Skip to contentSkip to conversation

ACPI: resources: add legacy irq override exception by DMI info

PATCH Superseded Hui Wang 3 participants 5 messages Sep 14, 2021 – Sep 15, 2021 ACPI

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Hui Wang

author
#0 lore raw

PATCH v2 1/1 ACPI: resources: add legacy irq override exception by DMI info

After the commit 0ec4e55e9f57 ("ACPI: resources: Add checks for ACPI IRQ override") is reverted, the keyboard of those Medion laptops can't work again.

To fix the keyboard issue, here adding an override check by DMI info, this will not affect other machines and this design refers to the prt_quirks[] in the drivers/acpi/pci_irq.c.

If we meet similar issues on other platforms, we could expand the table of skip_override_table[] or medion_laptop[].

---
 drivers/acpi/resource.c | 51 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 49 insertions(+), 2 deletions(-)
drivers/acpi/resource.c +49 -2
index ee78a210c606..676254ccac4d 100644
@@ -16,6 +16,7 @@
 #include <linux/ioport.h>
 #include <linux/slab.h>
 #include <linux/irq.h>
+#include <linux/dmi.h>
 
 #ifdef CONFIG_X86
 #define valid_IRQ(i) (((i) != 0) && ((i) != 2))
@@ -380,17 +381,63 @@ unsigned int acpi_dev_get_irq_type(int triggering, int polarity)
 }
 EXPORT_SYMBOL_GPL(acpi_dev_get_irq_type);
 
+static const struct dmi_system_id medion_laptop[] = {
+	{
+		.ident = "MEDION P15651",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "MEDION"),
+			DMI_MATCH(DMI_BOARD_NAME, "M15T"),
+		},
+	},
+	{ }
+};
+
+struct irq_override_cmp {
+	const struct dmi_system_id *system;
+	unsigned char irq;
+	unsigned char triggering;
+	unsigned char polarity;
+	unsigned char shareable;
+};
+
+static const struct irq_override_cmp skip_override_table[] = {
+	{ medion_laptop, 1, ACPI_LEVEL_SENSITIVE, ACPI_ACTIVE_LOW, 0 },
+};
+
+static bool acpi_dev_irq_override(u32 gsi, u8 triggering, u8 polarity,
+				  u8 shareable)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(skip_override_table); i++) {
+		const struct irq_override_cmp *entry = &skip_override_table[i];
+
+		if (dmi_check_system(entry->system) &&
+		    entry->irq == gsi &&
+		    entry->triggering == triggering &&
+		    entry->polarity == polarity &&
+		    entry->shareable == shareable)
+			return false;
+	}
+
+	return true;
+}
+
 static void acpi_dev_get_irqresource(struct resource *res, u32 gsi,
 				     u8 triggering, u8 polarity, u8 shareable,
-				     bool legacy)
+				     bool check_override)
 {
 	int irq, p, t;
+	bool override;
 
 	if (!valid_IRQ(gsi)) {
 		irqresource_disabled(res, gsi);
 		return;
 	}
 
+	override = check_override && acpi_dev_irq_override(gsi, triggering,
+							   polarity, shareable);
+
 	/*
 	 * In IO-APIC mode, use overridden attribute. Two reasons:
 	 * 1. BIOS bug in DSDT
@@ -401,7 +448,7 @@ static void acpi_dev_get_irqresource(struct resource *res, u32 gsi,
 	 * using extended IRQ descriptors we take the IRQ configuration
 	 * from _CRS directly.
 	 */
-	if (legacy && !acpi_get_override_irq(gsi, &t, &p)) {
+	if (override && !acpi_get_override_irq(gsi, &t, &p)) {
 		u8 trig = t ? ACPI_LEVEL_SENSITIVE : ACPI_EDGE_SENSITIVE;
 		u8 pol = p ? ACPI_ACTIVE_LOW : ACPI_ACTIVE_HIGH;
 
Signature · 1 line
2.25.1
#1 lore raw

Hi Rafael, Hui and all others,

thank you very much for your time, thoughts and work, again!

