XRT
XRT copied to clipboard
Port kernel driver to Linux 6.8
To be rebased on main once https://github.com/Xilinx/XRT/pull/8003 has landed.
I think the code would benefit from some review by DRM experts because I have 0 expertise. @superm1 @airlied
This is to have https://github.com/amd/xdna-driver working along
https://github.com/amd/xdna-driver/issues/3
keryell - is not a collaborator Can XRT admins please validate PR
/build
keryell - is not a collaborator Can XRT admins please validate PR
I have signed-off the commits now. This PR is not supposed to be squashed as its commits and their messages are independent.
I do not know if removing DRM_UNLOCKED which is no longer in use since Linux 6.3 is OK, since the driver is supposed to be compiled also on some antique kernels where it might be required.
I have no idea about the big picture of this driver.
I do not know if removing
DRM_UNLOCKEDwhich is no longer in use since Linux 6.3 is OK, since the driver is supposed to be compiled also on some antique kernels where it might be required. I have no idea about the big picture of this driver.
It should not be unconditionally removed. Maybe just add a Macro like #if KERNEL_VERSION(6, 8, 0) <= LINUX_VERSION_CODE #define DRM_UNLOCKED 0 #endif
It should not be unconditionally removed. Maybe just add a Macro
Yes I can do that.
This is not clear why the test on pcie-hw-tests (xilinx_u250_gen3x16_xdma_4_1_202210_1) is not working.
How to investigate this?
XOAH shows ERROR: Not enough host mem. Please check grub settings.: Cannot allocate memory on pcie-hw-tests (xilinx_u250_gen3x16_xdma_4_1_202210_1).
OK, the problem was that in af7a239 (#8005) I forgot to replace the << by *, so there was a UB operation due to the overflow, probably leading 0, so the allocation was always too big...
Ahem.
This PR is not supposed to be squashed as its commits and their messages are independent.
So, what is the problem on this PR?
@stsoe Thanks!