Created attachment 252771 [details]
Output from 'iw list'

Created attachment 252781 [details]
Output from 'modinfo brcmfmac'

Created attachment 252791 [details]
Output from dmesg after building brcmfmac with CONFIG_BRCMDBG=y and setting debug to 0x7fffffff

Same issue with MacBookPro13,2 as well (as it's obviously the same wifi chipset ;-)).

Same issues with MacBookPro13,3. Could be firmware related, like on the MBP 2015:

https://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/commit/brcm/brcmfmac43602-pcie.bin?id=ea1178515b8852a5ead23b165b4609ef2b49f36c

As I assume the same I contacted Arend van Spriel (who works at Broadcom and fixed it last time) a few weeks back with the request to help, but I got no response so far. :-(

revinfo shows a different chip revision compared to the chip in the MacBookPro12,1 (https://bugzilla.kernel.org/show_bug.cgi?id=100201#c21), whose proper function got fixed by a firmware update. So still certainly possible that it's something firmware related.

cat /sys/kernel/debug/brcmfmac/0000\:02\:00.0/revinfo

vendorid: 0x14e4
deviceid: 0x4318
radiorev: 0.13.32.105
chipnum: 43602 (aa52)
chiprev: 2
chippkg: 1
corerev: 49
boardid: 0x061b
boardvendor: 0x14e4
boardrev: P421
driverrev: 7.35.177.61
ucoderev: 0
bus: 0
phytype: 11
phyrev: 18
anarev: 0
nvramrev: 00000000

Created attachment 257709 [details]
dmesg output of "insmod brcmfmac debug=0xd416" with Linux 4.13-rc1

dmesg output of "insmod brcmfmac debug=0xd416" with Linux 4.13-rc1

During the captured trace multiple tries to connect to one of the Wi-Fi networks in range happened, but failed, even though the AP was only ~1m away. While the MacBookPro in question only detects a few Wi-Fi networks, a random, cheap Android device detects around a dozen.

(In reply to Daniel Roschka from comment #8)
> Created attachment 257709 [details]
> dmesg output of "insmod brcmfmac debug=0xd416" with Linux 4.13-rc1

So the dmesg shows:

[29270.808720] brcmfmac: brcmf_cfg80211_reg_notifier Enter: initiator=1, alpha=DE
[29270.809184] brcmfmac: brcmf_fil_iovar_data_get ifidx=0, name=country, len=12
[29270.809186] brcmutil: data
[29270.809189] 00000000: 58 30 00 00 00 00 00 00 58 30 00 ff              X0......X0..
[29270.809192] brcmfmac: brcmf_translate_country_code No country codes configured for device

The country code X0 is apparently stored in the device itself. Now this country needs to be supported by the firmware we released to linux-firmware and my guess is that it does not.

@Arend: So how to proceed? Can you provide a firmware with such support?

Dear @Arend,
Could you make a patch, please? All MacbookPro13 users will be extremely grateful to you.

Many thanks in advance.
All the best,
Luis.

Totally agree with @Luis!
@Arend, please make a patch, we all will be very thankful!
If you need any additional/debug info  - just give us a sign)

The same chip is also in the MacbookPro14,2.

@Arend, one more extra grateful Macbook Pro user here, should you be able to provide us with the firmware. You'd save us from using Macos when we could be using Linux. :D

Aaaand another one here willing to have decent wifi.

Another 14,2 user, who's loving running Linux kernel on mac, is also expecting this firmware update. Thanks and 谢谢

MacbookPro14,3 is also affected by this issue.

MacBookPro14,2 with 4.13.10 (fedora 26) also experiencing limited WIFI sensitivity and only wireless-n on BRCM43206 rev2 wireless-AC capable hardware. Please find a fix. 

Firmware installed but this does not fix problem.
https://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/plain/brcm/brcmfmac43602-pcie.bin has been installed in /lib/firmware/brcm/ and mac was rebooted. Problem persists.

Macbook Pro 14,2 user. Please fix this showstopper bug for the recent Macbook Pros.

Can you please share a timeframe in which you may fix that issue?
This would really help a lot! 

Thanks @Arend!

Is there anything we can do to push this forward? This is essentially the showstopper from running Linux on a pretty wide variety of Apple laptops.

