The WRT54GL: A 54Mbps router from 2005 still makes millions for Linksys | Hacker News Reader

The WRT54GL: A 54Mbps router from 2005 still makes millions for Linksys

arstechnica.com

138 pointsantr78 comments

Ten years ago I was studying CS in a conservative small religios city in central Iran. My girlfriend was living in a dormitory one kilometer away. Due to limitations I was unable to meet her freely. I borrowed a WRT54G and an outdoor antenna and made a WAN. The default output power was insufficient for my antenna and my intended target laptop in the other side of the city, however. With WRT54G having its firmware open sourced, I downloaded a custom built of it and hacked the device to increase the output power. Not only I circumvent the limitations, but I used an old 300mhz Intel PC as an internet gateway to let her access internet using her laptop. We were finally able to communicate freely at any time. My thanks to Linksys for crating this fantastic hardware and opening it to the world.

That's brilliant. Thank you for sharing.

And this is why the FCC requirements make little sense. Such products are used all over the world. Forcefully limiting the freedom in one locale is catering to the lowest common denominator.

Instead, the ability to use spectrum should be controlled by the end user, through some hardware measure (ex: jumpers) that is done at the user own responsibility.

"The government should not regulate the shared limited resource of radio bandwidth because it is useful to work around restrictions on free speech" is nonsensical.

If your government significantly curtails free speech, then it either is also so fascist that it doesn't care what you think about bandwidth regulation, or (as in the case of the GP) it is so technologically incompetent that it effectively does not regulate the airwaves. Either way your suggestion is useless.

You're misrepresenting what the GP post said and being obnoxious about it.

Given that there are different regulatory regimens, locking down hardware to a particular locale may be inappropriate. For example, Channel 14 is fine in Japan and 13 in the EU but not the US. Restricting the whole world to 11 channels to comply with US law would be silly.

Yes, I misread his post. My mistake.

This is exactly what Hacker News is all about. Thanks for sharing!

Do you hear that hardware manufacturers? People are willing to pay a lot, and you don't have to redesign/upgrade every year, if you make a simple, hackable, usable system!

I wish more hardware companies thought like this

The question is would you buy it at a higher price point, like say $300?

Yup. There's a lot of products or software I'd pay a lot more for if they were open. But I understand I'm in the extreme minority.

Absolutely. I actually have 2 >$200 routers that I use at home. I bought them specifically because I could put pfsense on them.

What's the hardware?

I don't know about the parent comment but you can run pfsense.org on a 120 pcengines board.

http://pcengines.ch/apu2c4.htm

Most companies don't want to make open/hackable solutions, because it makes the hardware a commodity. After all, they don't make the chips, they probably contract out the PCB design and manufacture or it's at least done via some sort of JDM process.

What's left at that point? Brand, mechanical design, and software... There's not a whole lot to differentiate yourself if you let go of the software piece.

I'm guessing these routers are an exception to that rule, but why is that?

We only know they "make millions from them" which doesn't say much.

Other routers done the bad way might make even more money -- which would easily explain why these are an exception.

> What's left at that point?

I'm not sure what motivates you, but making money selling things people want to buy seems to get a lot of people out of bed in the morning.

If that's true, why do they do such a horrible, useless, terrible job on the software?

> There's not a whole lot to differentiate yourself if you let go of the software piece.

And yet it's letting go of the software piece that differentiates the WRT54GL.

Really? DDWRT supports a ton of routers.

So then why do you think the WRT54GL has lasted so long? Is there something else about that router that keeps it selling (it's #28 in routers at Amazon)?

The best hardware in bad situations. They just work, no matter what, and never die.

The software is just a part of the success.

Network effect. Same reason no one else has had similar success despite many efforts to do so.

>Do you hear that hardware manufacturers? People are willing to pay a lot, and you don't have to redesign/upgrade every year, if you make a simple, hackable, usable system!

That's only half of an argument.

