Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Chromium blacklists Nouveau

lists.freedesktop.org

Post reply on HN
Print view View original

271–280 of 411 posts

Re: Chromium blacklists Nouveau

#271

Post by magicalist »

Dylan16807 wrote: post #257
a11595 wrote: post #175

Earlier quoted context omitted.

you're the only one saying the don't. they are saying the most common linux configuration causes issues so blacklist the buggy driver in it they are also saying their that driver may work fine on some less common linux configurations, and the chrome team does not have the resources to test those. you are literally saying they are saying the opposite of what they said, then blaming them for it.

They were using "not enough resources" to support the idea of blacklisting. If they meant it as you say, that means they're blanketly blacklisting the most common config despite having the resources to test it, which is even worse.

Huh? It appears they did test it, and for that matter (according to the linked post) so did the Nouveau devs who also found it failed the easily accessible test suite.

Re: Chromium blacklists Nouveau

#272

Post by vetinari »

AnssiH wrote: post #220
emersion wrote: post #31

Nouveau are doing their best to reverse-engineer cards. NVIDIA doesn't help at all, not providing any docs and not interacting with the rest of the community. On the other hand, Intel and AMD have high-quality drivers and are part of the open-source community.

> NVIDIA doesn't help at all, not providing any docs and not interacting with the rest of the community. That is simply not true. Various docs, provided based on requests by Nouveau developers: https://download.nvidia.com/open-gpu-doc/ Documentation released in the last year include e.g. https://download.nvidia.com/open-gpu-doc/MemoryTweakTable/1/... https://download.nvidia.com/open-gpu-doc/MemoryClockTable/1/... htt…

Isn't the reclocking support for GM20x and newer the biggest pain point for last several years?

Re: Chromium blacklists Nouveau

#273

Post by monocasa »

vetinari wrote: post #262
monocasa wrote: post #219

Earlier quoted context omitted.

It's way more complicated than that. AFAIU, Nvidia's legal argument is that their driver can't be 'derived from' the Linux kernel even though it's linked into the kernel since it's older than their Linux port and only is linked against their GPLed (and if push comes to shove, dual licensed GPL and proprietary) kernel abstraction layer. Judges don't care about linkers and the GPL doesn't actually say anything about li…

This is not the derivation argument (that would be forcing GPL on Nvidia, which they want to counter, of course). This is about GPL infrigment; you can combine GPL licensed code with differently licensed code, but only in your privacy, without distributing the result to others. Once you would distribute the result, you would be breaking GPL, thus having no right to distribute.

The GPLv2 literally defines itself in relation to trees of derived works:

> The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language.

The argument that I've is that even when linked together, the Nvidia binary driver isn't a 'derivative under copyright law' because it doesn't have Linux specific entry points, and is older than their Linux port, and therefore isn't part of "The 'Program'" as listed in the GPL.

Re: Chromium blacklists Nouveau

#275

Post by Dylan16807 »

magicalist wrote: post #271
Dylan16807 wrote: post #257

Earlier quoted context omitted.

They were using "not enough resources" to support the idea of blacklisting. If they meant it as you say, that means they're blanketly blacklisting the most common config despite having the resources to test it, which is even worse.

Huh? It appears they did test it, and for that matter (according to the linked post) so did the Nouveau devs who also found it failed the easily accessible test suite.

We agree Nouveau is completely blacklisted right? That's the main problem here.

And the test suite failed on a minor issue where the cause was uncertain, which gets in the way of checking for the bigger definitely-the-driver problem.

Re: Chromium blacklists Nouveau

#276

Post by robocat »

larkeith wrote: post #96

> One idea is to flip GL_VENDOR to some random string if chromium is running. What? Don't do this - never do this. Regardless of whether or not you agree with Chromium's decision, it is their decision to make as the project maintainers; trying to trick the software with falsified data is overstepping your bounds, and both corrodes trust in your driver and leads to whitelists. I thought we had moved past the days of f…

Totally agree.

