**** BEGIN LOGGING AT Sun Jul 06 02:59:59 2014 **** BEGIN LOGGING AT Sun Jul 06 04:03:50 2014 Jul 06 05:57:04 My Nokia N900 has been retired in favour of my Geeksphone Revolution. Jul 06 06:08:05 :( Jul 06 06:08:53 no keyboard Jul 06 06:08:55 no real linux Jul 06 06:09:30 The Revolution is not only more powerful than most, but also a first-of-its-kind. Geeksphone MultiOS technology allows you to choose your operating system. Jul 06 06:09:36 >first-of-its-kind Jul 06 06:09:37 u wot m8 Jul 06 06:22:34 booting another system that runs on top of the Android hardware adaptation doesn't count! Jul 06 07:05:14 Could someone give someone who's been away from the community for quite some time of maemo "state of the union"? Jul 06 07:24:04 kerio: hurrian: I used Debian Kit on the Android OS on my Geeksphone Revolution to install Debian GNU/Linux for x86-32 to an ext4 volume in a disk partition on a 32-GB microSDHC card. I can run the same Debian GNU/Linux installation on both my handheld computer and my desktop computer. You cannot do that with a Nokia N900 unless you use an ARM desktop computer. I want x86, not ARM. Jul 06 07:28:59 Yes, technically you can emulate an x86 computer on an ARM computer, such as the Nokia N900, but why bother with emulation when you can simply use an x86 mobile computer? Jul 06 07:33:07 My user experience with my highly-customised Maemo 5 installation on my Nokia N900 was very poor. Jul 06 07:34:16 Maemo 5 by default uses BusyBox crap instead of a proper GNU userland. Jul 06 08:24:06 nothing stops you from installing a real bash Jul 06 08:30:59 i really don't understand why the architecture's an issue Jul 06 08:31:17 :nod: I was about to say the same kerio Jul 06 08:31:31 the n900 has many problems, but the instruction set isn't one of them Jul 06 08:43:09 I don't get what's there to brag about running x86 on a phone unless you're running desktop Windows on it Jul 06 08:43:53 in this case, you're running Debian, which runs exactly* the same on about every architecture known to man Jul 06 08:44:05 [*] PowerPC Big-endian weirdness aside Jul 06 08:51:26 bencoh: I was replying to kerio’s reply of “no real linux”. Jul 06 08:55:50 there's quite a difference between a crappy busybox default and an android pile of crap Jul 06 08:56:45 and there "multios technology" is probably not better than jolla (if as good at all) Jul 06 08:57:11 (I guess you're tied to android/RIL stuff) Jul 06 08:57:32 hurrian: “phone” (sic) is a gross misnomer. Jul 06 08:57:33 I need an x86 desktop computer because of some of the software I need to use. If my desktop computer uses x86, then I want my embedded and mobile computers to use x86 too because I do not want to spend time on the cross-compilation chore when I can simply use the same architecture on both my host/development computer and my target computer. Jul 06 08:59:23 No, I do not want an ARM desktop computer. Jul 06 08:59:49 there's not much point in cross compiling when you use a binary distribution. Jul 06 09:02:21 other than that, I'm starting to resent the x86 monoculture, and would really like it if we had different chips adapted to their application, a side effect of which would promote and ensure interoperability. Jul 06 09:02:24 actually I wonder whether there thing is intimately tied to android or if one could have a "free" telephony stack (at least host-cpu-wise) Jul 06 09:02:56 hurrian: I too, but it feels like a dream Jul 06 09:03:12 bencoh: if you're using a Samsung device and run Cyanogenmod 10.1 and above on it, you're probably using a Free, Libre reverse-engineered Radio Interface Layer Jul 06 09:03:15 we had wintel, now we have androarm Jul 06 09:03:31 so yeah, host CPU wise, it's free. baseband could still be evil though. Jul 06 09:03:33 hurrian: hmm they still copy blobs afaik Jul 06 09:03:36 hurrian: I sometimes build software, including my own software, from source by myself. Jul 06 09:03:50 not sure about replicant though Jul 06 09:04:32 nah, building CM for my i9100 the only blobs needed now are camera, WiFi/BT (broadcom :/) and audio (super sekrit wolfson sauce) Jul 06 09:05:10 neat Jul 06 09:05:54 well, apart from the cpu/modem dma Jul 06 09:06:30 hmm Jul 06 09:06:31 brolin_empey: assuming you've got a similar phone that also dual-boots to Debian, but has an ARM core, there's nothing that stops you from running gcc natively on the phone Jul 06 09:06:36 hurrian: Modem XMM 6260 Linux kernel Secret protocol: Samsung-IPC, implemented in libsamsung-ipc and samsung-ril Jul 06 09:07:18 oh right, they got half of it