Whether "people are willing to pay a lot" for such systems is beside the point to hardware manufactures, if they can make much more by making non simple and no hackable systems they renew each year.

I suspect that most of the sales tend toward the B2B end of the spectrum. The reasons someone might buy a WRT54GL aren't really that attractive to a significant fraction of consumers. Even for someone like me who is susceptible to those rationales, I'd shop Buffalo. Open, 11N, and comparably priced.

Besides DD-WRT, which can be at times quite flaky a unreliable on WRT54GL with its outdated OS base and a slow UI (and also a real pain to keep track on the few working firmware revisions as time progresses over the years; many DD-WRT users and the DD-WRT site will sadly keep suggesting to use half a decade old firmware for it [1][2][3] and any 'official' activity seems to have stopped years ago) and the Open-WRT which sadly hasn't supported WRT54GL for the last several generations [4] despite its by now somewhat ironic name, there's also Tomato by Shibby [5] (note: This is a fork of the original 'well known' Tomato that stopped its development half a decade ago - though surprisingly people sill run even that one), which is amazingly supported today with regular and frequent updates [6], latest security fixes and modern revisions of base software and OS tools, with properly working kernel 2.6-generation, IPv6, QoS and not at the very least with the staple slick Tomato configuration interface. If you happen to run a WRT54GL, you must give Shibby's Tomato a try sometime to see what this piece of hardware can still do.

Disclaimer: I'm by no means affiliated, but run WRT54GL hardware actively and at times I'm still regularly amazed at what Shibby can squeeze out of the mere 4MB of flash space and a hardware that wouldn't properly run a smartwatch alarm applet in 2016.

[1] - http://www.dd-wrt.com/wiki/index.php/Linksys_WRT54GL

[2] - http://www.dd-wrt.com/phpBB2/viewtopic.php?t=51486

[3] - http://www.dd-wrt.com/phpBB2/viewtopic.php?t=288371

  If you're lucky, you may also end up with gems like this:
  > Malachi - DD-WRT Guru
  > Your router was made in the 1800's. Why do you need firmware made yesterday.

[4] - https://wiki.openwrt.org/toh/linksys/wrt54g

[5] - http://tomato.groov.pl/

[6] - http://tomato.groov.pl/?page_id=78

I always loved Tomato; ran it on my 54GLs which I still have (though one is pre-L branding).

I had a Cisco E3000 that I tried to love; bought it because it supported DD-WRT so once I got sick of the Cisco firmware I tried DD-WRT. It seemed like an endless headache and never worked quite right, so I flashed back to the stock Cisco stuff; I realized the stock OS supported everything I needed anyway, and was more stable. I wish consumer stuff had more cool graphs and the like.

Shibby Tomato supports E3000 fine, I've been using Tomato on it since I bought it new. It's very stable.

>It seemed like an endless headache and never worked quite right

The E3000 is based on a Broadcom SoC without free drivers. That's always a bad sign for third party support. Basically you are limited to some ancient kernel which may or may not be running stable. I know the E3000 and DD-WRT on it is indeed a mess.

"Supported" with proprietary drivers often means nothing more than that it boots somehow. The same goes for the new Linksys models with "official" OpenWRT support. The wifi isn't stable at all.

I installed Tomato on my old 54GL and gave it to my parents. It's served them reliably for years.

Got a WRT54GL here that runs on a recent DD-WRT build. Also run Shibby's tomato on a beefier Linksys. openWRT on yet another Linksys and plenty of pfSense around here as well. They are all quite wonderful.

> any 'official' activity seems to have stopped years ago

That's what I thought as well until I searched a bit better. DD-Wrt is under development and the recent DD-WRT v3.0-r29837 mini (06/06/16) does run quite well on the Linksys WRT54GL.

Installed it 12 days ago and it just runs and runs [1].

Yes it is a beta branch, but I've been on the beta track on a variety of builds for over a year and never had any issues. The trick was in finding out at the forum on what still worked on that old router.

