@vermaden
You can use `make makesum` to create / update distinfo.
This is how I update distinfo for nvidia GPU related ports.
And if you don't want to create and maintain local branches but work on "working tree" without committing, `git diff` would be your friend, unless you add any new files / directories. Adding option "-U999999" would allow you to generate diff for review (Phabricator).
If you add new files / directories, you can `git stash --include-untracked`, then, `git stash show -p --include-untracked` to generate diff including newly created files / directories and following `git stash pop` would restore your working tree.
Option "-U999999" would be wanted for `git stash show ...` part to generate diff for Phabricator.
My related post to freebsd-hackers ML here.
https://lists.freebsd.org/archives/freebsd-hackers/2025-November/005275.html
Thank You for adding this.
@vermaden
Precisely, -3{04|40|90} variants needs additional work to manually add entries for i386 versions when working on amd64 (if working on i386, it would be in reverse). Was annoying.
But it should no longer apply, as in-tree versions of them are exactly the final versions for 3xx series. 3xx series are already EoL'ed upstream and nvidia would never release updated versions even if any critical issues are found.
And all supported versions (Production, New Feature, Beta and legacy) dropped supports for i386 native environment (only supports 32bit emulation environments on 64bit). So `make makesum` works flawlessly now.
Note that current supported Legacy branch is 470 only. And 580 would appear once Production branch switches to 590 or later, as 590 drops supports for a bunch of old GPUs prior to Turing generations of architectures.
https://us.download.nvidia.com/XFree86/FreeBSD-x86_64/590.48.01/README/supportedchips.html
Here, 580 appears as "Legacy". So I believe 580 is promised to kept as Legacy after the transition.
So we'll add "-580" variant at the same time parent (aka master) ports are updated to 590 or later on FreeBSD. It would have an UPDATING entry just like the one dated 20260103.
To be honest ... NopeVidia does not exist for me ... unless they start providing proper open source drivers like Intel or AMD.
Systems I use/build are AMD based - both CPU and GPU - at least assuming I have a choice.
Here:
- https://vermaden.wordpress.com/2023/07/30/amd-based-freebsd-desktop/
- https://vermaden.wordpress.com/2024/08/04/perfect-nas-solution/
When I have no choice?
With devices such as ThinkPad W520 or my modified ThinkPad T480 (FrankenPad F25).
They come with Intel GPU and NopeVidia GPU - so I disable NopeVidia in the BIOS when I can - that is on W520.
... and as T480 does not have such option in the BIOS I jest load ppt(4) driver on it.
```
F25 # pciconf -lv | grep -B 1 -A 3 -i nvidia
vgapci1@pci0:1:0:0: class=0x030200 rev=0xa1 hdr=0x00 vendor=0x10de device=0x1d10 subvendor=0x17aa subdevice=0x225e
vendor = 'NVIDIA Corporation'
device = 'GP108M [GeForce MX150]'
class = display
subclass = 3D
F25 # pciconf -lv | grep -B 1 NVIDIA | awk '{print $1}' | head -1 | awk -F '@' '{print $2}' | sed '$s/.$//'
pci0:1:0:0
F25 # devctl detach pci0:1:0:0
F25 # devctl set driver pci0:1:0:0 ppt
```
... and on F25 I get a 'free' 0:30 in battery time because of that.
The only downside is that sometimes (W520) ALL the external monitors connectors are HARDWIRED to NopeVidia GPU ...
But that is me.