<Xogium>
marex: I looked into the code. But my level in c is newbie so I didn't understand 99% of what it does and how
liberated has joined #u-boot
srk_ has joined #u-boot
srk has quit [Ping timeout: 272 seconds]
srk_ is now known as srk
<Xogium>
btw is this the proper way to test that both boot_device=mmc and boot_instance=1 ? I.e: are both expressions evluated, or not ?
<Xogium>
if test $boot_device = "mmc" && test $boot_instance = "1"; then echo 2; fi
stipa has joined #u-boot
ikarso has joined #u-boot
<Xogium>
also, is device assignment stable in u-boot by default, i.e: will mmc1 always be mmc1, or do I also need to alias them like in the kernel dt ? If so, which version of u-boot do I need for this to work ?
stefanro has joined #u-boot
liberated has quit [Ping timeout: 264 seconds]
monstr has joined #u-boot
monstr has quit [Ping timeout: 246 seconds]
xroumegue has quit [Quit: WeeChat 2.3]
ldevulder has joined #u-boot
apritzel has joined #u-boot
indy_ is now known as indy
mckoan|away is now known as mckoan
apritzel has quit [Ping timeout: 255 seconds]
monstr has joined #u-boot
goliath has joined #u-boot
xroumegue has joined #u-boot
frieder has joined #u-boot
project10 has quit [Quit: Ping timeout (120 seconds)]
project10 has joined #u-boot
matthias_bgg has joined #u-boot
matthias_bgg has quit [Ping timeout: 255 seconds]
matthias_bgg has joined #u-boot
prabhakarlad has joined #u-boot
zear_ has joined #u-boot
zear has quit [Ping timeout: 255 seconds]
zear_ is now known as zear
Mis012- has quit [Ping timeout: 252 seconds]
GNUtoo has quit [Ping timeout: 252 seconds]
frieder has quit [Ping timeout: 240 seconds]
kingcake has joined #u-boot
Mis012 has joined #u-boot
GNUtoo has joined #u-boot
frieder has joined #u-boot
kingcake has quit [Quit: Quit]
naoki has joined #u-boot
naoki has quit [Quit: naoki]
sszy has joined #u-boot
ezulian has joined #u-boot
Mis012 has quit [Remote host closed the connection]
Mis012 has joined #u-boot
monstr has quit [Ping timeout: 260 seconds]
tnovotny has joined #u-boot
tnovotny_ has joined #u-boot
Clamor has joined #u-boot
Clamor has quit [Client Quit]
tnovotny has quit [Ping timeout: 252 seconds]
dsimic has quit [Ping timeout: 260 seconds]
dsimic has joined #u-boot
redbrain has quit [Read error: Connection reset by peer]
redbrain has joined #u-boot
GrantM11235 has quit [Quit: Client closed]
slobodan has joined #u-boot
frieder has quit [Ping timeout: 272 seconds]
prabhakarlad has quit [Quit: Client closed]
mmu_man has joined #u-boot
ungeskriptet6 has joined #u-boot
ungeskriptet has quit [Ping timeout: 260 seconds]
ungeskriptet6 is now known as ungeskriptet
frieder has joined #u-boot
frieder has quit [Ping timeout: 245 seconds]
sszy has quit [Read error: Connection reset by peer]
sszy has joined #u-boot
prabhakarlad has joined #u-boot
monstr has joined #u-boot
slobodan has quit [Read error: Connection reset by peer]
torez has quit [Remote host closed the connection]
<xypron>
Tartarus: The explanations by Andrey are fine. It is just weird why the kernel implemented that check in __reserved_mem_check_root(). But we will have to live with it
<dsimic>
Xogium: (re: is device assignment stable in u-boot by default) AFAIK, it's the DT that provides the stable device names
<Xogium>
dsimic: so I must alias the mmc devices too ?
<Xogium>
for u-boot
<dsimic>
AFAIK, yes
<Xogium>
do you know if I need a specific version of u-boot in order to alias mmc devices ? It wasn't possible until a specific linux version so why I ask
<dsimic>
there's a version that brought that feature for sure, but that must be a very old version
<dsimic>
anything newer than a few years should be fine
<Xogium>
right
<Xogium>
in linux, iirc it took until 5.10
<dsimic>
parhaps, I'm not really sure
<silurian_invader>
it worked earlier than 5.10 for sure
<Xogium>
hmm well somehow the info I got was that you couldn't have stable dev nodes based on this before some specific version, somehow
<Xogium>
maybe this person had it wrong
<dsimic>
well, all features were introduced at some point
<Xogium>
true :)
<marex>
Tartarus: did the CI hang ?
