i, Since I upgraded to 2.6.27-r7 OR 2.6.27-r8 (from 2.6.26-r4) I don't have neither net, nor usb, nor sound. For net in particular, when I bring up the interface I get an error something like this: SIOCCFLAGS ERROR: Device or ressource busy. Looking in dmesg, I've seen that after my ide2 takes IRQ11, all the modules for my network card (8139too), usb (uhci-hcd) and sound (snd-cs4236) complain that they can't assign an interrupt (IRQ11). I've included all files relevant to my computer in the bzip2 tar file attached (kernel config, dmesg, lspci, /proc/{interrupts,ioports,iomem,cpuinfo,irq/}. I've also tried to start with kernel parametters irqpoll and irqfixup, thinking it might help. It still didn't work, and I included the dmesg obtained in the attach. If anyone could give me a hint about what's going on, and how I can fix it, I'd be very grateful. Thanks for any help! Reproducible: Always Steps to Reproduce: 1. Turn computer on. 2. Check network interface 3. Check sound 4. Check usb. 5. Consult errors in dmesg, related to IRQ 11. Actual Results: Network interface doesn't come up, sound doesn't work, usb doesn't work. Expected Results: Network should be up, sound should work, usb should work. Portage 2.1.6.4 (default/linux/x86/2008.0, gcc-4.1.2, glibc-2.6.1-r0, 2.6.26-gentoo-r4 i586) ================================================================= System uname: Linux-2.6.26-gentoo-r4-i586-Pentium_MMX-with-glibc2.0 Timestamp of tree: Mon, 26 Jan 2009 13:35:01 +0000 app-shells/bash: 3.2_p39 dev-lang/python: 2.4.4-r13, 2.5.2-r7 dev-python/pycrypto: 2.0.1-r6 dev-util/cmake: 2.4.6-r1 sys-apps/baselayout: 1.12.11.1 sys-apps/sandbox: 1.2.18.1-r2 sys-devel/autoconf: 2.13, 2.63 sys-devel/automake: 1.4_p6, 1.5, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10.2 sys-devel/binutils: 2.18-r3 sys-devel/gcc-config: 1.4.0-r4 sys-devel/libtool: 1.5.26 virtual/os-headers: 2.6.27-r2 ACCEPT_KEYWORDS="x86" CBUILD="i586-pc-linux-gnu" CFLAGS="-O3 -march=pentium-mmx -pipe -fomit-frame-pointer" CHOST="i586-pc-linux-gnu" CONFIG_PROTECT="/etc" CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/revdep-rebuild /etc/splash /etc/terminfo /etc/udev/rules.d" CXXFLAGS="-O3 -march=pentium-mmx -pipe -fomit-frame-pointer" DISTDIR="/usr/portage/distfiles" FEATURES="clean collision-protect distlocks fixpackages parallel-fetch protect-owned sandbox sfperms strict unmerge-orphans userfetch" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo" LDFLAGS="-Wl,-O1" MAKEOPTS="-j1" PKGDIR="/usr/portage/packages" PORTAGE_COMPRESS="bzip2" PORTAGE_COMPRESS_FLAGS="-9" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/home/portage_tmpdir" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="16bit 16bittmp 7zip a52 aac aalib acl ads aim aimextras aio alias aliaschain alisp alsa amd amr amrnb amrr amrwb amuled ao aotuv apache2 apm ares asterisk astribank async asyncns audiofile auto-completion autoipd automount avahi bash-completion batch bcmath bcp beep berkdb bind-mysql bittorrent bjam bl bonjour branding bzip2 c++ cap caps capslib catalogs cdb cdda cddb cdio cdparanoia cdr cdrom cgi clamav clamd cluster clvm cman compat console crypt css cups cupsddk curl dbus deprecated device-mapper dhcp directfb discouraged disk-partition dmi dnsdb dri dts dv dvb dvbplayer dvbsetup dvd dvdnav dvdr dvdread dvi dynamic dynamicplugin ecaggressive eds elisp emacs embedded emboss enca encode eolconv epson