Broadcom Wifi Trisquel Etiona
- Inicie sesión ou rexístrese para enviar comentarios
Hi, I got a new computer and it doesn't work yet with trisquel. (i5-11500)
Oh no. I now don't have my original workstation PC that worked well with trisquel since my parents got it away so I just got out my laptop that would work.
The laptop is the Apple MacBook Early 2008 (4,1). Although the EFI firmware is proprietary most functions work out of the box on trisquel.
The camera doesn't work and needs proprietary firmware extracted from the Apple OS. I don't use it so it's fine.
The problem is that the wlan in the computer doesn't work.
It is a broadcom chip, in lspci it is shown as BCM4321, and previousely some years ago when I used the M$'s proprietary "window" operating system I had used the BCM4320 driver and it had worked.
The markings on the broadcom wlan card says BCM94321MC and some Apple Stuff. I think this is a remarked version of BCM4321 or something.
I searched the web, found Debian GNU/Linux's wiki and the "libre" operating system's wiki suggest me to install a proprietary driver...
Is there a way to use the card without installing proprietary firmware blobs?? Thanks.
(The good news is that at least the ethernet works!!)
Is there a way to install this on trisquel?
https://wiki.debian.org/rndis_wlan
Broadcom wireless NICs do have free/libre drivers (in the kernel). However, the free/libre drivers don't work without non-free firmware. And it's impossible to replace the Broadcom with Atheros, because of Apple's non-standard physical interface.
You may have to rely on a USB-based ath9k-htc.
Since we are here, what is the difference between "wireless LAN controller" and "wireless network adapter"?
https://h-node.org/wifi/view/en/201/Broadcom-Corporation-BCM4313-802-11b-g-n-Wireless-LAN-Controller--rev-01-/1/1/Broadcom-Corporation/undef/undef/undef/wifi-works/undef
EDIT: the difference is 801.11 vs. 802.11.
Uhhhh..
Then.. will this one work? http://prod.danawa.com/info/?pcode=1083704&keyword=pcie+%EB%AC%B4%EC%84%A0%EB%9E%9C%EC%B9%B4%EB%93%9C&cate=1131685#bookmark_product_information
It's Intel PRO/Wireless 3945ABG mini-PCIe card, it's the same size and stuff as the one in my machine now.
I think I have this lancard in one of my old laptops in the warehouse, so if it would work I would try finding that laptop and replacing the card.. Thanks.
The laptop only got two usb ports, which is awful, so I can't use an external wireless lancard...
No way. Anything other than Atheros ath5k/ath9k require non-free firmware to work.
Therefore you'll need to rely on USB ath9k-htc, which also requires firmware work, but there is free/libre firmware for ath9k-htc. If you don't have usable USB ports, you'll also need a USB hub.
Installing non-free firmware defeats all purposes of free/libre operating system, so don't do it.
I would not write that "installing non-free firmware defeats all purposes of free/libre operating system". After all, running little proprietary software is better for our freedom than running a lot of it.
The support of Wi-Fi is a significant problem we face. It will not disappear anytime soon: acquiring a supported USB adapter (or several: I bought five some years ago, to give to people I care about and who want to try a 100% free GNU/Linux distribution) is a lasting investment for your freedom.
To have certainty that a Wi-Fi adapter will work, you can buy it to some of the few vendors that guarantee that what they sell will perfectly work with 100% free GNU/Linux systems, e.g., on http://libre.thinkpenguin.com (whose CEO, Chris, is to be thanked for convincing Qualcomm to liberate the firmware for AR7010 and AR9271 chipsets) or on https://tehnoetic.com
If you refuse to support those free-software-activist vendors, paying the (high) price they sell their adapters, search for AR9271. For instance, on the site you previously linked to: http://search.danawa.com/dsearch.php?k1=AR9271
Even if I don't have ath5k/ath9k/ath9k-htc, I'd not "running a little proprietary software". If I allowed the non-free firmware for PCI/e wireless NICs to be loaded, my system would be in grave danger (due to the DMA capability of PCI/e peripheral). Without security there is no freedom at all. By contrast, there are many ways to obtain temporary Internet access. I could use ethernet NICs, or cellular modems, or even USB tethering function of an Android phone. Even if it seems that you have to tolerate "a little proprietary software", the non-free firmware of PCI/e wireless NICs is the most intolerable.
> (whose CEO, Chris, is to be thanked for convincing Qualcomm to liberate the firmware for AR7010 and AR9271 chipsets)
Thank you very much, Magic Banana. I got one more important source for
my teaching work.