reverse engineered Jul 06 09:07:41 they finished the reverse-engineering, and it works for most devices Jul 06 09:08:49 the ones where it's so far unstable are built with the proprietary RIL with a shim on top, but they're working on porting them all. Jul 06 09:09:09 where's all of this going on? Jul 06 09:10:14 all of what ? maemo ? Jul 06 09:10:39 bencoh: i9100 reverse engineering. sorry, should've made that clear. Jul 06 09:10:47 :) Jul 06 09:11:00 jaykru: check the replicant project and cyanogenmod gerrits. Jul 06 09:12:05 hurrian: ah, thanks! I don't own an i9100 but this sounds like a very exciting development :D Jul 06 09:12:45 a lot of the work done for samsung devices is portable to newer ones, don't just go rushing out and buy a i9100 :P Jul 06 09:12:58 it's getting really long in the tooth. Jul 06 09:13:21 lol I'd assume so at this point Jul 06 09:14:26 hurrian: hmm, wifi/bt are only missing firmwares Jul 06 09:14:30 seems quite nice Jul 06 09:14:45 pretty amazing actually Jul 06 09:15:02 hurrian: That may be true but my point was that I want x86, not ARM. Jul 06 09:17:21 brolin_empey: if that's what you need, then suit yourself, but I'll stick with ARM's high end for mobile devices. Jul 06 09:17:24 "multiOS technology" Jul 06 09:17:27 aka a bootloader Jul 06 09:17:28 Enjoy your SafariCom Yolo, brolin_empey Jul 06 09:17:54 did they really call it yolo Jul 06 09:17:59 hurrian: The Intel Atom x86 SoC is already adapted to the mobile application. Jul 06 09:18:02 yep... Jul 06 09:18:11 they did kerio -_- Jul 06 09:18:12 "introducing the yolo phone by intel" Jul 06 09:18:14 ROFL Jul 06 09:18:51 kerio: although I'm assuming it has more to do with an African word (it's from Kenya) than the acronym. Jul 06 09:19:00 still Jul 06 09:19:03 Z2560, TDP 3W Jul 06 09:20:04 that's a Tegra-sized TDP, and competes with the Snapdragon 600. Jul 06 09:20:09 brolin_empey: if you have any way to import from China, apparently Lenovo has a decent x86 android offering Jul 06 09:21:58 not exactly mobile-optimized, BUT, as we all know, Intel has fab technology about two nodes ahead of everyone. Jul 06 09:22:01 hurrian: If you want x86, a non-x86 SoC does not compete with an x86 SoC. Jul 06 09:22:20 well that just goes without saying Jul 06 09:22:39 brolin_empey: we were talking about mobile, unless you suddenly felt the need to strap a Core i7 Mobile in your phone... Jul 06 09:23:06 hurrian: how else will I use sony vegas on my phone? Jul 06 09:23:46 (interesting thought, it's definitely possible to supply the ~10-15W TDP a Core i7 mobile, low-voltage chip needs in a phone nowadays!) Jul 06 09:25:27 I think if Intel really wanted to make a play at phones, that's where they should focus. Jul 06 09:25:52 hurrian: I thought you were saying that an Intel Atom x86 SoC (“Z2560”) competes with an ARM SoC (“Snapdragon 600”). Jul 06 09:26:17 what did I say that contradicted that? Jul 06 09:26:51 I was just saying that if Intel wanted to turn heads in mobile, they should try and release a 10W Core i5/7 part. Jul 06 09:27:03 and get one of their ODMs to build a device based around that. Jul 06 09:27:41 maybe a tablet at first, because of the obvious power requirements. Jul 06 09:28:40 hurrian: “ brolin_empey: we were talking about mobile, unless you suddenly felt the need to strap a Core i7 Mobile in your phone...” Jul 06 09:28:40 I know we were talking about mobile technology. Why did you think I was not talking about mobile technology? Jul 06 09:29:09 At the moment x86 can't feasibly regarded as mobile. Jul 06 09:29:39 because you said that hey, we were talking about x86 right? seeing as there are no other x86 competitors for mobile (unless you count in VIA, and I'm pretty sure their x86 license expired), there was really nothing to compare it to Jul 06 09:31:00 jaykru: Based on my Real-World experience with the Geeksphone Revolution, your claim is false. Jul 06 09:32:24 I'll concede that, brolin_empey Jul 06 09:32:48 Still, I see little initiative on Intel's part to enter the mobile market on a large scale. Jul 06 09:34:23 jaykru: I see little reason to use (the) ARM when I can use an x86 SoC. Jul 06 09:39:05 To each their own. I probably wouldn't mind using x86 over ARM on mobile if there were a range of devices to choose from that offer it. Jul 06 09:40:05 Intel seems to usually make the documentation for their hardware publicly available. I do not know about Qualcomm but nVidia seems to expect an owner of their hardware to enter a non-disclosure agreement before providing documentation for their