The download location [2] is not as easy to find as official page only links back to the really really old firmware.

Just navigate up on that ftp-site and you can find newer builds as well.

From the different firmware builds on the beta tree that I tested, the mini build is the one that appears to work best on the WRT54GL. The normal build might work, but I had some troubles with it when I tried.

Of course I understand it if you prefer not to run beta software, but it's very stable and better then running a very outdated firmware in my opinion.

[1] http://www.dd-wrt.com/phpBB2/viewtopic.php?t=302224&postdays...

[2] http://www.dd-wrt.com/phpBB2/viewtopic.php?p=1036133&highlig...

> Of course I understand it if you prefer not to run beta software,

Oh not at all, this is good news that there is still some activity wrt 54GL (pun intended). Having alternatives to fall back to is always a good thing. I've been on DD-WRT originally since about 2007 and went through many, many years of digging through forum threads, tracking and collecting all the reasonably stable revisions, kept comparing results with nightlies with other GL owners, had some good long stable runs even with a few broken features here and there and a good number of near-brickings saved only by a handy TFTP bootstrap...

I've only dropped my last DD-WRT install roughly some two years ago when I discovered Shibby and replaced all the other alternatives with it (i.e. the other very good VicTek's RAF firmware is not usable in all deployments as it can't fit even IPv6 support into the 4MB build that the 54GL is limited with, making your options limited) and I stopped keeping much track on DD-WRT since. Some other current DD-WRT people I know weren't very happy still, even though some stay with DD-WRT for own obscure reasons.

But by the way, the forum threads you linked to don't paint the situation in the best light with regard to stability:

  > My WRT54GL is running since ages on the mini-generic 14929 build, you can find it here:
  > ftp://ftp.dd-wrt.com/betas/2010/...
  > Together with a daily scheduled reboot I tend to forget were the thing is located in our house

"2010 firmware..."

"daily scheduled reboot..."

Or:

  > I just upgraded to:
  > ftp://ftp.dd-wrt.com/betas/2016/...
  > I've tried a few different bin's and while I did get the standard version to work, it seemed a bit unstable.

"a bit unstable" is pretty much as good as simply "unstable".

Now with Shibby I currently run GL with average half a year uptimes (usually ended by a forced powerplant cycle in the location, never had a crash or a spontaneous reboot on it as far as I remember). Still, as long as recent versions of DD-WRT build and run on 54GL, there's some chance at a good revision that can possibly squeeze few more years out of the hardware, depending on what features will work. Will keep an eye on it, it may still become handy if something bad happens with Shibby's work, which at this point and the age of the hardware wouldn't be that surprising.

Hehe, that last reply in the thread is mine. Yes it _was_ unstable for me running the standard build. That was just an experiment though as I had been running on the mini build "forever". Figured I try the standard build, but after a few hours with weird errors I reverted back to the mini build.

As for the 2010 remark, well I don't know why somebody would run a firmware that old, especially if it isn't stable.

If I'm not mistaken then the last released -non beta- version is:

ftp://ftp.dd-wrt.com/others/eko/BrainSlayer-V24-preSP2/2014/12-22-2014-r25697/broadcom/

At least that's what is in my notes, but I'm happy on the beta trail.

FWIW, the mini build also doesn't support IPv6 which is exactly why I tried the standard version again. In the past the standard version did no longer fit into the 54GL anymore. At the moment it is not a real issue for me as the 54GL only serves a WiFi guest network, but if IPv6 ever gets adopted ;) it might become one.

As I noticed the current size of the standard build download was smaller I ended up trying it again. Didn't work well in my case, but the mini build is good.

The router with Shibby that I have here gives a similar experience it great uptime and really only ever gets rebooted when I decide to run a firmware upgrade (or when there's a power outage which is very rare down here).

Out of curiosity, what is your reason to run DD-WRT on a GL? Considering you have Shibby also deployed and are thus presumably familiar with its feature set, is there something in DD-WRT that makes it particularly important to have on the GL?