<Xogium>
guess if I want to know if it works, I could invert mmc0 and mmc1 on purpose using aliases
<marex>
Xogium: that stable mmc ID in Linux has been a mess
<marex>
Xogium: the device enumeration via DT aliases has been implemented very early in the DM days
<Xogium>
marex: how so ? Works good for me ?
<dsimic>
Xogium: perhaps you want to follow the kernel's DT assignments
<marex>
Xogium: there has been opposition to that on Linux side, and the suggestion was basically partuuid
ezulian has quit [Read error: Connection reset by peer]
ezulian_ has joined #u-boot
<Xogium>
which... Is absolutely no good if your partuuid is the same bloody one on all medium
<marex>
Xogium: U-Boot uses the aliases stuff to map DT <-> command line IDs
<Xogium>
seriously, don't people know that ?
<dsimic>
yeah, let's put (G)UUIDs everywhere
<marex>
Xogium: it is historical context, that's all
<Xogium>
right
<Xogium>
just surprised that people would just tell you to just use uuids :D
<Tartarus>
marex: Hum? CI doesn't fail immediately now on GitLab due to the rework from Toradex, I do need to follow up on that, or see what I need to add to fix that behavior change
<dsimic>
having the same UUIDs can be a huge issue
<dsimic>
e.g. with prebuilt disk images for various ARM devices
<Xogium>
indeed
<marex>
dsimic: I agree, I ran into that already, but with U-Boot you can update the UUIDs from the command line too
<dsimic>
I used to run into that very often, unfortunately
ezulian_ has quit [Ping timeout: 255 seconds]
qqq has joined #u-boot
<Xogium>
dsimic: would you happen to know if u-boot can be made to figure out the partuuid or dev node corresponding to the device it has to boot from and dynamically add to the bootargs ? I have this wiht barebox, where it finds both partuuid and can set /dev/mmcblk... provided the dt is stable
<Xogium>
but I couldn't find anything similar with u-boot
<silurian_invader>
marex: then general solution is to add /dev/disks/by-alias to systemd
<marex>
silurian_invader: isnt that in place already ?
<Xogium>
silurian_invader: that doesn't actually work. It could symlink the wrong partuuid or partlabel, as well :)
<silurian_invader>
that's be convenient
<marex>
silurian_invader: I usually used dev/disk/by-path since that's fixed full hardware path
<silurian_invader>
I haven't checked any recent systemds
<Xogium>
systemd has partuuid, partlabel, label, uuid
<dsimic>
Xogium: huh, I'm not sure that's possible
<Xogium>
and by path
<Xogium>
by-id as well
<silurian_invader>
marex: yeah, but by-path can also change depending on the configuration
<marex>
silurian_invader: how so ?
<Xogium>
but identical partlabel or uuid will always be problematic
<marex>
silurian_invader: if it is SD slot or eMMC slot, that by-path is tied to a controller, so should be stable-ish
<silurian_invader>
for example, we produced a board where the reference platform used "mmc 0"
<silurian_invader>
but our MMC was on mmc 1
<marex>
(yeah, so fix the DT, problem solved)
<silurian_invader>
so I adjusted the aliases so they were the same
<silurian_invader>
by-path would be different though
<Xogium>
yep dt aliases ftw ;)
<silurian_invader>
anyway, it doesn't seem like there is a by-alias right now
<silurian_invader>
I did this for netdevs; at some point maybe I will do it for block devs
<dsimic>
it could be very useful
<silurian_invader>
but of course, one of the main things we use block devs for but not net devs (usually) is specifying root=
<silurian_invader>
and systemd runs too late for that unless you have a (systemd-based) initramfs
<dsimic>
true
slobodan has quit [Read error: Connection reset by peer]
slobodan has joined #u-boot
deathmist has quit [Remote host closed the connection]
deathmist has joined #u-boot
slobodan has quit [Remote host closed the connection]
slobodan has joined #u-boot
goliath has joined #u-boot
prabhakarlad has joined #u-boot
zear has left #u-boot [Leaving]
zear has joined #u-boot
sakman has joined #u-boot
gsz has quit [Ping timeout: 245 seconds]
___nick___ has quit [Ping timeout: 258 seconds]
jclsn has quit [Quit: WeeChat 4.0.5]
jclsn has joined #u-boot
slobodan has quit [Read error: Connection reset by peer]
<sjg1>
dsimic: Xogium: Using aliases to select the device numbers was introduced in 2015.01 from what I can tell. The mmc uclass came in at 2015.07 and had this support enabled
<dsimic>
ah, thanks, that's good to know
<dsimic>
as assumed, it's been a while since the introduction
srk has joined #u-boot
<marex>
dsimic: DM is even older than that :)
<marex>
dsimic: the successful DM that is, there have been various failed attempts before that one ...
naoki has joined #u-boot
<dsimic>
marex: I wasn't involved with U-Boot back then, but I did read about the whole thing