hardware. Jul 06 09:41:12 That is true...I don't believe Qualcomm has a very good track record in being open either, to tell you the truth given issues we've had with video binary blobs on Android in recent years. Jul 06 09:41:25 brolin_empey: you've got that switched around - it's Qualcomm who's anal about NDAs, Nvidia cleaned up their act after Tegra 2 thoroughly bombed and Torvalds gave them a good scolding over having over 9000 patches to the kernel. Jul 06 09:41:55 we've actually got Qualcomm to blame for not being able to download flashable images for Jolla, and the delayed source releases for Nexus 4 and 5 Jul 06 09:43:19 TI was the best with open source, even if they couldn't open the ImgTec SGX drivers (if freaking Intel couldn't with GMA 950, no one can) Jul 06 09:44:21 I miss TI :'( Jul 06 09:44:46 we all do, nobody else ever had something as magical as OMAPPedia :( Jul 06 09:45:50 An Intel Atom x86 SoC uses the PCI or PCI Express bus; AFAIK, an ARM SoC usually does not use (the) PCI. Jul 06 09:46:44 brolin_empey: all the ARM chips have PCIe buses, but MIPI is more commonly used for peripherals instead Jul 06 09:47:24 I'm unsure if I /want/ PCIe on my phone, porting stuff over from x86pc like that leads to Windows RT tablets Jul 06 09:47:45 (hint: they straight up ported the PC architecture, complete with quirks, to ARM. Thanks, Microsoft.) Jul 06 09:49:45 I'm pretty sure Windows RT was the worst thing I've ever experienced Jul 06 09:50:04 allowing non-Store applications would've saved it. Jul 06 09:50:05 It's everything that sucks about PCs, without decades of x86 compatibility Jul 06 09:50:13 No, they were allowed. Jul 06 09:50:19 You could install PuTTY Jul 06 09:50:33 If it was compiled for ARM it worked. Jul 06 09:50:56 no, you had to install a hack to enable unsigned applications. Jul 06 09:51:27 plus, M$ crippled visual studio so you couldn't easily change architectures and target ARM. Jul 06 09:51:31 hurrian: If you dislike monocultures, then you should dislike Apple because Apple maximises the monoculture by having a software monoculture on a hardware monoculture. Jul 06 09:51:32 Huh. Maybe PuTTY had a signature. Jul 06 09:53:18 brolin_empey: pretty sure they didn't use x86 on the iPad Jul 06 09:53:57 hurrian: They did not. I did not say that they did. Jul 06 09:55:48 brolin_empey: their software stack is built in a way that avoids /having/ to become a monoculture. Jul 06 09:57:29 if Apple were to release an ARM Mac tomorrow, this would theoretically be possible: Jul 06 09:57:33 http://goput.it/vaev.png Jul 06 09:58:58 > 4chan Jul 06 09:59:04 > probably /g/ Jul 06 09:59:05 http://www.quickmeme.com/img/34/343de795dddb951ab89d8e9849457e6b0fa990b932c164f64dd4cec827d87f34.jpg Jul 06 10:00:11 uh oh, I revealed that I didn't use Our Lord and Savior's Truly Free OS :P Jul 06 10:00:24 don't worry, I've got a VM on my second monitor. Jul 06 10:01:08 Oh NOES Jul 06 10:01:28 b-but my thinkpad runs coreboot and gahnoo! Jul 06 10:01:49 I'm pretty sure we share the exact same computing setup Jul 06 10:02:04 ThinkPad running my gan00 loonix Jul 06 10:02:19 MacBook Air running hellfire spawn fischer-price OS Jul 06 10:02:20 hurrian: That screenshot contains oversimplification. x86-64 is not the 64-bit Intel architecture. x86-64 was created by AMD, not Intel. IA-64 AKA Itanium is the 64-bit Intel Architecture. Jul 06 10:03:38 brolin_empey, that was a screenshot off Xcode, where I added "armv7s", SDK OSX 10.8 as a build target. Jul 06 10:04:21 assuming that you're not running a hackintosh, 64-bit Intel means x86-64 plus extensions supported on Intel processors. Jul 06 10:05:36 to date, there have been no AMD Macs, but if Apple were to release a Macbook Air Light that ran on AMD APUs and was $10 cheaper, you bet they'd drop "Intel" from that asap. Jul 06 10:07:07 jaykru: close enough - liberated thinkpad and a hackintosh (mouse button 4/5 to trigger Mission Control is boss) Jul 06 10:07:17 is coreboot really that useful? Jul 06 10:07:47 hurrian: Intel has multiple 64-bit architectures: at least EM64T/Intel 64, which is the Intel version of x86-64, and IA-64/Itanium. Jul 06 10:08:09 kerio: OEM BIOS is awful, takes 5 seconds to POST. Coreboot gets me to /desktop/ in 5 seconds. Jul 06 10:08:18 i see Jul 06 10:08:42 i should time my boot Jul 06 10:09:10 brolin_empey: that's stepping into pedant territory, i386 and x86_64 are clearly marked to the right of the build configuration. Jul 06 10:09:10 also it's super liberated Jul 06 10:09:22 jaykru: I assume you've switched to Atheros wifi? Jul 