<karlp>
djfe: you can use 64 bit types in user space happily yes. you can do that on most 8 bit systems even, it's orthogonal to a 32bit/64bit kernel.
max_ has joined #openwrt-devel
xback has quit [Ping timeout: 480 seconds]
xback has joined #openwrt-devel
<djfe>
👍
srslypascal has quit [Ping timeout: 480 seconds]
robimarko has quit [Remote host closed the connection]
robimarko has joined #openwrt-devel
srslypascal has joined #openwrt-devel
max_ has quit [Remote host closed the connection]
max_ has joined #openwrt-devel
floof58 is now known as Guest8520
floof58 has joined #openwrt-devel
Guest8520 has quit [Ping timeout: 480 seconds]
xback has quit [Ping timeout: 480 seconds]
cbeznea has joined #openwrt-devel
cbeznea has quit [Quit: Leaving.]
djfe has quit [Ping timeout: 480 seconds]
cbeznea has joined #openwrt-devel
kenny has quit [Ping timeout: 480 seconds]
<rsalvaterra>
Is it just me, or does OpenSSL really dislike to be compiled with LTO too…?
bluew_ has quit [Ping timeout: 480 seconds]
ashkan has quit [Remote host closed the connection]
Guest8376 has quit []
Danct12 has joined #openwrt-devel
cbeznea has quit [Ping timeout: 480 seconds]
<hurricos>
Would anyone like to review a simple ath79 port? :^)
<hurricos>
looks like Hauke is doing review of these, at least. This one's probably the most overbuilt ath79 device OpenWrt will see ported -- https://github.com/openwrt/openwrt/pull/10794
Borromini has quit [Quit: Lost terminal]
valku has joined #openwrt-devel
f0g has joined #openwrt-devel
<f0g>
Hi all. I have a extra mtd and want to mount it as jffs2 to use. Where should I insert the mount command in openwrt souce? Is /lib/preinit/80_mount_root a good place?
<hurricos>
f0g: which board?
<hurricos>
or rather, why the extra mtd? I'm used to OpenWrt hacks to fuse separated MTDs
<f0g>
hurricos: it's a customized board, use mtk chips
<hurricos>
I see. I might be wrong, but it should be possible to indicate through the device tree that two discontinuous regions of storage are to be treated as a single block region, kind of like dm-linear does?
<f0g>
hurricos: not as complex as that. the kernel is 3.18 and I just want to add a command like 'mount /dev/mtdblock5 ...' to some where.
<hurricos>
-- no, maybe I'm misremembering. mtdsplit allows us to have a single rootfs partition which is then split into a read-only base image + read-write rootfs_data
minimal has joined #openwrt-devel
<f0g>
my openwrt has worked fine on the board. since the kernel already split the mtds and there still one partition is free, I want to mount it to use.
<f0g>
I noticed openwrt mount overlayfs in 80_mount_root script, so I am wondering is this a proper place to add new mount points
<hurricos>
My understanding is that modern OpenWrt wants to provide you a ro rootfs + rw rootfs_data, and apply the latter atop the former as an overlayfs
<f0g>
yes, it is. but my new mtd block has no relationship with the root partition, just a extra one to store some data
<hurricos>
if you're doing something downstream, it's "whatever works"
<hurricos>
You are probably safe adding your board-specific code to 80_mount_root :p
<f0g>
What's the standard method to add a mount point? I think this is what I really want to know
<f0g>
hurricos: I see.
<hurricos>
The standard way is, and I'm struggling to find a dts example, to combine the regions of flash
<f0g>
hurricos: actuall, the mtd partition has already been there. I just need to mount it.
<hurricos>
yeah, no mention of ubi.
<f0g>
No, it's a nand flash
floof58 is now known as Guest8534
floof58 has joined #openwrt-devel
<hurricos>
f0g: I know what you have intended -- just trying to relay what I'd do, based on what I've seen done + what I understand to be the standard way of using additional storage regions. It's a lot less like stock Linux
<f0g>
hurricos: Ok. thanks for the advice.
