4 ms·

If you look at https://en.wikipedia.org/wiki/Comparison_of_open-source_wireless_drivers https://en.wikipedia.org/wiki/Comparison_of_open-source_wire... you can

If you look at https://en.wikipedia.org/wiki/Comparison_of_open-source_wireless_drivers https://en.wikipedia.org/wiki/Comparison_of_open-source_wire... you can see that a few of the atheros chipsets have both OSS firmware and OSS drivers, the intel ones only seem to have OSS drivers. Perhaps they are using one of those that have the OSS firmware/drivers combination?

walrus01 9y ago
It really doesn't matter, the actual RF baseband inside the chip is still closed source. Atheros never released it for any model. The firmware interface to the OS driver, yes, but not what's going on under the hood.
- SahAssar 9y ago
  It's still better than a proprietery firmware though, right? The baseband does not have as deep access to the system (like DMA), so having the firmware be OSS should at least be better than the alternative.
  - walrus01 9y ago
    I do not see the logic in having a M.2 NVME SSD attached to the pci-express bus, which certainly has a totally closed source binary in its onboard controller... That's in the laptop, and it's OK? It has access to every disk read/write operation and manages a small RAM cache for the SSD flash, organizes the write wear leveling algorithm, etc. But they have to go with some weird, ancient 802.11n card instead of a modern 802.11ac 3x3 MIMO, dual band Intel chipset card, because they don't like the binary blob in the ROM of the Intel minipci-express card? At some point in time you have to trust the devices you're attaching to the pci-express bus in an x86-64 system or you won't have any useful functionality left.
    - SahAssar 9y ago
      I agree in principle, but there is also something to be said for not having a closed system being able to talk both to a wireless network and have DMA. Anyway, the more open components the system the better, and if this shows there is an interest for an more open system, the next one might be more open than this.
    - FreedomWarrior 9y ago
      The problem is that many wifi adapters require loading a binary blob at initialization time. "Burnt-in" firmware is practically the same as hardware, which you already trust since you bought it. Not so much for the firmware blobs the vendor maintains in the "linux-firmware" kernel.org repository and that are absolutely required for the device to function.

walrus01 9y ago

It really doesn't matter, the actual RF baseband inside the chip is still closed source. Atheros never released it for any model. The firmware interface to the OS driver, yes, but not what's going on under the hood.

SahAssar 9y ago
It's still better than a proprietery firmware though, right? The baseband does not have as deep access to the system (like DMA), so having the firmware be OSS should at least be better than the alternative.
- walrus01 9y ago
  I do not see the logic in having a M.2 NVME SSD attached to the pci-express bus, which certainly has a totally closed source binary in its onboard controller... That's in the laptop, and it's OK? It has access to every disk read/write operation and manages a small RAM cache for the SSD flash, organizes the write wear leveling algorithm, etc. But they have to go with some weird, ancient 802.11n card instead of a modern 802.11ac 3x3 MIMO, dual band Intel chipset card, because they don't like the binary blob in the ROM of the Intel minipci-express card? At some point in time you have to trust the devices you're attaching to the pci-express bus in an x86-64 system or you won't have any useful functionality left.
  - SahAssar 9y ago
    I agree in principle, but there is also something to be said for not having a closed system being able to talk both to a wireless network and have DMA. Anyway, the more open components the system the better, and if this shows there is an interest for an more open system, the next one might be more open than this.
  - FreedomWarrior 9y ago
    The problem is that many wifi adapters require loading a binary blob at initialization time. "Burnt-in" firmware is practically the same as hardware, which you already trust since you bought it. Not so much for the firmware blobs the vendor maintains in the "linux-firmware" kernel.org repository and that are absolutely required for the device to function.