The new [PATCH v2] works as well on here as the previous one. Keep my "Tested-by" attribute.

@Rafael: Do you see risks with this patch or the need for further improvements? If not, can you, please, push this patch towards kernel inclusion?

@Hui: Are you going to publish the updated patch on bugzilla or should I upload it? -- Many thanks for your recent quick actions!

Best regards,

Manuel

Quoted 25 lines · Hui Wang

After the commit 0ec4e55e9f57 ("ACPI: resources: Add checks for ACPI IRQ override") is reverted, the keyboard of those Medion laptops can't work again.

To fix the keyboard issue, here adding an override check by DMI info, this will not affect other machines and this design refers to the prt_quirks[] in the drivers/acpi/pci_irq.c.

If we meet similar issues on other platforms, we could expand the table of skip_override_table[] or medion_laptop[].

---
  drivers/acpi/resource.c | 51 +++++++++++++++++++++++++++++++++++++++--
  1 file changed, 49 insertions(+), 2 deletions(-)
drivers/acpi/resource.c +0 -0
index ee78a210c606..676254ccac4d 100644

[...]

author
#2 lore raw
Quoted 11 lines · Manuel Krause

Hi Rafael, Hui and all others,

thank you very much for your time, thoughts and work, again!

The new [PATCH v2] works as well on here as the previous one. Keep my "Tested-by" attribute.

@Rafael: Do you see risks with this patch or the need for further improvements? If not, can you, please, push this patch towards kernel inclusion?

@Hui: Are you going to publish the updated patch on bugzilla or should I upload it? -- Many thanks for your recent quick actions!

OK, will do it.

Thanks,

Hui.

Quoted 33 lines

Best regards,

Manuel

Quoted 25 lines · Hui Wang

After the commit 0ec4e55e9f57 ("ACPI: resources: Add checks for ACPI IRQ override") is reverted, the keyboard of those Medion laptops can't work again.

To fix the keyboard issue, here adding an override check by DMI info, this will not affect other machines and this design refers to the prt_quirks[] in the drivers/acpi/pci_irq.c.

If we meet similar issues on other platforms, we could expand the table of skip_override_table[] or medion_laptop[].

---
  drivers/acpi/resource.c | 51 +++++++++++++++++++++++++++++++++++++++--

  1 file changed, 49 insertions(+), 2 deletions(-)
drivers/acpi/resource.c +0 -0
index ee78a210c606..676254ccac4d 100644

[...]

#3 lore raw
In reply to Hui Wang
Quoted 86 lines · Hui Wang

After the commit 0ec4e55e9f57 ("ACPI: resources: Add checks for ACPI IRQ override") is reverted, the keyboard of those Medion laptops can't work again.

To fix the keyboard issue, here adding an override check by DMI info, this will not affect other machines and this design refers to the prt_quirks[] in the drivers/acpi/pci_irq.c.

If we meet similar issues on other platforms, we could expand the table of skip_override_table[] or medion_laptop[].

---
 drivers/acpi/resource.c | 51 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 49 insertions(+), 2 deletions(-)
drivers/acpi/resource.c +44 -1
index ee78a210c606..676254ccac4d 100644
@@ -16,6 +16,7 @@
 #include <linux/ioport.h>
 #include <linux/slab.h>
 #include <linux/irq.h>
+#include <linux/dmi.h>
 
 #ifdef CONFIG_X86
 #define valid_IRQ(i) (((i) != 0) && ((i) != 2))
@@ -380,17 +381,63 @@ unsigned int acpi_dev_get_irq_type(int triggering, int polarity)
 }
 EXPORT_SYMBOL_GPL(acpi_dev_get_irq_type);
 