Maybe there are "political reasons" why this is not moving forward? Arend: please do not reply to this bug report with more information if this is the case! :)

On the other hand, if something can be done outside the firmware, any hints or pointers  (other than reverse engineering the OSX drivers) would be highly appreciated!

Hello guys. I was in correspondence with Arend some time ago and got this answer (and permission to post his reply to this forum):

======= snip ==========

The firmware update is not an easy task as it updates regulatory data in firmware it means a full verification cycle is required to assure the power levels stay within regulatory limits. This is always a hard sell for us as there is no return of investment for the engineering costs involved. Hence we release fullmac firmware that has been verified in running projects. Sadly the 43602 in the MacBook does not use a fullmac driver architecture so there is no fullmac firmware available. I have forwarded your request, but no high hopes we can resolve this.

Regards,
Arend

======= snip ===========

I decided to sell my Macbook Pro and started using a Dell laptop instead with Linux working on it like a charm.

Are there any updates?

Is there any workaround?

Ok, I am really tired of using a usb wifi dongle, so I was digging a bit deeper and here is what I've got.

Changing ccode=X0 to 0 and regrev=15 to 0 in the firmware seems like improving things a little and makes the connection more usable:

- instantly connecting to the WiFi;
- link quality=70/70 (Max!);
- connections is stable, no disconnects;
- I've got stable 50 Mbps download and 20 Mbps upload (my ISP allows 200/20 Max as per my contract);
- The latency gets higher as I am downloading things, but otherwise it is as low as it should be;
- still no 5GHz ... not sure why, maybe some hints what can be changed in the FW ? I've tried many things already...;


--- 8.8.8.8 ping statistics ---
100 packets transmitted, 100 received, 0% packet loss, time 99154ms
rtt min/avg/max/mdev = 8.548/12.201/70.152/8.757 ms


Here is the FW diff:

# radiff2 -r /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig /lib/firmware/brcm/brcmfmac43602-pcie.bin
wx 3000 @ 0x0007c82f
wx 3000 @ 0x0007c839

# radiff2 -x /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig /lib/firmware/brcm/brcmfmac43602-pcie.bin
  offset     0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF    0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF
0x00000000  80f140b882f1a4b982f1b0b982f1bcb9 ..@.............   80f140b882f1a4b982f1b0b982f1bcb9 ..@.............
0x00000010  82f1cbb982f1dab982f1e9b982f1f8b9 ................   82f1cbb982f1dab982f1e9b982f1f8b9 ................
...
0x0007c820! 30643a66343a33650063636f64653d58 0d:f4:3e.ccode=X   30643a66343a33650063636f64653d30 0d:f4:3e.ccode=0
0x0007c830! 30007265677265763d31350061613267 0.regrev=15.aa2g   00007265677265763d30000061613267 ..regrev=0..aa2g
0x0007c840  3d3700616135673d370061676267303d =7.aa5g=7.agbg0=   3d3700616135673d370061676267303d =7.aa5g=7.agbg0=
0x0007c850  37310061676267313d37310061676267 71.agbg1=71.agbg   37310061676267313d37310061676267 71.agbg1=71.agbg
...


HW: MacBookPro14,3
Linux kernel: 4.16.8
WiFi card: Broadcom Limited BCM43602 802.11ac Wireless LAN SoC (rev 02)

Hope that workaround helps someone, until we get a proper FW, or a non-Apple replacement, finally :-)

Kind regards,
Andrey Arapov

Switching the txpower from 31 dBm down to 10 dBm (iwconfig wlp3s0 txpower 10dBm) appears to be also helping even with the non-modified FW.

I have also found the following posts which are still valid https://bugzilla.kernel.org/show_bug.cgi?id=103201

> Adrian Bjugård 2015-09-02 11:57:39 UTC:

> Also 5 GHz doesn't work, and the correct physical MAC address isn't
> recognised properly, everyone with this card (14e4:43ba) gets assigned the
> MAC address 00:90:4c:0d:f4:3e. Even by spoofing MAC address, it isn't
> possible to have two users of this card in the vicinity of each other.


> Arend van Spriel 2015-09-05 13:15:05 UTC:

> This issue has been root caused as a firmware issue. We are preparing a new
> firmware release to linux-firmware repo.

I am wondering whether we could extract the necessary FW from macOS, anyone has any pointers?

