[PATCH] powerpc: Fix booting P9 hash with CONFIG_PPC_RADIX_MMU=N

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot on
a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patching
in much of the asm code (ie. slb_miss_realmode)

This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from being
set from ibm.pa-features.

We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.

Signed-off-by: Michael Neuling <redacted>
---
 arch/powerpc/kernel/prom.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index 40c4887c27..f830562974 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c

@@ -161,7 +161,9 @@ static struct ibm_pa_feature {
 	{ .pabyte = 0,  .pabit = 3, .cpu_features  = CPU_FTR_CTRL },
 	{ .pabyte = 0,  .pabit = 6, .cpu_features  = CPU_FTR_NOEXECUTE },
 	{ .pabyte = 1,  .pabit = 2, .mmu_features  = MMU_FTR_CI_LARGE_PAGE },
+#ifdef CONFIG_PPC_RADIX_MMU
 	{ .pabyte = 40, .pabit = 0, .mmu_features  = MMU_FTR_TYPE_RADIX },
+#endif
 	{ .pabyte = 1,  .pabit = 1, .invert = 1, .cpu_features = CPU_FTR_NODSISRALIGN },
 	{ .pabyte = 5,  .pabit = 0, .cpu_features  = CPU_FTR_REAL_LE,
 				    .cpu_user_ftrs = PPC_FEATURE_TRUE_LE },

-- 
2.11.0

Michael Neuling [off-list ref] writes:

Instead can we do that feature removal in mmu_early_init_devtree. ie, something like

diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index 8f6f2a173e47..8f43f3827bac 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c

@@ -394,7 +394,7 @@ void __init mmu_early_init_devtree(void)
 {
 	/* Disable radix mode based on kernel command line. */
 	/* We don't yet have the machinery to do radix as a guest. */
-	if (disable_radix || !(mfmsr() & MSR_HV))
+	if (!IS_ENABLED(CONFIG_PPC_RADIX_MMU) || disable_radix || !(mfmsr() & MSR_HV))
 		cur_cpu_spec->mmu_features &= ~MMU_FTR_TYPE_RADIX;
 
 	/*

On Wed, 2017-05-24 at 14:26 +0530, Aneesh Kumar K.V wrote:

Michael Neuling [off-list ref] writes:
=20

quoted

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot on
a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patching
in much of the asm code (ie. slb_miss_realmode)
=20
This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from being
set from ibm.pa-features.
=20
We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.
=20
Signed-off-by: Michael Neuling <redacted>
---
=C2=A0arch/powerpc/kernel/prom.c | 2 ++
=C2=A01 file changed, 2 insertions(+)
=20

diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index 40c4887c27..f830562974 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c

@@ -161,7 +161,9 @@ static struct ibm_pa_feature {

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 3, .cpu_features=C2=A0=C2=

=A0=3D CPU_FTR_CTRL },

quoted

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 6, .cpu_features=C2=A0=C2=

=A0=3D CPU_FTR_NOEXECUTE },

quoted

=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 2, .mmu_features=C2=A0=C2=

=A0=3D MMU_FTR_CI_LARGE_PAGE

quoted

},
+#ifdef CONFIG_PPC_RADIX_MMU
=C2=A0	{ .pabyte =3D 40, .pabit =3D 0, .mmu_features=C2=A0=C2=A0=3D MMU=

_FTR_TYPE_RADIX },

quoted

+#endif
=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 1, .invert =3D 1, .cpu_fe=

atures =3D

quoted

CPU_FTR_NODSISRALIGN },
=C2=A0	{ .pabyte =3D 5,=C2=A0=C2=A0.pabit =3D 0, .cpu_features=C2=A0=C2=

=A0=3D CPU_FTR_REAL_LE,

quoted

=C2=A0				=C2=A0=C2=A0=C2=A0=C2=A0.cpu_user_ftrs =3D PPC_FEATURE_TRUE_L=

E },

quoted

--=C2=A0
2.11.0

=20
Instead can we do that feature removal in mmu_early_init_devtree. ie,
something like

It looks like mmu_early_init_devtree() gets called after parsing ibm,pa_fea=
tures
so that should work.

