IRC Logs for #crux-devel Wednesday, 2011-11-23

frinnstso, how are we looking?02:31
*** Romster has quit IRC02:44
*** acrux has quit IRC03:14
*** acrux has joined #crux-devel03:15
*** jue has quit IRC03:16
*** jue has joined #crux-devel03:17
pitillocould be related the ich vbox problem to a overload of modules? May be a conflict between them?03:18
frinnstit works in real life (tm) with ich903:19
pitillohas someone tried to make an install with piix3/4 and then change it to ich when the kernel is compiled? is this possible or can it give problems?03:19
frinnstor whatever i have in my desktop.. let me check03:19
pitilloyes, that's why I'm asking frinnst03:19
frinnstich9 should use ahci as sata03:20
frinnstunless virtualbox is retarded03:20
pitillomay be a conflict (I got it yesterday when I deployed 2.7.1 at home with piix ide support and mixed 2 modules (removing one of them made it work)03:20
pitillofrinnst: it uses in vbox too, but may be the overload of modules can cause a conflict. That's why I don't know if someone tried to install with piix3 and then changed to ich when the kernel is compiled with its support (ich)03:21
pitilloI think I'll make some tests here now which is in calm the enterprise03:21
*** mike_k has joined #crux-devel03:22
frinnstcorrection, ive tested with ich803:24
pitilloummm I'll try to make the install in that way and then, change it to ich and look what happens without all modules loaded03:24
*** sepen has joined #crux-devel03:57
*** Romster has joined #crux-devel04:36
*** Romster has joined #crux-devel04:36
pitilloI just built a kernel for ich, changed system to ich9 and ide to ich6 and now it's working right (I think the problem with media install is the conflict between modules)05:35
*** mike_k has quit IRC05:50
Romstercould it be having both modules loaded causes the issue?05:51
Romsteri've always had problems if you load too many controller modules at once.05:51
pitilloRomster: yes, that's what I meant. Seems that there is a conflict between modules in the install media (I got something similar yesterday deploying 2.7.1 and compiling my custom kernel with 2 "doubt" modules)05:52
pitilloI'm not sure if it's really that or if there is a fault in my side (but checking other's test seems a good reason, and more when I changed vbox setup and rebuild the kernel and it's working right)05:53
Romsterwell i think it's ok if 1 is not modprobed or does that still act up if you select 2 modules and only have one modprobed?05:54
pitilloyou haven't that option while booting the install media. In your own, you can do whatever you want. Here I used the right module (at home in my comp and here in vbox... I don't need to have more than I need)05:54
Romsteryes but the iso has mroe than 1 modules.05:56
pitilloyes, that's the point. And all are autoloaded. I don't know if you can force the use of one and blacklist the rest, but I don't think so. And I really didn't deep so much to look for the problematic modules to see if removing one of them solves the problem05:57
*** mike_k has joined #crux-devel06:02
*** sepen has quit IRC08:20
jaegerOne way to test would be to boot with "noata noscsi" and load the modules manually until a conflict hits08:23
jaegerUnless you've already identified the problem modules08:23
pitilloI think it's hard to know which ones conflict08:28
jaegerYou can also boot with "debug" to see the load order08:40
jaegerit freezes when loading ahci for me08:40
pitilloummm but first boot or when you recompiled the kernel to give ich support?08:41
jaegerI'm still referring to the boot of the install disk08:42
pitilloI didn't know about that option. I'll check with it too (should be the initramfs.lst order or it loads them in "random" order?)08:42
jaegerThey always load in the same order08:42
jaegeryou can also see it in initramfs/init in gitweb08:42
pitillook, sorry, I didn't understand it right. Here doesn't freeze, it boots but without detecting sata drives08:42
jaeger;a=blob;f=initramfs/init;h=b30145bd6a8a5ecff9948b0bc49c489a475c6c5a;hb=2.7 starting at line 23008:42
jaegerwell, for me it freezes for 30 seconds or so08:42
jaegerthen continues but without any drives08:43
jaegerif you use piix3 there's no 30 second delay or however long it really is08:43
pitilloah yes, it seems to be freezed. I thought you got stalled there08:43
jaeger(I didn't time it exactly)08:43
pitilloyes, here using piix3/4 it loads quite fast compared to ich08:43
jaegeranother data point: if I remove the SATA controller from virtualbox and use SAS instead, it still delays08:46
jaegerloading mptsas instead of ahci08:46
jaegerhrmm, the delay in this case has been far longer, it may actually be hung08:49
jaegerthe cursor is still flashing but it's not doing anything else08:49
pitilloummmm some minutes should be enought to know (not sure if I haven't enought patience to wait much more)08:50
pitilloI'm thinking in pata_mpiix which could be one module I loaded yesterday and gave me problems at home08:51
jaegerit just now finished booting08:53
jaegerso it took around 7 minutes08:54
jaegersame result, no drives found08:54
pitilloummmm I think there is a module making a mess08:54
jaegerI can't reproduce this in vmware esxi but esxi also doesn't let me change the chipset type08:56
jaegerhere's another data point:08:59
jaegerIf I load no modules aside from the default core stuff ("noata noscsi nofirewire nousb"), then load, in this order: sd_mod, mptsas, it still hangs09:00
pitillodoes dmesg show some undefined symbols doing that in that way?09:01
jaegerI don't think it did last time I checked but I'll check as soon as it finishes hanging09:02
jaegercan't do anything until it's done09:02
pitilloummmm here modprobing ahci got a freeze too09:02