Kind regards,
Andrey Arapov

This is very interesting! I played around with the ccode and regrev values a bit, and it looks like the ccode modification isn't necessary: just changing the regrev to any single-digit value is enough. Also, in the `strings` output I saw 'ccode=%c%c' so I'm thinking the ccode, if modified, should probably be set to something like 'US' rather than a number.

Secondly, the interface still isn't very stable for me: it seems to get stuck very often (e.g. a download will proceed in bursts), and (probably as a consequence) it ends up dropping the wireless connection every now and then.

As to using the FW from MacOS, I strongly doubt that'll work - the FW's tend to be very different for different OS'.

Hi! I'm new here, but I was wondering if someone could tell me the location of the firmware file I need to edit to get wifi working.

It seems like the easiest fix is either to change regrev in the firmware, or to just lower the tx-power to 10. Do I need to do both? Which is better? (And how can I tell?)

Thank you!

A(In reply to Arend van Spriel from comment #9)
> [29270.808720] brcmfmac: brcmf_cfg80211_reg_notifier Enter: initiator=1,
> alpha=DE
> [29270.809184] brcmfmac: brcmf_fil_iovar_data_get ifidx=0, name=country,
> len=12
> [29270.809186] brcmutil: data
> [29270.809189] 00000000: 58 30 00 00 00 00 00 00 58 30 00 ff             
> X0......X0..
> [29270.809192] brcmfmac: brcmf_translate_country_code No country codes
> configured for device
> 
> The country code X0 is apparently stored in the device itself. Now this
> country needs to be supported by the firmware we released to linux-firmware
> and my guess is that it does not.

Hey Arend,

Thank you so much for this example and explanation. I just wanted to add my plea of help to this bug as well, although I am sure you have been fighting for us as best you can.

Thank you!

The full path of the firmware file in question is: /lib/firmware/brcm/brcmfmac43602-pcie.bin . As to regrev vs power vs both: try them out for yourself and report back here.

Created attachment 276179 [details]
attachment-17819-0.html

I can confirm Andrey's workaround to set ccode=X0 to 0 and regrev=15 to 0
in the firmware gives stable connection.
The other option, "sudo iwconfig wlp2s0 txpower 10dBm" works well too with
the original firmware.

HW: MacBookPro14,2
Kernel: 4.15.0-22-generic

Satya

On 11 May 2018 at 06:56, <bugzilla-daemon@bugzilla.kernel.org> wrote:

> https://bugzilla.kernel.org/show_bug.cgi?id=193121
>
> Andrey Arapov (andrey.arapov@nixaid.com) changed:
>
>            What    |Removed                     |Added
> ------------------------------------------------------------
> ----------------
>                  CC|                            |andrey.arapov@nixaid.com
>
> --- Comment #25 from Andrey Arapov (andrey.arapov@nixaid.com) ---
> Ok, I am really tired of using a usb wifi dongle, so I was digging a bit
> deeper
> and here is what I've got.
>
> Changing ccode=X0 to 0 and regrev=15 to 0 in the firmware seems like
> improving
> things a little and makes the connection more usable:
>
> - instantly connecting to the WiFi;
> - link quality=70/70 (Max!);
> - connections is stable, no disconnects;
> - I've got stable 50 Mbps download and 20 Mbps upload (my ISP allows
> 200/20 Max
> as per my contract);
> - The latency gets higher as I am downloading things, but otherwise it is
> as
> low as it should be;
> - still no 5GHz ... not sure why, maybe some hints what can be changed in
> the
> FW ? I've tried many things already...;
>
>
> --- 8.8.8.8 ping statistics ---
> 100 packets transmitted, 100 received, 0% packet loss, time 99154ms
> rtt min/avg/max/mdev = 8.548/12.201/70.152/8.757 ms
>
>
> Here is the FW diff:
>
> # radiff2 -r /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig
> /lib/firmware/brcm/brcmfmac43602-pcie.bin
> wx 3000 @ 0x0007c82f
> wx 3000 @ 0x0007c839
>
> # radiff2 -x /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig
> /lib/firmware/brcm/brcmfmac43602-pcie.bin
>   offset     0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF    0 1 2 3 4
> 5 6
> 7 8 9 A B C D E F 0123456789ABCDEF
> 0x00000000  80f140b882f1a4b982f1b0b982f1bcb9 ..@.............
> 80f140b882f1a4b982f1b0b982f1bcb9 ..@.............
> 0x00000010  82f1cbb982f1dab982f1e9b982f1f8b9 ................
> 82f1cbb982f1dab982f1e9b982f1f8b9 ................
> ...
> 0x0007c820! 30643a66343a33650063636f64653d58 0d:f4:3e.ccode=X
> 30643a66343a33650063636f64653d30 0d:f4:3e.ccode=0
> 0x0007c830! 30007265677265763d31350061613267 0.regrev=15.aa2g
> 00007265677265763d30000061613267 ..regrev=0..aa2g
> 0x0007c840  3d3700616135673d370061676267303d =7.aa5g=7.agbg0=
> 3d3700616135673d370061676267303d =7.aa5g=7.agbg0=
> 0x0007c850  37310061676267313d37310061676267 71.agbg1=71.agbg
> 37310061676267313d37310061676267 71.agbg1=71.agbg
> ...
>
>
> HW: MacBookPro14,3
> Linux kernel: 4.16.8
> WiFi card: Broadcom Limited BCM43602 802.11ac Wireless LAN SoC (rev 02)
>
> Hope that workaround helps someone, until we get a proper FW, or a
> non-Apple
> replacement, finally :-)
>
> Kind regards,
> Andrey Arapov
>
> --
> You are receiving this mail because:
> You are on the CC list for the bug.
>