Familiarity, curiosity and not even realizing that the firmware from Shibby runs on the GL are some of the reasons. Besides that there's customers running on DD-WRT so it is good to have at least one form of it running inhouse. Those customers are not on the mini nor on a GL, but there it works and in that case cross flashing remote routers is something I'd rather not do.

If anyone is looking for a cheap up-to-date replacement that runs OpenWRT, I recently replaced my WRT54GL with a TP-Link Archer C7, and I've been quite happily running OpenWRT on it for the past few months.

Note that you have to do a bit of hackery to install OpenWRT, since the stock firmware refuses to install unsigned updates through the web interface: https://pappp.net/?p=1525

I don't know how my find compares to yours but I amassed a fleet of used Netgear WNDR3800's, and replaced their old stock OpenWRT with modern version.

Compared to the linksys boxes, they have GigE, better radios (speed and range) more memory and a more powerful CPU.

I might use the USB ports in the future for a disk or some serial thing I want to provide connectivity for.

The C7 has 802.11ac support. I don't have any devices that support 802.11ac yet, but I also don't want to have to buy another router in a long time, which is why I went for it. The C7 sells for under $100 on Amazon, while most other equal-speed routers from other companies are substantially more expensive, e.g. the Linksys WRT AC1200 which sells for $150 (despite actually being slower), or Buffalo's AirStation AC1750 which sells for $160.

The Archer C7 (and the equivalent D7) is a spectacular device and is frequently rated as the best buy for your dollar. It has 512mb for the NAT table so it can accommodate an office full of people skyping/downloading/torrenting

OpenWRT on the Archer C7 doesn't support the hardware NAT acceleration (aka cut through forwarding), so you won't be able to achieve more than ~150Mbps.

Do you know a similar speed device where this is supported?

Also I don't think this is an issue for most home router situations, since few home internet connections in the USA offer more than 150Mbps. But if you were shopping for routers to install in an office building or something, it might be an issue.

I have a WRT54G [not GL] that currently lives as the wireless interface an XBox 360. Most versions of the G are upgradable to to DD-WRT. That's what the "WRT" in DD-WRT was all about. Around the time I bought it, and in the years when the G's were still available, the GL's sold at a substantial premium.

Maybe there's something magical about $69 pricing. The HP12C lives on after 35 years.

https://en.wikipedia.org/wiki/HP-12C

http://store.hp.com/us/en/pdp/hp-12c-financial-programmable-...

What I want to know (and what the article doesn't address) is: Why are so many other wi-fi routers such trash?

I mean, I sort of know why: Because manufacturers keep making new hardware without the long-term support commitment that's necessary to get the firmware truly sorted out. But how did the market come to this?

I would have thought that the average home router customer, at this point, groans in dread when its time to buy a router. To heck with crazy speeds my "broadband" cannot realize, never mind bleeding-edge protocols and questionable NAS USB ports. Don't we all want something that doesn't need a manual reset every couple of days? Why isn't uptime the #1 selling point by now?

I am pretty happy with the stock software on the RT-N66U and it is a very stable router (I've had it for a few years)... but of course I wish that all hardware was open and that it's software was open source and that I could install DD-WRT, OpenWrt, etc if I wanted to.

Up until a just few years ago the Internet was fueled by extensive growth: router manufacturers (among others) made profits by selling to people who had never bought routers before.

This has all but ended now, and people shopping for their second or third router will be making different decisions.

> router manufacturers (among others) made profits by selling to people who had never bought routers before.

They also make a huge chunk of their profits from resellers such as ISP's who will rent you a router with your access.

You can sell a lot of garbage to consumers who have never bought that product type before.

If the full-performance product is crowded out, it's sub-type can be lost forever.

Once you enter the Garbaceous Period it can be the actual end of a higher quality era.

