[alsa-devel] [PATCH v6 1/2] PCI: Add a helper to check Power Resource Requirements _PR3 existence
A driver may want to know the existence of _PR3, to choose different runtime suspend behavior. A user will be add in next patch.
This is mostly the same as nouveau_pr3_present().
Reported-by: kbuild test robot lkp@intel.com Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com --- v6: - Only define the function when CONFIG_ACPI is set. v5: - Add wording suggestion from Bjorn. v4: - Let caller to find its upstream port device.
drivers/pci/pci.c | 18 ++++++++++++++++++ include/linux/pci.h | 2 ++ 2 files changed, 20 insertions(+)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index e7982af9a5d8..1df99d9e350e 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -5856,6 +5856,24 @@ int pci_set_vga_state(struct pci_dev *dev, bool decode, return 0; }
+#ifdef CONFIG_ACPI +bool pci_pr3_present(struct pci_dev *pdev) +{ + struct acpi_device *adev; + + if (acpi_disabled) + return false; + + adev = ACPI_COMPANION(&pdev->dev); + if (!adev) + return false; + + return adev->power.flags.power_resources && + acpi_has_method(adev->handle, "_PR3"); +} +EXPORT_SYMBOL_GPL(pci_pr3_present); +#endif + /** * pci_add_dma_alias - Add a DMA devfn alias for a device * @dev: the PCI device for which alias is added diff --git a/include/linux/pci.h b/include/linux/pci.h index f9088c89a534..1d15c5d49cdd 100644 --- a/include/linux/pci.h +++ b/include/linux/pci.h @@ -2310,9 +2310,11 @@ struct irq_domain *pci_host_bridge_acpi_msi_domain(struct pci_bus *bus);
void pci_msi_register_fwnode_provider(struct fwnode_handle *(*fn)(struct device *)); +bool pci_pr3_present(struct pci_dev *pdev); #else static inline struct irq_domain * pci_host_bridge_acpi_msi_domain(struct pci_bus *bus) { return NULL; } +static bool pci_pr3_present(struct pci_dev *pdev) { return false; } #endif
#ifdef CONFIG_EEH
Nvidia proprietary driver doesn't support runtime power management, so when a user only wants to use the integrated GPU, it's a common practice to let dGPU not to bind any driver, and let its upstream port to be runtime suspended. At the end of runtime suspension the port uses platform power management to disable power through _OFF method of power resource, which is listed by _PR3.
After commit b516ea586d71 ("PCI: Enable NVIDIA HDA controllers"), when the dGPU comes with an HDA function, the HDA won't be suspended if the dGPU is unbound, so the power resource can't be turned off by its upstream port driver.
Commit 37a3a98ef601 ("ALSA: hda - Enable runtime PM only for discrete GPU") only allows HDA to be runtime suspended once GPU is bound, to keep APU's HDA working.
However, HDA on dGPU isn't that useful if dGPU is not bound to any driver. So let's relax the runtime suspend requirement for dGPU's HDA function, to disable the power source to save lots of power.
BugLink: https://bugs.launchpad.net/bugs/1840835 Fixes: b516ea586d71 ("PCI: Enable NVIDIA HDA controllers") Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com --- v5, v6: - No change. v4: - Find upstream port, it's callee's responsibility now. v3: - Make changelog more clear. v2: - Change wording. - Rebase to Tiwai's branch. sound/pci/hda/hda_intel.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/sound/pci/hda/hda_intel.c b/sound/pci/hda/hda_intel.c index 240f4ca76391..e63b871343e5 100644 --- a/sound/pci/hda/hda_intel.c +++ b/sound/pci/hda/hda_intel.c @@ -1280,11 +1280,17 @@ static void init_vga_switcheroo(struct azx *chip) { struct hda_intel *hda = container_of(chip, struct hda_intel, chip); struct pci_dev *p = get_bound_vga(chip->pci); + struct pci_dev *parent; if (p) { dev_info(chip->card->dev, "Handle vga_switcheroo audio client\n"); hda->use_vga_switcheroo = 1; - chip->bus.keep_power = 1; /* cleared in either gpu_bound op or codec probe */ + + /* cleared in either gpu_bound op or codec probe, or when its + * upstream port has _PR3 (i.e. dGPU). + */ + parent = pci_upstream_bridge(p); + chip->bus.keep_power = parent ? !pci_pr3_present(parent) : 1; chip->driver_caps |= AZX_DCAPS_PM_RUNTIME; pci_dev_put(p); }
Dne 18. 10. 19 v 9:38 Kai-Heng Feng napsal(a):
Nvidia proprietary driver doesn't support runtime power management, so when a user only wants to use the integrated GPU, it's a common practice to let dGPU not to bind any driver, and let its upstream port to be runtime suspended. At the end of runtime suspension the port uses platform power management to disable power through _OFF method of power resource, which is listed by _PR3.