When try to open /lib/firmware/brcm/brcmfmac43602-pcie.bin in vim, it appears to be gibberish. Am I doing something wrong? Is this the right file? Is there another way to edit .bin files? Never done this before. Any guidance would be appreciated! Thanks!

(In reply to jacooba9 from comment #32)
> When try to open /lib/firmware/brcm/brcmfmac43602-pcie.bin in vim, it
> appears to be gibberish. Am I doing something wrong? Is this the right file?
> Is there another way to edit .bin files? Never done this before. Any
> guidance would be appreciated! Thanks!

You want to use an editor capable of handling binary files: one such editor is 'bvi'.

Thanks, Ill give that a try!

Hi everyone, i just installed ubuntu 18.04 on my macbook pro 15" 2017 with touchbar. I try "bvi /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig" in the terminal but the file seems empty. I don't  what to do.

(In reply to alex.zerah from comment #35)
> Hi everyone, i just installed ubuntu 18.04 on my macbook pro 15" 2017 with
> touchbar. I try "bvi /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig" in the
> terminal but the file seems empty. I don't  what to do.


The driver doesn't have the '.orig' extension, it is just named "brcmfmac43602-pcie.bin". And the '.orig' is used to back up the original driver.

Has anyone tried changing the country code to US or another country code. We had a similar issue with broadcom chips on Sony android devices that were only detecting 2.4Ghz networks that was resolved by setting a real country code. Unfortunately I am not in a position to test it for the firmware here but would certainly think it worth attempting.

https://review.carbonrom.org/#/c/6681/1/configs/wifi/bcmdhd.cal

I believe that 00 is the code for `world` a brief google isn't showing me the list I would hope for though

attempted to change X0 to 00 and it did not work out. all networking features have gone down now

I've done a bit of work in this area, but really to no avail.

I've ccode=0,regrev=0 works the best, but it isn't great. ccode=XV,regrev=1 has a stronger signal for me, but still has significant drop-out problems. I tried as many combos as I could as well as checking the macOS and Windows drivers/firmwares for specific pieces that may work.

I have not been able to get 5Ghz channels to activate no matter what I've chosen.

FWIW I've seen problems with the wifi of this machine when running Windows 10 on it installed through BootCamp - problems I've never had when running Mac OS X against the exact same wifi router at almost the same times. I'll document the problems more fully if it'll be useful. Given Apple appear to have thrown considerable weight behind users running Windows on this hardware, perhaps pushing for fixes to problems which can be also be demonstrated under Windows will have more success, while also fixing the problems on Linux.

It would be a big help if someone could post the modified binary file with regrev=0 and ccode=0 (or whatever you have working). If you cant upload the file, could you just open it in sublime and post the text encoding? I’m really struggling with editing the file and getting it to work at all. Please and thank you!

the code with regrev=0 and ccode=0  is strictly worse for me: wouldn't connect at all (regardless of the tx-power). 

