Nouveau - Aug 2013 - [PATCH 2/2] drm/nv84/therm: ack any pending IRQ at init

From: Martin Peres <martin.peres at labri.fr>

Some vbioses have extra useless entries after "the end" of the table.
This is
problematic since all of the vbios I found with this issue redefine the
pwm freq divider to insane levels (52750 Hz instead of 2500), thus breaking
fan management.

The first solution to solve this mess would be to change the length of the
table. The solution I choose was simply to avoid setting the pwm freq twice
as the other redefinitions are harmless with our current parser.

Signed-off-by: Martin Peres <martin.peres at labbri.fr>
Reported-by: Mariusz Bialonczyk <manio at skyboo.net>
Tested-by: Mariusz Bialonczyk <manio at skyboo.net>
---
 drivers/gpu/drm/nouveau/core/subdev/bios/therm.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
b/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
index 22a2057..22ac6db 100644
--- a/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
+++ b/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
@@ -184,7 +184,8 @@ nvbios_therm_fan_parse(struct nouveau_bios *bios,
 			cur_trip->fan_duty = value;
 			break;
 		case 0x26:
-			fan->pwm_freq = value;
+			if (!fan->pwm_freq)
+				fan->pwm_freq = value;
 			break;
 		case 0x3b:
 			fan->bump_period = value;
-- 
1.8.4

From: Martin Peres <martin.peres at labri.fr>

This is safe because ptherm hasn't been configured yet and will be a
little further down the initialization path. Ptherm should be safe
regarding to runtime reconfiguration.

Signed-off-by: Martin Peres <martin.peres at labri.fr>
---
 drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
b/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
index 42ba633..8615d62 100644
--- a/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
+++ b/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
@@ -209,13 +209,30 @@ nv84_therm_ctor(struct nouveau_object *parent,
 	return nouveau_therm_preinit(&priv->base.base);
 }
 
+static int
+nv84_therm_init(struct nouveau_object *object)
+{
+	struct nv84_therm_priv *priv = (void *)object;
+	int ret;
+
+	ret = nouveau_therm_init(&priv->base.base);
+	if (ret)
+		return ret;
+
+	/* ACK ptherm IRQs */
+	nv_wr32(object, 0x20100, 0xffffffff);
+	nv_wr32(object, 0x1100, 0x10000); /* PBUS */
+
+	return 0;
+}
+
 struct nouveau_oclass
 nv84_therm_oclass = {
 	.handle = NV_SUBDEV(THERM, 0x84),
 	.ofuncs = &(struct nouveau_ofuncs) {
 		.ctor = nv84_therm_ctor,
 		.dtor = _nouveau_therm_dtor,
-		.init = _nouveau_therm_init,
+		.init = nv84_therm_init,
 		.fini = _nouveau_therm_fini,
 	},
 };
-- 
1.8.4

From: Martin Peres <martin.peres at labri.fr>

This is safe because ptherm hasn't been configured yet and will be a
little further down the initialization path. Ptherm should be safe
regarding to runtime reconfiguration.

v2:
- do not limit this patch to nv84-a3 and make it nv84+

Signed-off-by: Martin Peres <martin.peres at labri.fr>
---
 drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c | 19 ++++++++++++++++++-
 drivers/gpu/drm/nouveau/core/subdev/therm/nva3.c |  4 ++++
 drivers/gpu/drm/nouveau/core/subdev/therm/nvd0.c |  4 ++++
 3 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
b/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
index 42ba633..8615d62 100644
--- a/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
+++ b/drivers/gpu/drm/nouveau/core/subdev/therm/nv84.c
@@ -209,13 +209,30 @@ nv84_therm_ctor(struct nouveau_object *parent,
 	return nouveau_therm_preinit(&priv->base.base);
 }
 
+static int
+nv84_therm_init(struct nouveau_object *object)
+{
+	struct nv84_therm_priv *priv = (void *)object;
+	int ret;
+
+	ret = nouveau_therm_init(&priv->base.base);
+	if (ret)
+		return ret;
+
+	/* ACK ptherm IRQs */
+	nv_wr32(object, 0x20100, 0xffffffff);
+	nv_wr32(object, 0x1100, 0x10000); /* PBUS */
+
+	return 0;
+}
+
 struct nouveau_oclass
 nv84_therm_oclass = {
 	.handle = NV_SUBDEV(THERM, 0x84),
 	.ofuncs = &(struct nouveau_ofuncs) {
 		.ctor = nv84_therm_ctor,
 		.dtor = _nouveau_therm_dtor,
-		.init = _nouveau_therm_init,
+		.init = nv84_therm_init,
 		.fini = _nouveau_therm_fini,
 	},
 };
diff --git a/drivers/gpu/drm/nouveau/core/subdev/therm/nva3.c
b/drivers/gpu/drm/nouveau/core/subdev/therm/nva3.c
index d11a7c4..f740eeb 100644
--- a/drivers/gpu/drm/nouveau/core/subdev/therm/nva3.c
+++ b/drivers/gpu/drm/nouveau/core/subdev/therm/nva3.c
@@ -60,6 +60,10 @@ nva3_therm_init(struct nouveau_object *object)
 	}
 	nv_mask(priv, 0x00e720, 0x00000002, 0x00000000);
 