erandom evo exif expat extensions extra-algorithms extraengine extras faillog fam fame fasttrack fat fax fbcon fbcondecor fbdev fbsplash festival ffmpeg fftw finger flac flash florz fluidsynth fontconfig foomaticdb fortran fr ftp fuse gadu gcj gd gdbm ggi gif gimpprint glibc-compat20 glibc-omitfp gnutella gnutls gopher gphoto2 gpm gs gsm gstreamer guile gzip-el h26x h323 hal hardware-carrier hddtemp howl-compat hpn http httpd iax iconv icq id3 id3tag idn ieee1394 ilbc imlib injection ipfilter iproute2 ipsec ipv6 irc isc isdnlog jabber jack jingle jpeg jpeg2k justify kate kerberos l7filter ladspa lame ldap ldapsam ldirectord libass libcaca libclamav libgcrypt libsamplerate libssh2 libsysfs libv4l2 live livebuffer lj login-watch logrotate logwatch lowmem lzo mad mail mailbox maildir maildrop masquerade mbox md5sum mdnsresponder mdnsresponder-compat meanwhile memlimit messages midi mikmod milter mime mixer mjpeg mmap mmx mmxext mng modplug mouse mp2 mp3 mpeg mpi mplayer msn msnextras multicall multislot multiuser musepack mysql mysqli nat ncurses nemesi net nethack netjack network network-cron networking networkmanager new-login nfconntrack nfqueue nfs nis nls nntp nptl nss ntfs ntp nvram oav objc objc++ objc-gc ogg ogg123 ogm openal openft openmp openntpd osp oss otr pam parport parse-clocks passwordsave pcre pdf perl png pnm portaudio posix ppds pppd pri ps pulseaudio pvr python quicktime quotas radio radius rar rdesktop readline realmedia recode regex reiser4 reiserfs remote remoteosd replaygain resolvconf rrdcgi rrdtool rss rtc rtsp samba sasl scanner screen sctp search-screen sendmail sensord serial session sftp sftplogging sguil sharedmem shout simplexml skey slang slp sndfile snmp sockets socks5 soundex speex spell sqlite srt ssl startup-notification stream subtitles subversion suid suidcheck svga svnserve swat symlink sysfs syslog sysvipc szip tcpd teletext tftp tga tgif theora threads tiff timezone timidity tivo truetype twolame unicode unzip upnp urandom usb v4l v4l2 vcd vcdinfo vcdx vhosts video vim vim-pager vim-syntax virus-scan vlm vorbis wav wavpack weak-algorithms wifi win32codecs winbind woomera wordexp x264 x86 xerces-c xfs xinetd xml xmlreader xmlrpc xmlwriter xsl xslt xvid yahoo yp zapnet zapras zeroconf zip zlib" ALSA_CARDS="snd-cs4236" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mmap_emul mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions alias auth_basic auth_digest authn_anon authn_dbd authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache dav dav_fs dav_lock dbd deflate dir disk_cache env expires ext_filter file_cache filter headers ident imagemap include info log_config logio mem_cache mime mime_magic negotiation proxy proxy_ajp proxy_balancer proxy_connect proxy_http rewrite setenvif so speling status unique_id userdir usertrack vhost_alias" ELIBC="glibc" INPUT_DEVICES="keyboard mouse ps2mouse evdev linuxinput" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" USERLAND="GNU" VIDEO_CARDS="fbdev vesa v4l vga mach64" Unset: CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, FFLAGS, INSTALL_MASK, LANG, LC_ALL, LINGUAS, PORTAGE_RSYNC_EXTRA_OPTS
Created attachment 179895 [details] bzip2 tar of dmesg, config, lspci, /proc/{cpuinfo,interrupts,ioports,iomem,irq/} all info about the system
So it's the piix driver for the "IDE interface: Intel Corporation 82371SB PIIX3 IDE [Natoma/Triton II]" that hogs IRQ 11?
