=== crimsun [n=crimsun@66.248.140.183] has joined #ubuntu-kernel === crimsun [n=crimsun@pdpc/supporter/silver/crimsun] has joined #ubuntu-kernel === rtcm [n=jman@213.146.193.173] has joined #ubuntu-kernel [01:00] BenC: are you around ? [01:06] yeah [01:07] BenC: i'm the guy from the sis performance problem [01:07] oh, hey [01:07] i've really found the problem now [01:07] and even sort of fixed it [01:07] at least for me [01:08] ok, let's hear it :) [01:08] is seems that loading vga16fb was killing my performance [01:08] do you recall the bug number? [01:08] loading vesafb with the .12 kernel is ok [01:09] but loading either vga16fb or vesafb with the .15 kernel is crap [01:09] #20545 [01:10] but, it happens that the author of the X driver is also the author of sisfb which works quite good [01:11] i changed the usplash hook and init script to load sisfb instead of vga16fb and all is good even on .15 :-) [01:12] and sisfb even autodetects the size os my LCD giving me a really nice 1400x1050 console [01:12] do you have UseFB in /etc/X11/xorg.conf? [01:12] or something similar [01:12] nope [01:13] I use the sis driver [01:13] I know you use sis [01:13] but that doesn't mean it doesn't use the framebuffer [01:13] please paste your the section called "Device" from xorg.conf [01:13] hmm, guess not, but Thomas should know about that [01:14] there's some crap commented there [01:14] i'll put it all here anyway [01:14] Section "Device" Identifier "Silicon Integrated Systems (SiS) 65x/M650/740 PCI/AGP VGA Disp Driver "sis" BusID "PCI:1:0:0" [01:15] Option "EnableSiSCtrl" "true" [01:15] # Option "UseSSE" "yes" Option "AccelMethod" "EXA" [01:15] # Option "MergedFBAuto" "on" [01:15] # Option "CRT2Position" "LeftOf" [01:15] # Option "MetaModes" "1024x768-1400x1050" [01:15] # Option "CRT2HSync" "30-67" [01:15] # Option "CRT2VRefresh" "50-75" [01:15] EndSection [01:15] but you get the idea [01:15] damn [01:15] I only have EXA and sisctrl enabled [01:16] what is sisctrl? [01:16] (brb) [01:16] is a utility developed by the driver's author to tweak various things in the driver [01:17] you can check the site: http://winischhofer.at/linuxsispart1.shtml#23 [01:21] any chance that any of that stuff is causing the problem? [01:21] maybe try without any extra options, just use the default config? [01:21] if the fb driver is causing this, then I really want to lean toward it being a sis X driver bug [01:21] because it should have no affect [01:23] actually i already tried without EXA, it is the same, only EXA feels faster, though the benchmarks are the same [01:23] for example, it has no affect on my ATI or nvidia systems [01:23] sisctrl I think is totally unrelated, but I can try [01:25] BenC: i'll report in the author's "bug tracker" aka the SIS Forum :-) [01:25] guess you can close the bug, or maybe i'll ask Thomas to comment there again [01:25] I've updated the ubuntu bug and reassigned to the sis driver [01:26] thanks for being so diligent in findin the cause :) [01:26] ok with me, thanks for the pointers === TheMuso [n=luke@dsl-202-173-132-131.nsw.westnet.com.au] has joined #ubuntu-kernel === crimsun [i=nobody@warped.bluecherry.net] has joined #ubuntu-kernel === cjb yawns. [08:31] BenC: please pull from my archive when you can. Changes: new OCFS2 from the splitted oracle git tree and a simple patch to make thermal_adt746x a bit more quite in dmesg (ppc only) === ubuntulog [n=ubuntulo@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel === Topic for #ubuntu-kernel: Ubuntu kernel development discussion ONLY | New git tree for dapper: https://wiki.ubuntu.com/KernelGitGuide | 2.6.15-9.14 uploaded (The "Adam Adulterated the Awesome Build Architecture" release) === Topic (#ubuntu-kernel): set by BenC at Wed Dec 21 14:58:27 2005 === fabbione [n=fabbione@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel === rikai [n=gtk2@pool-64-222-225-195.port.east.verizon.net] has joined #ubuntu-kernel === zul [n=chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel [02:51] morning [03:02] fabbione: ok [03:02] zul: good morning [03:02] hey guys [03:03] BenC: morning how was your christmas [03:03] was good except for getting a stomache virus about noon on christmas day [03:03] oh that sucks...hey fabbione [03:03] my wife, my brother-in-law, his wife, and I all got sick [03:03] BenC: suckage [03:03] hehe [03:04] BenC: do you have lrm for the