Please find attached ltmodem.tgz containing the new package net-dialup/ltmodem which provides the Lucent/Agere LT LinModem. Quoting from the package's original CHANGELOG: GPL'ed except for the binary module which is of unknown status. which is why I haven't filled in the LICENSE variable. The package works on my Dell Inspiron 8100 equipped with the Mini PCI Modem/Ethernet Combo and is the only way I can connect from home.
Created attachment 3840 [details] net-dialup/ltmodem Lucent/Agere LT LinModem
Hey Denys, could you bump that to the newest version please? -phoen][x-
of course, I'll attach my newest version. I wasn't sure whether I should bother pursuing this since someone checked in another ltmodem ebuild into the portage tree (I think mine does a rather better job, IMHO :-). I only mention x86; for any other architecture, someone else will have to try it and report on whether it works as expected.
Created attachment 6443 [details] net-dialup/ltmodem Lucent/Agere LT LinModem (version 8.26a9) as promised, here is the new version of my ltmodem ebuild. (by the way, I don't seem to be authorized to obsolete my own previous attachment).
Comment on attachment 3840 [details] net-dialup/ltmodem Lucent/Agere LT LinModem Obsolete. I shall add the new ebuild shortly. -phoen][x-
Created attachment 6933 [details] complete emerge log I can't emerge ltmodem-8.26_alpha9 & ltmodem-8.22_alpha5. I get the same error for both of them. Attached is the complete emerge log (via screen -L). I run the latest lolo-sources but had the same problem with vanilla-sources-2.4.20 as well.
Hi Manolis, thanks for your bug report (next time, though, give me a hint as to the format of your attachment - luckily "file" was able to guess it :-). I noticed that your compilation line is somehow missing all the command line options. It reads: gcc -c -o ltmodem.a ltmodem.c when it should be something like: gcc -D__KERNEL__ -I/lib/modules/2.4.19-gentoo-r9/build/include -Wall -Wstrict-prototypes -O3 -fomit-frame-pointer -fno-strict-aliasing -pipe -fno-strength-reduce -DMODULE -DLT_VERSION=8.26a9 -DLINUX -DNO_EEPROM -DLT_SERIAL_MODULE=lt_serial -c -o ltmodem.a ltmodem.c I don't understand how this happening as these options are explicitly put in by the Makefile. There is a line in Makefile.in that defines: CC = @CC@ $(KCFLAGS) I don't see how yours can mutate into just gcc. I would appreciate the following information: version info for GNU Make, your complete environment (env vars), make.conf and whatever. Also, invoke ebuild ltmodem-8.26_alpha9.ebuild unpack compile and send me the resulting Makefile in the source subdirectory. Thanks
Created attachment 6950 [details] ltmodem.tar.bz2 (make version,env vars,make.conf) Thanx for answering & sorry for not specifying the attachment's format. I included all the info u want on the ltmodem.tar.bz2 (I did right this time :) file I attached.
Hi Manolis, I have now discovered the cause of the problem you observed, but I am not entirely sure of the correct gentoo way of addressing it: the problem is caused by 2 unusual circumstances: 1. your environment exports CC=gcc CXX=g++ I don't understand why you are doing this, but this is surely not incorrect per se :-) 2. ltmodem's build_module invokes "make -e" which causes environment variables to override Makefile settings. This is also not incorrect, although extremely unusual (it should really only be invoked explicitly by an expert who actually knows what s/he is doing) I think, I'll remove the "-e" (using a patch) since gentoo is supposed to work properly without ever being aware of the details of makefiles. On the other hand, setting CC and CXX is probably an extremely bad idea when using gentoo. Cheers,
Here is the new ebuild fixing the "make -e" issue.
Created attachment 6959 [details] ltmodem.tgz fixing the "make -e" issue
Thanx for the new e-build. It works. About the CXX,CC vars now. I haven't set anything myself, they're included at /etc/env.d/05gcc.
> Thanx for the new e-build. It works. About the CXX,CC vars now. I haven't set > anything myself, they're included at /etc/env.d/05gcc. hmm, yes, I see. I have not upgraded to 3.2.1 yet and this setting of CC and CXX is a new thing. Thanks for bringing it to my attention.
