LinuxGizmos has an interesting article on how an Intel Engineer fixed up Linux's DisplayPort compliance, and got the kernel patch moved upstream.
At ELC 2017, Intel's Manasi Navare described how she patched Linux 4.12 for true DisplayPort compliance, and offered tips on pushing patches upstream.If you've ever hooked up a Linux computer to a DisplayPort monitor and encountered only a flickering or blank screen, we've got good news for you. A graphics kernel developer at Intel's Open Source Technology Center has solved the problem with a patch that will go into Linux Kernel 4.12. Manasi Navare's patch modifies Atomic Kernel Mode Setting (KMS) technology to gracefully drop down to a lower resolution to display the image."Someone had to fix this problem, so I said okay, I have the knowledge and I have the community to help me," said Navare at the recent Embedded Linux Conference.
At ELC 2017, Intel's Manasi Navare described how she patched Linux 4.12 for true DisplayPort compliance, and offered tips on pushing patches upstream.
If you've ever hooked up a Linux computer to a DisplayPort monitor and encountered only a flickering or blank screen, we've got good news for you. A graphics kernel developer at Intel's Open Source Technology Center has solved the problem with a patch that will go into Linux Kernel 4.12. Manasi Navare's patch modifies Atomic Kernel Mode Setting (KMS) technology to gracefully drop down to a lower resolution to display the image.
"Someone had to fix this problem, so I said okay, I have the knowledge and I have the community to help me," said Navare at the recent Embedded Linux Conference.
The only OS I've had issues with DisplayPort in has been Windows 10. Most Linux and BSD distros hate the hell out of my video card (Nvidia 1060) but once it's functional, stuff is good there. Windows, my second monitor is still giving me a full hall of mirrors like it does every time I wake from sleep. Only fix I've found for that is to open display settings (every time) and set monitors to clone, and then back to extended.
When I got my nVidia 1070, Linux Mint for some reason couldn't pull up the proprietary drivers for it. Manually installed the drivers and it worked fine. Found an article recently to update it to the latest version and it now responds to the proprietary driver selection menus properly. It used to take years for Linux to get drivers for new hardware up to snuff. Now you have to wait a few weeks. If you have some patience it isn't a big deal getting bleeding edge hardware to work properly in Linux. (I don't use displayport though so ymmv)
Well, when I say "hate the hell out of", I mean, "I had to manually install the proprietary driver". It's not something I've had to do in Linux for years. I'm toying with Sparky at the moment, which I believe is similar to Mint in the level of polish it has.
It's still leaps and bounds better than where Windows is. Seriously, if it weren't for games...
I'm already using DisplayPort and it's on a >60hz 3440x1440 monitor. No issues.
I use DisplayPort too, with a 4K monitor @60Hz. The kernel is 3.16.0
But I do have issues.
Sometimes the monitor won't wake up after sleeping, like there's no signal. I have to turn it off and on again, and sometimes change virtual consoles, or even restart X.
But what annoys me the most is that it freezes for some seconds in some situations. Like right after I login (it boots to the graphical login screen very fast, but takes a long time to load the desktop and unfreeze after I login), or after I connect or disconnect a HDMI cable to the TV.
I have a 4K HDMI monitor connected to a Broadwell NUC running Windows 10, and it still can't get itself right most of the time.
Watch a 2 hour movie on Netflix, at some point there will be at least 30 seconds of blank screen while the monitor and CPU are re-syncing with each other. MTBF on the HDMI sound interface seems to be around 12 hours, sometimes faster, sometimes slower, but eventually if you run it long enough, the sound will go away.
It's down to drivers, I'm sure - but even Windows 10 doesn't seem to be getting their HDMI drivers completely right. I know 4K is "challenging" - but, please, we've been doing HDMI for over a decade, can we get it right already?
That is probably the DRM fucking you over. (You using The Edge browser to get 4K instead of 720P?)
Chrome, and from across the room I don't really care if it's 4K or 720p.
Yeah, those could be DRM glitches on Netflix. When playing .mkv files back on Kodi, we get different glitches :(
Does it turn back on if you unplug and replug the displayport cable? For me, this forces it to come back to life. I'll look forward to it being fixed properly, since it's something which I've been annoyed by for years!
Is this a problem for OpenBSD, NetBSD, FreeBSD etc too?
(btw, the problem with NVidia is their closed policy, so avoid them)
Step 0: Make sure we have control of the situation.Step 1: Verify that if we do X, it will work.Step 2: Step 2 says X will work, so commit X.Step 3: Ooops, the commit did not work as planned.Step 4: Double oops, I don't have a plan for commit not working.
What can I say, with the real world, sometimes stuff happens.The patch added a recovery plan for the commit failing.
Probably a good lesson for folks that think they can do step 0.
> she patched Linux 4.12 for true DisplayPort compliance
And I'm sure she was met with nothing but hostility and vitriol because of the inherent misogyny in tech.
What's that? She wasn't?
How about some good 'ol racism because she isn't white?
None of that either? Damn it.
It's almost like they only cared about the code being good..
(On a serious note, thanks Manasi for your contributions!)
About a year ago, Intel's work force was "75 percent male and a combined 86 percent white and Asian." It's amazing how far they've come.
As a guess, if they've moved beyond that, and are hiring in North America, to vary much from those number means they're practicing discriminatory hiring. Shouldn't their 'diversity' numbers be pretty close to those of university's graduation rates?
Also, why lump "white and Asian" together?