k8 test kernel? [03:04] fabbione: not yet [03:04] ok [03:04] if you manage to build it, i can test it [03:04] but i can't without lrm [03:04] otherwise here nothing works [03:04] if you could just boot with it, I'd be happy [03:05] ok [03:05] sure [03:05] i can do that [03:06] BenC: that network bug with the linksys card uses the tulip module [03:06] which bug number is that? [03:07] 21487 [03:08] BenC: you should read the last comments on 20771 [03:08] fabbione: from adam? [03:08] BenC: no the one from the guy that did test one of the kernels (#6) and still fail.. where #6 is only a rebuild of -9- [03:08] with only the sparc specific patch applied. [03:09] i mean 100% sparc specific [03:09] so debian dropped sparc support? [03:09] right [03:09] zul: what?! [03:09] zul: uh? [03:09] i was reading in some guys blog [03:10] not that I've seen [03:10] fabbione: yeah, I saw that...I'm wondering if it's tool chain related [03:11] that's why i asked logs [03:11] for all the builds [03:11] if we can prove a discrepancies in the chroot [03:11] http://lists.debian.org/debian-devel-announce/2005/12/msg00013.html [03:11] i am gonna laugh my ass off [03:11] fabbione: the .deb's contain the build tools list don't they? [03:11] BenC: i don't think so [03:12] zcat /usr/share/doc/linux-image-`uname -r`/buildinfo.gz [03:13] no binutils though [03:13] lame [03:13] i want to see the full logs [03:14] I found a bug on amd64 that you might want to fix in breezy [03:14] I'm very surprised no one noticed it before me, I only found it by accident [03:17] BenC: what's that? [03:17] hold on a sec... [03:18] bah sparc has been taken out of Debian RC arches because of kernel stability problems [03:18] fabbione: fix it kthxbye [03:18] i am ready to bet that all debian's buildd are Sparc III+ [03:18] SMP [03:18] that is a known kernel bug [03:19] zul: we did try.. problem is that davem that can fix it, got he wrong hw :/ [03:19] and a III+ CPU is 700USD each [03:19] include/asm-x86_64/rwlock.h [03:19] and we need at least 2 [03:19] #define __build_write_lock_const(rw, helper) \ [03:19] - alternative_smp("lock; subl $" RW_LOCK_BIAS_STR ",%(0)\n\t" \ [03:19] + alternative_smp("lock; subl $" RW_LOCK_BIAS_STR ",%0\n\t" \ [03:19] holy crap! [03:19] fabbione: no, they are Ultra30 and Ultra80 [03:20] BenC: and why they crash building OOo2??? [03:20] so Sparc II [03:20] they didn't crash building OOo2, they just plain crashed [03:20] we had to go back to 2.6.8 kernels [03:20] ah [03:22] they were both running fine, but elmo wanted latest kernels, and that's when the trouble started [03:22] then we started getting problems with the hub that vore was on [03:23] yeah i remember the hub problem [03:23] not sure why auric is still down [03:23] well at least also hppa is out of the race :P === fabbione tickles lamont__ [03:23] I have serial console to it from vore, but it's too dead for even a serial break to get it back [03:24] crap [03:24] tbh [03:24] i have been running our kernel on my sparc for quite sometime [03:25] .12 i mean [03:25] no issues at all [03:28] BenC: 21487 the card is not in the pci_device_id table [03:29] hmmm...maybe we should ubuntu on the amiga.. [03:30] zul: we could.. so be careful :) [03:30] i dont have the hardware :( [03:30] zul: between infinity and I we have quite a bunch of m68k laying around [03:30] sucks to be you [03:30] hmmm...send me one [03:31] yeah but no one in canada uses an amiga or there isnt alot of people selling them [03:31] ebay [03:31] i could.. [03:31] wish I still had my mac IIci [03:32] I should try to port uclinux to my ] [e :) [03:32] ehehhe [03:32] zul: you can also get a cisco 25xx [03:32] they are m68k based [03:32] there has to be some way to hack it to be 32-bit [03:32] and you can run linux on it [03:32] except you need a mmu-less kernel [03:32] but that's almost mainline [03:33] fabbione: try convincing my wife about getting another computer ;) [03:33] cisco? [03:33] it's a router ;) [03:33] you know what i mean [03:33] yeah yeah [03:34] or port ubuntu to the palm pilot [03:36] or might be on crack [03:38] earlier ciscos are m68k based? [03:40] dilinger: yeps.. at least the c25xx [03:40] newer ones are ppc based [03:40] but they use some internal reduced ppc processors [03:42] zul: how do you know that linksys needs the tulip driver? [03:42] i looked it up..im pretty sure it does [03:43] linksys