Mine went dead after about 8 years of use, and really 11G is a bit slow these days when multiple devices started streaming. I wish Linksys can have a reasonably priced 11N version that can load open source code to continue this legacy.

You could look at the WRT1200AC. It's available as a certified refurb on amazon.com for $90, and as part of their WRT line is intended to support open source firmwares.

> as part of their WRT line is intended to support open source firmwares

Not sure about the 1200, but I have the WRT610N and it has never worked completely with Open/DD-WRT. The best I got (before leaving it and returning to the original firmware) was a functional router but no USB and no 5GHz band.

Wow that's really surprising. Is it one of the later revisions? I'm running the 610N with DD-WRT and I the 5GHz band running just fine @ 300 mbps

>and as part of their WRT line is intended to support open source firmwares.

The wireless driver is crap, though. The last time I checked it wasn't recommended due to wireless problems.

https://wiki.openwrt.org/toh/recommended_routers

I still have two. I'm really having a hard time making the decision to replace them, when they still work as well now as they did 8 years ago, running Tomato the whole time.

But the number of devices connecting and the bandwidth they require are only increasing, and 802.11n or 802.11ac would really help with some of them. I also have desire for a relatively inexpensive wireless router that does not attempt to prevent me from loading on it whatever software that pleases me to run. I will not pay a full purchase price for any device that I cannot own.

That doesn't necessarily mean I won't ever have such a device; it just means there has to be some way to jailbreak it before I will consider buying.

The AC68U is hackable and overall a pretty reasonable device. The only issue I've seen was that it had some weirdness with power levels with some revision of the Merlin firmware (I forget which one), in which it would report much higher output than a spectrum analyzer showed. I don't recall the fix, so it must have been fairly simple.

After the WRT54GL went away I bought a Routerstation Pro from UBNT and loaded with 2.4G/5G 11N radio cards, it's running Openwrt stably for years. Unfortunately I think Ubiquiti stopped selling that board.

I have a powerline adapter (made in 2014) that hangs every other day if I let it operate in 802.11n (and it hangs every day if in mixed mode). Pinging it from my Raspy mitigates but does not eliminate the problem. I even installed a wall switch in the wall plug so it would be easier to recycle! In the end, I had to put it in 802.11g mode to have a slower but dependable home network. There are 3 routers in the house, fortunately the others do work in 802.11n without needing a power cycle from time to time (the most stable is the one supplied by the cable company - that's really surprising).

This is the scenario of consumer-grade WiFi - way less reliable than 10 years ago. When I replace the powerline by a blue cable, I can choose between buying a very expensive AP, or a 'modern' AP that freezes, or a WRT54GL that works.

I have been curious, given all the echoes here about FCC wireless radio regulation, if these devices will in fact increase in price, and those like it, as this niche community covets them.

Kind of like as I watch the OpenGarages movement, the next time I buy a car I might pay premium for something vintage with the most minimal computerized components. Haha.

I have a WRT54G that I bought in 2004 or 2005. I haven't even replaced its firmware because it does what I need, and it still just works and always has. No lagging, no crashes, nothing but straight uptime year after year. I do wonder if I should buy a GL model as a "spare" in case it goes dead at some point.

Have you tried to log into the admin web page lately? The default firmware generates SSL certificates signed with algorithms that modern browsers have black listed and won't access the page.

Still have two originals in operation in older relatives' houses. God bless this little device.

I love these things. They were the basis for my first article in 2004. Its horrible to read now, but I had so much fun tinkering and researching and getting my first story as a tech journalist out to the readers.

Hm, it costs EUR 47,73 here in Amazon... How is it still at that high price?

Because people still buy it?

Same reason a TI-83 calculator still costs a hundred bucks.

The model's reputation for reliability probably plays a big role here. Read the reviews of almost any other modern wifi router, and you'll see lots of complaints about even expensive models frying themselves in the first year.

I still keep old wrt54g's around for backup duty. But I really do want a router that's actively kept up to date.