After commit b516ea586d71 ("PCI: Enable NVIDIA HDA controllers"), when the dGPU comes with an HDA function, the HDA won't be suspended if the dGPU is unbound, so the power resource can't be turned off by its upstream port driver.
Commit 37a3a98ef601 ("ALSA: hda - Enable runtime PM only for discrete GPU") only allows HDA to be runtime suspended once GPU is bound, to keep APU's HDA working.
However, HDA on dGPU isn't that useful if dGPU is not bound to any driver. So let's relax the runtime suspend requirement for dGPU's HDA function, to disable the power source to save lots of power.
This patch breaks the HDMI audio detection at least on some platforms (Lenovo P50 for example) with nouveau and the proprietary nvidia driver. Those laptops have the external HDMI/DP ports connected to dGPU instead the iGPU. The ACPI PR3 is set.
The runtime PM off fixes this problem:
echo on > /sys/bus/pci/devices/0000:01:00.1/power/control
But I don't think that it's the best solution. My proposal is to create a pr3 check blacklist to keep power for the HDMI audio for those machines. Also we may add a new module parameter for snd-hda-intel to control this. Other ideas?
Jaroslav
BugLink: https://bugs.launchpad.net/bugs/1840835 Fixes: b516ea586d71 ("PCI: Enable NVIDIA HDA controllers") Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com
v5, v6:
- No change.
v4:
- Find upstream port, it's callee's responsibility now.
v3:
- Make changelog more clear.
v2:
- Change wording.
- Rebase to Tiwai's branch. sound/pci/hda/hda_intel.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/sound/pci/hda/hda_intel.c b/sound/pci/hda/hda_intel.c index 240f4ca76391..e63b871343e5 100644 --- a/sound/pci/hda/hda_intel.c +++ b/sound/pci/hda/hda_intel.c @@ -1280,11 +1280,17 @@ static void init_vga_switcheroo(struct azx *chip) { struct hda_intel *hda = container_of(chip, struct hda_intel, chip); struct pci_dev *p = get_bound_vga(chip->pci);
- struct pci_dev *parent; if (p) { dev_info(chip->card->dev, "Handle vga_switcheroo audio client\n"); hda->use_vga_switcheroo = 1;
chip->bus.keep_power = 1; /* cleared in either gpu_bound op or codec probe */
/* cleared in either gpu_bound op or codec probe, or when its* upstream port has _PR3 (i.e. dGPU).*/parent = pci_upstream_bridge(p);chip->bus.keep_power = parent ? !pci_pr3_present(parent) : 1;
On Fri, 10 Jan 2020 10:43:26 +0100, Jaroslav Kysela wrote:
Dne 18. 10. 19 v 9:38 Kai-Heng Feng napsal(a):
Nvidia proprietary driver doesn't support runtime power management, so when a user only wants to use the integrated GPU, it's a common practice to let dGPU not to bind any driver, and let its upstream port to be runtime suspended. At the end of runtime suspension the port uses platform power management to disable power through _OFF method of power resource, which is listed by _PR3.
After commit b516ea586d71 ("PCI: Enable NVIDIA HDA controllers"), when the dGPU comes with an HDA function, the HDA won't be suspended if the dGPU is unbound, so the power resource can't be turned off by its upstream port driver.