If you want to use hardware acceleration then go to about://flags and search for blacklist

"Override software rendering list: Overrides the built-in software rendering list and enables GPU-acceleration on unsupported system configurations. – Mac, Windows, Linux, Chrome OS, Android"

Simple and if you have telemetry logging on then you are possibly advertising that nouveau works for you.

Re: Chromium blacklists Nouveau

#277

Post by Wowfunhappy »

larkeith wrote: post #96

> One idea is to flip GL_VENDOR to some random string if chromium is running. What? Don't do this - never do this. Regardless of whether or not you agree with Chromium's decision, it is their decision to make as the project maintainers; trying to trick the software with falsified data is overstepping your bounds, and both corrodes trust in your driver and leads to whitelists. I thought we had moved past the days of f…

> trying to trick the software with falsified data is overstepping your bounds.

Fwiw, I agree with this statement only because Chrome lets users override the default behavior.

In the hypothetical event that Chrome was closed-source and forcibly disabling acceleration under a driver, I'd consider fakery on the part of the driver to be perfectly warranted.

Re: Chromium blacklists Nouveau

#278

Post by Wowfunhappy »

The author, Ilia, noticed that this is on HN and responded in the mailing list thread. I don't know if I agree with the take, but I'll repost because I found it interesting.

---

I've glanced at the HN discussion about this situation [...] and it does seem like people are focusing on the wrong thing... the important bit isn't that nouveau crashes and burns in some situations—everyone already knew that, including the users of nouveau who continue to use it nonetheless. It's that if every piece of software feels free to ignore a system integrator's or user's wishes, then the user now has to know how to override that behaviour separately in every application. The situation is that Distro X has decided that nouveau is the right thing for its users. A user can disable that by uninstalling or otherwise disabling nouveau if they wish. But now chrome comes along with its own set of rules. What if every application starts doing that?

It should also be noted that outside of a few pathological cases, like creating 2GB+ textures which never happens in practice, nouveau works just fine for me. For other people, it dies at random intervals, irrespective of whether they're using chrome or not. While this is a non-ideal scenario, chrome shouldn't be in the business of worrying about things like that. It just confuses the situation for everyone.

Re: Chromium blacklists Nouveau

#279

Post by Sir_Cmpwn »

staktrace wrote: post #140
Sir_Cmpwn wrote: post #60

Earlier quoted context omitted.

Firefox supports GPU accelleration on nouveau and all Mesa drivers from Mesa 8.0 forward.

Not by default. GPU acceleration is off by default in Firefox on Linux. Distros may change that in their bundles, but the official builds have it turned off.

Do you have a source for this?

This page disagrees:

https://wiki.mozilla.org/Blocklisting/Blocked_Graphics_Drive...

My unmodified Firefox stable channel on Linux has WebGL enabled OOTB and my distro does not have any specific flags specified for WebGL (or GL in general) in the package.

Re: Chromium blacklists Nouveau

#280

Post by ajross »

mehrdadn wrote: post #224
ajross wrote: post #153

Earlier quoted context omitted.

No, Chromium is free in the GNU sense, which is what I meant. You don't whine at open source providers to support features you want, it's bad form. Go fix it yourself, or use hardware that doesn't require this kind of hacky reverse engineering.

> You don't whine at open source providers to support features you want, it's bad form. Go fix it yourself This just about sums up the problem with open source. The response is always "if you don't like it, fix it yourself". Because everybody is a developer with copious free time to learn how to fix the problem in their favorite product. It's not like cloning and building Chromium takes hours. Or like you would lose…

> Because everybody is a developer with copious free time to learn how to fix the problem in their favorite product

It takes copious free time to learn how to run chromium with an --ignore-gpu-blacklist argument? What kind of strawman are you trying here?

Look. The open source NVIDIA driver is terrible, and Google doesn't want it messing up the experience of the people using its product. Who are you to demand that Google fix someone else's terrible driver (which is terrible through no fault of Google's -- NVIDIA would prefer it not exist at all)?

If you really cared about this at all, you'd be upset with NVIDIA. But you're not. You apparently don't even like open source software, which presumably means you're using the binary drivers and unaffected, right?

Post reply on HN