<noneucat>
the pinephone may also need an extended cma allocation in order to switch cameras
<noneucat>
i'm getting an allocation fail when i try
<noneucat>
going to test with cma=64M
<samueldr>
would need to be fixed in the device tree imo, but good to test
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-aarch64
ajs124 has quit [*.net *.split]
grw1 has quit [*.net *.split]
ekleog has quit [*.net *.split]
Valodim has quit [*.net *.split]
dustinm has quit [*.net *.split]
wrmilling has quit [*.net *.split]
dustinm has joined #nixos-aarch64
ekleog has joined #nixos-aarch64
Valodim has joined #nixos-aarch64
grw1 has joined #nixos-aarch64
ajs124 has joined #nixos-aarch64
wrmilling has joined #nixos-aarch64
patagonicus7 has joined #nixos-aarch64
Valodim- has joined #nixos-aarch64
Amanda has joined #nixos-aarch64
Amanda is now known as Guest28275
wrmilling_alt has joined #nixos-aarch64
h0m1 has quit [Ping timeout: 264 seconds]
bigvalen_ has joined #nixos-aarch64
h0m1 has joined #nixos-aarch64
patagonicus has quit [*.net *.split]
AmandaC has quit [*.net *.split]
bridge[evilred] has quit [*.net *.split]
bigvalen has quit [*.net *.split]
delroth has quit [*.net *.split]
ajs124 has quit [*.net *.split]
grw1 has quit [*.net *.split]
Valodim has quit [*.net *.split]
wrmilling has quit [*.net *.split]
patagonicus7 is now known as patagonicus
bigvalen_ is now known as bigvalen
Valodim- is now known as Valodim
ajs124 has joined #nixos-aarch64
delroth1 has joined #nixos-aarch64
grw1 has joined #nixos-aarch64
red[evilred] has joined #nixos-aarch64
<red[evilred]>
sphalerite (IRC): Monday!
hexa- has quit [Quit: WeeChat 2.9]
hexa- has joined #nixos-aarch64
cole-h has quit [Ping timeout: 256 seconds]
<Ke>
note that using even bigger cma does not mean the memor is wasted, normal page cache pages can still be allocated in cma region for most filesystems
<Ke>
and I believe also user pages
<Ke>
some operations may require cma page to be migrated away from cma though
<samueldr>
about right according to what I could find
<samueldr>
but still, setting it on the kernel command-line should be left for the final end-user, according the kernel sources
<samueldr>
in order, it should be set by a kernel config, as a default, device tree, for what uses that device tree, or as a kernel command-line option
<samueldr>
though the latter will set the cma in ways that may break on some platforms
<samueldr>
one platform being the raspberry pi 4B with more than 1GB of ram
* samueldr
sighs
<samueldr>
tbf, we now can't assume there won't be another platform with such an issue
<Ke>
I definitely agree that base config should stay out of the way as much as possible
<Ke>
for what lies outside of nixpkgs I just cut open and mix to my personal configs as I see fit
<samueldr>
though, I'm not against increasing the default CMA in the default nixos kernel either, I left it to the default we were already setting, because it wasn't the goal of that PR to change cma globally
<samueldr>
(still, for the pinephone I would patch the kernel sources and configure it as needed)
<samueldr>
(as a device tree entry)
<Ke>
I guess the question is, how small devices do need to be supported
<Ke>
and are there too many of them to have individual overrides
v0|d has quit [Remote host closed the connection]
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
<clever>
samueldr: the pi problem that i remember, is that DT specifies an addr+size for the cma, but kernel cmdline giving just size, puts it at the wrong addr, and breaks everything
<samueldr>
yes
<samueldr>
exactly
<clever>
cmdline can set addr as well, but then its not one-size-fits-all
<samueldr>
the kernel, when going from the cmdline, uses a specific start address IIRC
<clever>
ive also recently been lookint at the pi4 gigabit ethernet, and it has some rather nasty problems
<samueldr>
something like from the end of the memory range
<samueldr>
but yeah, we do have the solution, and it's well understood... it's simply that we shouldn't be setting cma= anymore as a distro
<gchristensen>
flokli: gotcha, though I think the time sync is fine once it boots to nixos, but it can't boot since the cert is expired
FRidh has quit [Quit: Konversation terminated!]
<gchristensen>
" You should be aware that there is no security or authentication used for NTP, and so using the ntp command is effectively equivalent to ignoring the validity period of any X.509 certificates."
cole-h has quit [Ping timeout: 272 seconds]
srk has quit [Read error: Connection reset by peer]
jb55 has quit [Read error: Connection reset by peer]
srk has joined #nixos-aarch64
jb55 has joined #nixos-aarch64
orivej has joined #nixos-aarch64
<flokli>
Weeeel... No
<flokli>
It depends on your threat level of course
monk has joined #nixos-aarch64
<flokli>
But you could be a bit smarter, and look at some known-good last modified timestamp on the filesystem, and refuse to be thrown into past mir than that
<flokli>
There's also ntpsec I think (?)
<flokli>
But yeah, figuring out what time it is is a tough problem.
<noneucat>
DigitalKiwi: wonder what the epoch time is for that? may be some correlation :)
<DigitalKiwi>
flokli: funny story: i think my time was wrong when in installed nixos on that machine because i had a lot of files and problems with software being like...this is from the future!
orivej has quit [Ping timeout: 264 seconds]
<flokli>
Heh
<flokli>
Well, then some time was wrong for sure
orivej has joined #nixos-aarch64
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-aarch64
rajivr has quit [Quit: Connection closed for inactivity]
<noneucat>
hrmmm
<noneucat>
sxmo contains a workaround for reliable calls while on crust sleep
<noneucat>
which involves a rebind of the modem usb
<noneucat>
it doesn't play nice with the modem control driver which listens for udev events on the usb :(