Thread (56 messages) flat view 56 messages, 5 authors, 5h ago

Re: [PATCH v7 02/23] KVM: Make device name configurable

From: Steffen Eiden <seiden@linux.ibm.com>
Date: 2026-09-15 15:13:28
Also in: kvm, kvmarm, linux-s390, lkml

On Thu, Sep 03, 2026 at 01:58:17PM -0700, Sean Christopherson wrote:
On Thu, Sep 03, 2026, Paolo Bonzini wrote:
quoted
On Thu, Sep 3, 2026 at 5:54 PM Sean Christopherson [off-list ref] wrote:
quoted
quoted
I don't want KVM to be the. one that does things differently *once more*.
Too late :-D

arch/arm64/kvm/hyp/nvhe/Makefile:ccflags-y := -D__KVM_NVHE_HYPERVISOR__ -D__DISABLE_EXPORTS -D__DISABLE_TRACE_MMIO__
arch/arm64/kvm/hyp/vhe/Makefile:ccflags-y := -D__KVM_VHE_HYPERVISOR__
Sure, VHE/nVHE is fair game because the nVHE case is in the same boat
as vDSO, arch/*/boot, and so on.

Looking at a grep for (cflags|CFLAGS).*\-D most are -DDEBUG,
-DDISABLE_BRANCH_PROFILING, -DDEFAULT_SYMBOL_NAMESPACE, etc. and in
most cases they seem like a bad idea. Don't do it for KVM please;
requiring -D for all architectures and for random pieces of config,
when you can put it in asm/kvm_host.h, is just as bad if not worse.
Boo, it's never fun when a parent shows up and takes away your toys :-D
:) 

I will then create a new patch:

I will move config options to asm/kvm_host.h
It makes more sense to me as well to do this in the header.

Shean identified: 

KVM_MMU_LOCKLESS_AGING
KVM_GENERIC_PRE_FAULT_MEMORY
HAVE_KVM_MSI
HAVE_KVM_READONLY_MEM
HAVE_KVM_NO_POLL
HAVE_KVM_INVALID_WAKEUPS

I will check the Kconfig files if I find more.

Similary to Kconfig I want to implement them as opt-in.

While I am at it I would streamline the naming and remove the HAVE_
prefix.

<arch>/include/asm/kvm_host.h

#define KVM_GENERIC_PRE_FAULT_MEMORY
#define KVM_INVALID_WAKEUPS
#define KVM_MMU_LOCKLESS_AGING
#define KVM_MSI
#define KVM_NO_POLL
#define KVM_READONLY_MEM

(similar to the MMIO patch)

Any objections for doing this in one patch?

The patch changes the Kconfigs, asm/kvm_host.h and removes the
CONFIG_ prefixes of the configs in the code.

I would handle MMIO in a different patch as this is the only config that
also modifies the Makefile but I am defenitle not aggainst merging it
into the new conversion patch.


If there are no objections I will use the Makefilevariable approach for
KVM_DEV_NAME and KVM_DEV_MINOR. It feels wrong for me to define those in
a header. But defenitely, the devname and minor should be defined at
compile time. 

	Steffen
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help