| fling | What is the proper way of figuring out which patches are for shiftfs and onionfs and other lxd stuff? | 00:45 |
|---|---|---|
| fling | Is there a branch for this somewhere? Or am I supposed to grep git log | 00:45 |
| klebers | fling, we don't have feature branches in our git repos, for some of the changes you can run git log on specific locations, e.g. fs/shiftfs.c | 08:31 |
| fling | klebers: this is what I do but I also want to get related unionfs patches and any others | 08:32 |
| fling | I want the proper way of discovering them | 08:32 |
| fling | Are you saying it is impossible to filter out lxd related commits? | 08:33 |
| klebers | fling, it's hard to filter out everything that affects lxd because we also apply a lot of patches from the upstream stable branches, which could affect lxd without us including the changes as explicit feature request or bug fix for lxd | 08:34 |
| fling | I have stable upstream applied too | 08:36 |
| klebers | it's not impossible, but you need to know where's the code that are lxd related and run git log on them | 08:37 |
| fling | this is what I do all the time | 08:37 |
| fling | But it is not the proper way :D | 08:38 |
| klebers | yeah, unfortunately we don't have a better way to get this kind of information from the repo | 08:40 |
| luna_ | gpiccoli: my Ubuntu boots with the 5.9 kernel now and ZFS drivers :) | 08:55 |
| luna_ | it fixed itself with todays updates | 08:55 |
| luna_ | so closed the bug | 08:55 |
| sem2peie | can someone say something about my problem or do I have to be satisfied with workarounds for the next years? | 11:27 |
| sem2peie | https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1895333 | 11:27 |
| ubot5 | Ubuntu bug 1895333 in linux (Ubuntu) " cannot change the regulatory domain ath10k, QCA9984 (QNAP QWA-AC2600)" [Undecided,Confirmed] | 11:27 |
| sem2peie | maybe sfroschee ? | 11:28 |
| jeremy31 | sem2peie: sounds like https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=927710 | 11:31 |
| ubot5 | Debian bug 927710 in src:linux "ath10k locks to regulatory domain US on ACPI platforms" [Normal,Open] | 11:31 |
| sem2peie | yes exactly that seems to be the same | 11:33 |
| gpiccoli | Great luna_, thanks for the heads-up =) | 11:38 |
| sem2peie | the link also contains a matching patch from the openwrt git | 11:40 |
| jeremy31 | sem2peie: I can look at ath10k source later and see if the 5.4 can be patched using dkms | 11:41 |
| sem2peie | where the patch also enables "Atheros dynamic user regulatory testing" if I see it right | 11:55 |
| sem2peie | https://github.com/openwrt/openwrt/blob/master/package/kernel/mac80211/patches/ath/402-ath_regd_optional.patch | 11:55 |
| sem2peie | jeremy31 yes well i opened the ticket because i think it would be better if the patch was in the ubuntu kernel, there are some cards with 0x0 in the eeprom | 12:40 |
| sem2peie | 0x0 should be considered a widlcard and region free, no idea why Qualcomm is bitching (in the past it was no problem) | 12:40 |
| sem2peie | at the end of the day this only causes problems (frequencies that cannot be used or a too high transmission power) | 12:41 |
| sem2peie | or the cards cannot be used as AP because of the IR flag | 12:42 |
| sem2peie | they are well tested and in used worldwide | 12:43 |
| luna_ | no problem | 12:48 |
| fling | Which repository/branch for latest 5.4 and 5.10? | 13:03 |
| fling | Which tree for 5.10 ? | 15:18 |
| arighi | fling, https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/unstable/, branch 'master' | 15:33 |
| fling | arighi: thanks! | 15:35 |
| arighi | yw | 15:44 |
| fling | arighi: how to flip a symbol in defconfig for every architecture? | 16:08 |
| fling | arighi: for example to enable shiftfs by the default | 16:09 |
| arighi | fling, hm.. shiftfs should be already enabled for all the supported architectures | 16:10 |
| arighi | fling, but you can take a look at debian.master/config/config.common.ubuntu | 16:10 |
| arighi | change the option there and the config should be updated during the build | 16:11 |
| fling | arighi: sure but what is the right place to switch the default symbol value without debian directory? For example for enabling zfs or disabling reiserfs etc | 16:12 |
| arighi | fling, if you're going to use make directly (skipping the debian/ubuntu stuff) you can simply run a `make menuconfig`, then press '/' to search for a specific config name and then you can change it | 16:13 |
| arighi | oh wait, I think I understand what you mean | 16:14 |
| fling | arighi: sure but I want to change it in the sources | 16:14 |
| fling | just don't know the proper paths :> | 16:14 |
| arighi | fling, for example for SHIFT_FS, edit fs/Kconfig | 16:15 |