In worldwide IQ surveys white and Asian people tend to come up with top scores. Being thing oriented and taking initiative also favors male think styles, ie also persons with female sex but male brain. Which makes the current employee mix become quite clear why it is as it is.
"Also, why lump "white and Asian" together?"
Because Asian and European cultures are the hardest-working, most empirically, statistically, and demonstrably successful cultures on the planet who do not have obsessive victim complexes.
Exemplia gratis, nobody cares that the Chinese were slaves, because they do not bitch about it.Nobody cares that the Japanese in the US were thrown into camps, because they do not bitch about it.Nobody cares that millions more Russians were exterminated than anyone else in WWII, because they do not bitch about it.
Etc., etc. Whoops - guess reality is racist, better double up on my daily dose of cognitive dissonance and white guilt.
According to the article I linked, the company's goal is to "reflect the available talent pool." Right next to the part I quoted, the article says that they fell short of that and are making an effort to remedy the situation. The pool of actual applicants may be a better proxy for the "available talent" than is the pool of North American university graduates. If, as you appear to be saying, around 75% of those graduates are men and around 86% are white or Asian, that doesn't reflect the general population. I think you're mistaken about the first part of that: in 2009 in the United States, 41.3% of college degrees awarded (associate's through doctoral) went to men.
The situation, of course, may be different in Canada and Mexico. Whence comes your information?
According to the 2000 U.S. census,
The White non-Hispanic populationrepresented 71 percent of peoplewho reported exactly one race and70 percent of the total population.
-- https://www.census.gov/prod/2001pubs/c2kbr01-4.pdf [census.gov]
In 2010, it was less, around 66%:
[...] the number of Whites who reported onerace and identified as non-Hispanicnumbered 196.8 million, or 64 per-cent of the total population.Among the 7.5 million people whoreported White in combination withan additional race group(s), 2.4million were Hispanic. Multiple-race White respondents who wereof Hispanic origin represented1 percent of the total population.
The company does have a presence in Canada and in Mexico, among other countries.
https://www-ssl.intel.com/content/www/us/en/jobs/locations/canada.html [intel.com]https://www-ssl.intel.com/content/www/us/en/jobs/locations/mexico.html [intel.com]
> Also, why lump "white and Asian" together?
I would assume that, in the company's estimation, those are the groups which are adequately represented among its staff. Here's a link to their latest report, which includes figures going back to 2014.
To their credit, they are trying to encourage students in the under-represented groups, so that the pool of qualified workers may, in time, more closely resemble the general population.
I'm thinking of graduation rates specifically in technology, although I guess they do hire more than just engineers.
Or, maybe she succeeded despite all that. It's pure conjecture either way.
Exactly! Does Navare understand just how much damage she's done to the Misogynerd Narrative by focusing on her accomplishments?! We've got an AC modded to +4 because she didn't throw her hands up and scream sexual harassment the first question or concern some inferior assigned male scum dared to rape her with!
And I'm sure she was met with nothing but hostility and vitriol because of the inherent misogyny in tech.What's that? She wasn't?Well.. fuck.
Hooray! This ONE brown girl proved once and for all that racism and misogyny has been eradicated from the tech sector! Bravo insightful AC for pointing out.... exactly what?
Next up: Black man is assisted by friendly police officer in his search for his lost dog proving that racism and police corruption is no longer a problem.
Why did that even cross your mind?
Because he's sick and tired of being beat up by liberals for having a different opinion.He's been following all the rules, using politically correct speech in order to hide his bigotry and liberals just won't respect all the effort he puts in.They still call him a bigot anyway. Its not fair! Bunch of meanies. Why do liberals get to be mean but he doesn't?The white man just can't catch a break.
> Hooray! This ONE brown girl proved once and for all that racism and misogyny has been eradicated from the tech sector!
Funny, because I'm pretty sure I didn't say that. I was mocking the claim that racism/misogyny is inherent to tech (and thus, pervasive, happens to everyone of the affected groups). That doesn't require me to claim that there is none at all. Perhaps, and this is a bit crazy so bear with me.. I don't believe either of these extreme claims?
> Why did that even cross your mind?
I just checked down the list of oppressed groups and mentioned the ones I figured she could claim to be in. Pretty simple..
This was an article about a kernel patch. It had nothing to do with who submitted it.
Is there a problem to be solved? Do you understand the problem?
Have you examined the existing system sufficiently to understand how it behaves, how it should behave, the difference between its current behaviour and the new desired behaviour and any other parts of the system that depend on the existing behaviour being preserved? If so, what are you going to do about it?
Can you design a solution? OK, can you design several different solutions so that you can talk about them, appraise them and pick the best one (choose your definition of best carefully).
Can you implement your chosen solution? Did you have to fix any existing problems first? Did you report them? Did you analyse them enough to understand them thoroughly? Have you inadvertently changed any behaviour that something else depended on? Have you made it difficult to modify other parts of the system?
You implemented your chosen solution, great. Does it work? Can you prove it? Is it repeatable? Can other people understand the code and any changed existing code? Has someone else reviewed it and pointed out the mistakes you didn't find yourself?
Did you fix those mistakes and prove that things still work, or work better, and that you didn't accidentally break anything?
Did anyone who reviewed it come up with any insight into better designs or better solutions? Are they worth implementing?
Does the system as a whole behave as expected? Would the customer buy it?
Are you happy for your work to be published for the global peanut gallery to criticise...?
Some people choose to do this for fun. Others do it hoping to get paid.
You seem to like questions.