Announcement

Collapse
No announcement yet.

Broadcom Bids To Snatch Qualcomm For $103 Billion

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    Originally posted by pal666 View Post
    because you live in imaginary world where the only open arm vendor gpu driver is not developed by broadcom
    Hellooo, Broadcom makes wifi chips too, and I've had my fair share of dealing with Linux laptops sporting their hardware to know their stuff doesn't always "just work."

    Comment


    • #42
      Originally posted by pal666 View Post
      qualcomm trojan horse will force broadcom to stop developing open gpu drivers?
      i'd expect the opposite
      Broadcom did not develop open gpu drivers? AFAIK those were developed by a guy figuring it out on his own, then getting hired by Broadcom because of it. Broadcom only gave out specs not drivers themselves.

      Remembering the utter PITA Broadcom hardware has been on PC platform and in network appliances, open-source- and open-drivers-wise, it could burn in Hell for all I care.

      Comment


      • #43
        Originally posted by aht0 View Post

        Nvidia is active more or less only on PC GPU market.
        Doesn't change what I said or the truth of the matter.

        Broadcom put a lot more effort towards supporting open source than NVIDIA ever will.

        Comment


        • #44
          Originally posted by brad0 View Post

          Doesn't change what I said or the truth of the matter.

          Broadcom put a lot more effort towards supporting open source than NVIDIA ever will.
          Umm.. how exactly? Look at routers. DD-WRT guys had to pay broadcom for getting driver support and have to keep'em closed source because of NDA.

          I can't use bunch of broadcom chips in BSD but I can use Nvidia's GPU. Have always been able to. Nvidia at least made token effort of supporting my OS, be'em binary drivers or not. AMD, your love child and "good underdog", has not bothered even that much. Broadcom, even less.

          The fact that Broadcom made exception for a Raspberry, does not make it "good" in my eyes. It's a world of gray shades, where Broadcom is darker than most.
          Last edited by aht0; 08 November 2017, 12:51 PM.

          Comment


          • #45
            Originally posted by pal666 View Post
            what amount of Tegras will count as many?
            I hadn't noticed that that many devices were using Tegra now. Sorry about that, I guess Nvidia is even more of a problem than I realised.

            Originally posted by pal666 View Post
            Qualcomm trojan horse will force Broadcom to stop developing open gpu drivers?
            i'd expect the opposite
            Originally posted by pal666 View Post
            I don't see a singe OpenGL driver with open source(which is order of magnitude more important that firmware) from Qualcomm. unlike Broadcom
            Has this changed at all:

            Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite


            because it seems like while the driver is being worked on, there is no official involvement in open source VideoCore firmware/bootloader efforts, unless I'm missing something:

            Open source VPU side bootloader for Raspberry Pi. Contribute to christinaa/rpi-open-firmware development by creating an account on GitHub.


            Sure, if Qualcomm makes no open source drivers at all that's worse, but with two separate companies you can buy the most open source supported product from each and with enough combined purchasing power both companies might notice that there open source products are selling better.

            Comment


            • #46
              Broadcom and Qualcomm are very similar. Both are inclined on NDAs, both are proprietary-minded, both make ICs which are nearly impossible to buy on open markets, and overall both companies are quite hostile to opensource, tinkering, experiments and somesuch. Both are quite good at backdoors and all kinds of shady closed-source firmwares together with lack of documentation and all kinds of digital lockdowns, restrictions and patent trolling. So I do not think merger of two hostile entities would result in any improvement for Linux or opensource. "Two turkeys do not make an eagle". These two are real turkeys when it comes to Linux and opensource.
              Last edited by SystemCrasher; 11 November 2017, 09:11 AM.

              Comment


              • #47
                Originally posted by SystemCrasher View Post
                Broadcom and Qualcomm are very similar. Both are inclined on NDAs, both are proprietary-minded, both make ICs which are nearly impossible to buy on open markets, and overall both companies are quite hostile to opensource, tinkering, experiments and somesuch. Both are quite good at backdoors and all kinds of shady closed-source firmwares together with lack of documentation and all kinds of digital lockdowns, restrictions and patent trolling. So I do not think merger of two hostile entities would result in any improvement for Linux or opensource. "Two turkeys do not make an eagle". These two are real turkeys when it comes to Linux and opensource.
                Not picking on you this time. Agree with you 100%.
                Any examples on backdoors?