 MSR_HV))

What tree is this patch against?  Linus' tree doesn't look like that.

Mikey

=C2=A0		cur_cpu_spec->mmu_features &=3D ~MMU_FTR_TYPE_RADIX;
=C2=A0
=C2=A0	/*
=20

Michael Neuling [off-list ref] writes:

On Wed, 2017-05-24 at 14:26 +0530, Aneesh Kumar K.V wrote:

quoted

Michael Neuling [off-list ref] writes:
=20

quoted

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot on
a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patching
in much of the asm code (ie. slb_miss_realmode)
=20
This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from being
set from ibm.pa-features.
=20
We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.
=20
Signed-off-by: Michael Neuling <redacted>
---
=C2=A0arch/powerpc/kernel/prom.c | 2 ++
=C2=A01 file changed, 2 insertions(+)
=20

diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index 40c4887c27..f830562974 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c

@@ -161,7 +161,9 @@ static struct ibm_pa_feature {

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 3, .cpu_features=C2=A0=

=C2=A0=3D CPU_FTR_CTRL },

quoted

quoted

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 6, .cpu_features=C2=A0=

=C2=A0=3D CPU_FTR_NOEXECUTE },

quoted

quoted

=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 2, .mmu_features=C2=A0=

=C2=A0=3D MMU_FTR_CI_LARGE_PAGE

quoted

quoted

},
+#ifdef CONFIG_PPC_RADIX_MMU
=C2=A0	{ .pabyte =3D 40, .pabit =3D 0, .mmu_features=C2=A0=C2=A0=3D MM=

U_FTR_TYPE_RADIX },

quoted

quoted

+#endif
=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 1, .invert =3D 1, .cpu_f=

eatures =3D

quoted

quoted

CPU_FTR_NODSISRALIGN },
=C2=A0	{ .pabyte =3D 5,=C2=A0=C2=A0.pabit =3D 0, .cpu_features=C2=A0=

=C2=A0=3D CPU_FTR_REAL_LE,

quoted

quoted

=C2=A0				=C2=A0=C2=A0=C2=A0=C2=A0.cpu_user_ftrs =3D PPC_FEATURE_TRUE_=

LE },

quoted

quoted

--=C2=A0
2.11.0

=20
Instead can we do that feature removal in mmu_early_init_devtree. ie,
something like

It looks like mmu_early_init_devtree() gets called after parsing ibm,pa_f=

eatures

so that should work.

quoted

diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index 8f6f2a173e47..8f43f3827bac 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c

@@ -394,7 +394,7 @@ void __init mmu_early_init_devtree(v

oid)
=C2=A0{
=C2=A0	/* Disable radix mode based on kernel command line. */
=C2=A0	/* We don't yet have the machinery to do radix as a guest. */
-	if (disable_radix || !(mfmsr() & MSR_HV))

quoted

+	if (!IS_ENABLED(CONFIG_PPC_RADIX_MMU) || disable_radix || !(mfmsr() =

& MSR_HV))

What tree is this patch against?  Linus' tree doesn't look like that.

mpe/next at the time of this merge window. Do you want me to send a
complete patch against latest mpe/next ?

-aneesh

Michael Neuling [off-list ref] writes:

On Wed, 2017-05-24 at 14:26 +0530, Aneesh Kumar K.V wrote:

quoted

Michael Neuling [off-list ref] writes:
=20

quoted

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot on
a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patching
in much of the asm code (ie. slb_miss_realmode)
=20
This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from being
set from ibm.pa-features.
=20
We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.
=20
Signed-off-by: Michael Neuling <redacted>
---
=C2=A0arch/powerpc/kernel/prom.c | 2 ++
=C2=A01 file changed, 2 insertions(+)
=20

diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index 40c4887c27..f830562974 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c

