[02:52] more idiots tonight [02:53] ignore mode can get engaged mighty quick . [02:54] :D [03:11] time to leave the trolls to it i think, later folks o/ [03:13] nighty nite daftykins . [05:54] good morning to all [05:57] morning :) [05:58] hey TJ- [07:02] i hope more barebone companys will rise up [07:02] so winblows wont be attached to hardware [07:06] those clevo machines are nice, msi also [07:06] People for some reason keep letting MS off the hook, but they've never stopped doing what they started in the 1990s: Embrace, Extend, Extinguish" [07:08] yeah the sheep are so easy [07:08] i wonder why companys dont bet on more horses then MS? [07:08] its a good thing your hardware is wanted for multiple Os right? [07:09] kickbacks [07:10] Both MS and Intel have a long history of giving 'marketing rebates' to manufacturers for only shipping their stuff. Both Ms and Intel have been found guilty of it and made to pay large fines [07:10] I find it amazing why so many people still use Windows, on this basis alone: [07:10] Here, why don't you use this whizz-bang amazing software of mine? [07:11] lol [07:11] I don't think so - every month for the last 15 years you've issued security updates to fix vulnerabilities and you've still not fixed it, and meanwhile I hear millions of PCs in the world get infected with malware and become part of bot-nets as a result [07:11] i think users are trapped in 'a safe feeling' of the known [07:12] No one in their right mind would still drive an automobile with such a history [07:12] why goto another Os when you feel safe inside the windows [07:12] So, I think they get what they deserve [07:12] its like tradition right [07:12] a fake safe feeling of the known [07:13] feel safe... no... too lazy to try something different: yes (comfort zone trumps security) [07:14] but then you need to know someone who can support you, most users only find the start button [07:14] can't figure out my partitioning problem :( [07:14] Most modern Linux don't need support, no more than a similar Windows system [07:14] Ben64: whats up mate [07:14] Ben64: we're here to fix it :) [07:14] I have an external drive, 3TB Seagate. The SATA->USB bit on the bottom broke and can no longer access the drive with USB. I plugged it into SATA and was able to access it, but the partitions are screwed up. Testdisk finds an "Intel" MBR, as opposed to GPT, and so the partition it can recover is limited to 2.2TB of the 3TB. How can I fix this? [07:15] Ben64: ooo ooo ooo I know! I know! [07:15] :D [07:15] Ben64: can you get further with photorec? [07:15] well the data all seems to be intact [07:15] even finds the partition and I can browse it and everything [07:15] Ben64: the USB bridge was doing logical/physical sector size translation, direct SATA connection isn't [07:15] TJ-: boo, i was hoping not that [07:15] so the kernel is now seeing different sector sizes, so the relative LBA addresses are now wrong [07:16] You're the 2nd person with this issue this week [07:16] seagate? [07:16] makes no difference, the original problem was caused by the USB bridge translation [07:16] hmm.... so how to fix [07:16] instead of passing through what the drive actually reports, it 'munged' it [07:17] Ben64: can you show us "parted /dev/sdX unit s print" ? [07:18] http://sprunge.us/jiVP [07:18] you can see the 2.2TB it found [07:19] Ben64: how many partitions are on it? [07:19] one [07:21] Ben64: OK... "pastebinit <( dd if=/dev/sda count=16 bs=512 | hexdump -C )" [07:21] Ben64: this is t identify where the GPT starts [07:21] bah, living room computer turned off [07:21] hardly use it, so i have the power saving set to maximum [07:22] http://paste.ubuntu.com/13489742/ [07:23] Cripes! [07:23] ? [07:23] Ben64: I expected to see the GPT header at either offset 512, or 4096, since the logical/physical sizes are those [07:23] GPT should start in LBA 1 [07:23] We'd best look further! [07:23] Ben64: "pastebinit <( dd if=/dev/sda count=32 bs=512 | hexdump -C )" [07:24] i'm not sure gpt exists at all on this [07:24] http://paste.ubuntu.com/13489753/ [07:24] Ben64: No, not GPT there [07:25] why is it the same