Previousely when I used that machine I had installed Ubuntu (Yes, I was a beginner and I didn't even know what FOSS is) GNU/Linux on it and didn't check on install proprietary drivers, and the wifi worked out of the box. Then would the proprietary driver be installed when I did that to that computer in the past?
Also, If there is no wireless lan working other than Atheros, how does all those mediatek chips work on OpenWRT, a free operating system for routers?
Is there a way I can check if there's a FOSS driver for a wlan card? Maybe a website?
Thanks.
And, the wlan card slot seems to be just pcie, which is same with all laptops. Perhaps because this macbook is an old one. (I can replace many things like odd, hdd, ram, wlan card, fan, speaker etc but not cpu)
You still confused "driver" with "firmware". They are very different. The "driver" you try to install is usually a firmware loader.
There is an easy criteria. Atheros ath5k/ath9k doesn't require firmware. ath9k-htc requires firmware but there is free/libre firmware. Anything else require non-free firmware, although most of them do have free/libre drivers.
https://trisquel.info/en/forum/broadcom-wifi
How did this work??
There may be very few non-Atheros wireless NICs that could "work" on free/libre distribution. However, I'm afraid that those very rare NICs has copies of non-free firmware stored somewhere on the cards. The non-free firmware is still loaded and executed every time, and is equivalently malicious and dangerous.
If your system uses standard physical interface (e.g. mini PCI/e or M.2 NGFF) and doesn't have white-list restrictions. You should buy one Atheros ath5k/ath9k (but not Qualcomm Atheros ath10k/ath11k) model.
However, h-node still lists them as "working with free software":
https://h-node.org/wifi/catalogue/en/1/1/Broadcom-Corporation/undef/yes/undef/wifi-works/undef
NB: the listing is sometimes contradictory, as with BCM4313 (see links above).
EDIT: my bad, the non-working one is 801.11, while the other one is 802.11 (although both are named BCM4313).
Don't believe whatever one says without critical thinking...
Even though such models do work with free/libre distributions, we should still strongly against them. They are extremely malicious and dangerous, because the DMA capability of PCI/e devices. At this point, a free/libre operating system running on free/libre hardware is defenseless, as the non-free firmware is beyond the reach of any security mechanisms deployed on the host system.
I do believe you, with critical thinking. My point was that the h-node list is confusing, in this case.
My problem is that a BCM4313 on an eeepc is, in fact, spontaneously making wifi connections with both Trisquel 9 and Devuan Beowulf. I only noticed it was not an ath9 device, as I always had assumed it was, when I plugged in a Parabola live system and could not see any wireless network.
I was using ethernet anyway, but checking h-node led me to take the necessary steps to enable it. If something works with linux-libre-firmware, how can we guess it is not, in fact, fully free?
Some Asus eeePC sub-models have Atheros AR2427, which is essentially AR9285 with 802.11n disabled. It should be usable on any eeePC model even if there is white list like IBM/Lenovo or HP. (Although I haven't tested whether eeePC does have white lists.)
For your question: how can we guess is it fully free? This depends on the type of the peripheral.
For PCI/e-based NICs, they have DMA capability. Therefore non-free firmware is intolerable, otherwise your system is in grave danger. Atheros ath5k/ath9k doesn't have dedicated processor or memory on it. Even if you want to run any firmware on it, you have nowhere to load it and it has nothing to execute it. Any other wireless NICs are designed to only work when firmware is loaded. It's very unlikely that Realtek or Broadcom specifically chooses a few models and designs them in a way like ath5k/ath9k, i.e., not requiring firmware to work. You can try to search for the (leaked) schematics of such cards, and you'd probably find that they do have dedicated memory and processor to run firmware. Such models happen to work with free/libre distributions because they have copies of firmware stored on them. In such cases, we must consider them non-free, because the non-free firmware is still loaded and executed every time, and is extremely dangerous.
For USB-based NICs or bluetooth adapters. The non-free firmware is far less dangerous, because USB peripherals don't have DMA capability. For example, I have a Qualcomm Atheros ath10k USB wireless NIC, and use it to connect to WiFi hotspots that don't support 802.11b/g mode.
Thanks, I might then replace that evil BCM wifi card with a spare aht9k if I am to give that eeepc away. For the time being, I will stick to wired connection and use another laptop if I need wireless.
Maybe the h-node directory should take this flaw into account and consider de-listing the Broadcom adapters as "working with free software", or at least add a red banner to warn users about the situation.
Even if we don't talk about freedom or security issue, Broadcom's performance is also the worst, even worse than Realtek and Ralink/MTK.
