Luca, Would you please perform the simpler test of modem detection with wvdialconf: $ sudo wvdialconf which is a simpler test than fax usage. MarvS On Thu, Aug 4, 2011 at 6:08 PM, Luca Arzeni <l.arzeni@xxxxxxxxx> wrote: > Hi MavS, > I already compiled the driver, and installed it. And my problem is exactly that there was no port creation. > Anyway, looking at the source code ot tha different releases of the driver I spotted 2 differences > > 1) the old driver uses termios that is now obsolete, since the new kernel uses ktermios (so I cannot use the old driver) > 2) the old driver uses as pci_id 0x16ec (the USR Robotics pci_id), while the new driver uses 0x13a8 (the exar pci_id) > > From lspci, > > PCI slot PCI ID SubsystemID Name > ---------- --------- --------- -------------- > 09:02.0 16ec:0152 13a8:0129 Serial controller: > U.S. Robotics Device 0152 > > I can see that my modem uses the robotics pci_id, so I changed the driver sorce code e voila'! the driver installed AND NOW udev creates the devices ttyXR0 and ttyXR1: > > > #dmesg > [ 45.689418] Exar PCI 17c15x serial driver Revision: 4.7 > [ 45.689504] xrserial 0000:09:02.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18 > [ 45.689529] 0000:09:02.0: ttyXR0 at MMIO 0xfd7ffc00 (irq = 18) is a XR17c15x > [ 45.689728] 0000:09:02.0: ttyXR1 at MMIO 0xfd7ffe00 (irq = 18) is a XR17c15x > > > #setserial -ag /dev/ttyX* > > /dev/ttyXR0, Line 0, UART: 8250, Port: 0x0000, IRQ: 18 > Baud_base: 921600, close_delay: 50, divisor: 0 > closing_wait: 3000 > Flags: spd_normal skip_test > > /dev/ttyXR1, Line 1, UART: 8250, Port: 0x0000, IRQ: 18 > Baud_base: 921600, close_delay: 50, divisor: 0 > closing_wait: 3000 > Flags: spd_normal skip_test > > > Now the modem dials and seems to work, but when I try to send a fax (using hylafax) it hangs: > > Aug 5 00:44:07 tiberiade FaxGetty[2153]: /dev/ttyXR0: Can not initialize modem. > Aug 5 00:45:42 tiberiade FaxGetty[2153]: Unable to setup modem on /dev/ttyXR0; giving up after 2 attempts > Aug 5 00:45:42 tiberiade FaxGetty[2153]: CLOSE /dev/ttyXR0 > Aug 5 00:45:42 tiberiade FaxQueuer[2135]: MODEM /dev/ttyXR0 appears to be wedged > Aug 5 00:45:42 tiberiade FaxQueuer[2135]: MODEM WEDGED: bin/wedged "ttyXR0" "/dev/ttyXR0" > Aug 5 00:45:42 tiberiade FaxQueuer[2135]: NOTIFY exit status: 0 (2169) > > But I think that this is queston to put to the hylafax people! > > Thanks again for you help, > Luca > > >> ----- Original Message ----- >> From: Marvin Stodolsky >> Sent: 08/04/11 05:51 AM >> To: Luca Arzeni, Linmodems >> Subject: Re: exar PCI modem >> >> Luca, >> >> Under my x86_32 Ubuntu system, the driver compiles, installs and loads: >> [ 4876.681496] Exar PCI 17c15x serial driver Revision: 4.7 >> $ lsmod | grep xr >> xr17c15x 71061 0 >> >> But without the hardware, there is no spontaenous port creation >> $ ls /dev/ttyX* >> ls: cannot access /dev/ttyX*: No such file or directory >> >> The mknod_sh script has errors. >> The 3 instances of == must be replaced by = >> With that the script is OK through creation of >> $ ls -l /dev/exar >> crwxrwxrwx 1 root root 30, 32 2011-08-03 23:12 /dev/exar >> >> But there is some syntax error in the more arcane line: >> for (( port=0; $port<$PORTS; port++ )); do >> still blocking the creation of the needed /dev/ttyXRn ports >> >> So I did the edit per the attachments, finally