Red Hat Bugzilla – Bug 10735
kudzu 0.36-2 doesn't see my USR modem
Last modified: 2014-03-16 22:13:24 EDT
kudzu-0.27-1 detected my USR modem just fine: class: MODEM bus: SERIAL detached: 1 device: ttyS1 driver: ignore desc: "USR|0100 Courier V.Everything EXT" pnpmfr: USR pnpmodel: 0100 pnpcompat: PNPC107 pnpdesc: Courier V.Everything EXT When I upgraded to kudzu-0.36-2, it first informed me that my modem was a generic serial modem rather than the specific modem shown above, and then when I tried to duplicate that, it didn't find the modem at all. I will attach strace output from kudzu-0.36-2, in case that's helpful.
Created attachment 194 [details] strace output for kudzu run that doesn't detect the modem
It's not detecting it because the modem port is locked from another program (the file /var/lock/LCK..modem exists), so therefore kudzu won't probe that serial port. If you remove the /var/lock/LCK..modem file, does it then detect the modem correctly?
Created attachment 198 [details] strace of kudzu-0.38-1 not detectingmodem root@jik:/c/build/curl!67# ls -l /var/lock/LCK*; kudzu; rpm -q kudzu ls: No match. kudzu-0.38-1 root@jik:/c/build/curl!68# I'll attach another strace.out from kudzu-0.38-1 when the lock file does not exist.
What's the current /etc/sysconfig/hwconf when you're running this? (unfortunately, serial probe problems are very hard to pin down without access to the hardware in question... :( )
Created attachment 203 [details] hwconf
Eek, I've ignored this for *way* too long. I'm assuming the same behavior happens with the current version (the serial code hasn't changed...)
I have not seen this problem in a long time. I don't know what has changed to make it go away.
Hi, I was also having problems with kudzu-0.99.23 (RH7.2) and my USR modem. I took a look under the hood and found the problems... Older USR modems are not PnP but have ATI9 responses in the right format. The problem is in parse_pnp_string() - The string has <cr><lf> before the opening "(" and the PnP version number is in decimal rather than coded binary. The string also has extensions, but no checksum. The end result is that the valid ATI9 response gets rejected. I have hacked around with serial.c and produced something that works. The only problem left is that the modem refuses to dial after being probed. The problem must be with the state of the serial port as resetting the modem doesn't clear it. Once I work out how to post my modified serial.c, I'll pass it on. Sorry it got hacked around, but I needed to find out what was happening, so a diff might be diffucult! ('scuse the pun)
Created attachment 54524 [details] Hacked but functioning PnP serial.c
Closing out bugs from older, no longer supported releases. Apologies for the lack of response.