All work fails with "Error during OpenCL Kernel setup" (AMD HD5830 GPU/linux)

I thought I'd have a play with using an AMD GPU (HD5830) under linux, in place of my usual NVIDIAs. Unfortunately, it's not working at all so I'm wondering what I've missed.

Basically, although the driver has installed ok, all GPU work is failing with this message:-

[20:49:11][1427][ERROR] Error during OpenCL kernel setup: PS_R3 (error: -5)
[20:49:11][1427][ERROR] Demodulation failed (error: 2019)!

A quick poke at OpenCL docs gives:

#define CL_OUT_OF_RESOURCES -5

The card is 1024Mb and reports 968Mb available (with X running) so it seems unlikely to be out of memory.

Boinc startup and task messages:

06-Feb-2013 20:47:52 [---] No config file found - using defaults
06-Feb-2013 20:47:52 [---] Starting BOINC client version 7.0.44 for i686-pc-linux-gnu
06-Feb-2013 20:47:52 [---] log flags: file_xfer, sched_ops, task
06-Feb-2013 20:47:52 [---] Libraries: libcurl/7.21.0 NSS/3.12.10.0 zlib/1.2.5 libidn/1.18 libssh2/1.2.4
06-Feb-2013 20:47:52 [---] Data directory: /var/lib/boinc
06-Feb-2013 20:47:52 [---] Processor: 2 GenuineIntel Pentium(R) Dual-Core  CPU      E6700  @ 3.20GHz [Family 6 Model 23 Stepping 10]
06-Feb-2013 20:47:52 [---] Processor: 2.00 MB cache
06-Feb-2013 20:47:52 [---] Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm xsave lahf_lm dts tpr_shadow vnmi flexpriority
06-Feb-2013 20:47:52 [---] OS: Linux: 2.6.38.8
06-Feb-2013 20:47:52 [---] Memory: 1000.62 MB physical, 0 bytes virtual
06-Feb-2013 20:47:52 [---] Disk: 897.28 GB total, 297.06 GB free
06-Feb-2013 20:47:52 [---] Local time is UTC +0 hours
06-Feb-2013 20:47:52 [---] CAL: ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1741, 1024MB, 986MB available, 3584 GFLOPS peak)
06-Feb-2013 20:47:52 [---] OpenCL: ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (driver version 1016.4, device version OpenCL 1.2 AMD-APP (1016.4), 1024MB, 986MB available, 3584 GFLOPS peak)
06-Feb-2013 20:47:52 [Einstein@Home] URL http://einstein.phys.uwm.edu/; Computer ID 6570151; resource share 100
06-Feb-2013 20:47:52 [Einstein@Home] General prefs: from Einstein@Home (last modified 15-Dec-2012 17:24:36)
06-Feb-2013 20:47:52 [Einstein@Home] Computer location: home
06-Feb-2013 20:47:52 [Einstein@Home] General prefs: no separate prefs for home; using your defaults
06-Feb-2013 20:47:52 [---] Reading preferences override file
06-Feb-2013 20:47:52 [---] Preferences:
06-Feb-2013 20:47:52 [---]    max memory usage when active: 500.31MB
06-Feb-2013 20:47:52 [---]    max memory usage when idle: 900.56MB
06-Feb-2013 20:47:52 [---]    max disk usage: 10.00GB
06-Feb-2013 20:47:52 [---]    suspend work if non-BOINC CPU load exceeds 25 %
06-Feb-2013 20:47:52 [---]    (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
06-Feb-2013 20:47:52 [---] Not using a proxy
Initialization completed
06-Feb-2013 20:48:56 [Einstein@Home] task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 resumed by user
06-Feb-2013 20:48:56 [Einstein@Home] Starting task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 using einsteinbinary_BRP4 version 134 (opencl-ati) in slot 0
06-Feb-2013 20:49:13 [Einstein@Home] Computation for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 finished
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_0 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_1 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_2 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_3 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_4 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_5 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_6 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent
06-Feb-2013 20:49:13 [Einstein@Home] Output file p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1_7 for task p2030.20121001.G175.49-03.42.C.b2s0g0.00000_2056_1 absent

Here's an example task failure.

The machine involved has been working fine with an nvidia card, now it's got a fresh copy of Fedora, an HD5830 and Catalyst 12.10 drivers. I've tested with both boinc 7.0.28 and 7.0.44, and Catalyst 12.10/13.1, and I've tried resetting the project.

Also (although it's hardly a definitive test) the card is working fine under X; glxgears etc works, AMD control centre reports all ok, card is not overclocked and is nice and cool (29.5C). The same motherboard/PSU has been running solidly with a pair of GTX460s, so it seems unlikely to be a hardware issue.

All ideas and suggestions gratefully received, as I seem to be at a dead-end with this one!