That’s why I went for the /swapfile option. getting a proper swap set up manually was more work than I wanted to do to this machine to get Debian running, frankly.
I disagree that it is incorrect. swap is overspill for ram, and with 64 gb of ram you should never need any overspill (excluding memory leaks), so it chooses 1GB to avoid consuming masses of space on a potentially small nvme. as has always been the rule of thumb (2x your ram below $someramcapacity, ~4GB below $someramcapacity2, unnecessary over $someramcapacity2)
The fact that swap is used for hibernation on laptops appears not to come into consideration.
Memory leaks are a fundamental consideration. Firefox is particularly bad for this; and I don’t want all those unused allocations consuming actual RAM. I’m much happier with them out of the way, in swap.
Also, yes, hibernation; which is one reason why I picked the same swap as RAM.
So, I’ve been trying to make a bootable USB stick with Debian Bookworm, firstly as a means to dd backup my boot partition (which is inside an encrypted LVM), and secondly to see if vainfo is happy.
It looks currently like the Bookworm images are broken, except for netinst. When you put them on a USB stick and boot, you get th GRUB loader - not Debian!
A very recent bug was reported and fixed, which apparently led to this, but it’s not obvious to me that bug would be the cause in my case.
So I’m waiting for a while, I guess, to see if Bookworm sorts itself out; I’ve had enough writing images to USB and rebooting for now =-)
Would ventoy work for your case? If it does, it’ll save you a lot of hassle creating the USB over and over. You just have to copy over the new iso. I believe that I first heard about it on this forum. and I’d like to thank whoever noted it here, because it has worked very well for me.
In an effort to fix my vainfo, I installed intel-media-va-driver (21.1.1+dfsg1-1), both the regular and intel-media-va-driver-non-free. Now I get:
libva info: VA-API version 1.14.0
libva error: vaGetDriverNameByIndex() failed with invalid VADisplay, driver_name = (null)
vaInitialize failed with error code 3 (invalid VADisplay),exit
Does anyone get a different output when running vainfo? Is there a different driver/s I ought to install instead or in addition to that? Thank you in advance.
ETA: link, text, vainfo update.
I spent more than I would have liked trying to install Debian 11, including some attempts inspired from this thread and from Debian 11, Gen12th : Wifi working? - #9 by disposedtrolley … then luckily I found out that
https://wiki.debian.org/InstallingDebianOn/FrameWork/12thGen was apparently updated since @John_Grow posted it (?).
Following instructions there now works like a charm. In particular, you need a recent kernel and manual download of missing i915 firmware.
Thanks for sharing your experiences. For most people, we recommend Ubuntu 22.04 or Fedora 37 as the official recommendations.
@Pietro_Battiston how’s the system so far? I’m running Fedora 37 but I’ve had some annoying lockups lately, thinking about giving Debian a try.
After a week or so of normal use (with kernel 6.0.0-6, from backports), the system seems to work just great. Very responsive, I noticed no crashes or lockups whatsoever.
My only concern currently is that in suspend mode the laptop discharges much faster than I would have expected based on my former laptops. But I have yet to do a serious comparison with what other people posted ( e.g. Suspend on linux drains a lot of battery compared to other laptop - #26 by Matt_Hartley , Test results for standby battery use of Expansion Cards ) in order to find out if the OS is somehow to blame (or maybe the expansion cards). I will maybe do a test of Ubuntu 22.04 to see if it fares better.
By the way, the only other very minor annoyance I had was the somehow distracting automatic brightness regulation, but it can be easily disabled: "auto-brightness" aka ambient light sensor in Linux? Details? - #5 by devnull
Also make sure you have the latest beta BIOS.
Well, except that I hadn’t noticed hardware video acceleration is off (which does not visibly affect playback smoothness, but it does imply more heat and battery use). I experimented a bit with installing proprietary drivers (package intel-media-va-driver-non-free ) but no change: vainfo returns “vaInitialize failed with error code 3” (the error code was 5, I think, before I upgraded some packages to bullseye-backports).
No idea if this has to do with me using wayland rather than X11 (which is mentioned in the Debian page about this laptop). Or maybe with me using a later kernel than that described?
Thanks, I had missed this. Not sure when I will be brave enough to try, as it seems to be an irreversible experiment at the moment.
It’s what we recommend. Due to resources and time, we’re focused on Ubuntu and Fedora. And despite shared bits, Debian and Ubuntu differ on a number of areas.
@Matt_Hartley I think the threads on the forum are a type of Q&A or a type of specific topic-related discussion such as Debian 11. The “Something on the Framework Laptop” threads can be spaces where diverse people who have different preferences can stay in this community. For the thread with the type of specific topic discussion, it’s weird to see “[Answered]” prefix on the title. What do you think?
Edited x2: simplifying the sentences, and correcting English grammar.
Agreed for threads of this nature. Threads marked Solved are per request higher up the chain so we can see things are indeed being, resolved.
But yes, Answered can be removed from those threads.
Sure, I understand. I feel expert and risk lover enough to at least try, my aim in this thread is definitely not to complain or discredit Framework because the Debian experience is still not smooth enough By the way, Debian names itself “The Universal Operating System”, so its aim is to support even hardware that does not support Debian
I can understand the benefit of adding the prefix “[ANSWERED]” or “[SOLVED]” to the Q&A-type threads. I agree with the way. And thank you for removing the “[Answered]” from this specific topic-related thread.
I appreciate people staying and working to develop and maintain the Debian packages in the Debian community, and also their contributions to the upstream open source projects. Without people working for the greater good in Debian, and without a thriving Debian eco-system, Debian’s downstream Linux distros such as Ubuntu, Pop OS, and Mint cannot thrive.
Hardware acceleration worked (glxgears full screen) with X11, but it was buggy. I needed a compositor to have things look moderately right, and even then it was unreliable. I switched to Wayland (sway). I believe this is a(nother?) bug in the intel driver (and there are still some, even in Wayland). I would be surprised if this would be a kernel regression, but everything is possible!
That said, ake sure you have the latest linux-firmware
package, missing that might be why you don’t have hardware acceleration. See my section on firmware blobs.
Wow, great review, thanks, I had missed it. This said, I do have the last version of firmware-linux
(20221214-3), as well as firmware-linux-nonfree
(20221214-3), plus the missing blobs I manually downloaded as recommended in the “DebianOn” guide (and indeed, I don’t see firmware errors in dmesg
).
It might be worth mentioning that I do not have the last version of intel-media-driver
, as the version from unstable or even bullseye-backports (either of the free or non-free driver) wants me to upgrade stuff like libc6, something I currently would like to avoid. So I have 21.1.1+dfsg1-1 . What version do you have?