From what I see i8042's aux port grabs IRQ 12 at: 'serio: i8042 AUX port at 0x60,0x64 irq 12' After that ide2 grabs IRQ 11 at: 'ide2 at 0x1e8-0x1ef,0x3ee on irq 11' The first problem occurs when ide5 tries to grab IRQ 12 which is already taken by i8042, at: 'ide5: disabled, unable to get IRQ 12' Later 8139too tries to grab IRQ 11 which is already taken by ide2 at: 'eth0: RealTek RTL8139 at 0x1300, 00:50:ba:8d:00:6f, IRQ 11' Check out: http://www.pcguide.com/ref/mbsys/res/irq/numIRQ11-c.html http://www.pcguide.com/ref/mbsys/res/irq/numIRQ12-c.html and try checking out if any BIOS feature of yours helps out. Also, you can try booting with .i8042.noaux and it shouldn't hog IRQ 12 (and will let ide5 get it) to see what happens.
Hi, I tried to boot with i8042.noaux parameter. No error shows up in dmesg for the network module (as far as I can see), but the network still doesn't work, I get the same "Device or ressource busy" when I try to bring it up. For the usb, uhci-hcd gets errors at loading up (as per dmesg), so I didn't even try it. I'm attaching another file bellow (kernel-2.6.27-r8-i8042.noaux.tar.bz2) with the dmesg and some other /proc/* made while booting with 2.6.27-r8 and with i8042.noaux. Thanks for any help. PS: I've looked in the BIOS, but I didn't see anything relevant. I had no problem with this computer until 2.6.27-r7. Was working perfect with 2.6.26-r4, for example.
Created attachment 180414 [details] bzip2-ed tar of dmesg,/proc/{iomem,ioports,interrupts} for 2.6.27-r8 with i8042.noaux bzip2-ed tar of dmesg,/proc/{iomem,ioports,interrupts} for 2.6.27-r8 with i8042.noaux
(In reply to comment #4) > Hi, > I tried to boot with i8042.noaux parameter. > No error shows up in dmesg for the network module (as far as I can see), but > the network still doesn't work, I get the same "Device or ressource busy" when > I try to bring it up. > For the usb, uhci-hcd gets errors at loading up (as per dmesg), so I didn't > even try it. > > I'm attaching another file bellow (kernel-2.6.27-r8-i8042.noaux.tar.bz2) with > the dmesg and some other /proc/* made while booting with 2.6.27-r8 and with > i8042.noaux. > > Thanks for any help. > PS: I've looked in the BIOS, but I didn't see anything relevant. I had no > problem with this computer until 2.6.27-r7. Was working perfect with 2.6.26-r4, > for example. > Looking at the .noaux dmesg, I can see: 'ide5 at 0x160-0x167,0x366 on irq 12' which means that since i8042 didn't grab an IRQ for it's AUX interface (that's what .noaux was there for), ide5 managed to grab IRQ 12. There are no 8139too module error messages in the latest dmesg, because from what I see, it doesn't spit out errors regarding it's IRQs (I think that there were no errors in -.noaux dmesg with irqpoll and irqfixup enabled either). After further digging in the code, I found out that even tho i8042's AUX interrupt line (12) is flagged as shareable the ide5 one is not. Also, ide2's interrupt line is not shareable and that's why 8139too fails registering it's interrupt handler. I'll check out the related code changes between 2.6.26-r4 and 2.6.27-r7 soon, and try to come up with a fix. In the meantime, could you post the dmesg output (with irqpoll and irqfixup enabled) of a 2.6.26-r4 kernel? Thanks :)
Created attachment 180467 [details] dmesg, /proc/{iomem,ioports,interrupts,irq/} for 2.6.26-r4 with parameters irqpoll irqfixup dmesg,/proc/{iomem,ioports,interrupts,irq/} for 2.6.26-r4 with parametters irqpoll irqfixup
Please don't attach tarballs.. that just means that anyone who wants to help has to take an extra step. Please attach the files individually as text.