jaegerusing ICH9 still, I assume09:03
jaegerIt seems it HAS to be a problem with the ICH9 support09:03
jaegerboth ahci (SATA) and mptsas (SAS) controller modules cause it09:03
pitilloummm yes, still using it and loading with noata noscsi to load them by hand09:04
pitilloummm could be possible, but not sure about, because with a recompiled kernel it works, then this makes me missunderstand that09:05
jaegerWell, yeah, I'm sure there's a way to fix it because a compiled kernel later works well, but I can't figure out what the difference is09:05
jaegerI guess I could start over from scratch with the kernel config but that will be a pain09:06
jaegerok, no undefined symbols in dmesg related to mptsas but lot of mptsas reset messages09:06
jaegergonna switch it back to SATA since it's faster and both break09:07
pitilloummm the kernel config is right, it gives only problems with this ich9 under vbox. I'll continue deeping a bit to see if I can split the modules and look for the one which conflics (because this seems to be the problem IMHO)09:07
jaegerI'm not even 100% sure it's a module issue09:08
jaegerbecause if I load only sr_mod, sd_mod, and mptsas in the last test, it still fails09:08
pitilloummmmmm ok09:09
jaegerwhat if it's something else, like a CPU option, or memory, or SMP, or any number of other things09:09
pitillolot of things to check09:09
jaegerlots of "qc timeout" messages loading just ahci09:10
jaeger is interesting, though old09:12
jaegerthose error messages are exactly what shows up in dmesg09:13
jaegersome comments there do indicate it might be SMP related09:14
jaegerdisabling SMP entirely is on my list of things to try09:14
*** mike_k has quit IRC09:14
pitilloummm recomendation to don't use ich9....09:15
jaegerwell, shit, SMP isn't even enabled09:18
jaegerso that's out09:18
jaegerI'm going to try some different kernel options again09:23
jaegerIt's easy enough to build a new ISO with different options, the packages don't need to be rebuilt09:23
pitillothat sounds great then09:24
pitillocan you share the diff between actual and newer one?09:24
jaegersure, if I make any progress09:25
pitilloummmm jaeger I found something strange. In the config file on gitweb ext4 is supported as module but isn't included in lib/modules (I can't find it), is this right? sorry if I'm making noise with this kind of things09:40
jaegerIt's possible... because technically you don't need ext4 support to find disks09:41
jaegeronly later when you're trying to mount09:41
jaegerIt should be added, though, for people trying to install from an ext4 partition09:41
jaegerI will put that on my list to check09:41
pitilloyes, but I found this deeping in the undefined symbols for ext2/3 (nothing related to ide/sata controllers)09:41
pitillothe question is the kernel/modules were generated from the config file I'm looking atm09:42
pitillos/is the/is if the09:42
jaegerThere's a disconnect between the modules that the initramfs uses and the modules that are available after the initramfs is done loading09:42
jaegerthe initramfs only needs enough to get the media mounted09:42
jaegerfor example it doesn't have network modules09:43
jaegerwith that said, they're all generated from the same kernel config09:43
pitilloah ok, thank you for that explanation09:44
jaegernp, hope it helps a bit09:45
jaegerthe initramfs stuff was very confusing to me initially, took a lot of research and testing to get an understanding of it09:45
*** mike_k has joined #crux-devel09:48
jaegerno luck so far with my changes09:56
jaegerstill trying09:56
pitillohere I don't understand the undefined symbols error. Support is right for all those modules which reports them09:58
jaegerMy guess is that the autoloading of module dependencies is odd in busybox or just doesn't suppress those messages, because things still work09:59
jaegeralso, I think that's unrelated to the problem here but I'm not 100% sure09:59
mike_kboots on i915GM/ICH5, P55(at least USB boot), H55 here10:05
mike_klast two being SMP machines10:05
jaegerthanks for testing, appreciate it :)10:06
jaegerNone of my changes have had any effect whatsoever so far10:09
pitillois it worth trying to update bb? or does this imply an entire rebuild?10:10
jaegerThat would not be too hard10:11
pitillonot sure how it could affect to these ide/sata problems with ich. I'll take a look here when I arrive home. I can't test there but tomorrow I can continue with this10:24
jaegerbuilding one with busybox 1.19.3 now10:30
jaegerlooks like no change with the latest busybox10:50
*** mike_k has quit IRC11:07
juejaeger: tried to boot my desktop with 2.7.1 today, works with both, usb and cd11:52
jaegerGlad to hear it, thanks :)11:53
jaegerI've got to go eat lunch with coworkers, will be back in about an hour11:53
juenp, glad to help :)11:53
*** mike_k has joined #crux-devel12:45
teK_btw: anyone got irssi with alt + n (number) Window switching working? I updated nvidia and xorg-libpixman yesterday and then it suddenly stopped working15:33
teK_21:14 meta-1               change_window 115:33
*** acrux|G4 has joined #crux-devel15:46
*** acrux|G4 has joined #crux-devel15:46
*** acrux|G4 has quit IRC15:48
*** acrux|G4 has joined #crux-devel16:22
*** acrux|G4 has joined #crux-devel16:22
jaegerteK_: It works for me in xterms and putty, what terminal do you use?16:28
jaegerfor xterms I have to add "xterm*eightBitInput: false" and "xterm*metaSendsEscape: true" to .Xdefaults16:29
*** mike_k has quit IRC16:47
*** acrux|G4 has quit IRC16:56
*** ente has quit IRC18:03
*** jue_ has joined #crux-devel20:51
*** jue has quit IRC20:54
*** jue_ has quit IRC20:54
*** jue has joined #crux-devel20:54

Generated by 2.11.0 by Marius Gedminas - find it at!