I seem to be getting problems while compiling as well. I still get them when I try compiling by hand (./build_modules) but if i extract the sources.tar.gz and then run ./configure && make && make install it runs fine. Somebody else seems to have had that problem as well http://forums.gentoo.org/viewtopic.php?t=29528& maybe you want to change the ebuild? Another thing is that it would be nice that if the ebuild would automaticialy link /dev/modem to /dev/ttySLT0
are you using the ebuild that _I_ (denys) posted or those which have been checked into gentoo? I suspect the latter. Use mine. I don't understand why the gentoo devs choose to include an ebuild that doesn't quite do the job (e.g. /dev/modem issue). Maybe there is a reason, but no one is talking to me.
ok, it built succsesfully but my modem restarts whenever i try to connect using wvdial. I'm currently trying to resolve this on the linmodem mailing list but maybe you have ideas?
sorry, I exclusively use kppp which works like a charm. let me know what you find out.
hmm, kppp works fine, I guess this is a wvdial problem.
For those having problems with ltmodem - see bug 14528 and download the new ebuild. It has gentoo fixes, builds like a dream and will set devfsd to create links from /dev/modem to the ltmodem. Simply emerge this build and then point your kppp or wvdial to use /dev/modem. Note that a version of this ebuild is in portage but still has a bug because I can't get someone to update it. If you would like to use the portage versin do the following. Use this command line ACCEPT_KEYWORDS="~X86" emerge ltmodem Then this command line killall -HUP devfsd and all will be joyus!! If anyone reading this has CVS authority - please add the killall line to pkg_postinst()l per my comments in bug 14528.
*** Bug 14528 has been marked as a duplicate of this bug. ***
ok, lemme update everyone here ... i was asked sometime ago (in oct) to make an ltmodem ebuild ... this bug was being handled by phoenix and i really didnt know it existed so i made the current ebuild that is in Gentoo from scratch ... the reason the ebuild from Bug 14528 was added/looked at was because the bug was given to me ... phoenix has since gone inactive hence him not touching this bug (he needed to find a job, real life issues basically) ... ill toy with this version of the ltmodem ebuild and update the one in portage accordingly
ok, after reviewing the ebuilds you both have submitted, they do pretty much the same thing (although slightly different) putting a check in about the existence of devfs.d, although a complete solution, isn't truly needed since Gentoo is built off of devfs and sys-apps/baselayout provides /etc/devfs.d (i.e. it exists on everyone's system who uses Gentoo). the only outstanding issue is sending a HUP signal to the devfs daemon ... i talked to other -dev's and the general sentiment is 'no sending of signals automagically' ... so what we will now have is an einfo explaining that the user should send a kill signal if they wish ... and a few credit additions to the ChangeLog have i missed anything ?
in earlier gentoo devfs.d did not exist. It does now as per my suggestion in 8424. I wrote the ltmodem ebuild so as to accommodate both old and new baselayout. If that's no longer considered a concern, then sure the check can be removed. As for the HUP signal: I feel it is a bit sad to loose the fully transparent emerge. I really don't see what is gained: the user emerges ltmodem, is told to send HUP to devfsd and does so. s/he is just going to wonder why the ebuild couldn't bloody well do that last bit too :-) Perhaps this sort of issues could be addressed like with conf files, i.e. through explicit protection of selected system services e.g. using a variable DEMON_PROTECT set to a default in /etc/make.globals... just musing... thanks for everything. I am really glad that the ltmodem package is finally getting into portage in a reasonable shape.
