[Bug 295788] graphics/drm-kmod: i915kms Hardware acceleration not working for Intel Braswell since >drm-510-kmod

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=295788

--- Comment #58 from Roger Olofsson <[email protected]> ---
Some fun with lldb thread backtrace all and awk+sed.

First the output from the working drm-510-kmod (breakpoint at
i915_gem_create_context)

#0: libc.so.7`__sys_ioctl at ioctl.S:4     
#1: crocus_dri.so`intel_ioctl at intel_gem.h:83:15     
#2: crocus_dri.so`i915_gem_set_context_param at intel_gem.c:183:11     
#3: crocus_dri.so`intel_gem_set_context_param at intel_gem.c:92:11     
#4: crocus_dri.so`crocus_create_hw_context at crocus_bufmgr.c:1545:4     

Second the same output for drm-515-kmod with same breakpoint

#0: libc.so.7`__sys_ioctl at ioctl.S:4     
#1: crocus_dri.so`intel_ioctl at intel_gem.h:83:15     
#2: crocus_dri.so`i915_gem_create_context at intel_gem.c:35:8     
#3: crocus_dri.so`intel_gem_create_context at intel_gem.c:67:11     
#4: crocus_dri.so`crocus_create_hw_context at crocus_bufmgr.c:1525:9  

Frames #5-20 are identical.

I noticed here that when I singlestep drm-510 it actually breaks at the same
spot as drm-515 but when drm-510 runs on its own it works fine. This makes me
wonder if it's the si that breaks drm-515 or if it's the ioctl?

There's also the differences that drm-510 does gem_SET_context_param for intel
and i915 as opposed to drm-515 that does intel_gem_CREATE_context instead.

Roger

-- 
You are receiving this mail because:
You are the assignee for the bug.