06 10:09:35 nobody fucking uses itanium Jul 06 10:09:58 :] Jul 06 10:10:12 hurrian: Intel 6300 in my X220 at the moment Jul 06 10:10:22 kerio: sssh, you'll wake the HP-UX folks! Jul 06 10:10:39 haha Jul 06 10:10:43 It supports packet injection and promiscuous mode which is basically enough for me. Jul 06 10:12:37 hrmpf, 11 seconds to desktop :c Jul 06 10:12:53 hurrian: Let us step further into pedant territory: i686 != i386. Jul 06 10:13:40 s/further/farther/ ? Jul 06 10:13:48 ah. i'm pondering buying an ath9k card off eBay for maximum freedom, or waiting for the b43 team to reverse engineer the firmware for those sweet new 802.11ac cards. Jul 06 10:14:18 brolin_empey: unless you're using an Intel Galileo, you wouldn't target i386 anyway Jul 06 10:14:51 IIRC even the Linux kernel doesn't support the 386 anymore Jul 06 10:15:50 hurrian: AFAIK, the Linux kernel still does. FreeBSD dropped support for 80386 years ago, though. Jul 06 10:16:29 brolin_empey: linus said byebye to 386 as well Jul 06 10:16:31 Meaning that FreeBSD requires at least 80486. Jul 06 10:16:52 It's on Phoronix: http://www.phoronix.com/scan.php?page=news_item&px=MTI0OTg Jul 06 10:16:54 not that it really matters though Jul 06 10:20:33 hurrian: how do i make os x boot faster Jul 06 10:21:10 hurrian: I use GNU/Windows NT for x86-32 on my desktop client computers. Jul 06 10:22:04 kerio: SSD, enabling kernelcache (assuming it's a hackintosh) Jul 06 10:22:13 no, no Jul 06 10:22:25 kerio: Boot Mac OS X from an SSD instead of an HDD? Jul 06 10:22:34 it's already a ssd Jul 06 10:23:37 brolin_empey: unless you're using GNU libc as the C runtime, cygwin doesn't count :P Jul 06 10:24:12 kerio: check LaunchDaemons and StartupItems, if those are clean there's nothing much you could do. Jul 06 10:24:30 my hackintosh boots off a Fusion Drive, and gets to the desktop in 7 spins. Jul 06 10:24:38 how much is a spin Jul 06 10:24:50 the spinner under the apple boot logo, that is Jul 06 10:25:22 powerbutton-to-desktop, my laptop takes 11 seconds Jul 06 10:25:35 :3 Jul 06 10:26:23 kerio: The amount of spin depends on the strength of the Reality Distortion Field affecting the Marketing department at the time. :-P Jul 06 10:31:39 haha Jul 06 10:31:50 well 11s isnt that much anyawy Jul 06 10:32:01 beware, that kind of talk leads to systemd :> Jul 06 10:32:55 bencoh: in the ultimate sick irony, systemd is actually modeled after OS X's launchd :3 Jul 06 10:33:42 meh :) Jul 06 10:33:59 at least launchd doesnt force anything on you ... and isnt dep-bloated Jul 06 10:38:51 brb, taking an hour off for my nightly walk. Jul 06 10:39:13 Does someone want to “buy” (= license) Kermit 95 for me? Jul 06 10:43:30 I was reading [1] before I got pulled (back) into the IRC. Jul 06 10:43:30 [1] http://simpleconstraints.blogspot.ca/2009/04/what-is-strategy.html Jul 06 10:57:10 hurrian: So 2012 was apocalyptic for the Linux kernel on the 80386? :-D Jul 06 10:57:52 yeah, i actually quite like launchd Jul 06 10:59:23 * brolin_empey is only approximately 1.5 years late to learn that support for the 80386 has been dropped from the Linux kernel. Jul 06 11:02:32 kerio: Did you mean that using Itanium requires celibacy? Jul 06 11:03:01 “ nobody fucking uses itanium” Jul 06 11:03:06 :D Jul 06 11:06:18 My (interpersonal) sex life is not (literally) equal to zero: it is only asymptotically equal to zero. Jul 06 11:11:33 I actually printed to paper instead of only printing to (a) file at ${HOME} earlier this week but I never used the paper I printed because it is a ticket for an event I missed due to a literal road block. Jul 06 11:12:20 s/file/computer file/ Jul 06 11:12:22 brolin_empey meant: I actually printed to paper instead of only printing to (a) computer file at ${HOME} earlier this week but I never used the paper I printed because it is a ticket for an event I missed due to a literal road block. Jul 06 11:13:20 Both a literal and physical road block. Jul 06 11:14:33 infobot: seen infobot Jul 06 11:14:34 infobot is currently on #asterisk-dev (14h 2m) #harmattan (14h 2m) #utah (14h 2m) #magnia (14h 2m) #openwest (14h 2m) #pillbox (14h 2m) #opensuse-us-ut (14h 2m) #catroid (14h 2m) #wowprogramming (14h 2m) #va (14h 2m) #opie (14h 2m) #openzaurus (14h 2m) #mseide (14h 2m) #arm-netbook (14h 2m) #enlightenutah (14h 2m) #botpark (14h 2m) #orkut (14h 2m) #ldstech (14h 2m) #buglabs (14h ... Jul 06 11:15:16 infobot: seen root Jul 06 11:15:16 root <~root@83.149.46.153> was last seen on IRC in channel #kde, 144d 1h 29m 23s ago, saying: 'go to the scotland'. Jul 06 11:15:40 infobot: seen god Jul 06 11:15:40 god <~pawky@193.15.242.97> was last seen on IRC in channel #harmattan, 738d 7m 47s ago, saying: 'Am I still the only one who has noticed the Wifi HotSpot softy doesn't actually seem to use the encryption you set it to use?'