(In reply to comment #7) > Created an attachment (id=180467) [edit] > dmesg, /proc/{iomem,ioports,interrupts,irq/} for 2.6.26-r4 with parameters > irqpoll irqfixup > > > dmesg,/proc/{iomem,ioports,interrupts,irq/} for 2.6.26-r4 with parametters > irqpoll irqfixup > I'm wondering why the 2.6.27 kernel is trying to register 3 more ide interfaces than 2.6.26 does (ide2, ide3, ide4). Unfortunately, I'm not near a Linux box, so I'll be checking it out later today. PS: If you have nothing better to do, could you also give us the 'lspci -vvv' output of a 2.6.26 kernel?
Created attachment 180691 [details] lspci -vvv (on 2.6.26-r4) lspci -vvv (on 2.6.26-r4)
(In reply to comment #9) > I'm wondering why the 2.6.27 kernel is trying to register 3 more ide interfaces > than 2.6.26 does (ide2, ide3, ide4). > Unfortunately, I'm not near a Linux box, so I'll be checking it out later > today. > > PS: If you have nothing better to do, could you also give us the 'lspci -vvv' > output of a 2.6.26 kernel? You'r right, I only have 2 ide interfaces (3 harddisks and a cdrom). I don't know what ide{2,3,4,5} are, and why they grab an irq.
Created attachment 180697 [details] config file for kernel 2.6.26-r4
Created attachment 180698 [details] config file for kernel 2.6.27-r8
(In reply to comment #11) > (In reply to comment #9) > > I'm wondering why the 2.6.27 kernel is trying to register 3 more ide interfaces > > than 2.6.26 does (ide2, ide3, ide4). > > Unfortunately, I'm not near a Linux box, so I'll be checking it out later > > today. > > > > PS: If you have nothing better to do, could you also give us the 'lspci -vvv' > > output of a 2.6.26 kernel? > > You'r right, I only have 2 ide interfaces (3 harddisks and a cdrom). > I don't know what ide{2,3,4,5} are, and why they grab an irq. > I'm still not near a Linux box so I can't check out the code changes, but: i) Try using the ideX=noprobe parameter which I'm unsure whether it is actually removed from the kernel or not. (For example, ide2=noprobe) ii) Play around with ide_core's .noprobe boot options which is documented in Documentation/ide/ide.txt and try disabling the extra IDE interfaces and see how it goes. I'll try to find out what the ide{2,3,4,5} interfaces when I get myself near a Linux box.
So, I was looking at your .configs and their diffs and I noticed CONFIG_IDE_ATAPI in 2.6.27. Seems like CONFIG_IDE_ATAPI was introduced in 2.6.27 and that triggered me in taking a deeper look in your kernel configuration. What I noticed is that you have both old and deprecated CONFIG_IDE and the newer libata (CONFIG_ATA) enabled. I would like you to disable everything under the 'ATA/ATAPI/MFM/RLL support' menu (or well CONFIG_IDE*) and play solely with 'Serial ATA (prod) and Parallel ATA (experimental) drivers' (or well CONFIG_ATA*) which is the new and fresh blood as far as ATA/ATAPI/IDE stuff are concerned. You will also have to change your GRUB config and your fstab, because using libata is gonna change your hd* drivers to sd*. I recommend labeling your drives or using their UUID values in fstab/menu.lst, to avoid guessing their sd* names. Tutorials about the above operation are all over the internets. The mix of CONFIG_IDE* stuff and libata is the source of many conflicts and problems, and I would like to get this out of the way (let alone that the insertion of CONFIG_IDE_ATAPI in 2.6.27 may be the source of your problems). Thanks :)
(In reply to comment #15) > So, I was looking at your .configs and their diffs and I noticed > CONFIG_IDE_ATAPI in 2.6.27. Seems like CONFIG_IDE_ATAPI was introduced in > 2.6.27 and that triggered me in taking a deeper look in your kernel > configuration. > > What I noticed is that you have both old and deprecated CONFIG_IDE and the > newer libata (CONFIG_ATA) enabled. > I would like you to disable everything under the 'ATA/ATAPI/MFM/RLL support' > menu (or well CONFIG_IDE*) and play solely with 'Serial ATA (prod) and Parallel > ATA (experimental) drivers' (or well CONFIG_ATA*) which is the new and fresh > blood as far as ATA/ATAPI/IDE stuff are concerned. You will also have to change > your GRUB config and your fstab, because using libata is gonna change your hd* > drivers to sd*. I recommend labeling your drives or using their UUID values in > fstab/menu.lst, to avoid guessing their sd* names. Tutorials about the above > operation are all over the internets. > > The mix of CONFIG_IDE* stuff and libata is the source of many conflicts and > problems, and I would like to get this out of the way (let alone that the > insertion of CONFIG_IDE_ATAPI in 2.6.27 may be the source of your problems). > > Thanks :) > Hi George, Ok, I removed all the CONFIG_IDE* stuff from the kernel and I've started recompiling it. It will take a while to finish though, 1-2 days (my computer is not very fast). Could you please let me know what modules I need for the harddisks in this case? I usually don't have anything compiled in the kernel, and I have an init file in the initramfs that loads all the modules in the early boot stage. However, I need to know what modules I should put inside the initramfs file, because I have very little memory (64M). I can not include all the modules because if I do I can't load the initramfs in the memory, and I get stuck. So I need to know precisely what I should include and then load them up. I knew what modules I needed when I was using IDE. But now, using the SATA+PATA I don't. Could you tell me that, based on the lspci file ? Thanks!