Anyone have this happen and find a solution?

Thanks, FYI, "sudo iwconfig wlp2s0 txpower 10dBm" works well (2.4G only) with
the original firmware (Ubuntu 18.04 + v4.17 custom build kernel) on MacBook Pro 2018 with touchbar.

In Dell's reddit community to avoid BSODs people are encouraged using a bit more fresh firmware. Tried it already, but seems like it have some differences in init/settings with the linux one. Maybe it is possible to copy only the proprietary part of it.
Currently used version: 7.35.177.61 (r598657) FWID 01-ea662a8c
Attached version: 7.35.240.55 (r609579) FWID 01-89725025
https://www.dell.com/support/home/us/en/04/Drivers/DriversDetails?driverId=JKRJ9
The firmware itself resides in Drivers/WinThresh/WL/x64/43602a1rtecdc.bin

Not working with the default value of 31 dBm at home, I have to set it to either 10 or 11 dBm.
At work, where the AP signal strength is poor, it works with the default 31 dBm and only close to AP.
May it be like the driver's number-to-signalstrength value is not linear to real dBm?

Andrey,
Thank you for your works on this.  It seems like it works out well for you... However, when I change my ccode to 0 and regrev to 15 wifi does not start.  Some other people on this thread also reported the same issue.  I can only lower the txpower as the work-around.  However, I cannot connect to most of the wifi as it is too weak.  I really want to get your work-around (modifying the firmware) to work.
Do you have any ideas?
I'm using 
HW: MacBookPro14,3
Linux kernel: 4.17.x (Fedora 28)
WiFi card: Broadcom Limited BCM43602 802.11ac Wireless LAN SoC (rev 02)
Thank you again.


(In reply to Andrey Arapov from comment #25)
> Ok, I am really tired of using a usb wifi dongle, so I was digging a bit
> deeper and here is what I've got.
> 
> Changing ccode=X0 to 0 and regrev=15 to 0 in the firmware seems like
> improving things a little and makes the connection more usable:
> 
> - instantly connecting to the WiFi;
> - link quality=70/70 (Max!);
> - connections is stable, no disconnects;
> - I've got stable 50 Mbps download and 20 Mbps upload (my ISP allows 200/20
> Max as per my contract);
> - The latency gets higher as I am downloading things, but otherwise it is as
> low as it should be;
> - still no 5GHz ... not sure why, maybe some hints what can be changed in
> the FW ? I've tried many things already...;
> 
> 
> --- 8.8.8.8 ping statistics ---
> 100 packets transmitted, 100 received, 0% packet loss, time 99154ms
> rtt min/avg/max/mdev = 8.548/12.201/70.152/8.757 ms
> 
> 
> Here is the FW diff:
> 
> # radiff2 -r /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig
> /lib/firmware/brcm/brcmfmac43602-pcie.bin
> wx 3000 @ 0x0007c82f
> wx 3000 @ 0x0007c839
> 
> # radiff2 -x /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig
> /lib/firmware/brcm/brcmfmac43602-pcie.bin
>   offset     0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF    0 1 2 3 4 5
> 6 7 8 9 A B C D E F 0123456789ABCDEF
> 0x00000000  80f140b882f1a4b982f1b0b982f1bcb9 ..@.............  
> 80f140b882f1a4b982f1b0b982f1bcb9 ..@.............
> 0x00000010  82f1cbb982f1dab982f1e9b982f1f8b9 ................  
> 82f1cbb982f1dab982f1e9b982f1f8b9 ................
> ...
> 0x0007c820! 30643a66343a33650063636f64653d58 0d:f4:3e.ccode=X  
> 30643a66343a33650063636f64653d30 0d:f4:3e.ccode=0
> 0x0007c830! 30007265677265763d31350061613267 0.regrev=15.aa2g  
> 00007265677265763d30000061613267 ..regrev=0..aa2g
> 0x0007c840  3d3700616135673d370061676267303d =7.aa5g=7.agbg0=  
> 3d3700616135673d370061676267303d =7.aa5g=7.agbg0=
> 0x0007c850  37310061676267313d37310061676267 71.agbg1=71.agbg  
> 37310061676267313d37310061676267 71.agbg1=71.agbg
> ...
> 
> 
> HW: MacBookPro14,3
> Linux kernel: 4.16.8
> WiFi card: Broadcom Limited BCM43602 802.11ac Wireless LAN SoC (rev 02)
> 
> Hope that workaround helps someone, until we get a proper FW, or a non-Apple
> replacement, finally :-)
> 
> Kind regards,
> Andrey Arapov

I've had a bit more luck with this.