Last year i bought a xiaomi mini miwifi (less than 20usd at the time). With openwrt it ran mostly okay but the AC was occasionally locking up requiring a reboot. Thankfully with 15.03.1 that now seems to be fixed. Ordered another one of these for my parents who were having serious problems with netgear router that I hope to get soon. Not having gigabit is a bummer on these but they do well for low speed residential service and they are super cheap!

Could it be that the ability to more fully control a computer, e.g., a router, makes the computer more valuable?

What use are greater feature specifications if the user does not have anything close to full control over the computer?

How about a computer that where the seller makes it deliberately difficult to open, or impossible to add RAM, or threatens repair shops with litigation?

When the user powers on, the computer immediately starts phoning home to various endpoints controlled by the seller and the user cannot turn this off. Are users presumed to be inept?

Does it matter if this computer has a sexier form factor and better specifications, not to mention a higher price tag, than another one without the restrictions on tinkering?

I have a 11 year old WRTGL that got hit by lightning, fried two ports including the WAN port, but the other ports still work and it is still going strong.

I still have mine, going strong with Tomato firmware on it. Haven't touched the thing in years. Amazing piece of hardware, even at 50/5mbit service.

Doesn't the TI 83 calculator still sell quite well? It's a solid machine that does one thing and does it well.

I don't think that's quite an accurate comparison, considering schools almost nearly force you to purchase the Ti-83 (or they have been for the past few decades until very recently, when the Ti Nspire took over)

The WRT54GL is a purely market solution in a commoditized market where the consumer is not pressured or forced into buying it.

You may be right. Haven't thought of that.

I feel like the cost of materials has gone up, since the chips are now old enough to command a premium.

Obligatory comment/question: What routers do you use to run OpenWRT? What's the price like?

I have a DIR-505L ($15) I use for traveling. Has OpenVPN and SSIDs setup to bridge to the VPN (on layer 2). So when I connect to that special SSID it's as if I'm still onsite. It's also just a wallwart, so it's easy to travel with and doesn't look suspicious.

I also have a DIR-835 ($20). Also use it as a VPN router (both client and server). It's a lot more powerful than the DIR-505L and does 5ghz.

I used to use a WRT54GL, but then 802.11N came along and I also wanted 5ghz. So I got a Linksys E3000, but I later bricked it during a botched firmware upgrade. The uptime on that thing was like 400 days until I had a power outage.

Unfortunately, I'm having trouble finding a OpenWRT router that will support gigabit routing for the WAN. So now I use the AT&T supplied router/modem. Unfortunately with AT&T, bridging just does double NATing (and the box has a connection limit of ~2000 for ipv4).

I've got a GL.iNet 6416A ($23) for portable use -- it has two Ethernet ports along with 802.11n on 2.4GHz and a USB port. Pretty nifty for on-the-go use, but don't think about it if you need to hit speeds over 35Mbps.

That replaces my old modded TP-Link TL-WR703N ($35), since it beat it in almost everything. Now, I use it as a wireless bridge for an Ethernet-only device.

I've got two Xiaomi Mi WiFi Minis (150 CNY, or ~$24) as budget APs. They do 802.11ac, but the three Ethernet ports on it only does 100Mbps.

There's also a TP-Link TL-WDR3600 (free, but $60 on Amazon) that I use in another location. Works fine, and can easily hit the WAN's speed at 100Mbps. 802.11n over 5GHz also easily hits 90Mbps, which is nice. It's been incredibly stable, and hasn't went down in almost two years.

I've got a handful of other routers with OpenWRT on them, but I no longer use them for a variety of reasons. Another TL-WDR3600 I had was replaced by an EdgeRouter Lite, since it couldn't really saturate my current Internet connection (1000/500).

Great and just today it made me quite angry, wifi on mine is not very reliable.

I bought mine around 2007 and I have used it continuously.

my father is still using one which i bought 8-9 years ago. running dd-wrt, its unbelievably stable.

Reply on news.ycombinator.com