[PATCH] powerpc: move hmi.c to arch/powerpc/kvm/

hmi.c functions are unused unless sibling_subcore_state is nonzero, and
that in turn happens only if KVM is in use.  So move the code to
arch/powerpc/kvm/, putting it under CONFIG_KVM_BOOK3S_64_HANDLER
rather than CONFIG_PPC_BOOK3S_64.  The sibling_subcore_state is also
included in struct paca_struct only if KVM is supported by the kernel.

Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mahesh Salgaonkar <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: kvm-ppc@vger.kernel.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
	It would be nice to have this in 4.8, to minimize any 4.9 conflicts.
	Build-tested only, with and without KVM enabled.

 arch/powerpc/include/asm/hmi.h                     |  2 +-
 arch/powerpc/include/asm/paca.h                    | 10 +++++-----
 arch/powerpc/kernel/Makefile                       |  2 +-
 arch/powerpc/kvm/Makefile                          |  1 +
 arch/powerpc/{kernel/hmi.c => kvm/book3s_hv_hmi.c} |  0
 5 files changed, 8 insertions(+), 7 deletions(-)
 rename arch/powerpc/{kernel/hmi.c => kvm/book3s_hv_hmi.c} (100%)

diff --git a/arch/powerpc/include/asm/hmi.h b/arch/powerpc/include/asm/hmi.h
index 88b4901ac4ee..d3b6ad6e137c 100644
--- a/arch/powerpc/include/asm/hmi.h
+++ b/arch/powerpc/include/asm/hmi.h

@@ -21,7 +21,7 @@
 #ifndef __ASM_PPC64_HMI_H__
 #define __ASM_PPC64_HMI_H__
 
-#ifdef CONFIG_PPC_BOOK3S_64
+#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
 
 #define	CORE_TB_RESYNC_REQ_BIT		63
 #define MAX_SUBCORE_PER_CORE		4

diff --git a/arch/powerpc/include/asm/paca.h b/arch/powerpc/include/asm/paca.h
index 148303e7771f..625321e7e581 100644
--- a/arch/powerpc/include/asm/paca.h
+++ b/arch/powerpc/include/asm/paca.h

@@ -183,11 +183,6 @@ struct paca_struct {
 	 */
 	u16 in_mce;
 	u8 hmi_event_available;		 /* HMI event is available */
-	/*
-	 * Bitmap for sibling subcore status. See kvm/book3s_hv_ras.c for
-	 * more details
-	 */
-	struct sibling_subcore_state *sibling_subcore_state;
 #endif
 
 	/* Stuff for accurate time accounting */

@@ -202,6 +197,11 @@ struct paca_struct {
 	struct kvmppc_book3s_shadow_vcpu shadow_vcpu;
 #endif
 	struct kvmppc_host_state kvm_hstate;
+	/*
+	 * Bitmap for sibling subcore status. See kvm/book3s_hv_ras.c for
+	 * more details
+	 */
+	struct sibling_subcore_state *sibling_subcore_state;
 #endif
 };
 

diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile
index b2027a5cf508..fe4c075bcf50 100644
--- a/arch/powerpc/kernel/Makefile
+++ b/arch/powerpc/kernel/Makefile

@@ -41,7 +41,7 @@ obj-$(CONFIG_VDSO32)		+= vdso32/
 obj-$(CONFIG_HAVE_HW_BREAKPOINT)	+= hw_breakpoint.o
 obj-$(CONFIG_PPC_BOOK3S_64)	+= cpu_setup_ppc970.o cpu_setup_pa6t.o
 obj-$(CONFIG_PPC_BOOK3S_64)	+= cpu_setup_power.o
-obj-$(CONFIG_PPC_BOOK3S_64)	+= mce.o mce_power.o hmi.o
+obj-$(CONFIG_PPC_BOOK3S_64)	+= mce.o mce_power.o
 obj-$(CONFIG_PPC_BOOK3E_64)	+= exceptions-64e.o idle_book3e.o
 obj-$(CONFIG_PPC64)		+= vdso64/
 obj-$(CONFIG_ALTIVEC)		+= vecemu.o

diff --git a/arch/powerpc/kvm/Makefile b/arch/powerpc/kvm/Makefile
index 1f9e5529e692..855d4b95d752 100644
--- a/arch/powerpc/kvm/Makefile
+++ b/arch/powerpc/kvm/Makefile