| fling | ok, thanks | 16:15 |
| arighi | and add "default m" for example | 16:15 |
| arighi | below the tristate line | 16:15 |
| arighi | that will affect all the architectures | 16:15 |
| fling | got it! | 16:16 |
| fling | How to figure out which commits are related to lxd? Is not there a good list? | 16:16 |
| fling | I don't like grepping for UBUNTU and shiftfs and overlayfs | 16:16 |
| sem2peie | do I see it correctly that the patches are already included in the 5.8 kernel? https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/focal/log/?h=hwe-5.8&qt=grep&q=ath10k | 16:23 |
| arighi | fling, I think that information should be documented somewhere in lxd | 16:37 |
| arighi | fling, there's something here https://lxd.readthedocs.io/en/latest/requirements/, even if it doesn't tell you exactly which config options are required | 16:39 |
| fling | arighi: I'm not asking about config options but about ubuntu patches | 17:04 |
| fling | arighi: for example I know lxd can use shiftfs and ubuntu adds shiftfs patches to the kernel for this and overlayfs patches for it to work with shiftfs | 17:04 |
| sem2peie | gets better and better with 5.8.0-25-generic #26~20.04.1-Ubuntu is missing the complete midband > https://pastebin.com/D2NYBxBR | 17:06 |
| arighi | fling, hm... then I think the best way is to look at the git log and try to identify the SAUCE patches that are required by lxd (unless someone has a better idea...) | 17:06 |
| fling | this is what I did… | 17:07 |
| fling | just looking for the proper way :D | 17:07 |
| arighi | yup, that is the most "proper" way that I can see :) | 17:26 |
| fling | What does 'noup' mean in 'UBUNTU: SAUCE: (noup)' ? | 17:36 |
| arighi | fling, it seems that they're missing a reference to the upstream patch/commit, so maybe noup means something like "not upstream"? But I'm not really sure about that, personally I've never used that in an UBUNTU: SAUCE patch (and I haven't seen it used recently) | 17:59 |
| arighi | so I would say if they have 'noup' they're likely old patches (and probably you don't need them) | 17:59 |
| fling | arighi: but they are in the repo | 18:37 |
| tyhicks | fling: this isn't fool proof but you could look for patches with "SAUCE" in the subject line as those are typically out-of-tree patches that have been applied for some specific use case that Ubuntu is interested in | 19:21 |
| tyhicks | fling: https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/focal/log/?qt=grep&q=SAUCE | 19:22 |
| fling | tyhicks: this is what I've been doing from the beginning | 19:22 |
| fling | I picked ones with namespace, cgroup, shiftfs, overlayfs in the description | 19:23 |
| tyhicks | you might keep an eye on the ones with apparmor in the description, as well | 19:23 |
| tyhicks | LXD makes heavy use of AppArmor for container confinement | 19:24 |
| fling | hmmm ok, thanks | 19:24 |
| fling | n=0001 ; for c in whatever ; do git format-patch -1 --start-number $n $c ; n=$(($n+1)) ; done | 19:24 |
| fling | I'm extracting patches like this ^ | 19:24 |
| fling | ^ but what can I use instead of this n number? | 19:25 |
| fling | I want something more persistent | 19:25 |
| fling | so if the patch is going to get upstreamed | 19:25 |
| fling | all other patches will keep their numbers when I will be extracting them after | 19:25 |
| tyhicks | sorry, I don't understand what the number is signifying here | 19:25 |
| fling | just an incrementan number | 19:26 |
| fling | first patch is 0001, second patch is 0002 etc | 19:26 |
| tyhicks | the persistent part is the subject of the patch but that's also not very persistent because it could easily change during the upstreaming process | 19:26 |
| tyhicks | you're working with out-of-tree patches that are not yet upstream which, by definition, do not have persistence | 19:27 |
| fling | tyhicks: I thought I could use date instead of this number but dates are out of order too | 19:28 |
| tyhicks | fling: oh, you also want to keep an eye out for SAUCE seccomp patches as the container folks are making use of seccomp for various features | 19:29 |
| tyhicks | stuff like this is for LXD: https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/focal/commit/?id=965dde940df5444bdcff23dad0e7e14e185842a0 | 19:30 |
| fling | good catch | 19:30 |
| tyhicks | if Christian Brauner wrote it and it is applied to the Ubuntu kerenl tree, you need to take a look at it because there's a good chance that it is LXD related | 19:31 |
| fling | ubuntu-focal/master-next has way much more related SAUCE stuff in it compared to ubuntu-unstable/master | 19:37 |
| fling | I should investigate ubuntu-focal/master-next better | 19:37 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!