Commit 37a3a98ef601 ("ALSA: hda - Enable runtime PM only for discrete GPU") only allows HDA to be runtime suspended once GPU is bound, to keep APU's HDA working.
However, HDA on dGPU isn't that useful if dGPU is not bound to any driver. So let's relax the runtime suspend requirement for dGPU's HDA function, to disable the power source to save lots of power.
This patch breaks the HDMI audio detection at least on some platforms (Lenovo P50 for example) with nouveau and the proprietary nvidia driver. Those laptops have the external HDMI/DP ports connected to dGPU instead the iGPU. The ACPI PR3 is set.
The runtime PM off fixes this problem:
echo on > /sys/bus/pci/devices/0000:01:00.1/power/control
But this will keep the power of the graphics chip on, and that's what the patch was supposed to "fix".
But I don't think that it's the best solution. My proposal is to create a pr3 check blacklist to keep power for the HDMI audio for those machines. Also we may add a new module parameter for snd-hda-intel to control this. Other ideas?
For nouveau, the best fix is to merge the audio component patch. This will make things working without fiddling with the power up/down. The patch has been pending over months under review in DRM side, unfortunately... Please pinging them for driving ahead.
For Nvidia, though, it's no path a binary-only stuff can go with, due to the GPL symbol of the component framework. Those guys know of it well, and they seem adding the temporary power up/down procedure by poking the proc file from the user-space side at the HDMI connection.
About a module option: I don't think it's much better than the sysfs toggle. You can set up a simple udev rule if needed, too.
thanks,
Takashi
JaroslavBugLink: https://bugs.launchpad.net/bugs/1840835 Fixes: b516ea586d71 ("PCI: Enable NVIDIA HDA controllers") Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com
v5, v6:
- No change.
v4:
- Find upstream port, it's callee's responsibility now.
v3:
- Make changelog more clear.
v2:
- Change wording.
- Rebase to Tiwai's branch. sound/pci/hda/hda_intel.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/sound/pci/hda/hda_intel.c b/sound/pci/hda/hda_intel.c index 240f4ca76391..e63b871343e5 100644 --- a/sound/pci/hda/hda_intel.c +++ b/sound/pci/hda/hda_intel.c @@ -1280,11 +1280,17 @@ static void init_vga_switcheroo(struct azx *chip) { struct hda_intel *hda = container_of(chip, struct hda_intel, chip); struct pci_dev *p = get_bound_vga(chip->pci);
- struct pci_dev *parent; if (p) { dev_info(chip->card->dev, "Handle vga_switcheroo audio client\n"); hda->use_vga_switcheroo = 1;
chip->bus.keep_power = 1; /* cleared in either gpu_bound op or codec probe */
/* cleared in either gpu_bound op or codec probe, or when its* upstream port has _PR3 (i.e. dGPU).*/parent = pci_upstream_bridge(p);chip->bus.keep_power = parent ? !pci_pr3_present(parent) : 1;-- Jaroslav Kysela perex@perex.cz Linux Sound Maintainer; ALSA Project; Red Hat, Inc.
Dne 10. 01. 20 v 10:56 Takashi Iwai napsal(a):
On Fri, 10 Jan 2020 10:43:26 +0100, Jaroslav Kysela wrote:
Dne 18. 10. 19 v 9:38 Kai-Heng Feng napsal(a):
Nvidia proprietary driver doesn't support runtime power management, so when a user only wants to use the integrated GPU, it's a common practice to let dGPU not to bind any driver, and let its upstream port to be runtime suspended. At the end of runtime suspension the port uses platform power management to disable power through _OFF method of power resource, which is listed by _PR3.
After commit b516ea586d71 ("PCI: Enable NVIDIA HDA controllers"), when the dGPU comes with an HDA function, the HDA won't be suspended if the dGPU is unbound, so the power resource can't be turned off by its upstream port driver.
Commit 37a3a98ef601 ("ALSA: hda - Enable runtime PM only for discrete GPU") only allows HDA to be runtime suspended once GPU is bound, to keep APU's HDA working.
