Arch Linux User Repository

Note for anyone who has a Polaris GPU (Radeon RX 5xx) debugging issues with this package; Packages that use OpenCL like clinfo or davinci-resolve-studio will need you to downgrade opencl-amd to 1:5.7.1-1 as well as amdgpu-pro-oglp to 23.10_1620044-1 to avoid coredumps & segfaults.

DVR would not open unless these 2 packages were downgraded (along with their dependencies). Had to figure it out the hard way after hours using valgrind and rebooting over and over. Hopefully someone else will not have to pull their hair out trying to resolve their issue.

@Olympus593 Are you using the amdgpu kernel driver? You can try using the experimental support for your GPU series (https://wiki.archlinux.org/index.php/AMDGPU#Enable_Southern_Islands_.28SI.29_and_Sea_Islands_.28CIK.29_support)

From the official amdgpu-pro driver page (https://www.amd.com/en/support/kb/release-notes/rn-prorad-lin-18-30), I could see your GPU is supported but you might need to cross-check on the AMD forums.

Do I need to downgrade Xorg or do some configuration because mine always ends up with Error -6 in any app trying to use OpenCL. I have a GCN 1 card with support provided by the AMDGPU driver.

PSA : I remove my previous comment to avoid confusion as now that this package works with the current version (opencl-amd 18.3.641594).

@stoostranger Haha, seems like this package finally works although still needs the libdrm downgrade shenanigans. Thanks!

I tried downgrading without libdrm and lib32-libdrm downgrade but blender still segfault. then I downgrade libdrm to 2.4.93-1, then install opencl-amd 18.3.641594 again, then now Blender Circle renderer works perfectly. Hope this help

As some of you noticed 18.30.641594-1 does not seem to work with blender. Manual downgrading worked for me. Edit PKGBUILD to match the following values:

major='18.20'
minor='606296'

source=("https://www2.ati.com/drivers/linux/ubuntu/18.04/amdgpu-pro-18.20-606296.tar.xz")
sha256sums=('2a0716993e8efb1fadcb92d82e9328e344bdbc78769f5ff95298b82f49ff76f9')

also change amdgpu1_2.4.92 to amdgpu1_2.4.91 and as @Nightbane112 mentioned downgrade libdrm and lib32-libdrm to 2.4.93-1

I have same error too on Blender (portable, Manjaro), the only temporary solution I found is, in Blender folder run "blender-softwaregl" file, it uses software and is very slow depend on your system. then I can render with Opencl. I hope it gets sooner fixed.

@IMBJR I am getting roughly the same issue as you when this package is installed.

# Blender 2.79 (sub 0), Commit date: 2018-05-26 21:51, Hash 32432d91bbe

# backtrace
blender(BLI_system_backtrace+0x34) [0x55d95ecf13f4]
blender(+0xb7b562) [0x55d95e27d562]
/usr/lib/libc.so.6(+0x37e00) [0x7fd541b02e00]
/usr/lib/libdrm_amdgpo.so.1(amdgpu_get_marketing_name+0xc) [0x7fd4f8e5dbdf]
/usr/lib/libamdocl-orca64.so(+0x8d871e) [0x7fd4f993c71e]
/usr/lib/libamdocl-orca64.so(+0x8d8cbf) [0x7fd4f993ccbf]
/usr/lib/libamdocl-orca64.so(+0x8dbc71) [0x7fd4f993fc71]
/usr/lib/libamdocl-orca64.so(+0x8f5488) [0x7fd4f9959488]
/usr/lib/libamdocl-orca64.so(+0xc4bc7d) [0x7fd4f9cafc7d]
/usr/lib/libamdocl-orca64.so(+0x8cb799) [0x7fd4f992f799]
/usr/lib/libamdocl-orca64.so(+0x8cb80f) [0x7fd4f992f80f]
/usr/lib/libamdocl-orca64.so(+0x8cc597) [0x7fd4f9930597]
/usr/lib/libamdocl-orca64.so(+0xcf772e) [0x7fd4f9d5b72e]
/usr/lib/libamdocl-orca64.so(+0xcf8c9a) [0x7fd4f9d5cc9a]
/usr/lib/libamdocl-orca64.so(+0xcf8eb6) [0x7fd4f9d5ceb6]
/usr/lib/libamdocl-orca64.so(+0x8a9787) [0x7fd4f990d787]
/usr/lib/libamdocl-orca64.so(clIcdGetPlatformIDsKHR+0x8a) [0x7fd4f98f068a]
/usr/lib/libOpenCL.so(+0x5d1e) [0x7fd5010add1e]
/usr/lib/libOpenCL.so(clGetPlatformIDs+0x115) [0x7fd5010afc15]
blender(_ZN3ccl10OpenCLInfo17get_num_platformsEPjPi+0x1c) [0x55d95f2412dc]
blender(_ZN3ccl10OpenCLInfo13get_platformsEPNS_6vectorIP15_cl_platform_idNS_16GuardedAllocatorIS3_EEEEPi+0x39) [0x55d95f242389]
blender(_ZN3ccl10OpenCLInfo18get_usable_devicesEPNS_6vectorINS_20OpenCLPlatformDeviceENS_16GuardedAllocatorIS2_EEEEb+0x19b) [0x55d95f246fab]
blender(_ZN3ccl18device_opencl_infoERNS_6vectorINS_10DeviceInfoENS_16GuardedAllocatorIS1_EEEE+0x56) [0x55d95f233cd6]
blender(_ZN3ccl6Device17available_devicesEv+0xe2) [0x55d95f208852]
blender(+0x1a0216a) [0x55d95f10416a]
/usr/lib/libpython3.7m.so.1.0(_PyMethodDef_RawFastCallKeywords+0x148) [0x7fd548a42a88]
/usr/lib/libpython3.7m.so.1.0(_PyCFunction_FastCallKeywords+0x21) [0x7fd548a42d21]
/usr/lib/libpython3.7m.so.1.0(_PyEval_EvalFrameDefault+0x522c) [0x7fd548aba89c]
/usr/lib/libpython3.7m.so.1.0(_PyFunction_FastCallDict+0x11b) [0x7fd5489fbf3b]
blender(+0xfac373) [0x55d95e6ae373]
blender(RNA_property_enum_items_ex+0x61) [0x55d95eb5ea71]
blender(RNA_property_enum_items+0x14) [0x55d95eb5eb14]
blender(RNA_property_enum_value+0x33) [0x55d95eb5eeb3]
blender(+0xf9bf9a) [0x55d95e69df9a]
blender(+0xf9cc24) [0x55d95e69ec24]
/usr/lib/libpython3.7m.so.1.0(PyObject_SetAttr+0x88) [0x7fd548aaecf8]
/usr/lib/libpython3.7m.so.1.0(_PyEval_EvalFrameDefault+0xd7d) [0x7fd548ab63ed]
/usr/lib/libpython3.7m.so.1.0(_PyFunction_FastCallDict+0x11b) [0x7fd5489fbf3b]
blender(bpy_app_generic_callback+0xd5) [0x55d95e6a99d5]
blender(BLI_callback_exec+0x2d) [0x55d95ecb360d]
blender(WM_init+0x28a) [0x55d95e28dfda]
blender(main+0x3d1) [0x55d95e264841]
/usr/lib/libc.so.6(__libc_start_main+0xf3) [0x7fd541aef223]
blender(_start+0x2e) [0x55d95e27a01e]