Ultimately, why should you always believe whatever h-node says? They are so amateur and unskilled.
Some old Broadcom chips are supported by OpenFWWF (I don't think BCM4313 is one of them). This firmware is included in Trisquel.
Again, why not Atheros? Even if OpenFWWF can be considered free/libre and secure, such Broadcom cards still have dedicated processor and memory, and still require firmware to work. This is still an important attack surface.
Less importantly, the performance and reliability of Broadcom cards are among the worst, whereas Atheros is the best.
I never recommended buying Broadcom cards, I have an ath9k card and that is probably the best option. I'm just pointing out that a few Broadcom cards actually work without proprietary software (contrary to your statement that anything other than ath5k/9k requires proprietary firmware) and thus their listing on H-Node is not a mistake.
Thanks for the clarification. Then I may put it in such a way: anything other than ath5k/ath9k requires firmware, most of which is non-free, and is extremely dangerous for PCI/e based peripherals, because of the DMA capability.
Do you mean that the aforementioned BCM4313 wireless LAN controller might, in fact, work with free firmware, as the h-node page seems to indicate?
That would indeed explain why I can use it with both Devuan and Trisquel.
No, OpenFWWF is only for certain chips that are supported by the b43 driver. The driver for BCM4313 is brcm80211, and there is no free firmware. If it is actually a BCM4313, I'm not sure what is going on. Maybe there is firmware in the ROM and it doesn't actually need the external firmware to be loaded. I have realtek ethernet and it requests firmware but works fine without it, so this could be a similar situation. Could you post the relevant lines from dmesg?
> The driver for BCM4313 is brcm80211, and there is no free firmware.
The h-node page says the driver for BCM4313 is b43.
> Could you post the relevant lines from dmesg?
Not for the moment, I'm afraid.
The kernel.org page says otherwise: https://wireless.wiki.kernel.org/en/users/drivers/b43
Also, BCM4313 is not listed as supported on the OpenFWWF page: http://netweb.ing.unibs.it/~openfwwf/
The H-Node page says that it works with b43 and OpenFWWF, but on the other hand the output from lspci indicates that brcmsmac is being used.
I don't think we will have a good idea of what is going on without more complete dmesg logs. Which driver is being used, and does it request any firmware?
I see.
> brcmsmac is being used.
I missed that. So the h-node page might be misleading, after all.
> without more complete dmesg logs
We will get a chance, only I am not sure when. This is intriguing enough, I would like to know more. For now, I still plan to replace it if the occasion arises.
> Such models happen to work with free/libre distributions because they have copies of firmware stored on them.
I am getting exactly the same dmesg output as the one posted on the h-node page (which happens to be a patchwork from various systems, making things even more misleading). So the conclusion is that BCM4313 does sometimes work in 100% free software systems, but only because of the non-free firmware stored on device. I still think that this nuance should be made clear for all users, given the hard work which is put into freeing kernels and systems. The truth is that such device does not work without non-free firmware. Mabye h-node considers firmware to be hardware?
Using a different networking option is a good recommendation, but for this to happen users should first be made aware of the situation. If their 100% free software system is fine working with their non-free wireless firmware because it is stored on the device, how are they supposed to guess?
https://wiki.debian.org/brcm80211
https://packages.debian.org/buster/firmware-brcm80211
EDIT: I just noticed a couple of extra dmesg lines while running a Devuan system on the same laptop:
b43: probe of bcma0:1 failed with error -524
brcmsmac bcma0:1: mfg 4bf core 812 rev 24 class 0 irq17
brcmsmac bcma0:1: firmware: direct-loading firmware brcm/bcm43xx-0.fw
brcmsmac bcma0:1: firmware: direct-loading firmware brcm/bcm43xx_hdr-0.fw
EDIT: problem solved in Trisquel 10 Nabia, dmesg now says: brcmsmac: fail to load firmware /*DEBLOBBED*/
Thanks!
Hi, first thanks all for the replies, I learned many things.
I don't really have issue with the lan cable, so I guess I'll just stick with that.
But by the way, does marvell cards also not work? I remember using trisquel (? or maybe debian without any proprietary sws; I don't remember quite well) on a computer with marvell avastar (?) wlan card and the wifi worked there. Then I was using the evil proprietary sw again (...??)
Thanks..
Marvell's wireless NICs are almost as bad as Broadcom's, and they do require non-free firmware to work. Ask Microsoft Surface Pro users for more details.
- Inicie sesión ou rexístrese para enviar comentarios