Using kernel-smp-2.4.18-0.1, the paep driver won't load when there's an AEP SSL Accelerator board present. The relevant part of the lspci -v output: 00:03.0 Co-processor: Intel Corp.: Unknown device 1200 (rev 03) (prog-if 01) Subsystem: Unknown device 172a:0000 Flags: bus master, medium devsel, latency 66, IRQ 5 Memory at ec800000 (32-bit, prefetchable) [size=1M] I/O ports at 2000 [size=128] Memory at ee000000 (32-bit, prefetchable) [size=32M] Capabilities: [70] Power Management version 0 Running lspci with the -n flag: 00:03.0 Class 0b40: 8086:1200 (rev 03) (prog-if 01) Subsystem: 172a:0000 Flags: bus master, medium devsel, latency 66, IRQ 5 Memory at ec800000 (32-bit, prefetchable) [size=1M] I/O ports at 2000 [size=128] Memory at ee000000 (32-bit, prefetchable) [size=32M] Capabilities: [70] Power Management version 0
Almost forgot: the kernel message log: AEP Device Driver. Version: UNLABELLED Build Date: Feb 27 2002 PCI AEP driver: Error: no devices (returning -ENODEV) modprobe also complains that the driver has no license defined, while for the most part it looks like a standard BSD-without-advertising license (the statement at the top of driver_version.h is pretty scary, though).
Under 2.4.18-0.16smp, the driver loads, but applications will segfault when they attempt to open the device.
Works with kernel-2.4.18-0.18smp. The driver doesn't check for invalid minor numbers, and oopses instead of returning ENODEV, but that doesn't keep it from working with the proper minors.
any chance of getting that oops somewhere ? The code appears to check...
paep.c: line 576. The exit_if_too_many_cards() macro just exits if the card number is too high to be supported by the driver, not if it references a card which isn't installed.
Probably needs this inserted at line 574: if (unit_no >= nb_cards) return -ENODEV;
yup fixed