However, HDA on dGPU isn't that useful if dGPU is not bound to any driver. So let's relax the runtime suspend requirement for dGPU's HDA function, to disable the power source to save lots of power.
This patch breaks the HDMI audio detection at least on some platforms (Lenovo P50 for example) with nouveau and the proprietary nvidia driver. Those laptops have the external HDMI/DP ports connected to dGPU instead the iGPU. The ACPI PR3 is set.
The runtime PM off fixes this problem:
echo on > /sys/bus/pci/devices/0000:01:00.1/power/control
But this will keep the power of the graphics chip on, and that's what the patch was supposed to "fix".
But I don't think that it's the best solution. My proposal is to create a pr3 check blacklist to keep power for the HDMI audio for those machines. Also we may add a new module parameter for snd-hda-intel to control this. Other ideas?
For nouveau, the best fix is to merge the audio component patch. This will make things working without fiddling with the power up/down. The patch has been pending over months under review in DRM side, unfortunately... Please pinging them for driving ahead.
Adding Cc: to dri-devel. You probably mean this thread:
https://lists.freedesktop.org/archives/dri-devel/2019-July/thread.html#22742...
For Nvidia, though, it's no path a binary-only stuff can go with, due to the GPL symbol of the component framework. Those guys know of it well, and they seem adding the temporary power up/down procedure by poking the proc file from the user-space side at the HDMI connection.
Wow.
About a module option: I don't think it's much better than the sysfs toggle. You can set up a simple udev rule if needed, too.
Ok, it's a bit nightmare to maintain those extra settings in the distribution.
Jaroslav
thanks,
Takashi
JaroslavBugLink: https://bugs.launchpad.net/bugs/1840835 Fixes: b516ea586d71 ("PCI: Enable NVIDIA HDA controllers") Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com
v5, v6:
- No change.
v4:
- Find upstream port, it's callee's responsibility now.
v3:
- Make changelog more clear.
v2:
- Change wording.
- Rebase to Tiwai's branch. sound/pci/hda/hda_intel.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/sound/pci/hda/hda_intel.c b/sound/pci/hda/hda_intel.c index 240f4ca76391..e63b871343e5 100644 --- a/sound/pci/hda/hda_intel.c +++ b/sound/pci/hda/hda_intel.c @@ -1280,11 +1280,17 @@ static void init_vga_switcheroo(struct azx *chip) { struct hda_intel *hda = container_of(chip, struct hda_intel, chip); struct pci_dev *p = get_bound_vga(chip->pci);
- struct pci_dev *parent; if (p) { dev_info(chip->card->dev, "Handle vga_switcheroo audio client\n"); hda->use_vga_switcheroo = 1;
chip->bus.keep_power = 1; /* cleared in either gpu_bound op or codec probe */
/* cleared in either gpu_bound op or codec probe, or when its* upstream port has _PR3 (i.e. dGPU).*/parent = pci_upstream_bridge(p);chip->bus.keep_power = parent ? !pci_pr3_present(parent) : 1; chip->driver_caps |= AZX_DCAPS_PM_RUNTIME; pci_dev_put(p);-- Jaroslav Kysela perex@perex.cz Linux Sound Maintainer; ALSA Project; Red Hat, Inc.
On Fri, 18 Oct 2019 09:38:47 +0200, Kai-Heng Feng wrote:
A driver may want to know the existence of _PR3, to choose different runtime suspend behavior. A user will be add in next patch.
This is mostly the same as nouveau_pr3_present().