I was a little baffled by the fact that I could see some of my neighbors' 5GHz networks, but not mine.  Output of `sudo iw list | grep -A 15 Frequencies:` gives me a bunch of 5GHz channels, though fewer than I'd expect.

My router was set to automatically pick a channel, and happened to pick something (5580 MHz) not in the list the driver was reporting as supported.  So I manually forced my router to something in the range listed as supported by the driver, and now I can successfully connect to my 5GHz network.

I imagine regulatory domain does matter, so perhaps also ensure that 'crda' is installed, and then you can `iw reg get` and `iw reg set $COUNTRY_CODE` to set yours properly.

For reference, I'm running Debian buster with their stock 4.17.17-1 kernel, with firmware-brcm8-20180518-1 installed.

Actually -- and sorry for the noise -- I just had my grep command not printing out everything it should have been printing out, so the higher frequencies are indeed listed by the driver.

But still, maybe despite what the driver supports, it's still having trouble working with some frequencies, so I'd suggest playing around with the channel on the AP.

I found the issue I had with modifying the firmware directly.  I need to change the ccode from X0 to 0. and regrev from 15 to 0. 
So the string would look like this:
=========
ccode=0..regrev=0..aa2g
=========
Once I reboot, I can connect to my wifi without using "sudo iwconfig <> txpower 10dbm" anymore.

My link quality sadly is not 70/70 like Andrey reported. 

I hope this helps someone who wants to modify the firmware directly..

Regards,

(In reply to sxnguyen from comment #46)
> Andrey,
> Thank you for your works on this.  It seems like it works out well for
> you... However, when I change my ccode to 0 and regrev to 15 wifi does not
> start.  Some other people on this thread also reported the same issue.  I
> can only lower the txpower as the work-around.  However, I cannot connect to
> most of the wifi as it is too weak.  I really want to get your work-around
> (modifying the firmware) to work.
> Do you have any ideas?
> I'm using 
> HW: MacBookPro14,3
> Linux kernel: 4.17.x (Fedora 28)
> WiFi card: Broadcom Limited BCM43602 802.11ac Wireless LAN SoC (rev 02)
> Thank you again.
> 
> 
> (In reply to Andrey Arapov from comment #25)
> > Ok, I am really tired of using a usb wifi dongle, so I was digging a bit
> > deeper and here is what I've got.
> > 
> > Changing ccode=X0 to 0 and regrev=15 to 0 in the firmware seems like
> > improving things a little and makes the connection more usable:
> > 
> > - instantly connecting to the WiFi;
> > - link quality=70/70 (Max!);
> > - connections is stable, no disconnects;
> > - I've got stable 50 Mbps download and 20 Mbps upload (my ISP allows 200/20
> > Max as per my contract);
> > - The latency gets higher as I am downloading things, but otherwise it is
> as
> > low as it should be;
> > - still no 5GHz ... not sure why, maybe some hints what can be changed in
> > the FW ? I've tried many things already...;
> > 
> > 
> > --- 8.8.8.8 ping statistics ---
> > 100 packets transmitted, 100 received, 0% packet loss, time 99154ms
> > rtt min/avg/max/mdev = 8.548/12.201/70.152/8.757 ms
> > 
> > 
> > Here is the FW diff:
> > 
> > # radiff2 -r /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig
> > /lib/firmware/brcm/brcmfmac43602-pcie.bin
> > wx 3000 @ 0x0007c82f
> > wx 3000 @ 0x0007c839
> > 
> > # radiff2 -x /lib/firmware/brcm/brcmfmac43602-pcie.bin.orig
> > /lib/firmware/brcm/brcmfmac43602-pcie.bin
> >   offset     0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF    0 1 2 3 4
> 5
> > 6 7 8 9 A B C D E F 0123456789ABCDEF
> > 0x00000000  80f140b882f1a4b982f1b0b982f1bcb9 ..@.............  
> > 80f140b882f1a4b982f1b0b982f1bcb9 ..@.............
> > 0x00000010  82f1cbb982f1dab982f1e9b982f1f8b9 ................  
> > 82f1cbb982f1dab982f1e9b982f1f8b9 ................
> > ...
> > 0x0007c820! 30643a66343a33650063636f64653d58 0d:f4:3e.ccode=X  
> > 30643a66343a33650063636f64653d30 0d:f4:3e.ccode=0
> > 0x0007c830! 30007265677265763d31350061613267 0.regrev=15.aa2g  
> > 00007265677265763d30000061613267 ..regrev=0..aa2g
> > 0x0007c840  3d3700616135673d370061676267303d =7.aa5g=7.agbg0=  
> > 3d3700616135673d370061676267303d =7.aa5g=7.agbg0=
> > 0x0007c850  37310061676267313d37310061676267 71.agbg1=71.agbg  
> > 37310061676267313d37310061676267 71.agbg1=71.agbg
> > ...
> > 
> > 
> > HW: MacBookPro14,3
> > Linux kernel: 4.16.8
> > WiFi card: Broadcom Limited BCM43602 802.11ac Wireless LAN SoC (rev 02)
> > 
> > Hope that workaround helps someone, until we get a proper FW, or a
> non-Apple
> > replacement, finally :-)
> > 
> > Kind regards,
> > Andrey Arapov

