| 00:12.56 | *** join/#edev nullpuppy (~dustin@freematrix/staff/nullpuppy) |
| 00:26.36 | *** join/#edev nullpuppy (~dustin@freematrix/staff/nullpuppy) |
| 01:21.32 | *** join/#edev bizhanMona (~bizhan@nat/cisco/x-tfnziohnwaovysvz) |
| 03:19.10 | *** join/#edev TheSeven (~quassel@rockbox/developer/TheSeven) |
| 03:23.42 | *** join/#edev landley (~landley@cpe-72-177-21-136.austin.res.rr.com) |
| 03:23.42 | *** mode/#edev [+o landley] by ChanServ |
| 03:30.23 | *** join/#edev prpplague (~prpplague@ppp-70-249-149-181.dsl.rcsntx.swbell.net) |
| 03:47.14 | *** join/#edev aholler (~aholler@p57B20AF1.dip0.t-ipconnect.de) |
| 04:00.29 | *** join/#edev prpplague (~prpplague@ppp-70-242-125-74.dsl.rcsntx.swbell.net) |
| 05:28.50 | *** join/#edev clio (~andrej@85.159.109.222) |
| 06:23.21 | *** join/#edev prpplague (~prpplague@ppp-70-242-116-147.dsl.rcsntx.swbell.net) |
| 06:33.33 | *** join/#edev ChanServ (ChanServ@services.) |
| 06:33.33 | *** mode/#edev [+o ChanServ] by zelazny.freenode.net |
| 07:19.46 | *** join/#edev florian (~fuchs@port-217-146-132-69.static.qsc.de) |
| 07:19.46 | *** join/#edev florian (~fuchs@Maemo/community/contributor/florian) |
| 07:21.47 | *** join/#edev bizhanMona (~bizhan@nat/cisco/x-bjsmvnwrrwaxycxk) |
| 07:45.06 | *** join/#edev aholler (~aholler@p57B20AF1.dip0.t-ipconnect.de) |
| 08:21.30 | *** join/#edev ChanServ (ChanServ@services.) |
| 08:21.30 | *** mode/#edev [+o ChanServ] by zelazny.freenode.net |
| 09:20.46 | *** join/#edev grummund (~user@unaffiliated/grummund) |
| 09:30.58 | *** join/#edev mbuf (~user@49.248.85.146) |
| 09:36.29 | *** join/#edev grummund (~user@unaffiliated/grummund) |
| 10:41.11 | *** join/#edev itdnhr (~quassel@64.31.59.70) |
| 11:04.14 | *** join/#edev g1powermac (~g1powerma@unaffiliated/g1powermac) |
| 13:05.32 | *** join/#edev ccole (~cole@coledd.com) |
| 13:21.47 | *** join/#edev bizhanMona (~bizhan@nat/cisco/x-xussvagqoiiwqorv) |
| 13:25.28 | *** join/#edev mrcan_ (~mrcan@unaffiliated/mrcan) |
| 13:29.08 | *** join/#edev uwe_ (~uwe_@dslb-088-064-209-125.pools.arcor-ip.net) |
| 13:37.46 | *** join/#edev Crofton (~balister@12.207.17.3) |
| 13:55.22 | *** join/#edev landley (~landley@cpe-72-177-21-136.austin.res.rr.com) |
| 14:14.54 | *** join/#edev orbarron (~orbarron@nat/ti/x-daydvpvseftsemdp) |
| 14:32.55 | *** join/#edev mrcan__ (~mrcan@unaffiliated/mrcan) |
| 15:10.02 | *** join/#edev sndcrb (~sndcrb@dsl-hkibrasgw2-fef8de00-126.dhcp.inet.fi) |
| 15:14.24 | *** join/#edev GPSFan (~kenm@64.92.145.112) |
| 15:14.24 | *** mode/#edev [+o GPSFan] by ChanServ |
| 15:41.19 | *** join/#edev Crofton (~balister@12.207.17.3) |
| 15:52.40 | *** join/#edev NightMonkey (~NightrMon@pdpc/supporter/professional/nightmonkey) |
| 15:57.26 | *** join/#edev prpplague (~danders@192.94.92.11) |
| 15:58.49 | *** mode/#edev [+o prpplague] by ChanServ |
| 16:15.29 | *** join/#edev ohsix (ohsix@208.100.175.111) |
| 16:15.48 | *** join/#edev florian (~fuchs@sign-4d094488.pool.mediaWays.net) |
| 16:15.48 | *** join/#edev florian (~fuchs@Maemo/community/contributor/florian) |
| 16:20.42 | *** join/#edev landley (~landley@140.242.26.2) |
| 16:20.42 | *** mode/#edev [+o landley] by ChanServ |
| 16:24.33 | landley | slashbeast: I tried to reproduce your thing last night, and still hadn't managed it after a couple hours. |
| 16:25.46 | slashbeast | landley: unable to reproduce it? |
| 16:26.21 | *** join/#edev landley_ (~landley@140.242.26.2) |
| 16:26.31 | slashbeast | landley_: unable to reproduce? |
| 16:27.05 | slashbeast | be back in ~30min |
| 16:39.34 | landley | It's more that the test cases you sent me really don't show me how to reproduce a failure in my build. |
| 16:41.18 | landley | I emailed you about it, but my email's a bit unreliable these days thanks to thunderbird being "utter pants" as the british say... |
| 17:16.41 | slashbeast | landley: just run mdadm --assemble --scan and it should fail hard. |
| 17:17.08 | slashbeast | landley: as far as I can see you need system/env without udev to reproduce it, with udev running it somehow seems to be working. |
