markmi
commented
1 year ago Looks like there are later issues such as drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c: return drm_compat_ioctl(filp, cmd, arg); being in use with drm_compat_ioctl defined as NULL : drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c:43:26: error: called object type 'void *' is not a function or function pointer . It suggests that far more is just not present yet for aarch64 (from what I can tell).
So it appears my idea for trying to test reproducibility on aarch64 of what is reported on amd64 is just premature at this point.
It is not clear to me if this should be left open vs. closed (as aarch64-not-near-completed-yet?).
Describe the bug drivers/gpu/drm/drm_cache.c:362:10: error: call to undeclared function 'in_interrupt' (Note: linux/preempt.h provides in_interrupt when included )
FreeBSD version =>> Building graphics/drm-515-kmod build started at Tue Mar 7 11:27:15 UTC 2023 port directory: /usr/ports/graphics/drm-515-kmod package name: drm-515-kmod-5.15.25 building for: FreeBSD main-arm64-default-job-04 14.0-CURRENT FreeBSD 14.0-CURRENT 1400081 arm64 maintained by: x11@FreeBSD.org Makefile ident: Poudriere version: 3.2.8-23-ga7f8d188 Host OSVERSION: 1400073 Jail OSVERSION: 1400081
(I'm referencing the official package-builder failure. My private build attempt got the same but the official build attempt is a better context to reference, avoiding questions about what might be different in my personal context.)
PCI Info NA: Build time issue
DRM KMOD version package name: drm-515-kmod-5.15.25
To Reproduce Have poudriere(-devel) try to build graphics/drm-515-kmod on/for 14.0-CURRENT.
Screenshots NA
Additional context The official package-builder failure information was extracted from: http://ampere2.nyi.freebsd.org/data/main-arm64-default/p64e3eb722c17_s7fc82fd1f8/logs/errors/drm-515-kmod-5.15.25.log
I do not know if completing the
#include <linux/preempt.h>usage would be sufficient to allow the build vs. if there is some next issue also present. For reference, I found the file viasys/compat/linuxkpi/common/include/linux/preempt.hin FreeBSD.I'll note that I was attempting to set up an arm64 context somewhat analogous to an amd64 FreeBSD bug report that seems to involve amdgpu having been loaded. (I've a serial console context for arm64 that I do not have for amd64 --and have other reasons to avoid messing with the amd64 context that I've access to.) I do not normally use any drm-*-kmod . I was just going to see if I got some replication(s) analogous to the amd64 bug report vs. not.