Support building with Linux 6.1-rc1

antoineco / broadcom-wl

Broadcom Linux hybrid wireless driver (64-bit)

https://www.broadcom.com/support/download-search?pg=Wireless+Embedded+Solutions+and+RF+Components&pf=Legacy+Wireless&pa=Driver&dk=BCM4312&l=true

163 stars 47 forks source link

Support building with Linux 6.1-rc1 #33

Closed satmandu closed 2 years ago

satmandu commented 2 years ago

This was tested on a MacBookPro11,3.

Linux mbp113 6.1.0-rc1 #1 SMP PREEMPT_DYNAMIC Sun Oct 16 23:33:56 EDT 2022 x86_64 x86_64 x86_64 GNU/Linux

[    6.477752] wlan0: Broadcom BCM43a0 802.11 Hybrid Wireless Controller 6.30.223.271 (r587334)
[    6.483372] wl 0000:03:00.0 wlp3s0: renamed from wlan0

broadcom-wl/6.30.223.271, 6.1.0-rc1, x86_64: installed

antoineco commented 2 years ago

Thanks!

I need to update the CI runner to this version for good measure.

Also sharing the reference to the upstream patch that makes this change necessary; https://lore.kernel.org/all/20220730052643.1959111-4-quic_vjakkam@quicinc.com/

antoineco commented 2 years ago

@satmandu I just addressed all my comments directly on your branch, and the changes build successfully on Linux 6.1 rc1.

Summary:

Stick with the integer types from the original codebase
Use inline patching of function parameters to reduce additions/deletions
No variable renaming

I'm ready to click the merge button whenever you confirm this is also looking good on your side

antoineco commented 2 years ago

FYI here is another approach which is less explicit but requires less changes: https://gist.github.com/joanbm/94323ea99eff1e1d1c51241b5b651549

I personally don't have a strong preference, being explicit is often good as well. In that particular case, I would even say it makes future changes to these function signatures easier to introduce.

satmandu commented 2 years ago

I was waiting for his patch to show up!

That patch is so much simpler. I would just go with that...

antoineco commented 2 years ago

I've been thinking about it since my last message, and actually I think I like your approach better. It's not as "simple", but if anything in these function signatures changes again, Joan's patch needs to be reverted/revisited, whereas with your version it's just another if. Less chances to break something unintentionally.

I'm also happy to see new people contribute patches and show up in the commit history, appreciate your work! 🙌 If Joan's approach turns out to bring other advantages in the future, I'm happy to revisit.