Implement kvm_kerndisp

[luporl]

Summary

This change adds a new libkvm function, kvm_kerndisp(), that can be used to retrieve the kernel displacement (current/load address minus original base address).

This would be used by kgdb, to find out how to relocate kernel symbols (if needed).

[jhb]

Probably should add the new function to the lvm(3) manpage.

[luporl]

Add man page for kvm_kerndisp

[jhb]

lib/libkvm/kvm.h

127 ↗

(On Diff #67818)

Hmm, is kvaddr_t the better return type? That is, it's a delta in virtual addresses, not file offsets?

I see, it is signed and kvaddr_t is unsigned. I would be tempted to add a kvssize_t that is a int64_t so it's clear that it's a kernel virtual address delta rather than a file offset delta.

lib/libkvm/kvm_kerndisp.3

43 ↗

(On Diff #67818)

Perhaps:

.fn kvm_kerndsp
returns the number of bytes by which the kernel referenced by
.Fa kd
is displaced.

46 ↗

(On Diff #67818)

Possibly:

This is the difference between the kernel's base virtual address at run time
and the kernel base virtual address specified in the kernel image file.

(I think it's good to note that this is a virtual address displacement explicitly rather than a PA displacement)

51 ↗

(On Diff #67818)

If you use my suggestion for the first sentence, s/was/is/ to match the tense change.

[luporl]

Address review's comments

• change kvm_kerndisp() return type to kvssize_t
• improve man page text

[luporl]

lib/libkvm/kvm.h

127 ↗

(On Diff #67818)

Correct, the returned value is a delta in virtual addresses.
I've replaced the type by the new kvssize_t.

lib/libkvm/kvm_kerndisp.3

46 ↗

(On Diff #67818)

Right, this description is more clear and easier to understand.

[jhb]

sys/sys/types.h

263 ↗

(On Diff #67822)

My bad, I should have looked first! Since we already have ksize_t, I guess it would be best to call it kssize_t without the 'v'. You can commit it after fixing that without needing another round of review though.