"ac_rtld error: shdr->sh_size & 3 ELF error: invalid section index" on Linux OpenGL Mesa

braiam
braiam
Joined: 28 Mar 20
Posts: 2
Credit: 438699
RAC: 0
Topic 221520

On my Debian system with a Radeon 590, using Clover OpenGL provided by mesa, all WU assigned to the gpu fail with a Computation error. PRIMEGRID and Amicable Nubmers are the other projects that I've been able to confirm working.

<core_client_version>7.17.0</core_client_version>
<![CDATA[
<message>
process exited with code 11 (0xb, -245)</message>
<stderr_txt>
17:47:40 (650776): [normal]: This Einstein@home App was built at: Jan 16 2017 08:09:16
17:47:40 (650776): [normal]: Start of BOINC application '../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati'.
17:47:40 (650776): [debug]: 1e+16 fp, 6.3e+09 fp/s, 1658243 s, 460h37m22s60
command line: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati --inputfile ../../projects/einstein.phys.uwm.edu/LATeah1063L16.dat --alpha 1.41058464281 --delta -0.444366280137 --skyRadius 5.526880e-07 --ldiBins 30 --f0start 436.0 --f0Band 8.0 --firstSkyPoint 0 --numSkyPoints 1 --f1dot -1e-13 --f1dotBand 1e-13 --df1dot 2.512676418e-15 --ephemdir ../../projects/einstein.phys.uwm.edu/JPLEPH --Tcoh 2097152.0 --toplist 10 --cohFollow 10 --numCells 1 --useWeights 1 --Srefinement 1 --CohSkyRef 1 --cohfullskybox 1 --mmfu 0.1 --reftime 56100 --model 0 --f0orbit 0.005 --mismatch 0.1 --demodbinary 1 --BinaryPointFile ../../projects/einstein.phys.uwm.edu/templates_LATeah1063L16_0444_11144623.dat --debug 1 --device 0 -o LATeah1063L16_444.0_0_0.0_11144623_1_0.out
output files: 'LATeah1063L16_444.0_0_0.0_11144623_1_0.out' '../../projects/einstein.phys.uwm.edu/LATeah1063L16_444.0_0_0.0_11144623_1_0' 'LATeah1063L16_444.0_0_0.0_11144623_1_0.out.cohfu' '../../projects/einstein.phys.uwm.edu/LATeah1063L16_444.0_0_0.0_11144623_1_1'
17:47:40 (650776): [debug]: Flags: X64 SSE SSE2 GNUC X86 GNUX86
17:47:40 (650776): [debug]: glibc version/release: 2.30/stable
17:47:40 (650776): [debug]: Set up communication with graphics process.
boinc_get_opencl_ids returned [0x2a9e488 , 0x7f1c2f5f4260]
Using OpenCL platform provided by: Mesa
Using OpenCL device "Radeon RX 590 Series (POLARIS10, DRM 3.35.0, 5.4.0-4-amd64, LLVM 9.0.1)" by: AMD
Max allocation limit: 6871947673
Global mem size: 8589934592
OpenCL device has FP64 support
% Opening inputfile: ../../projects/einstein.phys.uwm.edu/LATeah1063L16.dat
% Total amount of photon times: 8950
% Preparing toplist of length: 10
% Read 1631 binary points
read_checkpoint(): Couldn't open file 'LATeah1063L16_444.0_0_0.0_11144623_1_0.out.cpt': No such file or directory (2)
% fft_size: 16777216 (0x1000000); alloc: 67108872
% Sky point 1/1
% Binary point 1/1631
% Creating FFT plan.
% fft length: 16777216 (0x1000000)
% Scratch buffer size: 136314880
% Starting semicoherent search over f0 and f1.
% nf1dots: 41  df1dot: 2.512676418e-15  f1dot_start: -1e-13  f1dot_band: 1e-13
% Filling array of photon pairs
ac_rtld error: shdr->sh_size & 3
ELF error: invalid section index
-- signal handler called: signal 1
4 stack frames obtained for this thread:
Frame 14:
	Binary file: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati (0x48b101)
	Source file: hs_boinc_extras.c (Function: sighandler / Line: 291)
Frame 13:
	Binary file: /usr/lib/x86_64-linux-gnu/gallium-pipe/pipe_radeonsi.so (0x7f1c27d2186e)
	Offset info: +0x15e86e
Frame 12:
	Binary file: /usr/lib/x86_64-linux-gnu/gallium-pipe/pipe_radeonsi.so (0x7f1c27d2186e)
	Offset info: +0x15e86e
Frame 11:
	Binary file: /lib/x86_64-linux-gnu/libMesaOpenCL.so.1 (0x7f1c2e2680db)
	Offset info: +0x3730db
Frame 10:
	Binary file: /lib/x86_64-linux-gnu/libMesaOpenCL.so.1 (0x7f1c2e268947)
	Offset info: +0x373947
Frame 9:
	Binary file: /lib/x86_64-linux-gnu/libMesaOpenCL.so.1 (0x7f1c2e265621)
	Offset info: +0x370621
Frame 8:
	Binary file: /lib/x86_64-linux-gnu/libMesaOpenCL.so.1 (0x7f1c2e265d7f)
	Offset info: +0x370d7f
Frame 7:
	Binary file: /lib/x86_64-linux-gnu/libMesaOpenCL.so.1 (0x7f1c2e25469d)
	Offset info: +0x35f69d
Frame 6:
	Binary file: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati (0x48fe01)
	Offset info: opencl_setup_photon_pairs_array+0x4c1
	Source file: unknown (Function: opencl_setup_photon_pairs_array / Line: 0)
Frame 5:
	Binary file: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati (0x47480d)
	Offset info: setup_photon_pairs_array+0x36d
	Source file: HSgammaPulsar.c (Function: setup_photon_pairs_array / Line: 2107)
Frame 4:
	Binary file: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati (0x47e28e)
	Offset info: MAIN+0x4dee
	Source file: HSgammaPulsar.c (Function: MAIN / Line: 4866)
Frame 3:
	Binary file: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati (0x46c06f)
	Offset info: main+0x5ff
	Source file: hs_boinc_extras.c (Function: worker / Line: 833)
	Source file: hs_boinc_extras.c (Function: main / Line: 1039)
Frame 2:
	Binary file: /lib/x86_64-linux-gnu/libc.so.6 (0x7f1c2f644e0b)
	Offset info: __libc_start_main+0xeb
Frame 1:
	Binary file: ../../projects/einstein.phys.uwm.edu/hsgamma_FGRPB1G_1.18_x86_64-pc-linux-gnu__FGRPopencl1K-ati (0x46e569)
	Source file: unknown (Function: _start / Line: 0)
End of stcaktrace
17:47:48 (650776): called boinc_finish
Warning:  Program terminating, but clFFT resources not freed.
Please consider explicitly calling clfftTeardown( ).
</stderr_txt>
]]>