@IMBJR Have you tried downgrading libdrm & lib32-libdrm from ver. 2.4.94 to ver. 2.4.93? AFAIK, that's the only workaround I found that works on my system.

EDIT: Seems like the workaround only worked for the previous version of this package (amdgpu-pro-18.30-633530). The current version (18.30-641594) needs to be downgraded to the previous version in order for opencl apps to work again.

Version 18.30.641594-1 is making blender segault:

Sep 05 18:05:35 pc audit[28523]: ANOM_ABEND auid=1000 uid=1000 gid=1000 ses=2 pid=28523 comm="blender" exe="/usr/bin/blender" sig=11 res=1 Sep 05 18:05:35 pc kernel: audit: type=1701 audit(1536167135.181:15): auid=1000 uid=1000 gid=1000 ses=2 pid=28523 comm="blender" exe="/usr/bin/blender" sig=11 res=1 Sep 05 18:05:35 pc systemd[1]: Started Process Core Dump (PID 28578/UID 0). Sep 05 18:05:36 pc systemd-coredump[28579]: Process 28523 (blender) of user 1000 dumped core.

                                        Stack trace of thread 28523:
                                        #0  0x00007f1eea241bdf amdgpu_get_marketing_name (libdrm_amdgpo.so.1)
                                        #1  0x00007f1eead2071e n/a (libamdocl-orca64.so)
                                        #2  0x00007f1eead20cbf n/a (libamdocl-orca64.so)
                                        #3  0x00007f1eead23c71 n/a (libamdocl-orca64.so)
                                        #4  0x00007f1eead3d488 n/a (libamdocl-orca64.so)
                                        #5  0x00007f1eeb093c7d n/a (libamdocl-orca64.so)
                                        #6  0x00007f1eead13799 n/a (libamdocl-orca64.so)
                                        #7  0x00007f1eead1380f n/a (libamdocl-orca64.so)
                                        #8  0x00007f1eead14597 n/a (libamdocl-orca64.so)
                                        #9  0x00007f1eeb13f72e n/a (libamdocl-orca64.so)
                                        #10 0x00007f1eeb140c9a n/a (libamdocl-orca64.so)
                                        #11 0x00007f1eeb140eb6 n/a (libamdocl-orca64.so)
                                        #12 0x00007f1eeacf1787 n/a (libamdocl-orca64.so)
                                        #13 0x00007f1eeacd468a clIcdGetPlatformIDsKHR (libamdocl-orca64.so)
                                        #14 0x00007f1ef3c23d1e n/a (libOpenCL.so)
                                        #15 0x00007f1ef3c25c15 clGetPlatformIDs (libOpenCL.so)
                                        #16 0x00005607619272fc _ZN3ccl10OpenCLInfo17get_num_platformsEPjPi (blender)
                                        #17 0x63c4f0885a279521 n/a (n/a)