Conversation

Notices

1. Embed this notice
   翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 08-Feb-2024 21:22:11 JST 翠星石

   • Dushman
   @dushman >amdgpu is a free driver but the firmware it interfaces with are all blobs. It does function properly though wdym?
   It's hardly a free driver if it's designed to refuse to work unless you load up proprietary software (although it could work somewhat without, considering that such GPUs when run under the radeonsi driver at least are able to output full resolution, too bad there's no 3D accel or suspend).
   
   Due to how less of it is proprietary than the typical proprietary driver, it's more reliable than such, but it does have quite a few bugs that can't be fixed.
   
   >Intel graphics stuff isn't fully free either though. It's just that the fw is shipped alongside the UEFI firmware and not handled by the OS in that case.
   It depends on the version of Intel Integrated grapics.
   
   Intel Integrated from the core 2 duo days is fully free, as GNUboot contains free graphics init (from coreboot) and there is a free Linux driver for those cards.
   
   Cards from a bit later don't require the user to load proprietary up software, although such is included in the BIOS or UEFI software.
   
   The newest integrated cards refuse to operate without proprietary peripheral software unfortunately.
   
   >What kind of GPU do you even use dawg?
   Intel Integrated in my thinkpad and a gtx 780 in my desktop (nouveau is a fully free driver for that card and has free peripheral software).
   
   Neither require me to load up proprietary software or agree to any proprietary software license - the only issue with the 780 is the proprietary VBIOS, but me or someone else will fix that eventually.
   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 08-Feb-2024 22:18:23 JST 翠星石

     in reply to

     • Phantasm
     @phnt >The new Ark and Xe graphics (iGPUs) don't require any proprietary drivers/software to run at all.
     >The only thing you need is a proprietary FW
     Make up your mind.
     
     Yes, they will refuse to operate usefully by running in 480p mode without 3D accel, even though full rez and 3D accel is proven easily possible with free software.
     
     >if that FW was burned on the chip, it would be considered a completely "Free" solution, no matter how you like it.
     Proprietary hardware isn't ever considered free, but if proprietary hardware works without proprietary software, then it's free from proprietary software.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Thursday, 08-Feb-2024 22:18:24 JST Phantasm

     in reply to

     • Dushman
     @Suiseiseki @dushman
     >The newest integrated cards refuse to operate without proprietary peripheral software unfortunately.
     Completely incorrect. The new Ark and Xe graphics (iGPUs) don't require any proprietary drivers/software to run at all. The i915 driver from Intel in kernel runs perfectly and is the official driver. Most of the graphics controlls available in the Windows software doesn't even change any settings for the HW level. Just enables some features on the Windows SW side. The only thing you need is a proprietary FW. Otherwise the GPU runs with nomodeset aka VGA (or probably SVGA). And because of the stupid FSF "Free" HW guidelines, if that FW was burned on the chip, it would be considered a completely "Free" solution, no matter how you like it.
   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 08-Feb-2024 22:23:47 JST 翠星石

     in reply to

     • condret :verified: :ancom:
     @condret >it caused my laptop to freeze sometimes and gave me bad perf in games.
     Nouveau is now stable on the 780 after a bugfix, but perf in games is still hit and miss, but slowly improving.
     
     The 780 does GNU emacs 1440p@120 just fine, so it works great.
     
     >i won't buy nvidia again, until they understand that their business is manufacturing gpus and selling them and not forcing me to use proprietary drivers.
     Based.
     
     I'm glad I've only ever bought second hand nvidia GPUs, so they've never directly got any money off me.
   • Embed this notice
     condret :verified: :ancom: (condret@shitposter.club)'s status on Thursday, 08-Feb-2024 22:23:48 JST condret :verified: :ancom:

     in reply to

     • Dushman
     @Suiseiseki @dushman good to read that noveau works for other people. last time (around 2016) i've tried it with my gtx765m, it caused my laptop to freeze sometimes and gave me bad perf in games.
     
     nvidia still hasn't released a free software version of their drivers. i won't buy nvidia again, until they understand that their business is manufacturing gpus and selling them and not forcing me to use proprietary drivers.
   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 08-Feb-2024 22:38:39 JST 翠星石

     in reply to

     • Phantasm
     @phnt >It may be not, but it's endorsed as if it was.
     >The "Respect Your Freedom" criteria don't ever mention hardware.
     How does RYF endorse hardware if it "doesn't even mention hardware"?
     