getting creation of: >> $ ls -l /dev/ttyXR* /dev/cuxr* >> crwxrwxrwx 1 root root 35, 1 2011-08-03 23:41 /dev/cuxr1,2 >> crwxrwxrwx 1 root root 30, 1 2011-08-03 23:41 /dev/ttyXR1,2 >> >> Hope this helps. >> >> Ahmed >> Is the error in the "for" line obvious to you. $port and $PORTS are >> properly defined earlier. See below >> >> MarvS >> >> ========== >> >> $ sudo ./mknod_sh >> [sudo] password for marv2: >> EXAR Multiport Make Node Utility for XR17C15X Family of Devices >> Please input Device Major Number (Default = 30) >> Please input Callout Device Major Number (Default = 35) >> Please input total number of ports (XR17C15X channels) for all boards >> (Default = 2) >> Creating new nodes for ports from Exar Board(s) >> crwxrwxrwx 1 root root 30, 32 2011-08-03 23:32 exar >> ./mknod_sh: 34: Syntax error: Bad for loop variable >> >> >> > On Wed, Aug 3, 2011 at 11:23 AM, Luca Arzeni <l.arzeni@xxxxxxxxx> wrote: >> >> MarvS, >> >> first of all, thank you very much for your support and time spent on my >> >> problems. >> >> >> >> Second, I cannot explain why the class is 700, I cannot believe that >> >> robotics is lying on their site! >> >> >> >> Regarding " code package", I'm not sure of what code package do you need. >> >> >> >> may be I explained poorly the situation. >> >> >> >>>Incidentally compiled drivers are only useful for the particular kernel >> >>> (resources), they were controlled with. >> >> >> >> Exar gave me the source code for the driver, not the binary. You can >> >> find the file "xr17c15x.c" in the zip that I previously sent to you, >> >> anyway I will attach here for reference. >> >> >> >> I compiled the exar code using my kernel-compiled tree and that went fine. >> >> >> >> It installed ok. >> >> >> >> So it should work. Am I missing something? >> >> >> >> Do you mean the driver? It was in the previous mail, anyway I attach here >> >> again. >> >> >> >> If you want something else, let me know! >> >> >> >> Thanks, Luca >> >> >> >> >> >> >> >> ----- Original Message ----- >> >> >> >> From: Marvin Stodolsky >> >> >> >> Sent: 08/03/11 04:15 PM >> >> >> >> To: Luca Arzeni >> >> >> >> Subject: exar PCI modem >> >> >> >> >> >> >> >> Luca, >> >> >> >> re: Serial controller: U.S. Robotics Device 0152 " >> >> CLASS=0700 >> >> PCIDEV=16ec:0152 >> >> SUBSYS=13a8:0129 >> >> IRQ=18 >> >> >> >> The CLASS=0700 suggests that it is a Winmodem >> >> as Controller Chipset modems are Class 0780 >> >> >> >> Please send me the code package to examine. >> >> >> >> Incidentally compiled drivers are only useful for the particular >> >> kernel (resources), they were controlled with. >> >> >> >> MarvS >> >> >> >> On Wed, Aug 3, 2011 at 6:22 AM, Luca Arzeni <l.arzeni@xxxxxxxxx> wrote: >> >>> Hi MarvS, >> >>> you will find attached the output of scanmodem. >> >>> >> >>> I think it's better to give you other infos to allow a better >> >>> understanding >> >>> of the situation. >> >>> >> >>> 1) This modem (USR Robotics 2891A) is NOT a soft(win)modem but a >> >>> controller >> >>> modem; at least, it's marketed this way, I attach modem specifications >> >>> for >> >>> reference. Actually this was the ONLY reason that made me choose this >> >>> modem! >> >>> >> >>> 2) I was able to find a driver from exar (modem chipset maker); it's >> >>> declared to work for kernel 2.6.37, anyway I compiled a