@@ -78,6 +78,7 @@ kvm-book3s_64-builtin-xics-objs-$(CONFIG_KVM_XICS) := \
 
 ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
+	book3s_hv_hmi.o \
 	book3s_hv_rmhandlers.o \
 	book3s_hv_rm_mmu.o \
 	book3s_hv_ras.o \

diff --git a/arch/powerpc/kernel/hmi.c b/arch/powerpc/kvm/book3s_hv_hmi.c
similarity index 100%
rename from arch/powerpc/kernel/hmi.c
rename to arch/powerpc/kvm/book3s_hv_hmi.c
-- 
1.8.3.1

Paolo Bonzini [off-list ref] writes:

hmi.c functions are unused unless sibling_subcore_state is nonzero, and
that in turn happens only if KVM is in use.  So move the code to
arch/powerpc/kvm/, putting it under CONFIG_KVM_BOOK3S_64_HANDLER
rather than CONFIG_PPC_BOOK3S_64.  The sibling_subcore_state is also
included in struct paca_struct only if KVM is supported by the kernel.

Ok. Initially I was concerned because there are a bunch of non-KVM
related HMI causes (e.g. the CAPP will raise an HMI if it loses the link
to the CAPI card.)
https://github.com/open-power/skiboot/blob/master/core/hmi.c lists lots
of HMIs created by hardware events.

Having said that, you're right that this particular file is KVM
specific.

Reviewed-by: Daniel Axtens <redacted>

Mahesh: is there a way to cause the TB to desynchronise and then test if
this resynchronisation works?

Regards,
Daniel

2016-08-03 13:36+0200, Paolo Bonzini:

hmi.c functions are unused unless sibling_subcore_state is nonzero, and
that in turn happens only if KVM is in use.  So move the code to
arch/powerpc/kvm/, putting it under CONFIG_KVM_BOOK3S_64_HANDLER
rather than CONFIG_PPC_BOOK3S_64.  The sibling_subcore_state is also
included in struct paca_struct only if KVM is supported by the kernel.

Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mahesh Salgaonkar <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: kvm-ppc@vger.kernel.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
	It would be nice to have this in 4.8, to minimize any 4.9 conflicts.
	Build-tested only, with and without KVM enabled.

If you agree with the change, would you prefer to have this patch merged
through the PowerPC tree?

(There are no expected conflicts with -rc2 KVM tree.)

Thanks.

Radim Kr=C4=8Dm=C3=A1=C5=99 [off-list ref] writes:

2016-08-03 13:36+0200, Paolo Bonzini:

quoted

hmi.c functions are unused unless sibling_subcore_state is nonzero, and
that in turn happens only if KVM is in use.  So move the code to
arch/powerpc/kvm/, putting it under CONFIG_KVM_BOOK3S_64_HANDLER
rather than CONFIG_PPC_BOOK3S_64.  The sibling_subcore_state is also
included in struct paca_struct only if KVM is supported by the kernel.
=20
Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mahesh Salgaonkar <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: kvm-ppc@vger.kernel.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
	It would be nice to have this in 4.8, to minimize any 4.9 conflicts.
	Build-tested only, with and without KVM enabled.

If you agree with the change, would you prefer to have this patch merged
through the PowerPC tree?

Yeah I'll pull it into my fixes branch.

cheers

Michael Ellerman [off-list ref] writes:

Radim Kr=C4=8Dm=C3=A1=C5=99 [off-list ref] writes:

quoted

2016-08-03 13:36+0200, Paolo Bonzini:

quoted

hmi.c functions are unused unless sibling_subcore_state is nonzero, and
that in turn happens only if KVM is in use.  So move the code to
arch/powerpc/kvm/, putting it under CONFIG_KVM_BOOK3S_64_HANDLER
rather than CONFIG_PPC_BOOK3S_64.  The sibling_subcore_state is also
included in struct paca_struct only if KVM is supported by the kernel.
=20
Cc: Paul Mackerras <redacted>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mahesh Salgaonkar <redacted>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: kvm-ppc@vger.kernel.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
	It would be nice to have this in 4.8, to minimize any 4.9 conflicts.
	Build-tested only, with and without KVM enabled.

If you agree with the change, would you prefer to have this patch merged
through the PowerPC tree?

Yeah I'll pull it into my fixes branch.

And then I pulled it out again:

  (.text+0x157fc): undefined reference to `.wait_for_subcore_guest_exit'
  (.text+0x15834): undefined reference to `.wait_for_tb_resync'
  make[1]: *** [vmlinux] Error 1

http://kisskb.ellerman.id.au/kisskb/buildresult/12775422/

cheers