El 22/6/25 a las 21:39, Jonathan Gray escribió:
On Sun, Jun 22, 2025 at 05:03:02PM -0400, Jose Maldonado wrote:
El 22/6/25 a las 16:20, Stuart Henderson escribió:
On 2025/06/22 15:50, Jose Maldonado wrote:
El 22/6/25 a las 13:13, Stuart Henderson escribió:
On 2025/06/22 15:35, Vlad Biley wrote:
Hello. An Intel Arc A310 GPU (not limited to; more:
https://www.reddit.com/r/openbsd/comments/
1lbecs2/intel_arc_gpu_support/) does not work:
OpenBSD 7.7 (GENERIC.MP) #625: Sun Apr 13 08:30:20 MDT 2025
...
ppb3 at pci3 dev 0 function 0 "AMD 500 Series PCIE" rev 0x00: msi
pci4 at ppb3 bus 4
ppb4 at pci4 dev 0 function 0 vendor "Intel", unknown product 0x4fa1 rev 0x01
pci5 at ppb4 bus 5
ppb5 at pci5 dev 1 function 0 vendor "Intel", unknown product 0x4fa4 rev 0x00
pci6 at ppb5 bus 6
"Intel Arc A310" rev 0x05 at pci6 dev 0 function 0 not configured
ppb6 at pci5 dev 4 function 0 vendor "Intel", unknown product 0x4fa4 rev 0x00
pci7 at ppb6 bus 7
azalia0 at pci7 dev 0 function 0 vendor "Intel", unknown product 0x4f92 rev
0x00: msi
azalia0: no supported codecs
The full dmesg: https://pastebin.com/7SgSw1n3
Related: https://marc.info/?l=openbsd-tech&m=167507478105125&w=2
The diff from the post in "related" won't apply directly but you could
try what is basically rhe same thing by removing
.require_force_probe = 1,
from struct intel_device_info dg2_info in sys/dev/pci/drm/i915/i915_pci.c
If it doesn't work with 7.7 it may be worth trying again with -current
Hi! How this patch is unrelated with this issue?
It's not unrelated. It is related, but it doesn't apply to the current
source tree.
Ya, sure! The patch is from 2023, I apply this change by hand. Work, the GPU
is detected, but not fully initialized.
I should add that, unlike @tfrohwein, my machine doesn't have an iGPU, and
yet it's not properly initialized.
In @jsg's debug, he mentions that the VGA arbiter may not have the necessary
logic to handle multiple inteldrm instances on a single machine, and that
this would require more work.
DG2 remains disabled as there is some kind of memory corruption.
I put a card in a machine today to check again with -current:
inteldrm0 at pci12 dev 0 function 0 "Intel Arc Pro A40/A50" rev 0x05
drm0 at inteldrm0
inteldrm0: msi, DG2, gen 12
...
intel_slicemask_from_xehp_dssmask: stub
intel_slicemask_from_xehp_dssmask: stub
i915_resize_lmem_bar: stub
Data modified on freelist: word 1474236 of object 0xffff800000e76010 size 0x10
previous type free (invalid addr 0x1f6bde33e90226f6)
intel_slicemask_from_xehp_dssmask: stub
kernel: protection fault trap, code=0
Stopped at malloc+0x4ed: movq 0x8(%r14),%rax
ddb{1}> tr
malloc(8,91,d) at malloc+0x4ed
ttm_tt_init(ffff800000e47800,ffff800000e38480,c,2,0) at ttm_tt_init+0x8f
i915_ttm_tt_create(ffff800000e38480,0) at i915_ttm_tt_create+0x12c
ttm_tt_create(ffff800000e38480,1) at ttm_tt_create+0x65
ttm_bo_handle_move_mem(ffff800000e38480,ffff800000e6dc00,0,ffffffff83073bc8,ffffffff83073aa0)
at ttm_bo_handle_move_mem+0x9f
ttm_bo_validate(ffff800000e38480,ffffffff82a6d6a0,ffffffff83073bc8) at
ttm_bo_validate+0x174
ttm_bo_init_reserved(ffff8000002499d0,ffff800000e38480,1,ffffffff82a6d6a0,1,ffffffff83073bc8,29484ceb0a3ab638,1,ffff800000e38480)
at ttm_bo_init_reserved+0x1bf
__i915_gem_ttm_object_init(ffff8000002fb800,ffff800000e38480,ffffffffffffffff,1000,1000,21)
at __i915_gem_ttm_object_init+0x166
__i915_gem_object_create_region(ffff8000002fb800,ffffffffffffffff,1000,1000,20)
at __i915_gem_object_create_region+0x130
alloc_pt_lmem(ffff800000e71c00,1000) at alloc_pt_lmem+0x35
gen8_ppgtt_create(ffff800000304000,20) at gen8_ppgtt_create+0x2c4
intel_gt_init(ffff800000304000) at intel_gt_init+0x1c2
i915_gem_init(ffff800000247000) at i915_gem_init+0x314
i915_driver_probe(ffff800000247000,0) at i915_driver_probe+0x90d
inteldrm_attachhook(ffff800000247000) at inteldrm_attachhook+0x50
config_process_deferred_mountroot() at config_process_deferred_mountroot+0x64
main(e) at main+0x861
end trace frame: 0x0, count: -17
I see the error is the same as the one shown previously when you spoke
with @tfrohwein, back in 2023. Back then, @tfrohwein showed that the
error appeared in:
xehp_load_dss_mask: chunk
xehp_load_dss_mask: chunk
intel_slicemask_from_xehp_dssmask: chunk
intel_slicemask_from_xehp_dssmask: chunk
i915_resize_lmem_bar: chunk
uvm_fault(0xffffffff825181f8, 0x10, 0, 1) -> e
--
**********************************************************
Dios en su Cielo, todo bien en la Tierra
