Posted: Sun Mar 21, 2021 20:47 Post subject: iOS Devices Dropping Wireless Internet
As of recently my two iPhones and an iPad periodically lose internet connection although still connected to the Wi-Fi network with a valid IP address. This happens both on the 2.4 and on the 5 GHz network. Switching the devices' Wi-Fi off and on again will cause them to reconnect and function correctly for a while but then the internet gets dropped again.
Nothing has been changed in the router settings (other than periodic firmware updates) and my other non-Apple devices do not have this problem. I even tried a second router (with the same settings) and had the same problem with that one.
Both routers are Linksys WRT1200AC and are running DD-WRT v3.0-r46130 std firmware.
I do realize that the problem might be caused with something on the client side but I have exhausted all my diagnostic knowledge on that side without success. I then tried loading a year-old DD-WRT firmware on one of the routers and the problem disappeared - all three devices remained correctly connected without any interruptions.
Maybe try the LDPC workaround, but the last known good build for Marvell WRT and Apple products is r44048.
Is there any firm evidence on why this is the case? I did fresh resets/flashes of two different firmwares today and used iPerf to evaluate network connectivity; sure enough, the two tested both had connectivity issues on my iOS devices.
This is still happening on my WRT1900ACS v2 with firmware 44715 (edit: the version recommended by the database--not necessarily the latest). I've reverted to 44048 in the meantime.
What causes this, and are there any known workarounds other than using an older version?
This topic is littered all over this forum. Some folks have gone as far as flashing upgrades to more recent builds and wiping out the stock Linksys firmware in the process and wiping out the nvram ('nvram erase && reboot) and re-configuring from scratch with some luck, apparently. Thanks for reviving this thread that is one of several that has the LDPC reference.
Great, glad it has some utility for being revived--didn't think to search first.
Full disclosure, if you search "dd-wrt iphone wrt1900acs" this is one of the first results, so for people in this narrow, specific use case with this specific hardware, this thread is going to be the one they will most likely find.
For anyone coming from Google, here's an extensive thread describing the problem in more depth.
Joined: 08 May 2018 Posts: 13274 Location: Texas, USA
Posted: Mon Aug 23, 2021 0:58 Post subject:
And xl600 balked on sending me (another WRT3200ACM) out of that deal so I could test at home. Plus, there is an Atheros interjection by ArminasK that was never removed, lol.
IRT your question as to 44048 vs newer builds vs workaround(s), this is an ongoing topic and everyone seems to have jumped ship on discussion; but you can read almost EVERY Marvell build thread from 44048 and up and a number of other threads about all of this ongoing issue.
Joined: 27 Apr 2016 Posts: 112 Location: Down down .... under
Posted: Mon Aug 23, 2021 3:00 Post subject:
Been running v3.0-r44048 for about a year and a half now, never been able to go forward due to the ios wifi problems. I check back here every now and again to see if its fixed yet but nothing.
Saying that, v3.0-r44048 is very stable and working perfectly, running many months without lockups or anything. Not sure if a newer version will actually give me better performance.
I will probably just leave it at this version for good. _________________ Linksys WRT1900ACv2 - My new go to build: DD-WRT v3.0-r44048 std (08/02/20) - STABLE
My WRT1900ACv2 has now been serving me for more than 5years Most reliable router I have ever owned!
Joined: 04 Aug 2018 Posts: 1384 Location: Appalachian mountains, USA
Posted: Mon Aug 23, 2021 23:16 Post subject:
I gave the LDPC change a careful try with 46069 on my main WRT1900ACSv2 to no avail ack in April. In fact I tried at least a dozen solutions du jour proposed in the forums, and none of them helped.
The most useful thing I did was to create an iOS shortcut to turn off wifi, wait 2s, and turn wifi back on. Then in iOS>Settings>Accessibility>Touch>BackTap, I assigned the double-tap action to that shortcut. Now a quick double tap on the back of my phone revives internet. Made a world of difference. _________________ Five Linksys WRT1900ACSv2 routers on 51530: VLANs, VAPs, NAS, client mode, OpenVPN client (AirVPN), wireguard servers (via AirVPN port forwarding) and clients (AzireVPN, AirVPN), 3 DNSCrypt providers via VPN clients.
