Check if mainline Linux has blobs for our targets
Check if torvalds/master Linux kernel includes blobs for our target devices (GT-I9300 and GT-I9305 so far). If so, we should change upstream to linux-libre.
Actually, linux-libre cannot be our upstream, because it is just a set of scripts to clean the source tree. linux-libre does not provide a git repo that we can follow.
We should try running linux-libre scripts on our kernel, check if they removing anything related to our targets, and then try it.
Updated by Denis 'GNUtoo' Carikli 4 months ago
The issue with linux-libre is also that it completely blocks the use of nonfree firmwares.
At this time I know no WiFi chip used in smartphone or tablets that work without nonfree firmware apart from two cases where the firmware has been put on a dedicated flash chip to not require the distribution to load the firmware. The WiFiResearch wiki page has more details on that.Replicant shall not redistribute nonfree software because:
- It's wrong and creates a ton of issues: Replicant wound't be fully free anymore
- The FSDG also forbids it
The Replicant project also supports people using ath9k_htc compatible WiFi dongles, and makes sure that even if nonfree firmware can work, people are not pushed into installing them at all. In fact any Replicant upgrade will result in the firmwares being removed.
However distributions still have some leeway on if we should completely block nonfree firmwares or not: we're not strictly obligated to use linux-libre nor to block the use of nonfree firmware.In my opinion it makes sense for FSDG compliant GNU/Linux distributions to block any nonfree firwmare:
- We have already several drivers that works without nonfree firmwares or with free firmwares for the internal WiFi on supported devices.
- It push people work to get free software WiFi firmwares: The firmware for the ath9k_htc compatible WiFi chips was also freed thanks to that.
However I think that we could actually use the Replicant situation in the opposite way: We could try to push people to work on getting a free WiFi firmware for an Internal WiFi chip and then switch to linux-libre once it works in linux-libre. In turn that could push more people in doing that kind of work as we'd have a recent successful example and a more strong push toward free firmwares.
We could try to push people to work on getting a free WiFi firmware for an Internal WiFi chip and then switch to linux-libre once it works in linux-libre.
Indeed, I do agree with you. The way that non-free firmware must be put into Replicant (copying the files, having to do it after each update, etc.) makes it really clear that the project does not support it. Let's keep torvalds/master as our upstream then.
Even so, I will leave this issue open until we have the opportunity to test Replicant 10 with linux-libre. All things point out that we are not missing any blob, but better double-check. Also, we must be careful with linux-libre because their process is (comprehensibly) highly automated and they can end up being too pedantic and removing blobs for which the source is available or that turn out to be data.
By the way, linux-libre just created a git repository with their releases, which is quite handy for us: