Not enable skb_recycler：
config skb_recycler 'opt'
option max_skbs '1024'
option max_spare_skbs '512'
option enable '0'
Enable skb_recycler：
config skb_recycler 'opt'
option max_skbs '1024'
option max_spare_skbs '512'
option enable '1'
skb_recycler seems to take up more memory.
Hi @profusely_fester259 you can use the issue tab on my repo so I can take a look 
There's nothing else to it. Latest commit just crashes when trying to use qmi_wwan_q with rmnet_nss.
call trace is just an nss coredump so nothing helpful there 
They beefed up ECM/SFE quite a bit too. A lot of the wifi offloading logic was brought over. Overall, it feels like a step in the right direction, but not sure if we're still beholden to a specific NSS firmware for features...
I couldn't find the typical *.bin for IPQ9574 in any of the images I examined. I'm still not 100% sure with the relationship on PPE with NSS. I've read that PPE sill loads NSS firmware, but that's it's embedded in the module and encrypted. If that's the case, we still end up having to chase down firmwares like before. Super annoying: https://boxmatrix.info/wiki/Property:qca-nss-ppe.ko
I would like just to thank all the community for the great work! Specially AgustinLorenzo
I installed my AX3600 with 5.5 and now updated to 6.6 and I could not be more happy. Everything works perfect and now I have full control over my router!
the newer sfe has technically nothing todo with nss or ppe. its still just a software offloading engine. with some modifications you can run it without nss or ppe.
there are 3 engines. nss/ecm, sfe and ppe. ecm and ppe are hardware offloading engines based on the acceleration cores . sfe is still pure software
I tested again latest commits of openwrt-ipq and nss-packages and it still crashes on rmnet_nss.
I cloned and build again. If you run quectel-cm in ssh it crash on line udhcpc: broadcasting discover
This is my second AW1000 from aliexpress so its not a hardware issue.
Can you try including luci-proto-quectel on your build and try adding your interface from there? I have no issues with the current commits even just by using luci-app-pcimodem and luci-app-usbmodem
You can also include this uci-defaults on your build
# Set 5G WAN Interface
uci batch <<EOF
set network.5g=interface
set network.5g.proto="quectel"
set network.5g.device="/dev/cdc-wdm0"
set network.5g.apn="internet"
set network.5g.auth="none"
set network.5g.pdptype="ipv4v6"
set network.wan=interface
set network.wan.proto="dhcp"
set network.wan.device="@5g"
set network.wan.metric="1"
set network.wan6=interface
set network.wan6.device="@5g"
set network.wan6.proto="dhcpv6"
set network.wan6.reqaddress="try"
set network.wan6.reqprefix="auto"
set network.wan6.metric="1"
EOF
uci commit
I have luci-proto-quectel which uses quectel-cm. It crashes anyway when a packet is sent or received. No diffrence. I dont have luci-app-usbmodem.
Are you using openwrt-ipq from qosmio to test? Can try?
RG500Q firmware is latest
quectel-cm log
requestQueryDataCall IPv4ConnectionStatus: DISCONNECTED
ip link set dev wwan0 down
ip addr flush dev wwan0_1
ip link set dev wwan0_1 down
requestSetupDataCall WdsConnectionIPv4Handle: -
ip link set dev wwan0 up
ip link set dev wwan0_1 up
busybox udhcpc -f -n -q -t 5 -i wwan0_1
udhcpc: started, v1.36.1
udhcpc: broadcasting discover
- dmesg nss coredump here
I have a question about my own build.
When I compile with the official openwrt source, I update to the latest version like this:
$ cd ~
$ cd openwrt
$ git pull
$ make package/symlinks
$ make defconfig
However, if I do this in the "openwrt-ipq" directory, I get the following message:
~/openwrt-ipq$ git pull
remote: Enumerating objects: 362, done.
remote: Counting objects: 100% (360/360), done.
remote: Compressing objects: 100% (157/157), done.
remote: Total 362 (delta 243), reused 247 (delta 203), pack-reused 2
Receiving objects: 100% (362/362), 87.14 KiB | 1.34 MiB/s, done.
Resolving deltas: 100% (243/243), completed with 20 local objects.
From https://github.com/qosmio/openwrt-ipq
+ bdd89bc66b...235db18a8e qualcommax-6.x-nss-wifi -> origin/qualcommax-6.x-nss-wifi (forced update)
+ f6402596a5...8753c0bf96 mac80211-fix-kconf-warning -> origin/mac80211-fix-kconf-warning (forced update)
hint: You have divergent branches and need to specify how to reconcile them.
hint: You can do so by running one of the following commands sometime before
hint: your next pull:
hint:
hint: git config pull.rebase false # merge (the default strategy)
hint: git config pull.rebase true # rebase
hint: git config pull.ff only # fast-forward only
hint:
hint: You can replace "git config" with "git config --global" to set a default
hint: preference for all repositories. You can also pass --rebase, --no-rebase,
hint: or --ff-only on the command line to override the configured default per
hint: invocation.
fatal: Need to specify how to reconcile divergent branches.
What should I do in this case?
I was able to replicate the issue using Qosmio's repo, I'll investigate. I've been using my own fork for some time now and it's in sync with his.
Can you confirm that you are using https://github.com/qosmio/nss-packages.git;NSS-12.5-K6.x for the NSS packages? Try refreshing your packages using ./scripts/feeds update -a and ./scripts/feeds install -a that worked for me. "Something" might have broke the patches, but I'm still looking into the issue.
Yes. I did a fresh clone of all repos today. If you can push your changes to your build I can also diff it myself.
If you have no local changes, you can git pull --rebase. Otherwise commit your changes first then run the same cmd.
I haven't made any changes to my fork's main branch yet as we are working on other quectel-cm issue and qmap-multiplexing, I wasn't able replicate the issue on those branches used. TBH we are struggling to seek support from Quectel themselves with regards to these issues.
No additional changes in your custom build? What other issue you have? I also want to look into qmap multiplex with offload so I was trying to get it working.
I'll help once I get it working.
Here are the reported issues so far,
quectel-cm won't reconnect or renew ip on cell handover on some casesUNKNOWN status on registration to some networksquectel-cm won't initiate virtual interfaces on some instancesOne of my colleagues was able to make qmap multiplex work and he's testing it right now. It will be pushed soon
UNKNOWNstatus on registration to some networks
I had this. It was just the modem in minmum function mode. Just need to send AT+CFUN=1 on boot.
Still crashing with
0b71120ef6b54a2a5c7604cad8668ed88bb1710f
235db18a8e38b4c6c90ffeb095196f5ec218d17c
@xHikari You said you also crash with this right? What commit hashes are you on that doesn't crash? I can bisect the issue