@@ -161,7 +161,9 @@ static struct ibm_pa_feature {

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 3, .cpu_features=C2=A0=

=C2=A0=3D CPU_FTR_CTRL },

quoted

quoted

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 6, .cpu_features=C2=A0=

=C2=A0=3D CPU_FTR_NOEXECUTE },

quoted

quoted

=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 2, .mmu_features=C2=A0=

=C2=A0=3D MMU_FTR_CI_LARGE_PAGE

quoted

quoted

},
+#ifdef CONFIG_PPC_RADIX_MMU
=C2=A0	{ .pabyte =3D 40, .pabit =3D 0, .mmu_features=C2=A0=C2=A0=3D MM=

U_FTR_TYPE_RADIX },

quoted

quoted

+#endif
=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 1, .invert =3D 1, .cpu_f=

eatures =3D

quoted

quoted

CPU_FTR_NODSISRALIGN },
=C2=A0	{ .pabyte =3D 5,=C2=A0=C2=A0.pabit =3D 0, .cpu_features=C2=A0=

=C2=A0=3D CPU_FTR_REAL_LE,

quoted

quoted

=C2=A0				=C2=A0=C2=A0=C2=A0=C2=A0.cpu_user_ftrs =3D PPC_FEATURE_TRUE_=

LE },

quoted

quoted

--=C2=A0
2.11.0

=20
Instead can we do that feature removal in mmu_early_init_devtree. ie,
something like

It looks like mmu_early_init_devtree() gets called after parsing ibm,pa_f=

eatures

so that should work.

But why is doing it later preferable?

Mikey's patch means it will never be set at any point during boot, which
seems obviously better to me.

cheers

On Thursday 25 May 2017 11:46 AM, Michael Ellerman wrote:

Michael Neuling [off-list ref] writes:

quoted

On Wed, 2017-05-24 at 14:26 +0530, Aneesh Kumar K.V wrote:

quoted

Michael Neuling [off-list ref] writes:

quoted

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot on
a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patching
in much of the asm code (ie. slb_miss_realmode)

This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from being
set from ibm.pa-features.

We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.

Signed-off-by: Michael Neuling <redacted>
---
  arch/powerpc/kernel/prom.c | 2 ++
  1 file changed, 2 insertions(+)

diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
index 40c4887c27..f830562974 100644
--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c

@@ -161,7 +161,9 @@ static struct ibm_pa_feature {
  	{ .pabyte = 0,  .pabit = 3, .cpu_features  = CPU_FTR_CTRL },
  	{ .pabyte = 0,  .pabit = 6, .cpu_features  = CPU_FTR_NOEXECUTE },
  	{ .pabyte = 1,  .pabit = 2, .mmu_features  = MMU_FTR_CI_LARGE_PAGE

},
+#ifdef CONFIG_PPC_RADIX_MMU
  	{ .pabyte = 40, .pabit = 0, .mmu_features  = MMU_FTR_TYPE_RADIX },
+#endif
  	{ .pabyte = 1,  .pabit = 1, .invert = 1, .cpu_features =
CPU_FTR_NODSISRALIGN },
  	{ .pabyte = 5,  .pabit = 0, .cpu_features  = CPU_FTR_REAL_LE,
  				    .cpu_user_ftrs = PPC_FEATURE_TRUE_LE },
-- 
2.11.0

Instead can we do that feature removal in mmu_early_init_devtree. ie,
something like

It looks like mmu_early_init_devtree() gets called after parsing ibm,pa_features
so that should work.

But why is doing it later preferable?

Mikey's patch means it will never be set at any point during boot, which
seems obviously better to me.

My suggestion was w.r.t consolidating different ways of clearing 
RADIX_MMU feature flag into one place. Also I was not sure we want to 
conditionally parse the pa-feature device tree. Instead we parse it 
without any #ifdef in there and clear things which are not supported by 
the kernel either via command line or via Kconfig.

