ppc64 hald "Rule is NULL on jump"

[Christian Schmitt]

Hi,

I'm using Gentoo on a G5 machine (userland 64bit as well). Sadly there 
is a problem with HAL. When starting hald it bails out with this error:

17:41:54.128 [I] coldplug.c:307: new event (no dev node) 
'/sys/class/i2c-adapter/i2c-5/5-004a'
17:41:54.128 [I] hotplug.c:138: /sys/class/i2c-adapter/i2c-5/5-004a is a 
device (subsystem)
17:41:54.128 [I] device.c:4406: add_dev: subsys=i2c 
sysfs_path=/sys/class/i2c-adapter/i2c-5/5-004a dev= parent_dev=0x00000000
17:41:54.128 [I] coldplug.c:307: new event (no dev node) 
'/sys/class/i2c-adapter/i2c-6/6-0046'
17:41:54.128 [I] hotplug.c:138: /sys/class/i2c-adapter/i2c-6/6-0046 is a 
device (subsystem)
17:41:54.128 [I] device.c:4406: add_dev: subsys=i2c 
sysfs_path=/sys/class/i2c-adapter/i2c-6/6-0046 dev= parent_dev=0x00000000
17:41:54.128 [I] coldplug.c:307: new event (no dev node) 
'/sys/devices/pci0000:f0/0000:f0:0b.0'
17:41:54.128 [I] hotplug.c:138: /sys/devices/pci0000:f0/0000:f0:0b.0 is 
a device (subsystem)
17:41:54.128 [I] device.c:4406: add_dev: subsys=pci 
sysfs_path=/sys/devices/pci0000:f0/0000:f0:0b.0 dev= parent_dev=0x00000000
*** [DIE] device_info.c:rules_match_and_merge_device():1087 : Rule is 
NULL on jump


After a longer IRC session with DannyK, I disabled the DIE() statment in 
device_info.c line 1087, HAL was at least starting and we tried to get 
some useful information.

greping for 'no match, skip to rule (0)' gives this: 
http://pastebin.com/d23dc376b

According to DannyK it might be a problem with writing a long (for the 
address with the jump_position) to the cache file with sizeoff (u_int32_t)

This issue is already in Gentoo bugzilla since over a year 
(http://bugs.gentoo.org/show_bug.cgi?id=176380) and occurs since version 
0.5.9

I hope to get some further help here.

Cheers
Chris