| 17:17.23 | slashbeast | but the glibc binary works well in initramfs enviroment. |
| 17:17.52 | slashbeast | landley: the eastiest way to reproduce would be replace mdadm binary i inintramfs and test it there, or if you need I can prepare fairly small gentoo w/ mdev where you can test it too. |
| 17:25.29 | *** join/#edev Crofton (~balister@157.22.28.10) |
| 17:29.50 | landley | slashbeast: I'm trying to reproduce it in dev-environment.sh |
| 17:30.10 | landley | I sent you email about it last night. |
| 17:38.24 | *** join/#edev NightMonkey (~NightrMon@pdpc/supporter/professional/nightmonkey) |
| 18:32.57 | *** join/#edev ChanServ (ChanServ@services.) |
| 18:32.57 | *** mode/#edev [+o ChanServ] by zelazny.freenode.net |
| 18:59.12 | *** join/#edev CosmicPenguin (~nobody@soa.codeaurora.org) |
| 19:00.44 | *** join/#edev CosmicPenguin (~nobody@soa.codeaurora.org) |
| 19:02.41 | *** join/#edev ChanServ (ChanServ@services.) |
| 19:02.41 | *** mode/#edev [+o ChanServ] by zelazny.freenode.net |
| 19:05.59 | *** join/#edev CosmicPenguin (~nobody@soa.codeaurora.org) |
| 19:05.59 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 19:10.29 | *** join/#edev CosmicPenguin (~nobody@soa.codeaurora.org) |
| 19:10.29 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 19:13.43 | *** join/#edev CosmicPenguin (~nobody@soa.codeaurora.org) |
| 19:13.43 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 19:21.47 | *** join/#edev bizhanMona (~bizhan@nat/cisco/x-xspqlleclqtzehwf) |
| 20:59.13 | *** join/#edev djerome (~djerome@ip68-2-20-108.ph.ph.cox.net) |
| 21:03.21 | *** join/#edev slashbeast (~slashbeas@unaffiliated/slashbeast) |
| 21:31.00 | slashbeast | landley: I am lost, is there anything I can provide to help you with it? |
| 21:31.36 | landley | slashbeast: what do I do, in my build environment, to reproduce your issue? |
| 21:31.50 | landley | So far your answer is "start with something other than that build environment". |
| 21:32.14 | landley | I've spent several hours reverse engineering your environment, attempting to extract the relevant bits so I can reproduce it under mine. |
| 21:32.19 | landley | So far, I haven't gotten all of them yet. |
| 21:33.00 | slashbeast | landley: a image with raid, a mdadm and run --assemble --scan |
| 21:33.12 | slashbeast | ensure that raid is not auto assembled by any udev hook. |
| 21:33.27 | landley | Aboriginal LInux system images haven't got udev. |
| 21:33.37 | landley | I built mdadm from source, using what I think are your instructions. |
| 21:33.51 | slashbeast | you can use one of the raid images I sent you, and hook it into your qemu (I believe you use it) |
| 21:33.53 | landley | I attached your hard drive image as /dev/hdc and ran the same line you're doing to create /etc/mdadm.conf |
| 21:34.09 | slashbeast | mdadm --assemble --scan should fail hard. |
| 21:34.14 | landley | It then complained that /dev/hdc didn't have the right signature to indicate it had raid stuff on it. |
| 21:34.37 | landley | Oh, and I had to rebuild my kernel to switch on md and raid support. |
| 21:34.50 | landley | I have no idea what kernel .config your build used. |
| 21:35.45 | slashbeast | landley: http://bpaste.net/show/2czzijiawwpHtJm8yT4c/ |
| 21:35.53 | slashbeast | from fresh vm I just created, with gentoo |
| 21:36.26 | slashbeast | https://paste.lugons.org/show/V12C66iSg8g5kK3LK29F/ and the kernel config from the VM. |
| 21:37.59 | slashbeast | landley: with glibc's mdadm mdadm --assemble --scan simply works. also, with the aboriginal's mdadm you can *manualy* assemble the raid |
| 21:38.26 | slashbeast | # ./mdadm --assemble /dev/md1 /dev/vdc1 |
| 21:38.28 | slashbeast | mdadm: /dev/md1 has been started with 1 drive (out of 2) |
| 21:38.45 | slashbeast | 'hope that help |
| 21:41.20 | slashbeast | the kernelconfig is from miniconfig so it should be fairly small. |
| 21:45.59 | landley | What's /dev/vdc1? |
| 21:50.13 | slashbeast | the virtio's block device |
| 21:50.26 | slashbeast | its the same image I sent you |
| 21:50.33 | slashbeast | I just use virtion instead of PATA in my qemu-kvm |
