Aggravation
Dec. 29th, 2006 14:49I've been fighting with my wireless drivers.
It feels like I'm always fighting with my wireless drivers.
It looks like I have a choice: use the native linux kernel drivers for my chip, which consistently register 25% less signal strength than I should be getting... or else stumble through patching and recompiling my kernel to use 16k stacks, then try to wrestle ndiswrapper into submission (for the nth time).
I have ndiswrapper installed, and it seems to be mostly okay... except that when I install the module, I get
I think this might be a 4k stack size problem, but I don't even know where to check to see whether my particular build of the 2.6.17 kernel uses 4k stacks or not. The net has been rather sketchy about this, and most suggestions I've seen have been telling Fedora users to use the Linuxant kernel patch to change to 16K stacks and rebuild. I assume I could do essentially the same in Ubuntu Edgy.
I really, really dislike using a custom kernel, though -- a stack size change is unlikely to break other stuff, but it does mean that I'll have to rebuild each kernel by hand instead of letting the package manager update it. (So I'm lazy.) I also haven't built a kernel in about four years, so I'd have to bring up a really good howto and hope I don't screw up.
Hatred and discontent.
It feels like I'm always fighting with my wireless drivers.
It looks like I have a choice: use the native linux kernel drivers for my chip, which consistently register 25% less signal strength than I should be getting... or else stumble through patching and recompiling my kernel to use 16k stacks, then try to wrestle ndiswrapper into submission (for the nth time).
I have ndiswrapper installed, and it seems to be mostly okay... except that when I install the module, I get
Dec 29 14:57:59 harkin kernel: [17188293.808000] ndiswrapper version 1.22 loaded (preempt=no, smp=yes) Dec 29 14:57:59 harkin kernel: [17188293.812000] ndiswrapper: driver bcmwl5 (Linksys,07/17/2003, 3.30.15.0) loaded Dec 29 14:57:59 harkin kernel: [17188293.812000] ACPI: PCI Interrupt 0000:02:06.0[A] -> Link [LNKC] -> GSI 11 (level, low) -> IRQ 11 Dec 29 14:57:59 harkin kernel: [17188293.824000] ndiswrapper (miniport_init:264): couldn't initialize device: C0000001 Dec 29 14:57:59 harkin kernel: [17188293.824000] ndiswrapper (pnp_start_device:428): Windows driver couldn't initialize the device (C0000001) Dec 29 14:57:59 harkin kernel: [17188293.824000] ndiswrapper (miniport_halt:327): device c17d8400 is not initialized - not halting Dec 29 14:57:59 harkin kernel: [17188293.824000] ndiswrapper: device eth%%d removed Dec 29 14:57:59 harkin kernel: [17188293.824000] ACPI: PCI interrupt for device 0000:02:06.0 disabled Dec 29 14:57:59 harkin kernel: [17188293.824000] ndiswrapper: probe of 0000:02:06.0 failed with error -22
I think this might be a 4k stack size problem, but I don't even know where to check to see whether my particular build of the 2.6.17 kernel uses 4k stacks or not. The net has been rather sketchy about this, and most suggestions I've seen have been telling Fedora users to use the Linuxant kernel patch to change to 16K stacks and rebuild. I assume I could do essentially the same in Ubuntu Edgy.
I really, really dislike using a custom kernel, though -- a stack size change is unlikely to break other stuff, but it does mean that I'll have to rebuild each kernel by hand instead of letting the package manager update it. (So I'm lazy.) I also haven't built a kernel in about four years, so I'd have to bring up a really good howto and hope I don't screw up.
Hatred and discontent.
no subject
Date: 2006-12-31 04:19 (UTC)