I'm researching an article about funding drivers & firmware for linux on mainstream hardware.
Looking for any and all feedback but tag in especially if you know about:
what skill-sets the community needs to develop these without industry support
budgets for driver development in-house at major shops, and whether it costs more or less in an OSS context
have OSS projects tried to support mainstream hardware on launch? are manufacturers willing to play ball?
reasons hardware companies don't support open software / ways to incentivize them to do so
I've owned a series of linux laptops and have an open-source router (librecmc, not openwrt), and am forever curious why manufacturers aren't more gung-ho about users like me.
Because a few enthusiasts don't bring in the big cheques; OEMs buying multiple thousands do. End users don't matter.
OEM doesn't care about FOSS either unless it results in cheaper operation. See what all the big boys do - they rely on FOSS for critical parts of their operation, but if they give back it's often very selective.
That being said, I'm sure some further reading would give you more info and insight. Asking OpenWrt users their opinion is not what I'd qualify as research, as knowledgeable as some people here might be.
True... Qualitative info is best gleaned from documented cases.
Nvidia being one. The key points being; Chipset driven based on perceived profit and to a lesser extent corporate ethics and direction.
Unsupported popular NICS being another one; key points being community driven X scale / complexity ( what is unknown )... ( or conversely Intel and your back in example one )
If you read between the lines... scale / featureset. As for costing more that's relative.
The NSS driver conundrum is particularly noteworthy in the SOHO router space. As the middleman ( the device manufacturer ) is perceived to be the "consumer"... Thus quantifying / qualifying realworld metrics of who a consumer is, and fostering accessible tooling / collaboration becomes paramount.
Modern hardware lifecycles for most but the largest manufacturers make such efforts unfeasible.
Top down driven "collectives" are possibly the only practicable broad ranging market influence, that can serve to provide balance where it is needed and realworld change.
( think schools adopting laptops / server farms and hp/ibm )...
The same things that makes c7v2/wndr3700v2 scarce on the second hand market and "half finished developer boards proliferate" ( RPi/Arduino etc ).
Products that are "too-popular/supported" outlive their lifecycle, which is a disincentive for producers as it drives down profit. For small producers ( developer boards example ) too long in development or too much development and profit takes a hit.
Why did google adopt non-removable batteries when they acquired motorola? Study the history of upstream acceptance of Android code base/s and you begin to understand the open source paradigm as it relates to producer incentive. linux support != linux support.
The closer the silicon is to the consumer... the more balance placed on the objective needs / result. The FOSS philosophy is not conducive to the "collectivism" approach.
Let's say my friend Linus Torvalds ( joking )... or similar top level linux foundation / consortium designated products as "linux certified" ( manufacturers opt in to adhere to principles of OSS, liase with regulators re: up to date drivers / toolchains etc. ).... this is possibly the simplest means of collectivism in the modern market. I'd happily pay $5-$10 more for every technical product I owned if it translated to bonafide Linux development and support.
That's right, one of the most critical pieces of information that can be used during development is the datasheet and it's errata.
For almost all devices, the datasheet is kept secret. For those that it's not (such as the raspberry pi), critical parts of the datasheet still being kept in secret (i.e. the GPU assembly and interface, the bootloader, many not "basic" hardware registers and it's errata is nothing near to updated).
Think also about Atheros, the firmware of the wireless chips is kept secret. The tool to calibrate the wireless radio is also secret. And many many examples more.
@lleachii by mainstream hardware I mean intel-based laptops other than apple. If you're a linux laptop user and your experience has been good, would be interested to know your laptop model & linux distro if you're willing to share.
I don't know, I'm not a device vendor. The thing is that is for real. That's one of the main reasons why driver's don't exist or are not at the same level of the OEM drivers; particularly when talking about embedded devices.
Purely guessing, protecting their supply chain and reputation against "end-of-reel" production of crap, as well as their "authorized" OEM/ODMs against competition from that crap.
Laptop model: All AMD/Intel-based x86_64 PCs and Laptops, with either Intel or Nvidia GPUs (I will admit other GPUs are not widely supported)
Disto: I prefer and regularly use Debian-based distros (i.e. Ubuntu); but also Red Hat-based ones too (e.g. CentOS)
I've only had issues using BSD-based distros; Linux works quite well. I'll also admit that non-mainstream devices, like: peripherals, USBs and expansion cards are what tend to have driver issues for me - hence why I inquired about your definition of "mainstream."
(BTW, I'm curious, have you posted in this forum before in the past; and perhaps lost your username/password?)