Skip to contentSkip to conversation

ALSA: hda: intel: Don't free interrupt when suspending

PATCH Applied Heiner Kallweit 2 participants 5 messages Oct 31, 2024 – Nov 4, 2024 Sound

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ALSA: hda: intel: Don't free interrupt when suspending

There's no need to free/re-request the interrupt on system suspend. PCI core takes care, using functions like pci_restore_msi_state().

---
 sound/pci/hda/hda_intel.c | 13 -------------
 1 file changed, 13 deletions(-)
sound/pci/hda/hda_intel.c +0 -13
index b4540c5cd..9fc5e6c5d 100644
@@ -1040,14 +1040,6 @@ static int azx_suspend(struct device *dev)
 	chip = card->private_data;
 	bus = azx_bus(chip);
 	azx_shutdown_chip(chip);
-	if (bus->irq >= 0) {
-		free_irq(bus->irq, chip);
-		bus->irq = -1;
-		chip->card->sync_irq = -1;
-	}
-
-	if (chip->msi)
-		pci_disable_msi(chip->pci);
 
 	trace_azx_suspend(chip);
 	return 0;
@@ -1062,11 +1054,6 @@ static int __maybe_unused azx_resume(struct device *dev)
 		return 0;
 
 	chip = card->private_data;
-	if (chip->msi)
-		if (pci_enable_msi(chip->pci) < 0)
-			chip->msi = 0;
-	if (azx_acquire_irq(chip, 1) < 0)
-		return -EIO;
 
 	__azx_runtime_resume(chip);
 
Signature · 1 line
2.47.0
#1 lore raw
Quoted 3 lines · Heiner Kallweit

There's no need to free/re-request the interrupt on system suspend. PCI core takes care, using functions like pci_restore_msi_state().

Well, it wasn't always true for hibernation in the past, e.g. snd-intel8x0 driver had to deal with it. I guess it's no longer needed for modern platforms for HD-audio, so I'm going to take this for 6.13.

thanks,

Takashi

Quoted 40 lines
---
 sound/pci/hda/hda_intel.c | 13 -------------
 1 file changed, 13 deletions(-)
sound/pci/hda/hda_intel.c +0 -13
index b4540c5cd..9fc5e6c5d 100644
@@ -1040,14 +1040,6 @@ static int azx_suspend(struct device *dev)
 	chip = card->private_data;
 	bus = azx_bus(chip);
 	azx_shutdown_chip(chip);
-	if (bus->irq >= 0) {
-		free_irq(bus->irq, chip);
-		bus->irq = -1;
-		chip->card->sync_irq = -1;
-	}
-
-	if (chip->msi)
-		pci_disable_msi(chip->pci);
 
 	trace_azx_suspend(chip);
 	return 0;
@@ -1062,11 +1054,6 @@ static int __maybe_unused azx_resume(struct device *dev)
 		return 0;
 
 	chip = card->private_data;
-	if (chip->msi)
-		if (pci_enable_msi(chip->pci) < 0)
-			chip->msi = 0;
-	if (azx_acquire_irq(chip, 1) < 0)
-		return -EIO;
 
 	__azx_runtime_resume(chip);
 
Signature · 1 line
2.47.0
#2 lore raw
Quoted 11 lines · Takashi Iwai
Quoted 3 lines · Heiner Kallweit

There's no need to free/re-request the interrupt on system suspend. PCI core takes care, using functions like pci_restore_msi_state().

Well, it wasn't always true for hibernation in the past, e.g. snd-intel8x0 driver had to deal with it. I guess it's no longer needed for modern platforms for HD-audio, so I'm going to take this for 6.13.

Interesting, do you have any link to what the issue with intel8x0 was? In the commit history I didn't find something related at a first glance. Maybe also PCI core simply is more mature meanwhile.

Heiner

Quoted 45 lines

thanks,

Takashi

Quoted 40 lines
---
 sound/pci/hda/hda_intel.c | 13 -------------
 1 file changed, 13 deletions(-)
sound/pci/hda/hda_intel.c +0 -13
index b4540c5cd..9fc5e6c5d 100644
@@ -1040,14 +1040,6 @@ static int azx_suspend(struct device *dev)
 	chip = card->private_data;
 	bus = azx_bus(chip);
 	azx_shutdown_chip(chip);
-	if (bus->irq >= 0) {
-		free_irq(bus->irq, chip);
-		bus->irq = -1;
-		chip->card->sync_irq = -1;
-	}
-
-	if (chip->msi)
-		pci_disable_msi(chip->pci);
 
 	trace_azx_suspend(chip);
 	return 0;
@@ -1062,11 +1054,6 @@ static int __maybe_unused azx_resume(struct device *dev)
 		return 0;
 
 	chip = card->private_data;
-	if (chip->msi)
-		if (pci_enable_msi(chip->pci) < 0)
-			chip->msi = 0;
-	if (azx_acquire_irq(chip, 1) < 0)
-		return -EIO;
 
 	__azx_runtime_resume(chip);
 
Signature · 1 line
2.47.0
#3 lore raw
Quoted 8 lines · Heiner Kallweit
Quoted 6 lines · Takashi Iwai
Quoted 3 lines · Heiner Kallweit

There's no need to free/re-request the interrupt on system suspend. PCI core takes care, using functions like pci_restore_msi_state().

Quoted 7 lines
Quoted 5 lines

Well, it wasn't always true for hibernation in the past, e.g. snd-intel8x0 driver had to deal with it. I guess it's no longer needed for modern platforms for HD-audio, so I'm going to take this for 6.13.

Interesting, do you have any link to what the issue with intel8x0 was? In the commit history I didn't find something related at a first glance.

I also don't remember exactly, sorry. It was decades ago.

The old bug entry
https://bugzilla.kernel.org/show_bug.cgi?id=219416
showed why the irq handler was freed at suspend on intel8x0. My vague memory says that this was the original fix, and there was another bug report of hibernation issue. I'm no longer sure whether it was intel8x0, though -- it might be another PCI sound driver, too. But the only certain thing is that it was tad old issue.

Quoted 1 line

Maybe also PCI core simply is more mature meanwhile.

Hopefully, and I bet for it by taking a risk to apply your cleanup :)

thanks,

Takashi

Quoted 47 lines

Heiner

Quoted 45 lines

thanks,

Takashi

Quoted 40 lines
---
 sound/pci/hda/hda_intel.c | 13 -------------
 1 file changed, 13 deletions(-)
sound/pci/hda/hda_intel.c +0 -13
index b4540c5cd..9fc5e6c5d 100644
@@ -1040,14 +1040,6 @@ static int azx_suspend(struct device *dev)
 	chip = card->private_data;
 	bus = azx_bus(chip);
 	azx_shutdown_chip(chip);
-	if (bus->irq >= 0) {
-		free_irq(bus->irq, chip);
-		bus->irq = -1;
-		chip->card->sync_irq = -1;
-	}
-
-	if (chip->msi)
-		pci_disable_msi(chip->pci);
 
 	trace_azx_suspend(chip);
 	return 0;
@@ -1062,11 +1054,6 @@ static int __maybe_unused azx_resume(struct device *dev)
 		return 0;
 
 	chip = card->private_data;
-	if (chip->msi)
-		if (pci_enable_msi(chip->pci) < 0)
-			chip->msi = 0;
-	if (azx_acquire_irq(chip, 1) < 0)
-		return -EIO;
 
 	__azx_runtime_resume(chip);
 
Signature · 1 line
2.47.0