| 22:31.57 | landley | Following your bpaste link: I got that last night, it's because the kernel .config didn't have MD or RAID support configured in. |
| 22:32.22 | landley | Once I pulled up menuconfig and build a kernel with CONFIG_MD and a bunch of stuff under it switched on, I got different errors. |
| 22:32.59 | landley | slashbeast: "md: hdc does not have a valid v1.0 superblock, not importing!" |
| 22:33.10 | landley | md: md_import_device returned -22 |
| 22:33.27 | landley | Possibly I need more config symbols to get 2.0 support or something, dunno... |
| 22:33.42 | landley | I haven't configured a kernel for raid in years. |
| 22:33.58 | slashbeast | landley: well, the image is 1.0, thats odd |
| 22:34.22 | landley | It could be that my qemu doesn't have qcow2 support compiled in. That's where I left off, anyway. |
| 22:34.22 | slashbeast | landley: what if you do mdadm /dev/md1 /dev/hdc1? |
| 22:34.27 | landley | Can't bang on it at work. |
| 22:34.59 | landley | Well, actually I might be able to ssh... |
| 22:36.05 | landley | work interrupt, back in a bit... |
| 22:43.14 | landley | \mdadm: an option must be given to set the mode before a second device (/dev/hdc1) is listed |
| 22:43.19 | landley | slashbeast: thingy. |
| 22:43.38 | landley | Ah. |
| 22:43.49 | landley | ./mdadm.static --assemble /dev/md1 /dev/hdc1 |
| 22:43.53 | landley | md: md1 stopped. |
| 22:43.55 | landley | md: bind<hdc1> |
| 22:44.00 | landley | bio: create slab <bio-1> at 1 |
| 22:44.09 | landley | md/raid1:md1: active with 1 out of 2 mirrors |
| 22:44.19 | landley | md1: detected capacity change from 0 to 32493568 |
| 22:44.27 | landley | mdadm: /dev/md1 has been started with 1 drive (out of 2). |
| 22:44.39 | landley | slashbeast: apparently that was the magic command line to do it, and it's working. |
| 22:44.57 | landley | All I changed was rebuilding my kernel with extra CONFIG_MD symbols. (I grabbed a half-dozen of them, anything that looked relevant.) |
| 22:45.01 | slashbeast | right, but the --scan --assemble die on kernel EINV as far as I remember |
| 22:45.20 | slashbeast | if you try --assemble --scan it will fail again. |
| 22:45.23 | landley | How do I un-mount it so I can go back to scan? |
| 22:45.38 | slashbeast | good question |
| 22:45.39 | landley | md1: unknown partition table |
| 22:45.48 | landley | mdadm: No arrays found in config file or automatically |
| 22:45.55 | slashbeast | it does not suppose to have any parititon table so ignore it |
| 22:46.56 | slashbeast | landley: mdadm -S /dev/md1 |
| 22:47.30 | slashbeast | thats what I googled |
| 22:47.58 | landley | slashbeast: ok, it's back to failing again. |
| 22:48.11 | landley | Complaining it's not importing because it doesn't have a valid superblock. |
| 22:48.23 | slashbeast | when it does have. |
| 22:48.33 | slashbeast | the mdadm bulit into buildroot also works |
| 22:48.42 | slashbeast | into/inside |
| 22:48.51 | slashbeast | or under, you are native, choose ;p |
| 22:49.29 | landley | Ok, so when you build mdadm.static under buildroot and then use it under this kernel, it works? |
| 22:50.05 | slashbeast | indeed |
| 22:50.21 | slashbeast | same goes for mdadm.static bulit under my glibc-based gentoo |
| 22:51.18 | slashbeast | landley: https://github.com/downloads/slashbeast/better-initramfs/mdadm-i586 bulit under buildroot |
| 22:51.21 | slashbeast | the working binary |
| 22:52.05 | landley | Ok, that's a uClibc config issue rather than a kernel config issue. |
| 22:52.32 | landley | Please email me the link to the working binary, and I"ll do a comparative strace of them when I get home. |
| 22:52.43 | landley | It's really awkward to do this from work, even after hours. :) |
| 22:52.53 | landley | I now know more than I did, though... |
| 22:53.24 | slashbeast | Roger that |
| 22:53.59 | slashbeast | sent |
| 23:00.09 | *** join/#edev prpplague (~prpplague@ppp-70-242-123-159.dsl.rcsntx.swbell.net) |
| 23:22.24 | *** join/#edev _Lucretia_ (~munkee@pdpc/supporter/active/lucretia) |
| 23:37.29 | *** join/#edev prpplague (~prpplague@ppp-70-249-147-169.dsl.rcsntx.swbell.net) |