data? [07:25] Ben64: "losetup /dev/loop1 --offset $( echo 2048*4096 | bc ) /dev/sda" then "blkid /dev/loop1" [07:26] no output [07:27] Ben64: OK, "losetup -d /dev/loop1 [07:28] Ben64: "losetup /dev/loop1 --offset $( echo 2048*512 | bc ) /dev/sda" then "blkid /dev/loop1" [07:28] (we're now trying 512-byte sector sizes) [07:28] /dev/loop1: LABEL="nessie" UUID="4d2222f6-3b59-4177-a586-2911d65a6ec1" TYPE="ext4" [07:28] That what you were looking for? [07:28] yep [07:29] OK, so what we did there was make the starting offset to partition 1 (which we know from the MBR starts at LBA 2048) different by multiplying the starting LBA by 4096 and 512, to find which was the correct one to use [07:30] so the USb bridge was using 512<>4096 translation whereas directly on SATA there is no translation [07:30] thats annoying [07:31] One of the 2 reasons I don't like USB bridges - the other being I've had several just fail (burn themselves out literally) when used with sustained data transfers [07:31] yeah i think thats what happened here [07:31] I prefer to use eSATA, on a laptop via an ExpressCard<>eSATA adapter [07:31] which is disappointing for a usb3.0 drive [07:31] think it could handle it [07:32] I have external cabinets that can do both eSATA and USB, and I use the eSATA always [07:32] the protocol can, but the chips don't have the cooling needed for sustained high-speed transfer [07:32] i guess my next drives are internal, and I'll use some old box as NAS [07:33] so if i get seagate to send me a new usb part, and i run testdisk again, it'll find all 3TB? [07:33] like I said - eSATA is the way, that also supports external port multipliers. Just ensure the SATA chipset in an ExpressCard is a recent one. older varieties can have disk capacity limitations [07:34] You've got it all now :) [07:34] honestly, the drive kind of sucks [07:34] the sata ports are too recessed, had to trim a cable to get it to fit [07:34] don't have a esata cable i'm willing to sacrifice for this [07:35] TJ-: is there pcie card to esata to use it internal? [07:36] loop1 7:1 0 2.7T 0 loop [07:36] :o it is all there [07:36] /dev/loop1 2.7T 2.6T 17G 100% /mnt [07:36] :D [07:37] wb seagate :p [07:37] If you notice, the ending sector number for partition 1 in the MBR 4294969342 isn't the same as the end of the drive, so that PT is wrong [07:38] Ben64: which suggests that USB bridge was doing something more than simple LBa translation, it may be actively using its own partitioning system somewhere else on the device [07:39] Ben64: might be an idea to create another loop device pointing to the sector following partition 1 and see what is there [07:39] i'll see if seagate wants to send me new bits [07:40] don't have a free sata on this computer to use [07:40] Ben64: "losetup /dev/loop2 -o $( echo 512*4294969343 | bc ) /dev/sda" [07:40] oh, i think thats because of testdisk [07:41] Oh! you allowed testdisk to write to the disk? [07:41] it originally had one partiton, around 340GB, unknown filesystem [07:41] yeah [07:41] it showed the partition, how many sectors, etc [07:41] Did you take a back-up of the original MBR before doing that? [07:41] i did the math and it was the full 3TB, but I forgot about the MBR [07:41] so it wrote 2.2TB :| [07:42] uh... nope :S [07:42] darn! first rule of recovery, *always* take a back-up of anything you change on a live system. The MBR is only 512 bytes :) [07:43] yeah, i messed up [07:43] i was just thinking, well it won't delete anything [07:43] you should be able to re-write that MBR though. delete P1, then create a new one, starting at 2048 and ending at the default (end of disk), set it's type and that should be it [07:43] i don't think that'd work [07:44] it'd end at 5860533167 then [07:44] i bet the 340GB partition was the whole thing [07:44] and with whatever seagate's sata-usb thing does it works out [07:44] but we know what you have now, and it's there on /dev/loop1, so you can write a partition entry that matches what is there [07:45] we know it starts at LBA (sector) 2048, and it uses the entire disk, so there's no risk of losing some later partition [07:47] well... usb