| Summary: | net-wireless/madwifi-ng-0.9.4 unresolved symbols if CONFIG_MODVERSIONS is set | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Pacho Ramos <pacho> |
| Component: | Current packages | Assignee: | Steev Klimaszewski (RETIRED) <steev> |
| Status: | RESOLVED FIXED | ||
| Severity: | minor | CC: | flaschenkopp, gengor, ikelos, jgascon, mmokrejs, pacho, patrizio.bassi, pva, wschlich |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | 237618 | ||
| Bug Blocks: | 232070 | ||
| Attachments: |
madwifi-ng-0.9.4-r1.ebuild.patch
kernel configuration madwifi-ng-0.9.4-symvers.patch |
||
|
Description
Pacho Ramos
2008-07-17 18:24:58 UTC
Same here :( Upstream bug tracker appears to have a patch: http://madwifi.org/ticket/1935 (In reply to comment #2) > Upstream bug tracker appears to have a patch: > > http://madwifi.org/ticket/1935 > Same bug here. It seems to be that upstream did fix it in their cvs, and this ticket here seems to be more recent/appropriate: http://madwifi.org/changeset/3696 At bottom of this link is an unified diff for download, I'll try and patch and will see what happens... Created attachment 160795 [details, diff]
madwifi-ng-0.9.4-r1.ebuild.patch
patch for ebuild to include the linux-2.6.26 patch.
also in my personal overlay i have removed sharutils from depend,
please someone check if this is needed or not.
Applied the patch from http://madwifi.org/changeset/3696?format=diff&new=3696 and now it compiles fine under linux-2.6.26. emerge --info Portage 2.2_rc1 (default/linux/amd64/2008.0/desktop, gcc-4.3.1, glibc-2.8_p20080602-r0, 2.6.26-gentoo x86_64) ================================================================= System uname: Linux-2.6.26-gentoo-x86_64-AMD_Athlon-tm-_X2_Dual_Core_Processor_BE-2400-with-glibc2.2.5 Timestamp of tree: Fri, 18 Jul 2008 20:00:01 +0000 app-shells/bash: 3.2_p33 dev-lang/python: 2.5.2-r5 dev-python/pycrypto: 2.0.1-r6 sys-apps/baselayout: 2.0.0 sys-apps/openrc: 0.2.5 sys-apps/sandbox: 1.2.18.1-r3 sys-devel/autoconf: 2.13, 2.62-r1 sys-devel/automake: 1.5, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10.1-r1 sys-devel/binutils: 2.18-r3 sys-devel/gcc-config: 1.4.0-r4 sys-devel/libtool: 1.5.26 virtual/os-headers: 2.6.25-r4 ACCEPT_KEYWORDS="amd64" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-march=native -O2 -pipe" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc" CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/revdep-rebuild /etc/splash /etc/terminfo /etc/udev/rules.d" CXXFLAGS="-march=native -O2 -pipe" DISTDIR="/var/tmp/distfiles" FEATURES="distlocks noinfo parallel-fetch preserve-libs sandbox sfperms strict unmerge-orphans userfetch userpriv usersandbox" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo" LANG="de_DE.UTF-8" LC_ALL="de_DE.UTF-8" LDFLAGS="-Wl,-O1,--hash-style=gnu,--sort-common,--as-needed" LINGUAS="de" MAKEOPTS="-j3" PKGDIR="/var/tmp/packages" 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="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage /usr/local/overlay/hugin /usr/local/layman/desktop-effects" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="3dnow 3dnowext X aac acpi alsa amd64 berkdb branding bzip2 cairo cdr cli cracklib crypt cups dbus dga djvu dri dvd dvdr dvdread emboss encode evo expat fam firefox flac fortran gdbm gif gmp gnome gstreamer gtk hal iconv isdnlog jpeg lcms libnotify mad midi mikmod mmx mmxext mp3 mpeg mudflap multilib ncurses nls nptl nptlonly ogg opengl openmp pcre pdf perl png ppds pppd python qt3support quicktime readline reflection sdl session spl sse sse2 ssl ssse3 startup-notification svg threads tiff truetype unicode vorbis xcb xml xorg xv zlib" ALSA_CARDS="hda-intel" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions alias auth_basic authn_alias authn_anon 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 deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" ELIBC="glibc" INPUT_DEVICES="keyboard mouse evdev" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="de" USERLAND="GNU" VIDEO_CARDS="vesa nv nvidia" Unset: CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS But why do you need this drive with 