. Jul 06 11:16:45 god uses Harmattan? Jul 06 11:48:36 infobot: little spy :] Jul 06 11:57:03 Good fucking job, Qualcomm http://www.androidpolice.com/2014/07/05/qualcomm-backtracks-after-issuing-dmca-takedown-notices-for-116-github-repos-including-some-belonging-to-cyanogen-sony-and-qualcomm/ Jul 06 12:14:07 I hate the DMCA Jul 06 12:14:11 Ever so much Jul 06 12:29:41 hi! I have observed a very strange thing when tethering my n900 Jul 06 12:30:19 when I connect to gprs, I can't ping n900 over usb Jul 06 12:31:06 I can still ssh, I can ping the computer from phone, but pings and DNS requests don't go through Jul 06 12:32:46 I would be glad if someone else had that happen Jul 06 12:37:20 hurrian: good way to "improve" your brand image in the technical community :) Jul 06 12:37:56 M4rtinK: while I'm glad that they retracted the DMCA, they should've kicked that firm to the curb for doing that. Jul 06 12:38:01 well it was not directly Qualcom but some rather trigger happy loosers they hired Jul 06 12:38:18 but yeah, it is their fault for paying them Jul 06 12:38:38 and not checking what those fools are actually doing Jul 06 12:39:03 (also, Jolla system images can't be redistributed because of QCOM binaries. jeebus christ, CAF source is available EVERYWHERE) Jul 06 12:39:16 CAF ? Jul 06 12:39:40 Code Aurora Forum, their version of OMAPPedia Jul 06 12:40:44 I have heard about it - what is it actually ? Jul 06 12:41:10 looks like a semi legal binary blob dumping ground from a quick look Jul 06 12:41:38 well, it's technically Qualcomm's blob dumping ground, for both the Linux kernel and Android Jul 06 12:41:42 rhn: that is a route issue Jul 06 12:42:15 rhn: make sure you have a route for the subnet you use for the IPs on the usb connection Jul 06 12:42:16 they do send out patches that legitimately improve Linux performance on ARM, so it's not all bad Jul 06 12:43:03 although, it boggles the mind why QCOM hasn't resolved glaring issues with their legal department yet - JBQ quit AOSP because they weren't able to clear something as simple as posting rootfs ZIPs for the Nexus 4 and 5 Jul 06 12:43:51 I really wonder what are they trying to achieve with non-redistributable binaries Jul 06 12:44:12 the above is the exact same reason everybody with a Jolla is deathly scared to soft-brick them. Jul 06 12:44:26 anyone who wants to reverse engineer them (or find if they are infringing on GPL or something similar) Jul 06 12:44:33 _will_ get them Jul 06 12:44:50 yeah, I know the no-image story Jul 06 12:45:46 it is a big shame - considering all the awesome low level development than happened on the N900 Jul 06 12:45:48 and N9 Jul 06 12:45:51 I'm already missing TI :( Jul 06 12:46:07 there aren't working firmware blobs for whatever chip is in jolla phones somewhere in the linux kernel tree? Jul 06 12:47:37 honestly: while ath9k is incredibly open for a WiFi chipset, Qualcomm Prima stuff is top sekrit, apparently. Jul 06 12:47:48 ah. Jul 06 12:48:25 I'm following ath10k development, watching from the distance, and it's really weird. Jul 06 12:49:17 ath10k needs proprietary FW blobs, which is why I'm not buying them until an open FW is released. Jul 06 12:49:21 yeah. Jul 06 12:49:33 honestly: I think the route is ok - I have the usb entry. New observation: I can't ping localhost... Jul 06 12:49:47 what do you mean by "open firmware"? Jul 06 12:50:13 the b43 team seems a little further along porting the open firmware for 802.11ac cards, luckily Jul 06 12:51:19 * dos1 wonders what's so different about closed wifi firmware and closed HDD firmware Jul 06 12:51:43 HDD firmware implements a standardised API Jul 06 12:51:58 wifi usually... doesn'ty Jul 06 12:52:01 -y Jul 06 12:52:15 but you usually have free driver provided Jul 06 12:52:35 you have a strange conception of "usually" Jul 06 12:52:37 and when you don't, you simply ignore such chipset Jul 06 12:53:20 dos1: not much difference from the security standpoint :) Jul 06 12:53:23 http://cyberarms.wordpress.com/2013/08/04/hard-drive-hacking-hardware-backdoor-even-if-drive-wiped/ Jul 06 12:53:31 a few months ago, I was supposed to do research evaluating the potential of 802.11ac cards for use in a research mesh network implementation Jul 06 12:53:32 http://hackaday.com/2013/08/12/hacking-transcend-wifi-sd-cards/ Jul 06 12:53:40 M4rtinK: yup, that's my point Jul 06 12:54:03 my advisor saw that there was a driver for ath10k and got a few QCA98whatever cards Jul 06 12:54:04 I can't see why one would require WiFi to use "open firmware", and don't require HDDs, SSDs and SD cards to do so as well Jul 06 12:54:37 but the whole thing was a nonstarter because nothing works properly Jul 06 12:54:49 especially not the low-level packet injection needed for the project Jul 06 12:54:56 maybe except things like implementing missing features and/or bugs by yourself - that might be a valid point Jul 06 12:55:04 there are no such problems with hard disks Jul 06 12:55:36 honestly: but is that firmware or driver issue? Jul 06 12:55:38 you plug them into the SATA bus and they follow the protocol Jul 06 12:55:48 dos1: I imagine security. devices on PCI usually have much wider access to system memory than USB or others Jul 06 12:56:14 dos1: have you ever checked how many hard drive drivers there are in the kernel? Jul 06 12:56:25 and then checked how many wireless drivers are in the kernel? Jul 06 12:56:32 honestly: that's not the point Jul 06 12:56:37 honestly: I don't remember there being any cards capable of injection yet, though Jul 06 12:56:40 that is exactly the point Jul 06 12:57:02 "open firmware" doesn't change anything in this regard Jul 06 12:57:16 wireless cards aren't standardised to the point that firmware simply implements a fixed driver interface Jul 06 12:57:19 hard drives are Jul 06 12:57:30 unless you reimplement it all by yourself following some standard you choose Jul 06 12:57:44 for each of card you want to use separately Jul 06 12:58:09 so, while it's a valid issue, it's unrelated to firmware openness and freedom Jul 06 12:58:16 I don't know what point you're trying to make anymore Jul 06 12:58:32 [14:49] ath10k needs proprietary FW blobs, which is why I'm not buying them until an open FW is released. Jul 06 12:58:35 [14:51] * dos1 wonders what's so different about closed wifi firmware and closed HDD firmware Jul 06 12:58:37 http://feedback.zlauncher.com/knowledgebase/articles/377088-what-sort-of-data-are-you-collecting Jul 06 12:58:52 but my point is that the lack of open specifications and open firmware for wireless cards is a practical issue Jul 06 12:59:10 why buying HDD, SDD or SD with closed firmware is ok, and buying WiFi with closed firmware is no-go? Jul 06 12:59:15 SSD* Jul 06 12:59:27 I'm not going to say it a third time Jul 06 12:59:43 you can either choose to read and understand what I'm sayting or not Jul 06 12:59:46 not my problem Jul 06 13:00:01 honestly: I'm *not* refering to *your problem*, I'm refering to *hurrian's one* Jul 06 13:01:26 dos1: it is to a point acceptable - sure they could hide evil stuff on reserved areas, but they don't have DMA access to CPU, nor can their firmware be easily updated ;) Jul 06 13:02:18 hurrian: a) their firmware can be updated b) you don't have to give WiFi module DMA access to CPU Jul 06 13:02:48 so without DMA, it would be all ok? ;) Jul 06 13:02:59 > you don't have to give WiFi module DMA access to CPU Jul 06 13:03:01 what Jul 06 13:03:06 do you know how PCI works? Jul 06 13:03:14 why are you insisting on PCI modules? Jul 06 13:03:21 in mobile phones there's hardly any PCI Jul 06 13:03:22 for instance Jul 06 13:03:39 but the same problem with "muh freedoms" applies there as well Jul 06 13:04:02 easily updated being the key word here, you'd have to have your box already rooted or physically accessible, plus you'd notice a drive going off to rewrite firmware, needing a reboot ;) Jul 06 13:05:54 hurrian: isn't it same with wifi? and how can you be sure that any rogue firmware update requires device to go off? Jul 06 13:06:46 wifi has an RF antenna listening to the outside world. Jul 06 13:06:50 maybe only some EEPROM is flashed, while old firmware stays 100% in RAM until next reboot? Jul 06 13:06:58 as for phones, I think it'd be reasonably sane to assume DMA between host CPU and peripherals is done (.dts trees could be a good hint) Jul 06 13:08:33 HDDs use private ATA or SCSI commands to flash, these most certainly require exclusive access to the drive else you end up with a bad flash and a bricked drive Jul 06 13:09:11 hurrian: as is between the CPU and modem, but there are projects like Neo900 where proper separation is a key design factor Jul 06 13:09:49 yet still people often insist on free wifi firmware, and I don't really understand why Jul 06 13:10:24 unfortunately, the Neo900 is the exception rather than the rule - it's