This topic is littered all over this forum. Some folks have gone as far as flashing upgrades to more recent builds and wiping out the stock Linksys firmware in the process and wiping out the nvram ('nvram erase && reboot) and re-configuring from scratch with some luck, apparently. Thanks for reviving this thread that is one of several that has the LDPC reference.
I’m sorry but I think the problem about this topic all over the forum is because the running misinformation that this “is not a dd-wrt problem”. Reverting to the golden build always does the trick means that something broken was merged, and newer build became frustration for apple client users and I’ve seen several non-apple client users reporting the same problem.
Clearing everything a trillion of times doesn’t fix this magically. _________________ Linksys WRT3200ACM
Joined: 04 Aug 2018 Posts: 1384 Location: Appalachian mountains, USA
Posted: Tue Aug 24, 2021 14:10 Post subject:
To say what everyone who has followed this over the months knows, the wifi software subsystem, i.e. the wifi driver, for the Linksys/Marvell routers is "black box" code, a "binary blob," from Linksys. The dd-wrt maintainers do not have its source code, and Linksys will not share. All dd-wrt can do is tweak driver controls incessantly, hoping for a miracle. But if Linksys got something important wrong in the driver, there is nothing dd-wrt can do.
Sometime near the beginning of this year Linksys released an updated driver for these routers, at a point in their history when drivers had long been stable. Why does a company do that? Software is expensive, and updated software for products long out of support does not happen for no reason. My suspicion is that a security problem was identified that required immediate patching. What if the "fix" amounted to switching off or bypassing some functionality that turned out to contain problems? Turning something off is certainly cheaper than deep investigation and reworking of old code.
And to ask the perhaps obvious, a question that practically answers itself, can the dd-wrt staff simply ignore a new driver when the obvious reason for its existence is an important security patch?
You can see that there is much conjecture in the above. But the bottom line is that it is indeed very reasonable to suspect the problem is not dd-wrt, not the code the dd-wrt maintainers control. _________________ Five Linksys WRT1900ACSv2 routers on 51530: VLANs, VAPs, NAS, client mode, OpenVPN client (AirVPN), wireguard servers (via AirVPN port forwarding) and clients (AzireVPN, AirVPN), 3 DNSCrypt providers via VPN clients.
To say what everyone who has followed this over the months knows, the wifi software subsystem, i.e. the wifi driver, for the Linksys/Marvell routers is "black box" code, a "binary blob," from Linksys. The dd-wrt maintainers do not have its source code, and Linksys will not share. All dd-wrt can do is tweak driver controls incessantly, hoping for a miracle. But if Linksys got something important wrong in the driver, there is nothing dd-wrt can do.
Sometime near the beginning of this year Linksys released an updated driver for these routers, at a point in their history when drivers had long been stable. Why does a company do that? Software is expensive, and updated software for products long out of support does not happen for no reason. My suspicion is that a security problem was identified that required immediate patching. What if the "fix" amounted to switching off or bypassing some functionality that turned out to contain problems? Turning something off is certainly cheaper than deep investigation and reworking of old code.
And to ask the perhaps obvious, a question that practically answers itself, can the dd-wrt staff simply ignore a new driver when the obvious reason for its existence is an important security patch?
You can see that there is much conjecture in the above. But the bottom line is that it is indeed very reasonable to suspect the problem is not dd-wrt, not the code the dd-wrt maintainers control.
If the blame is on the Marvell drivers, it's easy to think that the solution is to revert them to the working version when 44048 came out. No need to have the source code for that. It just works, we keep that driver version, the rest of the system can be updated. Then everyone's happy.
AND security update that breaks the entire connection sounds like cutting your ethernet cable to prevent hacking. The driver revert until a well working new one comes, is a must to keep at least functionality. _________________ Linksys WRT3200ACM
Last edited by 7heblackwolf on Tue Aug 24, 2021 15:50; edited 2 times in total