With ccode and regrev set to 0 I can see some networks, with ccode set to EU and regrev to 66 I think I can see all 2.4GHz networks. However, still with a very weak signal and changing txpower does nothing to fix that.
I also can't get 5GHz to work at all.

Is it possible to extract some usable things from the Windows driver?
If I remember correctly there was a project to support Windows wifi drivers a long time ago.

I had some success on a 13,3 Macbook using the bcm43602 chipset by adding the 
brcmfmac43602-pcie.txt to /lib/firmware/brcm on kernel 5.4 .

I first used files from different chips as a basis for the content, enabling the device to see 5ghz and in the end to establish a stable connection with my 5Ghz network.

I used the textfiles found in the windows driver package for Broadcom BCM94350ZAE DW1820A, changing device ids and macaddress to the values of my device.

Created attachment 285753 [details]
brcmfmac43602-pcie.txt working on macbook 13.3

The mentioned text file. I removed the macaddress of my device and replaced it with xx:xx:xx:xx:xx:xx . For usage I think you have to replace it with the macaddress of your device.

Hi Simon,

I have tried what you have suggested and voila!
Everything is working, 2GHz, 5GHz wireless networks.
Now I can finally get rid of my USB Wi-Fi dongle!

Much thank you for sharing this finding and making life easier :-)


I have also figured that the brcmfmac43602-pcie.bin driver has a hardcoded mac address in Linux:

# strings /lib/firmware/brcm/brcmfmac43602-pcie.bin |grep macaddr=
macaddr=%s
macaddr=00:90:4c:0d:f4:3e

So brcmfmac43602-pcie.txt file can be unique and everyone can start using their wifi card by simply placing it to their /lib/firmware/brcm/ directory, reloading the brcmfmac driver.

I guess the next step would be figuring a legal way of shipping that file along with the linux-firmware package so everyone can finally get their wifi adapter working.

Does anyone know who is able making this into a linux-firmware package?


Kind Regards,
Andrey Arapov

While 5Ghz WiFi networks seem to work with this txt file, I can't get 2.4Ghz networks to work with it on my MacBookPro13,2.

I also experienced some 2,4Ghz not connecting successfully, while others work fine.

I think there is still some modification of the parameters needed to find an optimal configuration and I will definitely do this at some point.

Unfortunately documentation for the parameters is not that good and the only method currently available to me is trial and error.

@Andrey: Great to hear that this works for you

I created an account just to say thanks to you Simon and Andrey. It is working great for me (on 5GHz, haven't tried 2.4GHz).

Thanks for the driver config, it does seem to help, but I have one problem I’m hoping ya‘lol can help with:

I can get the config to apply with an installed OS by restarting the machine, but that’s not very graceful. Plus, I also want to apply this config during an install, so restart isn’t viable.

How exactly do I go about “ reloading the brcmfmac driver”. Please ELI5

Thanks for your help :)

@domhnall.tilliere@gmail.com

rmmod brcmfmac && modprobe brcmfmac

I tried the txt file you attached, however I could not get 5Ghz to work at all..