2.6.26 kernel? That kernel has ath5k driver which is supposed to be better and will be the only supported driver in future. I'm interested are there any regressions in ath5k comparing to madwifi that affects you so you can't use ath5k instead of madwifi? *** Bug 232342 has been marked as a duplicate of this bug. *** I haven't recently verified this, but for the 2.6.25 kernel, the ath5k driver was still not able to enter master mode (AP mode for madwifi-ng). I'm also not certain of the state of injection on the ath5k (whereas madwifi-ng's injection capabilities are well tested). As such the madwifi-ng driver is still invaluable in certain situations... 5:) i tried to use the ath5k but seems totally unstable. i can just ping some machines but no bigger activities. on my configuration it's totally broken and wpa doesn't work at all... patch applies but during linking some object are undefined and... ath_pci: no symbol version for ath_hal_detach ath_pci: Unknown symbol ath_hal_detach ath_pci: no symbol version for ieee80211_cipher_none ath_pci: Unknown symbol ieee80211_cipher_none ath_pci: no symbol version for ieee80211_note ath_pci: Unknown symbol ieee80211_note ath_pci: no symbol version for ieee80211_crypto_delkey ath_pci: Unknown symbol ieee80211_crypto_delkey ath_pci: no symbol version for ieee80211_media_change ath_pci: Unknown symbol ieee80211_media_change ath_pci: no symbol version for ieee80211_beacon_miss ath_pci: Unknown symbol ieee80211_beacon_miss ath_pci: no symbol version for ieee80211_beacon_alloc ath_pci: Unknown symbol ieee80211_beacon_alloc ath_pci: no symbol version for ieee80211_getcfframe ath_pci: Unknown symbol ieee80211_getcfframe ath_pci: no symbol version for ieee80211_iterate_nodes ath_pci: Unknown symbol ieee80211_iterate_nodes ath_pci: no symbol version for ieee80211_vap_attach ath_pci: Unknown symbol ieee80211_vap_attach ath_pci: no symbol version for ieee80211_ibss_merge ath_pci: Unknown symbol ieee80211_ibss_merge ath_pci: no symbol version for ath_hal_probe ath_pci: Unknown symbol ath_hal_probe ath_pci: no symbol version for ieee80211_rate_attach ath_pci: Unknown symbol ieee80211_rate_attach ath_pci: no symbol version for ieee80211_rate_detach ath_pci: Unknown symbol ieee80211_rate_detach ath_pci: no symbol version for ieee80211_send_qosnulldata ath_pci: Unknown symbol ieee80211_send_qosnulldata ath_pci: no symbol version for ieee80211_create_vap ath_pci: Unknown symbol ieee80211_create_vap ath_pci: no symbol version for ieee80211_wme_acnames ath_pci: Unknown symbol ieee80211_wme_acnames ath_pci: no symbol version for ieee80211_input_all ath_pci: Unknown symbol ieee80211_input_all ath_pci: no symbol version for ieee80211_media_status ath_pci: Unknown symbol ieee80211_media_status ath_pci: no symbol version for ieee80211_start_running ath_pci: Unknown symbol ieee80211_start_running ath_pci: no symbol version for ieee80211_vap_detach ath_pci: Unknown symbol ieee80211_vap_detach ath_pci: no symbol version for ieee80211_announce ath_pci: Unknown symbol ieee80211_announce ath_pci: no symbol version for ieee80211_mark_dfs ath_pci: Unknown symbol ieee80211_mark_dfs ath_pci: no symbol version for ieee80211_chan2ieee ath_pci: Unknown symbol ieee80211_chan2ieee ath_pci: no symbol version for ieee80211_state_name ath_pci: Unknown symbol ieee80211_state_name ath_pci: no symbol version for ath_hal_init_channels ath_pci: Unknown symbol ath_hal_init_channels ath_pci: no symbol version for ieee80211_dturbo_switch ath_pci: Unknown symbol ieee80211_dturbo_switch ath_pci: no symbol version for ieee80211_crypto_encap ath_pci: Unknown symbol ieee80211_crypto_encap ath_pci: no symbol version for ieee80211_chan2mode ath_pci: Unknown symbol ieee80211_chan2mode ath_pci: no symbol version for ieee80211_getrssi ath_pci: Unknown symbol ieee80211_getrssi ath_pci: no symbol version for ieee80211_find_txnode ath_pci: Unknown symbol ieee80211_find_txnode ath_pci: no symbol version for ath_hal_getwirelessmodes ath_pci: Unknown symbol ath_hal_getwirelessmodes so it's still unusable Committed the patch from upstream, didn't realize this bug existed (was conversating with Flameeyes on IRC) - this should be fixed now. Please test (no rev-bump) Yep here same problem with ath_pci :/ Created attachment 161061 [details]