thing is alive [07:52] [423695.404476] sde: p1 size 34359738360 extends beyond EOD, truncated [07:55] the partitioner should set it to 5860533167 [07:56] doesn't seem like its letting me write to the disk [07:56] what error? [07:56] are you talking about the warning that the kernel's partition table couldn't be updated? [07:56] Partition: Write Error (from testdisk) [07:57] cant you scan the whole hd with photorec and recover the data? [07:57] Ben64: don't use testdisk. [07:57] just use fdisk [07:57] lotuspsychje: the data doesn't need recovering, we have it fine [07:57] lotuspsychje: yes, but the filesystem and everything is fine [07:57] ah kk [07:57] good job [07:57] don't really use fdisk a lot [07:58] TJ-: delete partition, create partition? [07:58] Ben64: run 'fdisk /dev/sda' ... delete any existing partitions. create a (n)ew, it will likely suggest the default starting sector as 2048, if not, type it, then for the ending sector just press enter to accept the default (will be end of disk), set the type 8e, and then check the table with (p)rint. Then if it looks good, (w)rite it [07:59] start is at 256 now [07:59] Ben64: 256 in what units? [07:59] uh.... sectors? [08:00] delete anything thats there, create the partition you need using the instructions above ^ [08:01] ooh maybe it does need to be 2048 still [08:01] using your loop trick, shows up 2048*512 [08:02] yes, it HAS to be 2048 [08:02] First sector (256-732566644, default 256): [08:02] ahhh, that's because it's looking at the physical sector size 4096 [08:03] oh duh, thats why 2048*512 worked still... because that doesn't care about sectors [08:03] echo '256*4096; 2048*512' | bc => 1048576 [08:05] created partition, wrote to disk, no error, mounted partition [08:05] Ben64: if the kernel is using the 4096 byte sectors you can redo the start is 256, or you can pull a neat trick. Create 2 partitions, one starting at 256 and the other at 2048... you have to do that editing manually of course on-disk using 'hexedit' but it's a clever way of being able to access the file-system no matter what translation is operating [08:05] Ben64: so you now have a working /dev/sda1 ? [08:05] well its /dev/sde1 over here, but yep :D [08:05] thanks [08:06] dunno how the usb part came back to life [08:06] tried it days later and it still didn't work, now it does [08:06] bad connection on the cabling/sockets - possibly cracked solder where the sockets are soldered to the PCB [08:07] so I should still see if it's under warranty [08:07] I often see that, and just reflow fresh solder/flux on the pins and ensure the socket itself is well soldered down on its lugs [08:07] OK, well, I have to go out now. Don't go breaking it again :) [08:07] lol [08:07] i'll try not to write too hard to it anymore [08:08] i also have a 1TB seagate with double usb ports cable, works like a charm [08:08] double usb ports? [08:08] yeah to give it enough power [08:09] its from few years ago [08:09] now its all 1 usb cable [08:09] mine is still split [09:00] devices are much lower power now, so they don't need more than the 0.5A per USB2 port they could expect [10:45] hi cfhowlett :p [10:47] http://arstechnica.com/security/2015/11/dell-apologizes-for-https-certificate-fiasco-provides-removal-tool/ [10:47] a lot of big companys apologiys recently :p [10:49] maybe they shouldn't pull dumb crap like that [10:49] yeah indeed [10:50] MS forced updates, VW cheating system,... [10:50] its all going bad :p [10:54] it's because 'developer' does NOT equal 'engineer' - Engineer implies a professional, with ethics, standards and so forth [10:55] If bridge Engineers worked the same way software developers did there would be uproar over all the bridge failures [10:57] TJ-: you remember how many devs worked on windows7 [10:57] and after a year MS sacced them all [11:01] I don't recall seeing that [11:04] TJ-: http://www.zerohedge.com/news/2015-07-08/microsoft-fires-7800-second-biggest-mass-layoff-its-history [11:09] That's this year, nothing to do with Windows 7 [11:10] Microsoft also announced the reduction of up to 7,800 positions, primarily in the phone business [11:10] yeah, not windows [11:10] no surprise. hard to get transaction against iPhone [11:11] iphone? [11:11] you mean android [11:11] both [11:12] 82.8% android [11:12] 13.9% ios [11:12] 2.6% windows [11:12] eh? that's seems rather high. source for the stats? [11:12] thought it'd be lower actually [11:12] thats units shipped 2015 q2 [11:13] says who? [11:13] http://www.idc.com/tracker/showproductinfo.jsp?prod_id=37 [11:16] I'm very surprised MS/Elop weren't charged criminally for the way they destroyed Nokia. Even more amazing is the shareholders sat back and took the destruction of value. From this side of history it looks like MS set out to destroy Nokia, any way they could [11:17] microsoft was in the business of destroying competition [11:17] was? [11:18] 1. Get MS executive onto Nokia board. Elop appointed CEO - CHECK. 2. 'pivot' from existing dominance of mobile market into making only MS Windows devices - CHECK. 3. Destroy value of Nokia - CHECK. 4. Buy remains cheaply - CHECK. 5. Fire the remaining inherited staff and close down the business - CHECK [11:19] One hell of a way to run a railroad! [11:19] cfhowlett: well they don't seem to be doing it like they were before [11:21] They're still funding partisan 'industry associations' to attack competitors both politically and legally [11:22] And providing 'marketing rebates' for Windows-only devices for those manufacturers that toe the line [11:22] and for all that, their phone division is failing. [11:22] right, both on 'phone and search they really don't understand the market drivers [11:24] They're used to dictating how users behave I guess, and have problems visualising how to leap-frog what is now the incumbant: Google [11:24] the only reason people use windows is because its what most people use [11:25] true [11:25] they're not used to people having choice [11:25] we were on about this earlier :) [11:25] "who in their right mind would choose a product that over 15 years has needed security patches every month?" [11:25] some people still imagine linux being like the lines of code in the matrix [11:26] lol [11:26] i like how they don't even push patches until a certain day [11:26] neo took over my computer [11:26] eh whatever, stay vulnerable till next patch wednesday [11:26] haha [11:26] I'm glad it does, that way you get an idea what it is doing [11:27] windows 10 is increasingly scary sounding, so thats good for linux [11:28] i agree on that Ben64 [11:28] will be good for my ubuntu shop :p [11:28] you have an ubuntu shop? [11:28] windows has been scary for YEARS and linux is still stickied at ... what? 2 - 4% market penetration? Hell, even OSX has more than that. [11:29] its scaring more normal people now [11:29] Ben64: not yet, next year ill start [11:29] forced updates, privacy stuff spelled out [11:29] Ben64: ubuntu + ssd + tweaked [11:29] selling ubuntu computers? [11:30] yes also [11:30] and phones and tablets [11:30] do you need to do anything with canonical [11:30] ill contact them before i start to ask that [11:30] not sure the licensing stuff for ubuntu [11:31] but ubuntu is free right [11:31] "Ubuntu" isn't free [11:31] Ubuntu is free [11:31] If I'm remembering properly [11:31] you mean the name mentioning on someone elses website then [11:32] I think its like the iceweasel thing [11:32] i could also mention 'a linux distro' [11:33] Ben64: well, you know system76, ill do a bit liek them [11:33] Ben64: [11:33] https://system76.com/ [11:33] or just ask permission [11:33] yes i will contact them [11:34] "To use the Ubuntu trademark, you must secure the Canonical's OEM services team's permission. You can contact them to see if what you're doing is okay. This should not be too difficult, as several small vendors like System76 and ZaReason have secured an agreement." [11:34] Sounds pretty easy [11:34] yeah [11:36] Ben64: https://system76.com/disclaimer [11:36] last 3 lines [11:49] hey EriC^^ [11:49] hey lotuspsychje [11:49] what's up? [11:49] oh nothing much, easy chatting here this morning with the lads here :p [11:50] cool [11:51] TJ-: yesterday someone was trying to convert his legacy install