Reported-by: kbuild test robot lkp@intel.com
It's confusing -- this particular change isn't reported by the test bot, but only about the lack of the CONFIG_ACPI ifdef.
thanks,
Takashi
Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com
v6:
- Only define the function when CONFIG_ACPI is set.
v5:
- Add wording suggestion from Bjorn.
v4:
- Let caller to find its upstream port device.
drivers/pci/pci.c | 18 ++++++++++++++++++ include/linux/pci.h | 2 ++ 2 files changed, 20 insertions(+)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index e7982af9a5d8..1df99d9e350e 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -5856,6 +5856,24 @@ int pci_set_vga_state(struct pci_dev *dev, bool decode, return 0; }
+#ifdef CONFIG_ACPI +bool pci_pr3_present(struct pci_dev *pdev) +{
- struct acpi_device *adev;
- if (acpi_disabled)
return false;- adev = ACPI_COMPANION(&pdev->dev);
- if (!adev)
return false;- return adev->power.flags.power_resources &&
acpi_has_method(adev->handle, "_PR3");+} +EXPORT_SYMBOL_GPL(pci_pr3_present); +#endif
/**
- pci_add_dma_alias - Add a DMA devfn alias for a device
- @dev: the PCI device for which alias is added
diff --git a/include/linux/pci.h b/include/linux/pci.h index f9088c89a534..1d15c5d49cdd 100644 --- a/include/linux/pci.h +++ b/include/linux/pci.h @@ -2310,9 +2310,11 @@ struct irq_domain *pci_host_bridge_acpi_msi_domain(struct pci_bus *bus);
void pci_msi_register_fwnode_provider(struct fwnode_handle *(*fn)(struct device *)); +bool pci_pr3_present(struct pci_dev *pdev); #else static inline struct irq_domain * pci_host_bridge_acpi_msi_domain(struct pci_bus *bus) { return NULL; } +static bool pci_pr3_present(struct pci_dev *pdev) { return false; } #endif
#ifdef CONFIG_EEH
2.17.1
On Oct 18, 2019, at 16:18, Takashi Iwai tiwai@suse.de wrote:
On Fri, 18 Oct 2019 09:38:47 +0200, Kai-Heng Feng wrote:
A driver may want to know the existence of _PR3, to choose different runtime suspend behavior. A user will be add in next patch.
This is mostly the same as nouveau_pr3_present().
Reported-by: kbuild test robot lkp@intel.com
It's confusing -- this particular change isn't reported by the test bot, but only about the lack of the CONFIG_ACPI ifdef.
Hmm, because the test bot asked to add the tag. If it's not appropriate will you drop it? I can also send a v7.
Kai-Heng
thanks,
Takashi
Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com
v6:
- Only define the function when CONFIG_ACPI is set.
v5:
- Add wording suggestion from Bjorn.
v4:
- Let caller to find its upstream port device.
drivers/pci/pci.c | 18 ++++++++++++++++++ include/linux/pci.h | 2 ++ 2 files changed, 20 insertions(+)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index e7982af9a5d8..1df99d9e350e 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -5856,6 +5856,24 @@ int pci_set_vga_state(struct pci_dev *dev, bool decode, return 0; }
+#ifdef CONFIG_ACPI +bool pci_pr3_present(struct pci_dev *pdev) +{
- struct acpi_device *adev;
- if (acpi_disabled)
return false;- adev = ACPI_COMPANION(&pdev->dev);
- if (!adev)
return false;- return adev->power.flags.power_resources &&
acpi_has_method(adev->handle, "_PR3");+} +EXPORT_SYMBOL_GPL(pci_pr3_present); +#endif
/**
- pci_add_dma_alias - Add a DMA devfn alias for a device
- @dev: the PCI device for which alias is added
diff --git a/include/linux/pci.h b/include/linux/pci.h index f9088c89a534..1d15c5d49cdd 100644 --- a/include/linux/pci.h +++ b/include/linux/pci.h @@ -2310,9 +2310,11 @@ struct irq_domain *pci_host_bridge_acpi_msi_domain(struct pci_bus *bus);
void pci_msi_register_fwnode_provider(struct fwnode_handle *(*fn)(struct device *)); +bool pci_pr3_present(struct pci_dev *pdev); #else static inline struct irq_domain * pci_host_bridge_acpi_msi_domain(struct pci_bus *bus) { return NULL; } +static bool pci_pr3_present(struct pci_dev *pdev) { return false; } #endif
#ifdef CONFIG_EEH
2.17.1
On Fri, 18 Oct 2019 10:30:11 +0200, Kai-Heng Feng wrote:
On Oct 18, 2019, at 16:18, Takashi Iwai tiwai@suse.de wrote:
On Fri, 18 Oct 2019 09:38:47 +0200, Kai-Heng Feng wrote:
A driver may want to know the existence of _PR3, to choose different runtime suspend behavior. A user will be add in next patch.