kernel configuration
I get the same error using the patch from [1], which suggests it's an issue with some changes that went into 2.6.26. I've also noticed some problems with other modules when using modprobe -f (invalid format, although without -f they load just fine), so I think their symbol versioning must've had an overhaul. [2] seems to suggest there's been some changes around there also... [1] http://madwifi.org/attachment/ticket/1935/madwifi-dfs-ieee80211-skb-update.patch [2] http://www.linuxhq.com/kernel/v2.6/26/kernel/module.c Please try turning off module versioning - I'd asked Diego to, but he was tired of compiling his kernel for the day and didn't get a chance to do it. Please let me know if that does fix it - if so, then we have a place to go - I don't have module versioning turned on and everything compiles and works fine here, but I'd like to know for sure that this could possibly be the cause. If I remember correctly, this happened for me before and its why I ended up turning off versioning in the first place. It would have to be something in there though as ieee80211 itself hasn't changed at all in the kernel Hiya Steev, Yep, that got rid of the dmesg errors. I'm afraid I can't fully confirm it all works (left the PCMCIA card at work), but previously modprobe ath_pci caused the errors and with module versioning turned off there were no errors. Seems a bit of a shame, it had been working fine up until this kernel even with module versioning. Anyway, hope that helps, let me know if there's any other tests I can run... 5:) (In reply to comment #6) > But why do you need this drive with 2.6.26 kernel? That kernel has ath5k driver > which is supposed to be better and will be the only supported driver in future. > I'm interested are there any regressions in ath5k comparing to madwifi that > affects you so you can't use ath5k instead of madwifi? > here last test using ath5k created an device but i wasn't able to use it. chipset is: 01:00.0 Ethernet controller: Atheros Communications, Inc. AR5006EG 802.11 b/g Wireless PCI Express Adapter (rev 01) so only chance is to use madwifi at the moment. (In reply to comment #11) > Committed the patch from upstream, didn't realize this bug existed (was > conversating with Flameeyes on IRC) - this should be fixed now. Please test > (no rev-bump) > I've rebuild the madwifi-ng-0.9.4 with your committed patch, and here it works good with the kernel-2.6.26. Ealier I'd took the patch from comment #3 which also worked. My emerge --info is in comment #5. (In reply to comment #6) >But why do you need this drive with 2.6.26 kernel? That kernel has ath5k driver >which is supposed to be better and will be the only supported driver in future. >I'm interested are there any regressions in ath5k comparing to madwifi that >affects you so you can't use ath5k instead of madwifi? Sure, I went ahead and gave the new and shiny ath5k a try (again, like when the 2.6.25 arrived), and it even is able to create the device, but then I've found out that it is still not ready for use with my WPA2 a.k.a wpa_supplicant. So I am looking forward for the further development of the ath5k. But for now, users want to have something that they know of if it is working or not, and that is still the madwifi-ng modules. Please can someone explain to me for what the DEPEND="app-arch/sharutils" is needed? I see no use of that, maybe I'd overlook something? it compiles but still problems with undefined versions and i don't want to disable versioning in 2.6.26 (in 2.6.25 works like a charm) (In reply to comment #10) > patch applies but during linking some object are undefined and... > ath_pci: no symbol version for ath_hal_detach > ath_pci: Unknown symbol ath_hal_detach I hit the same problem. You have to load a kernel driver yourself before loading ath_pci. It seems to me like broken dependency in modules so that modprobe doesn't load the deps first. That's before you have the following in your .config: # # Wireless # CONFIG_CFG80211=m CONFIG_NL80211=y CONFIG_WIRELESS_EXT=y CONFIG_MAC80211=m are u aware of any progress here? Currently going through some things in real life - I haven't forgotten about this at all, I am looking into it still, just, real life has really been kicking me about lately. I promise I will get to the bottom of this as soon as I can. (In reply to comment #19) > Please can someone explain to me for what the DEPEND="app-arch/sharutils" > is needed? Seems that previously it was used for uudecode program, but now uudecode is a part of madwifi-ng sources so... DEPEND is dropped. Thank you. Another workaround for this bug is to enable CONFIG_MODULE_FORCE_LOAD (together with CONFIG_MODVERSIONS). It's good idea to enable that anyway and makes this issue less critical. Created attachment 165387 [details, diff] madwifi-ng-0.9.4-symvers.patch This patch (together with patch for kernel from bug #237616) fixes this issue for me. We have to wait for kernel bug to be closed before applying this one. Well and patch above just workaround problems in our build system. Seems that real problem was reported in bug 237618... But let's wait for comments of our kernel team. I've added workaround in madwifi-ng-0.9.4-r1 and now it should work. But, please, keep this bug open, until kernel team handles this situation. unfortunatly bug 242238 blocks my test on .27 http://madwifi.org/wiki/UserDocs/KernelConfig check the kernel configuration Can someone please confirm when the messages in comment #10 appear (compile time?) and if they actually prevent something from happening, or if they are just spamful warnings? (In reply to comment #31) > Can someone please confirm when the messages in comment #10 appear (compile > time?) and if they actually prevent something from happening, or if they are > just spamful warnings? Just did that. If I build this modules without following hack: # Workaround our build system, bug #232099 (bug #237618 describes details) touch Module.symvers for ms in ath net80211 ath_hal ath_rate/{amrr,minstrel,onoe,sample}; do ln -s "${S}/Module.symvers" "${ms}" done which was added in madwifi-ng-0.9.4-r1 attempt to load module fails: camobap ~ # modprobe ath_pci FATAL: Error inserting ath_pci (/lib/modules/2.6.26-openvz.git-35f41f1/net/ath_pci.ko): Unknown symbol in module, or unknown parameter (see dmesg) camobap and in dmesg I have: ath_hal: 0.9.18.0 (AR5210, AR5211, AR5212, RF5111, RF5112, RF2413, RF5413) wlan: 0.9.4 ath_pci: no symbol version for ieee80211_encap ath_pci: Unknown symbol ieee80211_encap ath_pci: no symbol version for ieee80211_input ath_pci: Unknown symbol ieee80211_input ath_pci: no symbol version for ieee80211_ifattach ath_pci: Unknown symbol ieee80211_ifattach ath_pci: no symbol version for _ath_hal_attach ath_pci: Unknown symbol _ath_hal_attach ath_pci: no symbol version for ieee80211_beacon_update [and so on] So no, this is not a cosmetic issue. when I do "modprobe ieee80211" FATAL: Error inserting ieee80211 /lib/modules/2.6.25-gentoo-r8/kernel/net/ieee80211/ieee80211.ko: Unknown symbol in module, or unknown parameter (see dmesg) dmesg shows: ieee80211_crypt: registered algorithm 'NULL' iee80211: Unknown symbol request_module Higher up in dmesg, when the system was booting it shows this: wlan: Unknown symbol request_module and then: ath_pci: Unknown symbol ieee80211_check_mic ath_pci: Unknown symbol ieee80211_encap ... ... ... and so on.. This problem appears when updating from 2.6.25-r7 to 2.6.25-r8. I still depend on the madwifi driver since I have an Atheros AR5007EG, which has no range with the ath5k driver. More people with this problem: http://www.linuxquestions.org/questions/linux-wireless-networking-41/madwifi-doesnt-like-to-work-problem-might-start-at-emerge.-679865/ Leander, you pointed to the thread with different solutions presented there. Have you tried them? Have you double checked that everything required is enabled in the kernel? If the answer is yes but the problem persists, please, open *new* bug. This bug documents different problem. This bug was long time ago fixed both in stable and unstable branches. Closing. |