It gave me error saying clm blob is not found and hence it disabled some bands.

I am not sure how I need to extract them from Catalina..

@mail@simonsiebert.de

I found the problem.. It seems the problem is because of Region.. Many 5GHz channels are disabled and my Wi-Fi Network was part of that, so could not even see it.

I tried setting iw reg set 00, and even though it enables it I was not able to see the networks. I guess I need to turn it off and on again ?

I’m not sure that it’ll work. You may have better luck modifying the driver file.

Thank you all so much for the research and solutions!
I was finally able to connect to my WiFi with my MacBookPro14.2

Distribution: 

$ uname -a
Linux caribou-MacBookPro 5.0.0-37-generic #40~18.04.1-Ubuntu SMP Thu Nov 14 12:06:39 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux

$ lspci -nn -d 14e4:
02:00.0 Network controller [0280]: Broadcom Inc. and subsidiaries BCM43602 802.11ac Wireless LAN SoC [14e4:43ba] (rev 02)

____________________
First section of the text file: brcmfmac43602-pcie.txt

boardrev=0x1101

sromrev=11
boardtype=0x073e
vendid=0x14e4
devid=0x43ba

macaddr=00:90:4c:0d:f4:3e

ccode=0
regrev=1
_______________________

So for new 2020 comers, here's what I did to solve the issue (I first had to reinstall the WiFi driver)
I solved my initial issue by reinstalling the WiFi driver for my chip using this article:
https://askubuntu.com/questions/55868/installing-broadcom-wireless-drivers

1. How I was able to reinstall the proper WiFi driver if you experience the No WiFi Adapter is available issue in Settings>WiFi
sudo apt-get purge bcmwl-kernel-source
sudo apt update
sudo update-pciids
sudo apt install firmware-b43-installer
sudo reboot


2. How I resolved connectivity issues
- Downloaded brcmfmac43602-pcie.txt here: https://bugzilla.kernel.org/attachment.cgi?id=285753
- Opened in Text Editor
- Changed the following lines:
macaddr=xx:xx:xx:xx:xx:xx
ccode=X3
regrev=15

to

macaddr=00:90:4c:0d:f4:3e
ccode=0
regrev=1

In Terminal:
sudo cp ~/Downloads/brcmfmac43602-pcie.txt /lib/firmware/brcm/
sudo reboot


Et voila le travail! I'm now writing this from my MacBook Pro connected on WiFi :)
So far great speeds and latency:

Download: 270Mbps
Latency Unloaded: 2ms
Latency Loaded :15 ms
Upload Speed: 180 Mbps

Happy New Year everyone!

The WIFI drivers worked just fine for me on MacBookPro13,3 with Linux Mint 19.3 and Linux kernel version 5.4.6; I needed to download the brcmfmac43602-pcie.txt from https://bugzilla.kernel.org/attachment.cgi?id=285753 and setup macaddr properly with help of the command 'ifconfig -a wlp3s0' and copy the modified file to /lib/firmware/brcm/brcmfmac43602-pcie.txt destination; after an 'sudo rmmod brcmfmac && sudo modprobe brcmfmac' it worked so I could connect to an 5G router with good speed and latency.

For more system info go to https://termbin.com/ozy4

Created attachment 287497 [details]
NVRAM captured via VFIO trace of Windows BRCM driver

I have a MacBookPro14,3 which I have been using with Ubuntu 18.04 and 5.3 kernel. WiFi has poor performance even with brcmfmac43602-pcie.txt from https://bugzilla.kernel.org/attachment.cgi?id=285753.

I validated Broadcom drivers shipped with Apple's BootCamp on a Windows VM running on this Macbook. The Broadcom WiFi device was assigned to Windows VM via PCIe pass through and it worked great in that environment. I then setup vfio_region_write tracing to capture FW and NVRAM from Windows environment which yielded the attached files. Sadly they do not seem to work with Linux brcmfmac drivers. I see FW initialization error. 

I am attaching them here in case someone who is knowledgeable about BRCM WiFi drivers would like to take a look.

Created attachment 287499 [details]
FW captured via VFIO trace of Windows BRCM driver

If you would like to learn more about the setup I used to capture this let me know.