> Hi George, > Ok, I removed all the CONFIG_IDE* stuff from the kernel and I've started > recompiling it. It will take a while to finish though, 1-2 days (my computer is > not very fast). > > Could you please let me know what modules I need for the harddisks in this > case? > I usually don't have anything compiled in the kernel, and I have an init file > in the initramfs that loads all the modules in the early boot stage. However, I > need to know what modules I should put inside the initramfs file, because I > have very little memory (64M). I can not include all the modules because if I > do I can't load the initramfs in the memory, and I get stuck. So I need to know > precisely what I should include and then load them up. > I knew what modules I needed when I was using IDE. But now, using the SATA+PATA > I don't. Could you tell me that, based on the lspci file ? > > Thanks! > The module you need for your hard disks is called 'ata_piix' :)
> The module you need for your hard disks is called 'ata_piix' :) ok, that's instead of the piix.ko that I had when using IDE. But before I had to load some modules related to the ide: ide-core, ide-disk, ide-generic. I don't know much about this...but what do I have now if I don't have IDE? What do I need now instead of the modules above? Thanks for your help :)
(In reply to comment #18) > > The module you need for your hard disks is called 'ata_piix' :) > > ok, that's instead of the piix.ko that I had when using IDE. But before I had > to load some modules related to the ide: ide-core, ide-disk, ide-generic. > I don't know much about this...but what do I have now if I don't have IDE? What > do I need now instead of the modules above? > > Thanks for your help :) > Okay, let's not mess with module loading through initramfs and compile all the required parts into the kernel. I noticed that you have everything compiled as modules and that not only is quite pointless but also contributes to the huge compilation time. What I want you to do is enable the following options: CONFIG_ATA_GENERIC CONFIG_SATA_AHCI CONFIG_ATA_SFF CONFIG_PATA_MPIIX CONFIG_PATA_OLDPIIX CONFIG_BLK_DEV_SD CONFIG_BLK_DEV_SR built into the kernel. The rest from 'Serial ATA (prod) and Parallel ATA (experimental) drivers' and 'SCSI device support' can be disabled. Also, since you have a slow machine, try optimizing your kernel for the stuff you really need instead of compiling essentially everything. Anyway, I find the above solution of compiling only the nessecary stuff into the kernel much better and cleaner than module loading through irfs.
Hi George, I confirm that now is working perfectly. The network is up, all right, I've checked the usb too and the sound as well. All up and running, problem solved. Thanks for you help! Much appreciated! PS: I have some reason for which I keep everything modularised in the kernel and everything compiled. Sometimes I add some device or another, and I don't like to wait on the spot for it to compile the driver in that particular moment. Whenever I need any module of the kernel, it is already there. On the other hand, I keep the computer on all the time. So I don't really mind it takes 2 days to compile the kernel -- I do it only when there is a release of the kernel which is marked stable by gentoo, so that's not very often.