<hurricos>
in the case of NAND, you "should be" using UBI, because blocks go bad. With UBI, you have much more flexibility
<hurricos>
Where that doesn't help you is if your u-boot doesn't have ubi support. If you're not using ubi, it probably doesn't :^)
<hurricos>
not sure what to tell you there. I've never seen mtdsplit -> ubi
<hurricos>
usually vendor u-boot supports ubi if it's booting from NAND.
<hurricos>
f0g: And now that I know what I'm looking for, I have examples that use mtdsplit -> ubi, with a separate kernel partition (so no u-boot ubi support): target/linux/ramips/dts/mt7621_jcg_q20.dts
<hurricos>
If you're still not worried about jffs2-on-NAND, you can subtitute ubi for just plain jffs2, which is fine.
<hurricos>
for readers interested in experimenting with ubi without a real MTD device, try the nandsim module; e.g. `sudo modprobe nandsim first_id_byte=0x20 second_id_byte=0xaa third_id_byte=0x00 fourth_id_byte=0x15 parts=64,64,64,64`, c.f. https://elinux.org/File_Systems#Mounting_UBI_Image_on_PC_using_nandsim
<f0g>
hurricos: it's interesting.
<f0g>
But as I said, I don't need to change anything of the mtd partition, just mount it to use is enough.
<hurricos>
:thumbsup:
<hurricos>
I'm assuming you don't have the flexibility to go back to OpenWrt upstream. OpenWrt really closely associates soldered parts with devicetree and thus kernel, detachable ones with conf files and thus userspace.
<hurricos>
block-mount + /etc/config/fstab is the way I deal with additional storage coming from disks
<f0g>
hurricos: great, that's what I want
floof58 is now known as Guest8539
floof58 has joined #openwrt-devel
<hurricos>
f0g: good. I didn't want to suggest it directly, but ...
<hurricos>
you can also pick a more flash-friendly fs at that point
Guest8539 has quit [Ping timeout: 480 seconds]
<f0g>
hurricos: got it, thank you.
srslypascal has joined #openwrt-devel
aparcar has quit [Remote host closed the connection]
_0x4a6f has quit [Remote host closed the connection]
aparcar has joined #openwrt-devel
_0x4a6f has joined #openwrt-devel
Ansuel has joined #openwrt-devel
<Ansuel>
hello
<f0g>
hurricos: I tried to mount my mtdblock9 with jffs2 in /etc/config/fstab, but it didn't work. my openwrt version is 19.07 and I didn't find fstype option in block.c. Does that version fstab not support mouting mtdblock as jffs2?
<stintel>
so shibboleth's TP-Link WPA8630P v2.0 logs mount_root: no usable overlay filesystem found, using tmpfs overlay
<stintel>
before it inits the jffs2
cbeznea has joined #openwrt-devel
<stintel>
the log also contains OF: Bad cell count for /ahb/spi@1f000000/flash@0/partitions and doesn't end on an erase/write block -- force read-only but recently it was discussed here these are harmless
<Ansuel_>
doesn't end on an erase/write block this may be problematic
<stintel>
Ansuel_: no, because this is about read-only partitions
<stintel>
so since with small flash devices there is no way to put the config backup in a /boot partition like on rpi and restore it after boot -- do we just not erase the jffs2 partition during sysupgrade and simply reuse it ?
<f00b4r0>
stintel: no. If I read your question correctly, the answer is that the config is put in a jffs2 stub located after squashfs and the deadcode marker marks the beginning of the segment that should be erased
djfe has quit [Ping timeout: 480 seconds]
Ansuel has joined #openwrt-devel
<f00b4r0>
i had written a detailed summary of this process in an 2017 lede-dev RFC where I suggested doing differently, titled "[RFC] filesystem-agnostic sysupgrade backup system". Sadly it seems no longer available online. I can forward you the email tho, if you'd like
Ansuel has quit []
Ansuel has joined #openwrt-devel
Ansuel has quit []
Ansuel has joined #openwrt-devel
Ansuel has quit []
Ansuel has joined #openwrt-devel
Ansuel has quit []
Ansuel has joined #openwrt-devel
<Ansuel>
(sorry for the spam... finally setup znc irc bouncer so i can be online... next step setup push notification so i can receive ping even if i'm not connected to the bouncer)