This is mostly the same as nouveau_pr3_present().
Reported-by: kbuild test robot lkp@intel.com
It's confusing -- this particular change isn't reported by the test bot, but only about the lack of the CONFIG_ACPI ifdef.
Hmm, because the test bot asked to add the tag.
Yes, but it's only if you add a new fix patch on top of it. You can write some their credit, too, but basically it'd be enough to point to the Link tag to the thread.
If it's not appropriate will you drop it? I can also send a v7.
I can modify in my side.
thanks,
Takashi
Kai-Heng
thanks,
Takashi
Signed-off-by: Kai-Heng Feng kai.heng.feng@canonical.com
v6:
- Only define the function when CONFIG_ACPI is set.
v5:
- Add wording suggestion from Bjorn.
v4:
- Let caller to find its upstream port device.
drivers/pci/pci.c | 18 ++++++++++++++++++ include/linux/pci.h | 2 ++ 2 files changed, 20 insertions(+)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index e7982af9a5d8..1df99d9e350e 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -5856,6 +5856,24 @@ int pci_set_vga_state(struct pci_dev *dev, bool decode, return 0; }
+#ifdef CONFIG_ACPI +bool pci_pr3_present(struct pci_dev *pdev) +{
- struct acpi_device *adev;
- if (acpi_disabled)
return false;- adev = ACPI_COMPANION(&pdev->dev);
- if (!adev)
return false;- return adev->power.flags.power_resources &&
acpi_has_method(adev->handle, "_PR3");+} +EXPORT_SYMBOL_GPL(pci_pr3_present); +#endif
/**
- pci_add_dma_alias - Add a DMA devfn alias for a device
- @dev: the PCI device for which alias is added
diff --git a/include/linux/pci.h b/include/linux/pci.h index f9088c89a534..1d15c5d49cdd 100644 --- a/include/linux/pci.h +++ b/include/linux/pci.h @@ -2310,9 +2310,11 @@ struct irq_domain *pci_host_bridge_acpi_msi_domain(struct pci_bus *bus);
void pci_msi_register_fwnode_provider(struct fwnode_handle *(*fn)(struct device *)); +bool pci_pr3_present(struct pci_dev *pdev); #else static inline struct irq_domain * pci_host_bridge_acpi_msi_domain(struct pci_bus *bus) { return NULL; } +static bool pci_pr3_present(struct pci_dev *pdev) { return false; } #endif
#ifdef CONFIG_EEH
2.17.1
On Fri, 18 Oct 2019 10:45:43 +0200, Takashi Iwai wrote:
On Fri, 18 Oct 2019 10:30:11 +0200, Kai-Heng Feng wrote:
On Oct 18, 2019, at 16:18, Takashi Iwai tiwai@suse.de wrote:
On Fri, 18 Oct 2019 09:38:47 +0200, Kai-Heng Feng wrote:
A driver may want to know the existence of _PR3, to choose different runtime suspend behavior. A user will be add in next patch.
This is mostly the same as nouveau_pr3_present().
Reported-by: kbuild test robot lkp@intel.com
It's confusing -- this particular change isn't reported by the test bot, but only about the lack of the CONFIG_ACPI ifdef.
Hmm, because the test bot asked to add the tag.
Yes, but it's only if you add a new fix patch on top of it. You can write some their credit, too, but basically it'd be enough to point to the Link tag to the thread.
If it's not appropriate will you drop it? I can also send a v7.
I can modify in my side.
Now I merged both patches to for-next branch.
thanks,
Takashi
participants (3)
-
Jaroslav Kysela -
Kai-Heng Feng -
Takashi Iwai