1. 4

    Jared McNeill has been been putting in tremendous effort in improving Allwinner SoCs, the whole ARM tree in general and even porting bwfm(4). I’m really happy about him and I hope we can work more together to improve the *BSD/ARM ecosystem!

    1. 12

      This is a work in progress project and there is still a lot do.

      If you want to get involved in OpenBSD wifi development, you could get started by helping us with this effort.

      Eventually, this driver could support wireless on the rpi3 and some macbooks.

      1. 10

        Thanks, stsp, for all your help getting me into the wireless world! And for reaping all that karma now. ;)

        This driver might be an easier way of getting into OpenBSD wifi development, since all the wifi madness is done in the firmware, so one doesn’t have to understand everything right away.

        1. 2

          Will it support BCM43225?

          1. 10

            BCM43225

            Nope, that’s a SoftMAC chip. The problem with these chips is that the Linux driver code is too insane and large by OpenBSD’s standards: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/net/wireless/broadcom/brcm80211/brcmsmac/phy/phy_n.c

            With FullMAC, this junk is hidden in the firmware so we don’t have to maintain it.

            1. 5

              30k LOC, no further questions :)

            2. 1

              Are these ones the ones on macbook/macbook airs?

          1. 6

            It‘s been pretty cool doing that interview with jcs, sitting somewhere in a hallway in the corner. He left plenty of space for me to talk into, or maybe I just like to talk a lot. ?

            1. 0

              Given that you’re the first person to have commented on the post, I’d say it’s the latter… Just teasin’ :^P

              1. 1

                Given that I’m the person that was interviewed… I’m quite squirrelly seeing this being posted here! ?

            1. 15

              Best birthday present ever.

              1. 4

                I had been using SixXS for quite a while, and it’s been a real pleasure. At some point I gained IPv6 support at my ISP and servers so I stopped using them, but that’s already been a few years. Thank you for your service.

                1. 6

                  ?

                  1. 4

                    Ouch. This is a disaster. At any time and without warning, your CPU may ‘lose’ functionality you were just using.

                    1. 2

                      Apparently Samsung fucked up when they paired a low-cost standard ARM with one of their own creations. All other ARMs seem to be 64b by default. Properly configured big.LITTLE should have all cores reports the same size.

                    1. 1

                      It was about time actually! Theo finally got some armv7 hardware as a proper build machine. So now he can retire armish. Next up zaurus?

                      1. 1

                        I kinda want to see the armv7 port’s OMAP support take it to the Nokia N-series tablets. Those are fun, but don’t have a supported OS!

                      1. 1

                        Right. That stuff already reached u-boot mainline. OpenBSD uses this feature on all supported ARMs to boot up like a x86 PC would. It kinds feels like a normal PC now with that stuff, it’s quite nice.

                        1. 7

                          Awesome! This seemed, at one point, to be one of those “it’ll be a cold day in hell before openBSD supports RaspberryPI”–what changed?

                          1. 9

                            @cr1901: WOW, that was unexpected. Why the change of heart?

                            @phessler: there is no more binary blob required in the kernel, and now ppl care about armv7.

                            https://twitter.com/phessler/status/762346582117326849

                            1. 6

                              I think the boot situation is a lot better now. OpenBSD uses UEFI, so the kernel on the ARMv7 is universal between each board supported on that port. It’s just a matter of loading the UEFI stub. I’m not sure if Broadcom’s firmware can do this directly (as Windows 10 runs on Pi, and it’s also ARM on UEFI) or if it’s using U-Boot (and presumably the firmware can handle it better) like the other boards.

                              Graphics will be fun though.

                              1. 4

                                I have two of these devices sitting unused. I had ambitions of doing things with them that never happened. I am sure that installing openBSD on one of them is another, but one can dream!

                              2. 4

                                I would say it’s because there was nobody who wanted to invest time in it. No developer really cared about armv7. The requests to support the Raspberry Pi were just blocked with the “too many blobs” argument. There actually are blobs, but they are not kernel drivers. The blobs are run on the GPU, parallel to the CPU. They are loaded and run even before the CPU is turned on. So yeah, they are blobs, but it’s just firmware. Firmware is ok. Some time ago the 3D graphics stack actually had closed-source linux kernel drivers. Still, they were not needed to boot the device. Now the whole graphics stack is BSD-licensed anyway, so it’s all good.

                              1. 14

                                I have the final exam of my degree tomorrow (AI & Game theory), then I’m organising moving to Munich which happens on Sunday for my internship at Google. I don’t have time to write more because I need to revise auction theory!

                                1. 2

                                  Oh! Welcome to Munich then!

                                1. 4

                                  This specific ransomware actually sounds like a fun task to write oneself. I still remember the iOS filesystem encryption layers of a few years ago, where you had different encryption keys for the two NAND flash layers, the HFS metadata and the per-file encryption. Similarly the OS X FDE is also a bunch of layers of encryption. I bet whoever wrote this had a bit of fun on that task. That said, I really dislike the programmer’s intention in writing that piece of software.