     >The HW runs in a failsafe mode, that limits it's capabilities, because it can't be ensured everything works as expected.
     The 480p mode is only so legacy OS's without drivers at least get an image.
     
     There are hardware flags to toggle the base features like the resolution and framerate, but rather than providing documentation on how to do so, Intel provides proprietary software that toggles the hardware flags and tells you to shut up.
     
     >It's the same thing why CPUs don't support HW accelerated virtual machines without microcode. They literally don't know how without it.
     Lol, lmao.
     
     Virtualisation is a hardware feature and I have no proprietary microcode software installed on my thinkpad or KGPE-D16 and virtualization works just fine.
     
     Intel is just so incredibly incompetent, that they now release broken CPU hardware and rather than releasing free software to fix such, they instead release proprietary software that unbreaks virtualization and does????.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Thursday, 08-Feb-2024 22:38:41 JST Phantasm

     in reply to
     @Suiseiseki
     >Proprietary hardware isn't ever considered free
     It may be not, but it's endorsed as if it was. The "Respect Your Freedom" criteria don't ever mention hardware. All it wants is that the hardware must run free software in all layers, that are user upgradeable. All the Intel GPU team has to do, in order to comply with this, is burn the FW in ROM, that can't be upgraded and then ask for a certification.
     
     https://libreplanet.org/wiki/Group:Hardware/Certifications/Respect_Your_Freedom/Criteria
     
     >Make up your mind.
     >Yes, they will refuse to operate usefully by running in 480p mode without 3D accel, even though full rez and 3D accel is proven easily possible with free software.
     
     That's literally what I said. The HW runs in a failsafe mode, that limits it's capabilities, because it can't be ensured everything works as expected. It's the same thing why CPUs don't support HW accelerated virtual machines without microcode. They literally don't know how without it.
     
     
   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 08-Feb-2024 23:32:55 JST 翠星石

     in reply to

     • Phantasm
     @phnt >That's how your computer and many of it's other subsystems start up/bootstrap [1]
     Only laptops have EC's and the boot process of the thinkpad is;
     EC detects power button press->global reset signal->CPU init into real mode->coreboot init and switches to 32-bit mode->RAMinit and remaining hardware init->GRUB loaded
     
     Desktops and servers generally use a SuperIO
     
     >You are welcome to disassemble the firmware and with a custom lowlevel OS find out what the separate flags do.
     The proprietary license on the peripheral software attempts to forbid reverse engineering, because there's certainly not malware in there.
     
     >I seriously doubt Intel VT-X and AMD-V run without microcode.
     Both VT-x and AMD-V with KVM work just fine running GNU/Hurd for me without proprietary software.
     
     Software emulation works too, but it's slower.
     
     >if they do, it's broken, because the microcode that is on the CPU die itself is weeks/months old even when the manufacturing starts.
     If the manufacturer is half competent, like AMD was during the Opetorn 62XX days, the original microcode is plenty stable, as it's not like there's anything special about a slightly faster CPU.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Thursday, 08-Feb-2024 23:32:57 JST Phantasm

     in reply to

     • Dushman
     @Suiseiseki
     >How does RYF endorse hardware if it "doesn't even mention hardware"?
     Read it and don't try to spin my words without context to fit your narrative.
     
     >There are hardware flags to toggle the base features like the resolution and framerate, but rather than providing documentation on how to do so, Intel provides proprietary software that toggles the hardware flags and tells you to shut up.
     
     That's how most firmware initializes hardware. You load up firmware, run it, the firmware sets hardware flags usually by setting values in the hardware's memory space, pulls a few signals and presses the go button. At that point, the main processor for the specific hardware wakes up, usually with the new firmware already loaded and with the settings set by the flags. That's how your computer and many of it's other subsystems start up/bootstrap [1]. Of course they won't document hardware flags. What do you think all the magic numbers in the drivers for AMDGPU do? You are welcome to disassemble the firmware and with a custom lowlevel OS find out what the separate flags do. Just like Apple ARM hardware is being booted up with Linux.
     
     >Virtualisation is a hardware feature and I have no proprietary microcode software
     You are running virtualization without a lot of HW acceleration. I seriously doubt Intel VT-X and AMD-V run without microcode. And if they do, it's broken, because the microcode that is on the CPU die itself is weeks/months old even when the manufacturing starts.
     
     The reason why it's semi-fast is because you are still running x86/x86_64 instructions on an x86/x86_64 CPU.
     
     cc @dushman Literally sorry.
     