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4972
Credit: 18773535752
RAC: 7213665

Instead of the Mesa OpenCL

Instead of the Mesa OpenCL drivers, try the native ones.

 

braiam
braiam
Joined: 28 Mar 20
Posts: 2
Credit: 438699
RAC: 0

What do you mean by natives?

What do you mean by natives? These are "natives" (don't need further configuration) to Debian.

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3965
Credit: 47230672642
RAC: 65397019

native by getting the

native by getting the proprietary AMD drivers, not the opensource Mesa ones.

_________________________________________________________________________

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5872
Credit: 117792278568
RAC: 34675716

braiam wrote:On my Debian

braiam wrote:
On my Debian system with a Radeon 590, using Clover OpenGL provided by mesa ...

As a long term user of multiple AMD GPUs at this project, the Clover OpenCL implementation doesn't work here, unfortunately.  You need the proprietary implementation contained in the "Radeon Software for Linux" package, formerly known as AMDGPU-PRO.

I don't even use one of the supported distros (Red Hat, Ubuntu, OpenSUSE) so I extract the OpenCL bits from the full Red Hat package, since my distro is RPM based.  I don't know the situation with Debian but if someone has repackaged AMDGPU-PRO to work with Debian, maybe that's what you need.  I have lots of Polaris series GPUs from RX460 to RX580 and they work very well for me with the extracted bits from the Red Hat package.

Cheers,
Gary.

cecht
cecht
Joined: 7 Mar 18
Posts: 1537
Credit: 2915638625
RAC: 2110169

I'm a little confused with

I'm a little confused with the discussion bouncing between OpenGL and OpenCL, but I'll assume that all that you want is to have OpenCL working to get E@H crunching. As the others have said, get the AMD driver stack (from https://www.amd.com/en/support). The included documentation will instruct how to install OpenCl from an install script.
In brief, the command line, from within the extracted stack directory (e.g. amdgpu-pro-19.50-967956-ubuntu-18.04), will be,
$ ./amdgpu-pro-install -y --opencl=legacy --headless
then reboot.
This will install:
    Only base kernel
    Pro OpenCL (headless mode)

If the --headless option is omitted, then these components will be installed:
    Base kernel
    Accelerated graphics
    Mesa multimedia
    Pro OpenGL
    Pro Vulkan
    Pro OpenCL

I always install it on my Ubuntu systems using the --headless option.
In either case, before running the script, I'd recommend shutting down BOINC.

This previous discussion forum may also be of help.

Ideas are not fixed, nor should they be; we live in model-dependent reality.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5872
Credit: 117792278568
RAC: 34675716

cecht wrote:I'm a little

cecht wrote:
I'm a little confused with the discussion bouncing between OpenGL and OpenCL ...

Mesa, also called Mesa3D and The Mesa 3D Graphics Library, is an open source software implementation of OpenGL, Vulkan, and other graphics API specifications.  Mesa also contains an OpenCL implementation which goes by the name of 'Clover'.  The OP did refer to 'Clover OpenGL' but I assumed that was just a typo or perhaps a misunderstanding.  The message I wanted to convey was that people have had problems trying to use the OpenCL implementation that comes with Mesa.

Of course, if Clover is being actively developed (I don't know the status of that) it could well be that at some point Einstein apps might start working with a new version of Mesa/Clover.

Since Ubuntu is a Debian derivative, maybe a package designed for Ubuntu could be used to provide the PRO OpenCL for Debian.  I suspect maybe not directly without some creative modifications to the install script :-).  I would guess it's more likely that someone has already created a package tailored to Debian - so no mods needed to install it successfully.

Cheers,
Gary.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.