (hardware drivers) problem. > external monitor resolutions being wrong, and monitor scaling working properly when you have two monitors with different resolutions I suspect this is an X11 problem, which can be considered "Linux", although I'm not sure exaclty. Scaling is a bloody mess, and I think it's not related to drivers. To summarize, the fault is a mixed bag. for hardware/driver cases (and they're many), there's nothing to do; some distros adopt kernels earlier, but that's not necessarily a good thing. For display scaling, I suspect there's nothing to do, either - this may still take some time.

reisse 4y ago
> In such cases, I actually don't know how Windows works (for example, if they patch the ACPI tables). It uses Modern Standby (aka S0 Low Power Idle). https://learn.microsoft.com/en-us/windows-hardware/design/device-experiences/modern-standby-vs-s3?source=recommendations https://learn.microsoft.com/en-us/windows-hardware/design/de... > if there is no S3, it's the producers fault Or it's Linux fault for not supporting the S0LP? Sure, there are a lot of problems with the S0LP implementation on Windows, but the idea itself is good - that's what all the phones have utilized for at least previous 10 years. Maybe Linux actually could do it better than Windows, if some effort was put into it?
- pizza234 4y ago
  > Or it's Linux fault for not supporting the S0LP? That Linux should support all the crap in existence is a fair argument, however... > the idea itself is good - that's what all the phones have utilized for at least previous 10 years Phones are different from PCs, in the hardware design, but also in the usage pattern. If something makes perfect sense on a phone, it doesn't imply it does it on the PC as well. I owned in the past two hybrid (Microsoft) laptops, which I carried and used on the way frequently, and still, the idea of a connected standby was evidently nonsensical (even ignoring the horribly broken implementation). If one wants to stay connected, they use a smartphone, and/or a smartwatch. A laptop can't be kept in a pocket and pulled it out in a second; it's also often awkard to use on the way. There are no measurements around; embarrassingly, a presentation by an Intel engineer¹ has zero data, but has photos of cute animals. It certainly consumes more than S3, but how much, is carefully ignored (I bet considerably more). The modern standby, and the convergence narrative in general, is pushed by the companies that couldn't set a foot in the mobile market (Intel and Microsoft), and are trying to appear relevant. I hope it dies, although with M$ behind, it's unlikely that it will. It's tragic that a useless idea clearly invented by business suits, is affecting the hardware landscape so significantly.

reisse 4y ago

> In such cases, I actually don't know how Windows works (for example, if they patch the ACPI tables). It uses Modern Standby (aka S0 Low Power Idle). https://learn.microsoft.com/en-us/windows-hardware/design/device-experiences/modern-standby-vs-s3?source=recommendations https://learn.microsoft.com/en-us/windows-hardware/design/de... > if there is no S3, it's the producers fault Or it's Linux fault for not supporting the S0LP? Sure, there are a lot of problems with the S0LP implementation on Windows, but the idea itself is good - that's what all the phones have utilized for at least previous 10 years. Maybe Linux actually could do it better than Windows, if some effort was put into it?

pizza234 4y ago
> Or it's Linux fault for not supporting the S0LP? That Linux should support all the crap in existence is a fair argument, however... > the idea itself is good - that's what all the phones have utilized for at least previous 10 years Phones are different from PCs, in the hardware design, but also in the usage pattern. If something makes perfect sense on a phone, it doesn't imply it does it on the PC as well. I owned in the past two hybrid (Microsoft) laptops, which I carried and used on the way frequently, and still, the idea of a connected standby was evidently nonsensical (even ignoring the horribly broken implementation). If one wants to stay connected, they use a smartphone, and/or a smartwatch. A laptop can't be kept in a pocket and pulled it out in a second; it's also often awkard to use on the way. There are no measurements around; embarrassingly, a presentation by an Intel engineer¹ has zero data, but has photos of cute animals. It certainly consumes more than S3, but how much, is carefully ignored (I bet considerably more). The modern standby, and the convergence narrative in general, is pushed by the companies that couldn't set a foot in the mobile market (Intel and Microsoft), and are trying to appear relevant. I hope it dies, although with M$ behind, it's unlikely that it will. It's tragic that a useless idea clearly invented by business suits, is affecting the hardware landscape so significantly.