custom kernel >> >>> 2.6.32-5 disabling XR17C152 support, as requested in the readme, and then >> >>> compiled their driver. I installed the custom kernel, and inserted the >> >>> driver in the kernel without problems. I attach the driver if someone >> >>> should >> >>> need in the future. >> >>> >> >>> 3) There were now problems with IRQ sharing, as reported from dmesg: >> >>> root@tiberiade:~# dmesg | grep serial >> >>> [ 5.436146] serial 0000:09:02.0: PCI INT A -> GSI 18 (level, low) -> >> >>> IRQ >> >>> 18 >> >>> [ 5.436153] serial 0000:09:02.0: PCI INT A disabled >> >>> >> >>> I solved this issue poking in the bios, reserving an IRQ for the modem and >> >>> then passing to the kernel the acpi=ht parameter >> >>> >> >>> now dmesg shows: >> >>> >> >>> [ 5.571445] serial 0000:09:02.0: PCI->APIC IRQ transform: INT A -> IRQ >> >>> 18 >> >>> 4) At this point iinstalled the module xr17c15x, and from syslog it seems >> >>> that it's all working fine: >> >>> tiberiade kernel: [ 468.310191] Exar PCI 17c15x serial driver Revision: >> >>> 4.7 >> >>> The main issue, now, is that I cannot see the devices in /dev. >> >>> >> >>> Udev does not detect the driver? >> >>> wvdial does not find the serial. >> >>> >> >>> Can I create manually the device? >> >>> >> >>> I'm stuck at this point. >> >>> >> >>> scanmodem was run with this configuration: custom kernel and custom module >> >>> inserted in the kernel. >> >>> >> >>> That's all. Let me know if you need more info. >> >>> >> >>> From my point of view I am very disappointed with USR Robotics that >> >>> markets >> >>> this modem as fully supported under linux. >> >>> >> >>> I've always rear good reviews of their modem but it seems to me that this >> >>> is >> >>> not true for this model. >> >>> >> >>> Thanks for your help, Luca >> >>> >> >>> >> >>> ----- Original Message ----- >> >>> >> >>> From: Marvin Stodolsky >> >>> >> >>> Sent: 07/29/11 04:15 PM >> >>> >> >>> To: Luca >> >>> >> >>> Subject: Re: cmc, UK kernel 2.6.32-22-generic >> >>> >> >>> Luca >> >>> >> >>> >> >>> Get scanModem from >> >>> http://linmodems.technion.ac.il/ >> >>> and send us the output ModemData.txt >> >>> >> >>> MarvS >> >>> >> >>> On Thu, Jul 28, 2011 at 3:04 PM, Luca <l.arzeni@xxxxxxxxx> wrote: >> >>>> >> >>>> >> >>>> Craig McLaren <craigmc35 <at> googlemail.com> writes: >> >>>> >> >>>> > >> >>>> > Hello All, Please can someone advise what steps are necessary to get >> >>>> > the USR802981-OEM modem working. I have two of these which I plan on >> >>>> > using with Hylafax if they ever work. At the moment the serial port >> >>>> > does not appear to have been created although the device is listed >> >>>> > when I run lspci. I'm assuming they are compatible with Linux in some >> >>>> > way as this is what it says on USR website:- >> >>>> > >> >>>> > http://www.usr.com/support/product-template.asp?prod=oem#2980 >> >>>> > >> >>>> > Any help greatly appreciated. >> >>>> > >> >>>> >> >>>> >> >>>> Hi there, I know that exar released a driver for kernel 2.6.21, but then >> >>>> no >> >>>> other driver come from them. >> >>>> >> >>>> >> >>>> I'm using debian squeeze under AMD-64 (2.6.32-5-amd64) architecture. >> >>>> >> >>>> Have you any success with your attempt? >> >>>> >> >>>> Thanks, Luca >> >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> >>> >> >>> >> >> >> >> >> > > >