to uefi, ( booting the usb in uefi mode gave him a grub shell ), i tried to modprobe efivars but grub still couldn't find the efivars [11:51] EriC^^: Ben64 suggested me to contact canonical for my shop and i found this: https://system76.com/disclaimer [11:52] was wondering if there was anything else needed to be done to get grub-efi to install from a legacy booted install, in the end he turned off secure boot and that somehow didn't drop him to a grub shell anymore and he installed in uefi mode ( kind of odd isn't it? ) [11:54] EriC^^: the way I do it is to install grub-efi, but also copy /efi/EFI/grub/grubx64.efi to /efi/EFI/BOOT/BOOTX64.EFI ... that is the EFI simple (removable media) boot path the firmware checks on a device you try to boot in EFI mode. That way the system will still be bootable, and once booted in EFI mode, you can redo the "grub-install" which can successfully use efibootmgr to add the Ubuntu boot menu [11:54] entry [11:55] i see [11:55] lotuspsychje: what does that mean to you? [11:55] EriC^^: i also have to mention ubuntu's logo etc trademark [11:56] TJ-: how come it didn't drop to a grub shell after disabling secure boot? isn't secureboot supposed to not let grub even be loaded in the first place? [11:56] lotuspsychje: oh ok, that's a good heads up then [11:56] yeah [11:58] EriC^^: if the PC is booting in legacy mode its reading LBA 0 from the boot device. If it starts in EFI mode, and there is no existing boot menu entry, it'll look at the bootable devices for an EFI system partition, and then look for the simple boot path. Without either of those the firmware won't even be able to load GRUB [12:00] yeah, he said when he chose UEFI usb in the boot menu it gave him a grub shell, legacy worked though [12:00] when he disabled secureboot, he then chose UEFI usb again, and it booted into ubiquity [12:01] maybe he had a bad shimx64.efi or something? it's odd no? [12:03] The grub shell in UEFI mode tells you it loaded /EFI/grub/grubx64.efi - otherwise known as GRUB's core.img [12:03] So it sounds like GRUB simply couldn't find its root file-system (/boot/) where the grub files are [12:04] in that case always use "set" to check the values of "root" and "prefix" variables, and then look in the "prefix" location for the grub.cfg and other files/dirs to check the expected files are there [13:07] HI folks [13:07] hey BluesKaj [13:08] hi lotuspsychje [13:09] BluesKaj: you said yesterday something about nvidia depraced? [13:09] nouveau or current? [13:10] no more nvidia-current [13:10] BluesKaj: whats the replacement for it? [13:11] or do users need to specifi nvidia-340 etc [13:11] if one tries to install the driver using nvidia-current, it fails [13:11] yes, specific afaik [13:11] any reason? [13:12] dunno [13:13] BluesKaj: ok tnx [13:14] the last nvidia-current was the 304 driver [13:15] !info nvidia-current [13:15] which is was used on 13.10 iiirc [13:15] nvidia-current (source: nvidia-graphics-drivers-304): Transitional package for nvidia-current. In component restricted, is optional. Version 304.131-0ubuntu0.15.04.1 (vivid), package size 4 kB, installed size 36 kB (Only available for i386; amd64) [13:15] right [13:15] !info nvidia-current xenial [13:15] nvidia-current (source: nvidia-graphics-drivers-304): Transitional package for nvidia-current. In component restricted, is optional. Version 304.131-0ubuntu2 (xenial), package size 4 kB, installed size 19 kB (Only available for i386; amd64) [13:21] so the 304 driver is the last nvidia-current [13:23] BluesKaj: so users think they gonna get latest and get the 304 [13:23] yup [13:23] devs know about this? [13:23] lemme go ask in devel [13:25] dunno , never seen it discussed on #kubuntu-devel, but maybe the devs at #ubuntu-devel have an answer [13:32] BluesKaj: is nvidia-current going to be depraced? [13:32] shows still nvidia-graphics-drivers-304 [13:32] it's already deprecated, as it's just a transitional package [13:34] ok [13:34] thanks [13:34] BluesKaj: it will vanish after 304 is dead [13:34] makes sense [13:34] BluesKaj: and additional drivers section now manages right driver for the system [13:35] yeah in kubuntu it's the driver manager [13:35] good to know [13:35] yeah, it's important [14:08] his pc might have autorun worm and error on the stick TJ- [14:09] lotuspsychje: nah, the while thing has been wiped with the dd of the ISO [14:10] This is the person that was banned for being a pain not so many weeks ago, in fact, the nickname got notoruious [14:10] oh [14:10] cant recall that nick [14:11] 2015-11-04 18:12:07 TJ- Eh? No - I was commenting on pikapi managing to destroy Ubuntu 3 times a day apparently [14:11] 2015-11-05 12:54:47 daftykins TJ-: well it's been all go here too ;) pikapi is one of those bad-advice trolls [14:11] 2015-11-05 12:55:30 TJ- pikapi, the 3-a-day Ubuntu kid? [14:12] right [14:13] I think pikapi is one of those people that does all sorts of random stuff 'read on the forums' without any concept of what it does, and then comes here wanting to be dug out of the hole [14:13] thats not the same as that pikachu guy that gave false advise? [14:13] TJ-: the on you tought to have asperger lol [14:13] Right! that's the one [14:13] looool [14:13] I'd forgotten... thankfully :D [14:14] no wonder its stressing me [14:14] TJ-: but ikonia banned that guy [14:14] i was there [14:16] ahh, yes, ddaftykins told me about that [14:17] doesnt make sense, that guy gave false advise was more agressive to other users [14:17] lol soon we are investigators instead of volunteers :p [14:21] this is what I mean about him doing random things: [14:21] 2015-11-03 19:41:16 pikapi i did sudo apt-get remove initramfs-tools and it was uninstalling just about everything [14:21] lol [14:22] * TJ- is rolling eyes like bowling balls :) [14:22] hehe [14:22] someone has a mental disease :p [14:24] when you've got users randomly doing things like that there could be any number of things wrong with the system causing knock-on effects - like this booting issue, we still don't know what the PC is I don't think, but I wouldn't be surprised to learn its been 'tuned' and overclocked in the firmware setup [14:25] true [14:25] thats why support is so hard [14:25] never know what the user did really [14:28] I had a remote ssh session on Sunday, to recover a RAID-1 array on a live virtual machine Xen host that had 'lost' access to a pair of disks after the datacentre remote-hands had killed the power to the server. Took about 3 hours to fix it up whilst the system was live. It made me realise how much better it is to have hands/eyes-on [14:29] yeah [14:30] TJ-: bt you did a great job on my ralink remote :p [14:30] I did? we did? I don't remember that! [14:30] lol [14:31] stay off the dog food :p [14:31] oh... you mean that WMP600N PCI ID issue [14:31] ralink 2800pci [14:31] yep [14:31] ahh, see, I don't think of it as ralink at all [14:31] yeah linksys as main :p [14:33] My brain remembers visuals and concepts and procedures, but not precise facts - not for long anyhow. I constantly have to 'refresh' by searching for the precise way of doing something. One reason I prefer precise accurate documentation and exact instructions. [14:34] TJ-: yeah youre a very practical thinker and i like that [14:38] bbl driving lessons [14:51] TJ-: uefi conversion [14:51] robotti says he can't boot in uefi mode on his mac [14:52] how do we go about getting grub installed? [15:02] TJ-: you there? [15:45] sorry, was out and about in the fields [15:50] TJ-: np, it apparently installed fine [15:50] sudo modprobe efivars might have done it [22:34] on macs it's fine to just install grub to the partition that ubuntu is on; then suggest installing rEFInd as a boot loader [22:34] amusingly i'm attempting to fix a boot issue on my old HTPC today :D [22:35] a dist-upgrade murdered a 14.04 install o0 it claims it can't find a UUID at boot but they're all correct, very odd indeed [22:35] anywho i'm not gonna give up just yet, giving it a look over now via chroot :) [22:41] daftykins: I do not know Macs or rEFInd , not much direct help there .. BUT I do boot an (L)ubuntu install with grub installed onto the partition . [22:42] oh the mac thing was just a delayed comment to EriC^^ - i was scrolled up a bit