![]() |
Hi, a little off-topic, but.... with RH7.0 Red Hat deffinitly ships a broken version of gcc. they decided to use the newest non-official bleeding-edge-compiler from the development-tree without asking the developers for problems and without reading the documentation and therefor got many flames from the developer's. If somebody needs to compile programs under RH7.0 it is a good idea first to get a working version of GCC from one of the gnu-sites all aroundthe world. Christoph Lee Howard schrieb: > > At 06:42 PM 12/12/00 -0500, Douglas Younger wrote: > >Hello, > > I am trying to set up a fax server machine with four internal PCI > >modems. I have USR/3Com 3CP5610A modems which are not winmodems. > > You are very much correct, it is not a winmodem. AFAIK, there are three > chipsets that control hardware-based PCI modems: USR/TI Kermit (as on the > 3CP5610), Lucent Venus, and Topic TP560i. I have only ever used the USR > one as I have not been able to find any of the other two. For an internal > modem, and if you don't mind the nuances of USR fax firmware, the 3CP5610A > is a very good modem. I've used many of them. With many new motherboards > coming out with PCI-only options, PCI is the only means of internalizing a > modem, although I will side with Jay's opinion about multi-modem systems > preferring external units. I've *never* had to reboot my single-fax > systems, though (all internal modems), due to fax problems. > > > I tried at > >first to just get one modem set up, and using the standard kernel supplied > >with RH7.0 (2.2.16 & patches), I was having no luck at all. I also tried to > >compile 2.2.17 (which is a feat in and of itself with RedHat shipping a > >beta gcc that won't compile kernels), and was still unable to access the > >modem. So, I got the 2.4.0-test12 which is the latest beta kernel (the > >modem box says you need 2.3.x or better, which is funny, because 2.2.x is > >the current "stable" revision). Anyway, I finally got a modem setup & > >communicating with minicom. So, I put in the other 3 to see if I could > >configure them. Well no luck. I seem to get a whole lot of resource > >conflicts in the boot sequence. Looking at /proc/pci it only listed 2 of > >the modems, And /etc/sysconfig/hwconf (from kudzu) lists 3 ?!? I tried > >dropping to only 2 modems and I still get conflicts. With more than 1 > >modem, I can't access any of them. > > These modems *do not* require kernel 2.3.x or greater. I've run them with > most of the 2.2.x flavours and have yet to use any 2.3.x or 2.4.x kernel. > RedHat does a lot of interesting things with kudzu and isapnp during boot, > and I'm not sure what they've done in 7.0 to make PCI modems work in a > plug-n-play fashion, but it's not doing anything more than using kudzu, > isapnp, and setserial. I like RedHat, but the two aspects of the > distribution that I dislike most are linuxconf and kudzu. I've had more > trouble because of those two things making unseen and unasked-for changes > on their own... Anyway, the essence of setting up a hardware PCI modem is > setserial. For example, in /etc/rc.d/rc.sysinit I have the following line > at the top: > > /bin/setserial /dev/ttyS2 uart 16550A port 0xb800 irq 9 > > read the setserial manpage for better info on how do do this. You need to > make sure, though, that the setserial command is run *before* the serial > port is accessed by the HylaFAX daemon (so rc.local comes too late). > > As for your kernel compilation issues with gcc... I believe that the > official verdict is that the bugs are in the kernel source and not in the > compiler. Apparently gcc is significantly more stringent about syntax and > code structure than kgcc or egcs are. But yes, it is beta, yet it compiles > HylaFAX source just dandy. > > As for not having enough manageable IRQs to accomplish your desires, well, > you and your motherboard will need to argue about that. Don't expect any > sympathy from Dell, though. > > If you dont have any ISA slots, and if you are interested in using external > modems, you may be interested in some multi-port PCI I/O cards like the one > at: > > http://www.siig.com/io/pci_serial_8000_plus.php > > I installed one of these the other day in a Windows system, and it went > very well. It doesn't say anything about Linux compatiblity, but it does > say that they are DOS compatible, so I suppose that is a good sign for > Linux-use, but I didn't have the time to test it on a Linux system. I > still imagine that a hefty use of setserial would be needed with such a > device. > > Lee Howard. > > ____________________ HylaFAX(tm) Users Mailing List _______________________ > To unsub: mail -s unsubscribe hylafax-users-request@hylafax.org < /dev/null ____________________ HylaFAX(tm) Users Mailing List _______________________ To unsub: mail -s unsubscribe hylafax-users-request@hylafax.org < /dev/null