+	/* ACK ptherm IRQs */
+	nv_wr32(object, 0x20100, 0xffffffff);
+	nv_wr32(object, 0x1100, 0x10000); /* PBUS */
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/nouveau/core/subdev/therm/nvd0.c
b/drivers/gpu/drm/nouveau/core/subdev/therm/nvd0.c
index 54c28bd..d9b5103 100644
--- a/drivers/gpu/drm/nouveau/core/subdev/therm/nvd0.c
+++ b/drivers/gpu/drm/nouveau/core/subdev/therm/nvd0.c
@@ -114,6 +114,10 @@ nvd0_therm_init(struct nouveau_object *object)
 	}
 	nv_mask(priv, 0x00e720, 0x00000002, 0x00000000);
 
+	/* ACK ptherm IRQs */
+	nv_wr32(object, 0x20100, 0xffffffff);
+	nv_wr32(object, 0x1100, 0x10000); /* PBUS */
+
 	return 0;
 }
 
-- 
1.8.4

On 31/08/2013 01:58, Martin Peres wrote:
> From: Martin Peres <martin.peres at labri.fr>
>
> Some vbioses have extra useless entries after "the end" of the
table. This is
> problematic since all of the vbios I found with this issue redefine the
> pwm freq divider to insane levels (52750 Hz instead of 2500), thus breaking
> fan management.
>
> The first solution to solve this mess would be to change the length of the
> table. The solution I choose was simply to avoid setting the pwm freq twice
> as the other redefinitions are harmless with our current parser.
>
> Signed-off-by: Martin Peres <martin.peres at labbri.fr>
Sorry, the email should be ammended to martin.peres at labri.fr (only one b).

On Sat, Aug 31, 2013 at 9:58 AM, Martin Peres <martin.peres at free.fr>
wrote:
> From: Martin Peres <martin.peres at labri.fr>
>
> Some vbioses have extra useless entries after "the end" of the
table. This is
> problematic since all of the vbios I found with this issue redefine the
> pwm freq divider to insane levels (52750 Hz instead of 2500), thus breaking
> fan management.
>
> The first solution to solve this mess would be to change the length of the
> table. The solution I choose was simply to avoid setting the pwm freq twice
> as the other redefinitions are harmless with our current parser.
I'm still *really* not convinced we're using the thermal table
correctly....

However, this makes sense for our current handling, so, merged.

Thanks,
Ben.
>
> Signed-off-by: Martin Peres <martin.peres at labbri.fr>
> Reported-by: Mariusz Bialonczyk <manio at skyboo.net>
> Tested-by: Mariusz Bialonczyk <manio at skyboo.net>
> ---
>  drivers/gpu/drm/nouveau/core/subdev/bios/therm.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
b/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
> index 22a2057..22ac6db 100644
> --- a/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
> +++ b/drivers/gpu/drm/nouveau/core/subdev/bios/therm.c
> @@ -184,7 +184,8 @@ nvbios_therm_fan_parse(struct nouveau_bios *bios,
>                         cur_trip->fan_duty = value;
>                         break;
>                 case 0x26:
> -                       fan->pwm_freq = value;
> +                       if (!fan->pwm_freq)
> +                               fan->pwm_freq = value;
>                         break;
>                 case 0x3b:
>                         fan->bump_period = value;
> --
> 1.8.4
>
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau

On 04/09/2013 03:55, Ben Skeggs wrote:
> On Sat, Aug 31, 2013 at 9:58 AM, Martin Peres <martin.peres at
free.fr> wrote:
>> From: Martin Peres <martin.peres at labri.fr>
>>
>> Some vbioses have extra useless entries after "the end" of
the table. This is
>> problematic since all of the vbios I found with this issue redefine the
>> pwm freq divider to insane levels (52750 Hz instead of 2500), thus
breaking
>> fan management.
>>
>> The first solution to solve this mess would be to change the length of
the
>> table. The solution I choose was simply to avoid setting the pwm freq
twice
>> as the other redefinitions are harmless with our current parser.
> I'm still *really* not convinced we're using the thermal table
correctly....
>
> However, this makes sense for our current handling, so, merged.
>
> Thanks,
> Ben.
Well, I do agree that the first versions of the vbios table really felt 
like a script but this isn't the case anymore (not sure how I tested 
this though as it was more than a year ago).