by accident [22:42] my one is a standard PC [22:43] try to update-initramfs -u maybe [22:45] :) .. I multi boot 'buntus .. took me weeks of trial and effort and a steep learning curve to be able to boot any I desire, and a update on one not messing up grub on the others .. Hey, I feel good with what I did learn in this endevour . [22:45] EriC^^: yeah that seems like the ticket now, i had some udevadm errors - wireless keyboard wouldn't call up GRUB so i'm going to do it from live session again :> [22:46] Bashing-om: sounds good indeed, i don't really know where to start with boot most of the time [22:46] and i don't know about you guys but i found those boot repair logs a nightmare to read =| [22:47] daftykins: I still get lost with the hooks that are called into the kernel . Not real sure to this time what is taking place in kernel land . [22:51] :) [22:51] cat trying to sneak on my lap whilst i'm busy D: [22:57] ah, i love the native resolution boot logo up on my large TV... shame the nvidia proprietary drivers ruin that [22:58] not half as much as i love the native resolution TTYs mind ;) [22:58] ok that's mine all fixed ^_^ [22:59] I've had to fix a GCC 5 induced bug in syslinux that prevents systems booting, and it affects USB startup disk creator and anything else that uses syslinux for boot loader services, in order to build a new plopkexec floppy. Then I had to figure out how to build a kernel small enough to fit on a floppy, and now the resulting darned plopkexec kernel image can't find the plop /init exectuable! [22:59] >_ why are we even moving to GCC 5 when it seems to have screwed over so many things? [23:01] Because it provides some great improvements [23:01] ah just painful to get there :> [23:01] the problems were caused by programmers using unsupported functionality, rather than sticking to the standards [23:01] oh ok sort of like when things become deprecated? [23:02] not even that... things GCC just happened to do but didn't match the language/POSIX specifications, so when later versions of the compiler are used to build that code, and have tightened up the observence of the rules, things break [23:05] ah har [23:05] oh btw guys i read AMD have moved a lot more graphics cards to legacy status now, including some APUs [23:05] this is mostly for the Windows drivers but i'm sure that'll trickle down into the Linux driver(s) at some point too [23:06] I'm having to force using gcc 4.8 for this 3.1.4 version that plop uses: "make ARCH=i386 CC=gcc-4.9 bzImage" [23:06] Yes, they seem to do that very aggressively - not the best way to keep your loyal customers happy [23:09] http://anandtech.com/show/9815/amd-moves-pre-gcn-gpus-to-legacy [23:09] here's more [23:10] so before it was up to and including HD 4000 that was useless on anything newer than 12.04.1; now HD 5000, HD 6000, HD 7000 and HD 7600 are gone [23:10] also pre-Kaveri APUs [23:13] oink [23:15] o/ [23:18] ahah! The kernel's initramfs/ directory had a symlink to the built version of the /init exectuable I'd built in another location, but the kernel was literally including the symlink, not resolving it to the exe! [23:19] boots and works in a VM... now to try it again on that obstinate mobo [23:19] * daftykins sighs at the channel [23:26] nope, doesn't like the hardware. Grrr, so frustrating! [23:26] that the embedded one again? [23:28] TJ-: The bright side . Fighting the motherboard . slim chance you will find trouble elsewhere and land in Jail . [23:29] :D [23:29] haha! can reproduce it in the VM now... qemu-system-x86_64 -cpu 486 ... [23:29] So, the compiler is building the /init exe for a later 32-bit CPU [23:30] easily remedied? [23:38] yes, alter the compiler options to target the earlier CPU [23:41] \o/ [23:46] so, it dies in qemu VM with 486 or pentium, but not pentium2. I built the /init exe with gcc -march=c3-2 (which is the specific VIA C3 Samuel 2 CPU on the mobo) [23:47] oof those little things :) [23:50] hmmm, it's looking like it's not the /init options so much as the kernel options [23:54] ^^ That ole song " What does it take to keep a kernel like you, Satisfied " . [23:54] :D