     [1] EC detects power up signal->global reset signal->ME/PSP CPU init->CPU bring up into real mode->BIOS init from a set memory location->BIOS peripherals init->Boot loader/EFI binary->Kernel takes over control from BIOS->Protected mode->User space. In a very simplified fashion.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 00:15:21 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest
     I seriously doubt Intel VT-X and AMD-V run without microcode. And if they do, it's brokenIt's not necessarily broken, it's just older. Older doesn't mean broken...
     
     If Intel or AMD wants me to run their updated microcode, why don't they give me the source code? I would gladly run an updated version if they made it free.

     翠星石 likes this.
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Friday, 09-Feb-2024 00:34:46 JST Sexy Moon

     in reply to

     • SuperDicq
     • Dushman
     • Phantasm
     @phnt @dushman @Suiseiseki @SuperDicq they should release microcode for cpus out of support, that would be the ethical thing to do. but that would compete with their current products. this all goes back to vendors being unethical, which is the central and continually-relevant point of the FSF.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 00:34:47 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest it would break licensing for the ISA itselfThat's completely arbitrary and I think we all agree that licensing issues like this shouldn't exist.
     they would blow their specific CPU architecture wide openThat's a good thing and they should make it free.
     other smaller issuesSuch as?

     翠星石 likes this.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 00:34:48 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman
     >It's not necessarily broken, it's just older. Older doesn't mean broken...
     In this case it usually does, because microcode does in fact have security vulns. It's stupid, I know. For example without a proper microcode update, some speculative execution vulns cannot be fully patched. Kernel level mitigations do work, but some rely on microcode to do it's things properly.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 00:34:49 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman
     >If Intel or AMD wants me to run their updated microcode, why don't they give me the source code? I would gladly run an updated version if they made it free.
     
     Few reasons, it would break licensing for the ISA itself (some x86 instructions are handled by microcode and not by the CPU circuitry itself without help), they would blow their specific CPU architecture wide open and other smaller issues.
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Friday, 09-Feb-2024 00:37:00 JST Sexy Moon

     in reply to

     • SuperDicq
     • Dushman
     • Phantasm
     • Johnny Peligro
     @mischievoustomato @dushman @phnt @Suiseiseki @SuperDicq they don't ahve to support it, if they release the source for the microcode then someone else can potentially fix the security bugs that are discovered
   • Embed this notice
     Johnny Peligro (mischievoustomato@rebased.taihou.website)'s status on Friday, 09-Feb-2024 00:37:02 JST Johnny Peligro

     in reply to

     • Sexy Moon
     • SuperDicq
     • Dushman
     • Phantasm
     > support stuff out of support
     what
   • Embed this notice
     ロミンちゃん (romin@shitposter.club)'s status on Friday, 09-Feb-2024 00:37:51 JST ロミンちゃん

     in reply to

     • Sexy Moon
     • SuperDicq
     • Dushman
     • Phantasm
     • Johnny Peligro
     @Moon @mischievoustomato @Suiseiseki @SuperDicq @dushman @phnt are they releasing the signing keys as well? :l_think:
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 00:38:10 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest
     The fact that something contains security vulnerabilities doesn't mean it is broken. It is only broken at the point when these vulnerabilities actually get exploited.
     
     There are many (albeit a bit hacky) workarounds for the stability and security issues you described caused by no microcode updates. A lot of them are included in projects like GNU Boot.
     
     I fully advocate to do activism to make Intel or AMD give us the ability to fix these issues in a better way without their proprietary code.

   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Friday, 09-Feb-2024 00:38:43 JST Sexy Moon

     in reply to

     • SuperDicq
     • Dushman
     • ロミンちゃん
     • Phantasm
     • Johnny Peligro
     @romin @Suiseiseki @SuperDicq @dushman @mischievoustomato @phnt you release one final microcode update that unlocks it.
   • Embed this notice
     Johnny Peligro (mischievoustomato@rebased.taihou.website)'s status on Friday, 09-Feb-2024 00:42:09 JST Johnny Peligro

     in reply to

     • Sexy Moon
     • SuperDicq
     • Dushman
     • ロミンちゃん
     • Phantasm
     that seems reasonable, wont happen ever, but reasonable
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Friday, 09-Feb-2024 00:42:09 JST Sexy Moon

     in reply to

     • SuperDicq
     • Dushman
     • ロミンちゃん
     • Phantasm
     • Johnny Peligro
     @mischievoustomato @dushman @phnt @Suiseiseki @SuperDicq @romin I just popped into the thread to remind everyone that proprietary software is inherently unethical for this and other reasons.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 00:53:07 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest You mean an old version of LibreBoot.GNU Boot is a fork that is based on a previous version of Libreboot from before they started including proprietary blobs. GNU Boot is not just an old version of Libreboot, it has many changes that are not currently present in Libreboot.