                Comment


                • #48
                  Originally posted by aht0 View Post

                  Not picking on you this time. Agree with you 100%.
                  Any examples on backdoors?
                  Qualcomm CPUs with cell modem have quite fancy system architecture. On power up, modem CPU boots first using all these scary blobs and secure boot kind of thing. Then it brings up secondary CPU, one that runs Linux. So Linux is basically fancy GUI to Qcom modem but it mostly live without it. Fancy, isn't it? Of couse primary CPU of cellular modem enjoys by full access into system, runs unknown huge proprietary blobs all the time and its hard to do something meaningful about it. So to put it directly it appears like supercharged version of Intel ME kind of thing. And you see, these days even if you point finger on someone, they would insist its "engineering login" they've "forgot" or just a "bug". So it has became rather difficult to get idea where bugs end and backdoors start. Say, Broadcom wireless firmware has recently scored a bug, leading to code execution. Radio takeover is bad on its own, but it could also be escalated to full system takeover by means of doing DMA "right" way. And it has been hidden in proprietary firmware, more difficult to analyze and fix. It could be unintentional. Or not. Who knows? This is shady blob developed behind closed doors, we have no access to commits history and so on. How do we know if this bug is intentional or not? Especially when everybody caught on backdoors pretends its a some unintentional mistake. Oh yea, code execution in firmware is mistake. System takeover is a mistake. Engineering login is a mistake. RRLP implementation in cell modem and its ability to access GPS hardware is a "mistake" (oh yeah, that's a backdoor, if you will - it would send your GPS coordinates to cell net without even asking your consent, of course "for your own good"). I would say it is bit too much of "mistakes" around of some shady companies. These companies proven to be secretive and uncooperative, which does not adds up either. When someone refuses to publish datasheets it raises some questions as well. What they are hiding? Lists of registers? Highly unlikely, since registers give competitors very little knowledge, its just an external api to HW, which hardly gives details on inner working. So it usually implies there're more nasty things to hide. Somehow Qcom and Bcom proven to be damn good at this. Even worse than Intel and AMD. OFC their crap eventually leaks, despite of NDAs and somesuch. Then, honestly, I wouldn't live long enough to fully reverse cell modem firmware to know all its dirty secrets (most cell things have hell a lot of undocumented stuff, btw. Qcom modems as well). So if something looks like backdoor and behaves like backdoor, I would assume it is a backdoor. Qcom system design looks like backdoor. Bcom wifi radios are looking like backdoor. They could claim it just unintentional set of decisions taken, but since everyone tells it "unintentional", get the idea.
                  Last edited by SystemCrasher; 19 November 2017, 02:54 AM.

                  Comment


                  • #49
                    Originally posted by aht0 View Post
                    Not picking on you this time. Agree with you 100%.
                    Any examples on backdoors?
                    If you don't want to read his less-coherent ramblings with no proof whatsoever, here is an article on Replicant wiki describing an actual backdoor in Samsung smartphones where the modem can read/write stuff where it wants in the smartphone's filesystem. https://redmine.replicant.us/project...GalaxyBackdoor

                    Comment


                    • #50
                      Originally posted by starshipeleven View Post
                      If you don't want to read his less-coherent ramblings with no proof whatsoever, here is an article on Replicant wiki describing an actual backdoor in Samsung smartphones where the modem can read/write stuff where it wants in the smartphone's filesystem. https://redmine.replicant.us/project...GalaxyBackdoor
                      Yeah, I am familiar with this particular backdoor. I used Replicant myself with Galaxy S3 i9300 and tried to keep myself current with what they do. That's pretty much only solid proof about backdoors in the phones though. That's why I asked. I was hoping he had something more concrete besides conspiracy theories.

                      Comment

                      Working...
                      X