has their own tulip driver i think as well [03:45] tulip_core.c has the vendor/device id's [03:45] whoops [03:45] in the device table [03:46] i suck [03:46] er...which line [03:46] 215 [03:48] don't make the same mistake I did and look at the subsystem id :) [03:48] 0x0985 is the device id [03:51] meh.. [04:15] BenC: So now there's /two/ versions of the broadcom wireless driver, one for softmac and one for devscape [04:16] mjg59: yeah, it's getting a little rediculous for a driver to fork when it isn't even ready yet :/ [04:16] devscape gives them the advantage of working encryption [04:17] But has an insanely ugly userland interface right now [04:17] so it doesn't work with iwconfig and friends? [04:18] would be nice if these new ieee80211 stacks would atleast be compatible with current userland tools [04:18] it does already handle private ioctl's [04:18] so it's not like it can't be extended [04:19] The softmac stuff is much closer to using the normal wireless extensions [04:19] The devscape stuff involves having a wifi device and then creating station devices from that using mad ioctl shit [04:20] sounds like a power user system [04:21] I thought I read that bcm43xx+softmac had working encryption using the kernel crypto? [04:22] Uhm. I don't think the kernel has any 80211 crypto [04:22] no, I mean using the crypto functions (AES and such) [04:22] Oh, right [04:22] Not that I'd seen, but... [04:22] maybe I confused that with the devscape email I saw [04:33] BenC: OSDL are having a linux wireless meeting early next year, so with luck things will get better [04:34] let's hope [04:35] They're also having a power management one, so I'm off to San Jose in April [05:01] i386 95.40% 6456 of 6767 [05:01] amd64 94.23% 6211 of 6591 [05:01] powerpc 94.20% 6240 of 6624 [05:01] ia64 93.38% 6132 of 6567 [05:01] hppa 92.00% 6018 of 6541 [05:01] sparc 89.42% 5884 of 6580 [05:01] hmmm? === lamont wonders which race fabbione meant.... [05:06] lamont: Debian RC arche [05:07] +s === lamont would rather just drop kde. :0)( [05:07] ehehhe [05:08] just for giggles, /me tries qt-x11-free with unaligned trap support turned on === BenC [n=bcollins@debian/developer/bcollins] has joined #ubuntu-kernel === lamont pushes sparc and hppa logs to people.u.c [05:28] ENOSPC on buildd.mmjgroup.com. sigh [05:37] Subject: LiveCD: Successful dapper:ubuntu build on hppa [05:37] yeah! [05:59] sweet === ..[topic/#ubuntu-kernel:BenC] : Ubuntu kernel development discussion ONLY | New git tree for dapper: https://wiki.ubuntu.com/KernelGitGuide | 2.6.15-10.15 uploaded (The "Howdy Ho!" release, AKA amd64-alternative-smp + rc7) === rikai [n=gtk2@pool-64-222-225-195.port.east.verizon.net] has joined #ubuntu-kernel [08:07] hmm that's something i could do in the next days.. [08:07] livecd and daily d-i [08:07] BenC: meh dude.. upload is pointless today. [08:07] there are no ftp.master to process new === crimsun [n=crimsun@cpe-066-056-187-199.triad.res.rr.com] has joined #ubuntu-kernel [08:42] lamont: -9.14 is still no go on my hppa [08:42] lamont: do you want the backtrace? [09:20] what sort of hppa do you have? [09:20] I'm running the hppa64-smp on my A500 box [09:22] i have a j5000 [09:22] smp [09:22] i did try the 32-smp [09:22] is it hppa64 or hppa32? [09:22] hppa64 [09:22] according to lamont it's better to run a 32bit kernel [09:23] it does an OOPS initializing pci [09:23] i am gonna try 64bit kernel and see if it still happens [09:23] odd, I've been running the hppa64-smp since I started 2.6.15 devel [09:23] otherwise i will capture the OOPS and bitch a bit the parisc guys :) [09:23] what's the main cause of the oops? [09:24] NULL deref, or something else? [09:24] the scsi driver load and it calls some pci_enable_ [09:24] i would need to read it again [09:24] too long to remember [09:24] pci_enable_device()? [09:25] if it crashes there, then that's pretty damn low level [09:25] ok.. let me boot it up again [09:29] DOH! [09:29] it just booted [09:29] WTF [09:29] 32-bit? [09:29] yes [09:29] weird [09:30] hmmm [09:30] ahhh i see [09:30] it did boot -8- before [09:30] and -9 now [09:31] oh well [09:31] fuck === fabbione tries again [09:36] baldios login: module ipv6 relocation of symbol xfrm_state_alloc is out of rang) [11:21] fabbione: looks like binutils problem there === rikai [n=gtk2@pool-64-222-225-195.port.east.verizon.net] has joined #ubuntu-kernel