-aneesh

On Thu, 2017-05-25 at 08:23 +0530, Aneesh Kumar K.V wrote:

Michael Neuling [off-list ref] writes:
=20

quoted

On Wed, 2017-05-24 at 14:26 +0530, Aneesh Kumar K.V wrote:

quoted

Michael Neuling [off-list ref] writes:
=20

quoted

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot =

on

quoted

quoted

quoted

a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patc=

hing

quoted

quoted

quoted

in much of the asm code (ie. slb_miss_realmode)
=20
This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from be=

ing

quoted

quoted

quoted

set from ibm.pa-features.
=20
We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.
=20
Signed-off-by: Michael Neuling <redacted>
---
=C2=A0arch/powerpc/kernel/prom.c | 2 ++
=C2=A01 file changed, 2 insertions(+)
=20

diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.=

c

quoted

quoted

quoted

index 40c4887c27..f830562974 100644

--- a/arch/powerpc/kernel/prom.c
+++ b/arch/powerpc/kernel/prom.c

@@ -161,7 +161,9 @@ static struct ibm_pa_feature {

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 3, .cpu_features=C2=

=A0=C2=A0=3D CPU_FTR_CTRL },

quoted

quoted

quoted

=C2=A0	{ .pabyte =3D 0,=C2=A0=C2=A0.pabit =3D 6, .cpu_features=C2=

=A0=C2=A0=3D CPU_FTR_NOEXECUTE

quoted

quoted

quoted

},
=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 2, .mmu_features=C2=

=A0=C2=A0=3D

quoted

quoted

quoted

MMU_FTR_CI_LARGE_PAGE
},
+#ifdef CONFIG_PPC_RADIX_MMU
=C2=A0	{ .pabyte =3D 40, .pabit =3D 0, .mmu_features=C2=A0=C2=A0=3D=

 MMU_FTR_TYPE_RADIX

quoted

quoted

quoted

},
+#endif
=C2=A0	{ .pabyte =3D 1,=C2=A0=C2=A0.pabit =3D 1, .invert =3D 1, .cp=

u_features =3D

quoted

quoted

quoted

CPU_FTR_NODSISRALIGN },
=C2=A0	{ .pabyte =3D 5,=C2=A0=C2=A0.pabit =3D 0, .cpu_features=C2=

=A0=C2=A0=3D CPU_FTR_REAL_LE,

quoted

quoted

quoted

=C2=A0				=C2=A0=C2=A0=C2=A0=C2=A0.cpu_user_ftrs =3D
PPC_FEATURE_TRUE_LE },
--=C2=A0
2.11.0

=20
Instead can we do that feature removal in mmu_early_init_devtree. ie,
something like

=20
It looks like mmu_early_init_devtree() gets called after parsing
ibm,pa_features
so that should work.
=20

quoted

diff --git a/arch/powerpc/mm/init_64.c b/arch/powerpc/mm/init_64.c
index 8f6f2a173e47..8f43f3827bac 100644
--- a/arch/powerpc/mm/init_64.c
+++ b/arch/powerpc/mm/init_64.c

@@ -394,7 +394,7 @@ void __init mmu_early_init_devtree(v

oid)
=C2=A0{
=C2=A0	/* Disable radix mode based on kernel command line. */
=C2=A0	/* We don't yet have the machinery to do radix as a guest. */
-	if (disable_radix || !(mfmsr() & MSR_HV))

quoted

+	if (!IS_ENABLED(CONFIG_PPC_RADIX_MMU) || disable_radix ||
!(mfmsr() & MSR_HV))

=20
What tree is this patch against?=C2=A0=C2=A0Linus' tree doesn't look li=

ke that.

quoted

=20

=20
mpe/next at the time of this merge window. Do you want me to send a
complete patch against latest mpe/next ?

If you can test the broken case to make sure it fixes it... Sure!

Thanks,
Mikey

On Wed, 2017-05-24 at 07:03:26 UTC, Michael Neuling wrote:

Currently if you disable CONFIG_PPC_RADIX_MMU you'll crash on boot on
a P9. This is because we still set MMU_FTR_TYPE_RADIX via
ibm,pa-features and MMU_FTR_TYPE_RADIX is what's used for code patching
in much of the asm code (ie. slb_miss_realmode)

This patch fixes the problem by stopping MMU_FTR_TYPE_RADIX from being
set from ibm.pa-features.

We may eventually end up removing the CONFIG_PPC_RADIX_MMU option
completely but until then this fixes the issue.

Signed-off-by: Michael Neuling <redacted>

Applied to powerpc fixes, thanks.

https://git.kernel.org/powerpc/c/d957fb4d173647640a2b83e7c7e56a

cheers