just soooooooo much easier for OEMs to poke at random memory regions. Jul 06 13:11:12 tl;dr dumb block devices, with no means of communication with the outside world except for SATA/M.2/MMC/etc don't represent a credible threat - with no physical access you'd need to be 0wned, and if you had physical access (evil maid, probably) you'd be 0wned either way Jul 06 13:11:32 well, I *can* understand why free firmwares are good in general. Just can't understand why allowing/forbidding to update it changes if people accept it or not Jul 06 13:11:57 if it's closed, you most likely already can't trust it Jul 06 13:12:02 as for free as in freedom, all firmwares should be free of course Jul 06 13:12:15 well, it boils down to trust. Jul 06 13:12:18 so if manufacturer says "you can't update it", why would you trust him? Jul 06 13:12:20 and all hardware should be free too Jul 06 13:13:37 if you had a ROM that you trust, and could verify, you can use that as the basis to trust software that it executes. Jul 06 13:13:49 FSF treats any non-updatable firmware as a "circuit" and is fine with it Jul 06 13:14:16 it's a bit ironic how that kind of security model sounds like trusted computing, but if the end-user gets to make that decision, it's alright. Jul 06 13:14:24 and I can't see how it's any different from updatable firmware - without trusting the manufacturer, you have no idea if the ROM you verified is really the firmware that gets executed Jul 06 13:15:04 dos1: the fsf is fine with proprietary circuits? in what context? Jul 06 13:15:24 honestly: for instance for Respects Your Freedom certification Jul 06 13:15:39 right Jul 06 13:15:47 the fsf is the free *software* foundation Jul 06 13:15:53 they care about software Jul 06 13:16:00 honestly: IIRC Stallman himself said that he doesn't need to know what his microwave runs, as long as there isn't an "update software" button on it. Jul 06 13:16:12 yet it's running a certification program for hardware Jul 06 13:16:15 for all intents and purposes, it is an appliance or something Jul 06 13:16:51 wait, wasn't RYF about "hardware that can run 100% FLOSS on all application processors"? Jul 06 13:17:14 richard stallman isn't the fsf Jul 06 13:17:39 hurrian: yep, but a device with WiFi chipset where firmware needs to be uploaded on each boot counts for them as "application processor" as well Jul 06 13:18:00 err, WiFi chipset counts as one Jul 06 13:18:08 yep, which is why iirc the WiFi chipset is replaced with an ath9k one Jul 06 13:18:17 *on the RYF certified X60 Jul 06 13:18:40 and when hardware is sealed inside the chipset, it's all fine Jul 06 13:19:17 and in my opinion, it has no implication in my privacy and it's worse for my freedom Jul 06 13:19:42 as with such device I can't even RE the firmware and replace it with my open one... Jul 06 13:20:02 it's a black box I can't even touch Jul 06 13:20:57 closed firmware is a black box as well, but at least I can work on replacement Jul 06 13:21:14 for microwave I might be fine with it being a black box appliance Jul 06 13:22:34 that's actually a good point, what if tomorrow broadcom decides to seal the firmware inside the chip, and release a driver that just interfaces with it? Jul 06 13:22:36 for WiFi - not really, especially if there are some bugs or missing features, like honestly said before Jul 06 13:23:12 hurrian: then FSF would be happy with it, but for me it would be step backwards Jul 06 13:23:23 hurrian: they won't, because they couldn't come up with a working firmware on the first attempt if they tried >_> Jul 06 13:24:06 yeah, that's one workaround that'd get them RYF, and it certainly makes it easier for driver developers, but that'd be a shockingly opaque black box. Jul 06 13:24:48 hurrian: doesn't the Raspberry GPU work like this already ? Jul 06 13:25:02 and it's funny, cause what I consider a step forward the freedom is at the same time condemned by FSF Jul 06 13:25:21 it does seem that the FSF stopped at software though, as it could be construed that blown PROM fuses could be construed as part of the "hardware", and demanding that all chip makes release VHDL to their cores would be insane. Jul 06 13:25:39 they accept either "shockingly opaque black box", or "everything free" Jul 06 13:26:10 hurrian: I don't think that's insane. Jul 06 13:26:17 so some "shockingly opaque" hardware can get certification, while "as free as feasible" devices like Neo900 can't Jul 06 13:26:20 it's not going to be popular any time soon though. Jul 06 13:26:29 which kinda breaks the whole idea about certification