My goal is simply a working ltmodem ebuild. Denys, I see now you had a working ebuild but I missed it (I especially like the slot idea!). Spanky, I agree that sending signal automagically is not the best way to do things. In reality the requirement to send the -HUP is because there is no "/etc/init.d/devfsd restart" to cause devfsd to reread it's config file. The fact that Gentoo is based on devfsd, but gentoo doesn't have the ability to control devfsd is probably a bug. In the current ebuild, I don't think the einfo is clear because most user don't know what devfs is - I think it should say something like You must reboot or "killall -HUP devfsd" to complete the installation. Then your LTModem will be at "/dev/modem" It seems others have used killall to send signals automagically - because sometimes there is no other way when dealing with daemons. #grep -H -R killall /usr/portage/* /usr/portage/app-admin/sysklogd/files/syslog-cron:/bin/killall -HUP syslogd /usr/portage/app-sci/foldingathome/files/folding-init.d: killall foldingathome /usr/portage/app-sci/setiathome/files/seti-init.d-r1: killall setiathome /usr/portage/app-text/ggv/ggv-1.1.94.ebuild: killall gconfd-1 2>/dev/null >/dev/null /usr/portage/app-text/tetex/files/teTeX-1.0.dif:+ killall -HUP rpc.mountd rpc.nfsd /usr/portage/dev-db/postgresql/files/postgres: killall -TERM ${EXE} /usr/portage/dev-util/devhelp/devhelp-0.4-r1.ebuild: killall gconfd-1 2>/dev/null >/dev/null /usr/portage/media-video/realone/realone-1.ebuild: killall ${INSTALL_BINARY} /usr/portage/net-analyzer/portsentry/files/portsentry.rc6: killall portsentry /usr/portage/net-analyzer/arpwatch/files/arpwatch.init: killall arpwatch > /dev/null /usr/portage/net-dialup/ltmodem/ltmodem-8.26_alpha9-r1.ebuild: einfo "killall -HUP devfsd" /usr/portage/net-fs/nfs-utils/files/nfs: killall -0 rpc.statd &>/dev/null && return 0 /usr/portage/net-fs/nfs-utils/files/nfs: killall -0 rpc.statd &>/dev/null || return 0 /usr/portage/net-fs/nfs-utils/files/nfsmount: killall -0 rpc.statd &>/dev/null && return 0 /usr/portage/net-fs/nfs-utils/files/nfsmount: killall -0 nfsd &>/dev/null && return 0 /usr/portage/net-fs/nfs-utils/files/nfsmount: killall -0 rpc.statd &>/dev/null || return 0 /usr/portage/net-fs/openafs/files/afs.rc:# killall inetd.afs /usr/portage/net-fs/openafs/files/afs.rc: killall -HUP bosserver /usr/portage/net-fs/openafs/files/afs.rc.rc6:# killall inetd.afs /usr/portage/net-fs/openafs/files/afs.rc.rc6: killall -HUP bosserver /usr/portage/net-fs/samba/files/samba-init: killall -HUP smbd &>/dev/null /usr/portage/net-fs/samba/files/samba-init: killall -HUP nmbd &>/dev/null /usr/portage/net-ftp/glftpd/glftpd-1.27.ebuild: -e "s:killall -USR2 xinetd:0:" \ /usr/portage/net-ftp/glftpd/glftpd-1.28.ebuild: -e "s:killall -USR2 xinetd:0:" \ /usr/portage/net-im/jabber-server/files/jabber.rc6-r1: killall jabberd /usr/portage/net-irc/unrealircd/files/unrealircd.rc: killall unrealircd /usr/portage/net-mail/sendmail/ChangeLog: Now the init script sends 'killall ${KILL_OPTS} /usr/sbin/sendmail' /usr/portage/net-mail/sendmail/files/sendmail-r4: killall ${KILL_OPTS} /usr/sbin/sendmail /usr/portage/net-mail/sendmail/files/sendmail-r5: killall ${KILL_OPTS} sendmail /usr/portage/net-mail/vm-pop3d/files/vm-pop3d.rc3: killall -15 /usr/sbin/vm-pop3d /usr/portage/net-misc/fakeidentd/files/fakeidentd.rc: killall fakeidentd /usr/portage/net-misc/omniORB/ChangeLog: Bumped the init.d file to omniORB.305; removed notifd from and fixed the killall command. /usr/portage/net-misc/omniORB/files/omniORB.305: killall omniNames 1>&2 /usr/portage/net-misc/omniORB/files/omniORB.rc5: killall $EXE1 1>&2 /usr/portage/net-misc/omniORB/files/omniORB.rc5: killall $EXE2 1>&2 /usr/portage/net-misc/omniORB/files/omniORB.rc6: killall /usr/bin/omniNames 1>&2 /usr/portage/net-misc/omniORB/files/omniORB.rc6: killall /usr/bin/notifd 1>&2 /usr/portage/net-misc/omniORB/files/omniORB-4.0.0: killall omniNames 1>&2 /usr/portage/net-misc/vtun/files/vtun.rc: killall vtund /usr/portage/net-www/apache/files/apache.rc6: killall -USR1 apache &> /dev/null /usr/portage/net-www/apache/files/2.0.40/apache2.initd: killall -USR1 apache2 &> /dev/null /usr/portage/sys-apps/baselayout/baselayout-1.8.5.5.ebuild: sed -e 's:PROGS\t= init halt shutdown killall5 runlevel sulogin:PROGS\t= init halt shutdown killall5 runlevel:g' \ /usr/portage/sys-apps/baselayout/baselayout-1.8.5.5.ebuild: dosbin init halt killall5 runlevel shutdown sulogin /usr/portage/sys-apps/baselayout/baselayout-1.8.5.5.ebuild: dosym killall5 /sbin/pidof /usr/portage/sys-apps/baselayout/baselayout-1.8.5.8.ebuild: sed -e 's:PROGS\t= init halt shutdown killall5 runlevel sulogin:PROGS\t= init halt shutdown killall5 runlevel:g' \ /usr/portage/sys-apps/baselayout/baselayout-1.8.5.8.ebuild: dosbin init halt killall5 runlevel shutdown sulogin /usr/portage/sys-apps/baselayout/baselayout-1.8.5.8.ebuild: dosym killall5 /sbin/pidof /usr/portage/sys-apps/baselayout/baselayout-1.8.6.2.ebuild: sed -e 's:PROGS\t= init halt shutdown killall5 runlevel sulogin:PROGS\t= init halt shutdown killall5 runlevel:g' \ /usr/portage/sys-apps/baselayout/baselayout-1.8.6.2.ebuild: dosbin init halt killall5 runlevel shutdown sulogin /usr/portage/sys-apps/baselayout/baselayout-1.8.6.2.ebuild: dosym killall5 /sbin/pidof /usr/portage/sys-apps/lufs/files/lufs-0.7.0-init: killall lufsd /usr/portage/sys-apps/lufs/files/lufs-0.8.0-init: killall lufsd /usr/portage/sys-apps/psmisc/psmisc-19-r2.ebuild: dobin killall pstree /usr/portage/sys-apps/psmisc/psmisc-19-r2.ebuild: dosym killall /usr/bin/pidof /usr/portage/sys-apps/psmisc/psmisc-19-r3.ebuild: dobin killall pstree /usr/portage/sys-apps/psmisc/psmisc-19-r3.ebuild: dosym killall /usr/bin/pidof /usr/portage/sys-apps/psmisc/psmisc-21.2.ebuild: dosym killall /usr/bin/pidof /usr/portage/sys-apps/sysvinit/sysvinit-2.83-r1.ebuild: dosbin halt killall5 runlevel shutdown sulogin /usr/portage/sys-apps/sysvinit/sysvinit-2.83-r1.ebuild: dosym killall5 /sbin/pidof /usr/portage/sys-apps/sysvinit/sysvinit-2.83.ebuild: dosbin halt killall5 runlevel shutdown sulogin /usr/portage/sys-apps/sysvinit/sysvinit-2.83.ebuild: dosym killall5 /sbin/pidof /usr/portage/sys-apps/xinetd/files/2.3.9/xinetd.rc6: killall -HUP xinetd &>/dev/null /usr/portage/sys-apps/xinetd/files/2.3.9/xinetd.rc6: killall -USR1 xinetd &>/dev/null /usr/portage/sys-apps/xinetd/files/2.3.9/xinetd.rc6: killall -IOT xinetd &>/dev/null /usr/portage/sys-apps/usbd/files/usbd: killall /usr/sbin/usbd /usr/portage/x11-base/xfree/files/4.2.99.4/startDM.sh: killall -9 ${EXE##*/} /usr/portage/x11-base/xfree/files/4.1.0-r6/xdm.start: killall X /usr/portage/x11-base/xfree/files/4.2.1/startDM.sh: killall -9 ${EXE##*/} /usr/portage/x11-base/xfree/files/4.2.1-r1/startDM.sh: killall -9 ${EXE##*/} /usr/portage/x11-base/xfree/files/4.2.1-r2/startDM.sh: killall -9 ${EXE##*/}
actually, all init.d scripts have a restart target ... if you dont define one it'll just run the stop/start functions there is another bug about 'restarting daemons when changes/ebuilds have been emerged' you can disable devfs actually, but /etc/devfs.d/ will always exist in current systems, so there isnt really a need to check for it ill update the einfo to say like 'in order to access your modem via /dev/modem without rebooting, issue this kill command'