hald Segfaults on startup

[Martijn vdS]

When I try to start hald (or actually, dpkg tried it on upgrade), hald crashes
with a Segmentation Fault.

The last few lines of hald --daemon=no --verbose=yes follow:

08:38:15.319 [I] physdev.c:1019: phys_add: subsys=pcmcia
sysfs_path=/sys/devices/pci0000:00/0000:00:1e.0/0000:02:06.0/0.0, parent=0x080c3458
08:38:15.325 [I] physdev.c:944: Add callouts completed
udi=/org/freedesktop/Hal/devices/pcmcia__1__1
08:38:15.326 [I] hald.c:90: Added device to GDL;
udi=/org/freedesktop/Hal/devices/pcmcia__1__1
08:38:15.328 [I] classdev.c:884: class_add: subsys=tty
sysfs_path=/sys/class/tty/ttyS14 dev=/dev/ttyS14 physdev=0x080c0b78
Segmentation fault

The backtrace is quite long (239 items), I'll attach it.

ttyS14 is my PCMCIA GPS adapter (ttyS14 at I/O 0x400 (irq = 3) is a 16C950/954),
which works fine otherwise.

[Martijn vdS (martijn)]

Created an attachment (id=2468)
hald backtrace (gdb)

This is the hald backtrace. After it sefaults, a hald-addon-acpi is still
running.

[Martijn vdS (martijn)]

When I remove my GPS, hald works fine and doesn't crash. When I plug it in
later, it crashes again, in the same location.

cardctl info tells me:
PRODID_1="CF CARD"
PRODID_2="GENERIC"
PRODID_3=""
PRODID_4=""
MANFID=0279,950b
FUNCID=2

(this is exactly the same as the contents of
/sys/class/pcmcia_socket/pcmcia_socket0/device/0.0)

[Martin Pitt (pitti)]

 hal (0.5.2-0ubuntu5) breezy; urgency=low
 .
   * Added debian/patches/property-null-values.patch:
     - hald/hal_property_new_string(): If the given value is NULL, create a
       property with value "" instead of NULL to prevent all sorts of segfaults
       at a very low level. (Ubuntu #11060)

Since you could not confirm in IRC: please reopen the bug if that does not help
to fix that bug.