for me Jul 06 13:27:06 honestly: well, considering you have entire massive companies based upon developing hardware designs (fabless semis), for them that'd be financial suicide. Jul 06 13:27:25 dos1: so the only reason we've been having this conversation is so you can explain that you don't think the FSF's methodology is entirely self-consistent? Jul 06 13:27:36 maybe in a few years, 3D printing would be HP/Canon's domain and the hip new thing would be home photolithography of chips :D Jul 06 13:27:39 hurrian I don't think you quite understand. If I may, consider the following: 04m07u08h03 12f02r06e00e07d00o03m12s Jul 06 13:27:45 hurrian: so what? Jul 06 13:28:30 jaykru: I'd just like to interject for a moment... yeah, I think it'd be bad for the logs if I pasted it here Jul 06 13:28:58 hurrian: maybe we have more important things to worry about these days, like that consumer hardware these days has active backdoors in it when it comes out of the fab Jul 06 13:29:01 honestly: I think that open hardware could be the next frontier, seeing as open software is already so prolific Jul 06 13:29:09 hurrian: higher order kek indeed, old chum. Jul 06 13:29:20 hurrian: so maybe some business models stop being viable if we try to fix that, well, humanity will survive that Jul 06 13:30:20 hurrian: I don't think we'll be able to get by in the long run *without* pervasive free hardware Jul 06 13:30:22 honestly: nah, I just wanted to see the reasoning behing using lots of devices with closed firmware, but giving wifi special requirements on that Jul 06 13:30:35 honestly: I think open hardware will compete with the bigger boys just like OSS competes with proprietary nowadays. Jul 06 13:30:54 if someone insists of everything being 100% free hardware, then it's at least consistent :P Jul 06 13:31:07 just take the currently available RepRaps Jul 06 13:31:25 the barrier to entry TODAY is staggeringly high - either you buy a fab, or a FPGA kit, but who knows how it'll pan out in the future? Jul 06 13:31:27 they are open source all the way to hardware & firmware Jul 06 13:31:50 hurrian: huh? fpga devkits are cheap Jul 06 13:31:53 and the rest of the workflow from model creation down is also OSS Jul 06 13:31:58 so it is certainly doable Jul 06 13:32:22 honestly: it's like developing a C project in the open in the early 90s - Jul 06 13:32:30 some time ago good compilers weren't easy to come by, nowadays GCC eats everybody's lunch. Jul 06 13:32:43 and if you have a market that wants what you're developing, they will pay for the absence of economy of scale Jul 06 13:33:10 maybe they will even be happy they get a free FPGA with whatever appliance they wanted. Jul 06 13:34:09 the kids playing with a custom CPU architecture for blinking a LED at impossible fast speeds today might pose a threat to x86-64 tomo... maybe a bit further from that, but you get the point Jul 06 13:35:01 I know someone who is making specialised hardware for processing / distributing / transmitting RF signals / video streams / stuff like that and it's all fpga-based Jul 06 13:35:20 of course it's a boutique thing, but it's economically viable and the barrier to entry isn't dramatic Jul 06 13:36:55 if you can get the whole idea of how HDLs work into your head Jul 06 13:37:04 which 90% of people exposed to it don't Jul 06 13:37:26 but that's the same as for software programming (: Jul 06 13:39:35 milkymist? Jul 06 13:42:05 anyways, i'm heading off for the night, been nice talking to y'all! Jul 06 16:56:42 now my n900 stops charging without me seeing a disconnect or an error in the usb host computer Jul 06 16:56:45 's log :/ Jul 06 16:56:54 sigh. Jul 06 19:09:04 (( as for phones, I think it'd be reasonably sane to assume DMA between host CPU and peripherals)) NOPE! not at all Jul 06 19:15:52 (([2014-07-06 Sun 15:26:29] which kinda breaks the whole idea about certification for me)) right on, right on! Jul 06 19:21:03 (( now my n900 stops charging)) a quite fuzzy claim. What exactly is it that "stopped"? The blinking LED, the icon in systray, the 'charging' status of the charger chip, the current flowing into battery cell? Jul 06 19:21:36 LED and icon Jul 06 19:21:46 that doesn't mean anything yet Jul 06 19:21:49 I can't diagnose anything else Jul 06 19:21:58 also, the phone isn't warming up Jul 06 19:22:03 oh? you can't? why? Jul 06 19:22:20 ~bq27k Jul 06 19:22:30 ~bq27k-detail Jul 06 19:22:31 methinks bq27k-detail is http://maemo.cloud-7.de/maemo5/usr/local/sbin/bq27k-detail2 Jul 06 19:23:04 cat /sys/*/*/*/*/*/charger Jul 06 19:23:14