I see, the issue is the generated ifunc trampolines only have BTI landing pads when the note is in all object files.

The change is correct. I think it is more likely that rebuilding the kernel with a consistent toolchain is the fix for the BTI exception.

How does it fix the Branch Target exception? The kernel doesn't care about the note, it's just there to ensure all source files are built with BTI enabled.

Should the dma sync operations be BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD and BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE given virtio may read from & write to the buffer.

Where is the thunderbolt device defined? I don't see it in main.

Should the alignment for the rx & tx DMA tags be VTNET_ETHER_ALIGN?

In D54959#1259073, @jrtc27 wrote:

virtio_blk calls bus_dma_tag_set_iommu on PowerPC; would that not be needed for these new tags?

Do we want to zero the CPU for PPIs & IPIs? I expect that check should be before the handlers == 0 check as they may be sent to all CPUs.

These should be the same as Linux. Some userspace software defines these values if they are not already defined.

This appears to only be a problem for binutils for mov with :abs_g0_nc:

It looks like this is missing the pmap changes for the sanitizer kernels.

Add a comment

What is the overhead form all the extra wakeups this adds? I'd prefer we add a way for userspace to tell the kernel to perform an update.

In D55075#1258873, @markj wrote:

Does D55079 fix the problem for you?

The only user of get_kernel_reg_iss_masked that I know about is vmm, however get_kernel_reg_iss is used via the get_kernel_reg macro in many places, see D55105 for the full list.

D55103 moves it earlier, along with other cleanups.