   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 00:53:09 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman
     >The fact that something contains security vulnerabilities doesn't mean it is broken.
     The vulnerabilities have already been exploited. It _is_ broken.
     >GNU Boot
     You mean an old version of LibreBoot.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 00:57:41 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest Annoy Intel lawyers about that. And when they open x86, go to AMD for x86_64.That's indeed what the free software movement is trying to achieve.
     Issues with competition stealing your parts of your architectureThat's not an issue. There is no such thing as stealing here, there's only sharing.
     custom optimizations for selected workloadsWhat's wrong with this? It would be great if we could buy computers that are optimized for your specific workload.
     Generally it would be a lawsuit madness for anyone involved.It doesn't have to be if things were free as in freedom in the first place.

     翠星石 likes this.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 00:57:42 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman
     >That's completely arbitrary and I think we all agree that licensing issues like this shouldn't exist.
     We don't live in a perfect world, and never will. Truth is, you can't do anything about it, because x86/x86_64 is cross-licensing hell.
     >That's a good thing and they should make it free.
     Annoy Intel lawyers about that. And when they open x86, go to AMD for x86_64.
     >Such as?
     Issues with competition stealing your parts of your architecture, custom optimizations for selected workloads. Generally it would be a lawsuit madness for anyone involved.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 01:08:52 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest
     The discussion about Libreboot is very similar to the discussion on why Debian is not a FSF endorsed distribution I think.
     
     You can indeed install a fully freedom respecting Debian system if you configure it to be that way.
     
     Same with Libreboot, it can be fully free if you configure it like that. The existence of Libreboot or Debian isn't a bad thing, as some users clearly want this configuration and optional proprietary blobs.
     
     However for an organization like the FSF it would be very hypocritical to recommend people to endorse projects like Libreboot or Debian when those projects recommend to their users that they should install proprietary software.
     
     So that's why projects like GNU Boot or distros like Trisquel exist instead. Because their goals align fully and completely with the FSF, so they share the same message.

   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 01:08:53 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman
     >they started including proprietary blobs
     They don't include it, they allow them to be included and loaded. That's something completely different. Both build versions exist and you can choose from both.
     
     You get a choice, that's the whole point of it. You can either include/exclude the binary blobs for platforms where they aren't needed for better stability and some features, or you include them for platforms where they are needed for at least some free software support.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 01:14:50 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest A common misunderstanding about the RYF trademark is that it should imply that all products without the RYF trademark are not freedom respecting.
     
     This is not true.
     
     The point of RYF certification is that for those specific products the entire user experience from the website, payment method, ordering, getting support and using the product aligns with the ideals of the FSF. It's a full package deal.
     
     That's why the FSF also maintains h-node.org. A repository with information on a lot more hardware and their freedom status than the RYF certification program will ever provide.

     翠星石 likes this.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 01:17:45 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest On the specific drama that happened between the Libreboot and GNU Boot project I will not go in to because I am probably biased because one of the maintainers of the GNU Boot project is a personal friend of mine.

   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 01:17:46 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman
     >The discussion about Libreboot is very similar to the discussion on why Debian is not a FSF endorsed distribution I think.
     I agree with most thinks. Another point is that the lead developer and GNU/FSF don't really like each other. Libreboot was for some time a GNU project, then the lead developer quit that agreement, had some politically charged discussions, applied for it to be a part of GNU again after some time and never got back in. There was a whole blog post about it on the Libreboot website and maybe still is, if it survived the redesign.
   • Embed this notice
     SuperDicq (superdicq@minidisc.tokyo)'s status on Friday, 09-Feb-2024 01:21:43 JST SuperDicq

     in reply to

     • Dushman
     • Phantasm

     @phnt@fluffytail.org @Suiseiseki@freesoftwareextremist.com @dushman@den.raccoon.quest Don't worry, I am very well aware of everything that happened between Leah, the GNU Project and all the other Libreboot contributors in that timeframe.

   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Friday, 09-Feb-2024 01:21:44 JST Phantasm

     in reply to

     • SuperDicq
     • Dushman
     @SuperDicq @Suiseiseki @dushman That drama predates GNUBoot by about 6 years. You can find some articles about it on the libreboot website on the Wayback machine around 2017.