+static const struct dmi_system_id medion_laptop[] = {
+       {
+               .ident = "MEDION P15651",
+               .matches = {
+                       DMI_MATCH(DMI_SYS_VENDOR, "MEDION"),
+                       DMI_MATCH(DMI_BOARD_NAME, "M15T"),
+               },
+       },
+       { }
+};
+
+struct irq_override_cmp {
+       const struct dmi_system_id *system;
+       unsigned char irq;
+       unsigned char triggering;
+       unsigned char polarity;
+       unsigned char shareable;
+};
+
+static const struct irq_override_cmp skip_override_table[] = {
+       { medion_laptop, 1, ACPI_LEVEL_SENSITIVE, ACPI_ACTIVE_LOW, 0 },
+};
+
+static bool acpi_dev_irq_override(u32 gsi, u8 triggering, u8 polarity,
+                                 u8 shareable)
+{
+       int i;
+
+       for (i = 0; i < ARRAY_SIZE(skip_override_table); i++) {
+               const struct irq_override_cmp *entry = &skip_override_table[i];
+
+               if (dmi_check_system(entry->system) &&
+                   entry->irq == gsi &&
+                   entry->triggering == triggering &&
+                   entry->polarity == polarity &&
+                   entry->shareable == shareable)
+                       return false;
+       }
+
+       return true;
+}
+
 static void acpi_dev_get_irqresource(struct resource *res, u32 gsi,
                                     u8 triggering, u8 polarity, u8 shareable,
-                                    bool legacy)
+                                    bool check_override)
 {
        int irq, p, t;

drivers/acpi/resource.c unsigned int acpi_dev_get_irq_type(int triggering, int polarity) · line 440

Quoted 10 lines
drivers/acpi/resource.c +4 -0
+       bool override;
 
        if (!valid_IRQ(gsi)) {
                irqresource_disabled(res, gsi);
                return;
        }
 
+       override = check_override && acpi_dev_irq_override(gsi, triggering,
+                                                          polarity, shareable);
+

This can be still made more straightforward.

The local bool variable is not necessary and check below can be rearranged as follows below.

drivers/acpi/resource.c static void acpi_dev_get_irqresource(struct resource *res, u32 gsi, · line 451

Quoted 9 lines
drivers/acpi/resource.c +1 -1
        /*
         * In IO-APIC mode, use overridden attribute. Two reasons:
         * 1. BIOS bug in DSDT
@@ -401,7 +448,7 @@ static void acpi_dev_get_irqresource(struct resource *res, u32 gsi,
         * using extended IRQ descriptors we take the IRQ configuration
         * from _CRS directly.
         */
-       if (legacy && !acpi_get_override_irq(gsi, &t, &p)) {
+       if (override && !acpi_get_override_irq(gsi, &t, &p)) {

if (check_override &&

    acpi_dev_irq_override(gsi, triggering, polarity, shareable) &&
    !acpi_get_override_irq(gsi, &t, &p)) {
Quoted 6 lines
drivers/acpi/resource.c +0 -1
                u8 trig = t ? ACPI_LEVEL_SENSITIVE : ACPI_EDGE_SENSITIVE;
                u8 pol = p ? ACPI_ACTIVE_LOW : ACPI_ACTIVE_HIGH;
 
--

2.25.1

author
#4 lore raw
Quoted 1 line · Rafael J. Wysocki

On Tue, Sep 14, 2021 at 3:34 PM Hui Wang <[email protected]> wrote:

[...]

Quoted 19 lines

drivers/acpi/resource.c unsigned int acpi_dev_get_irq_type(int triggering, int polarity) · line 440

Quoted 4 lines
drivers/acpi/resource.c +3 -0
+       override = check_override && acpi_dev_irq_override(gsi, triggering,
+                                                          polarity, shareable);
+

This can be still made more straightforward.

The local bool variable is not necessary and check below can be rearranged as follows below.

drivers/acpi/resource.c static void acpi_dev_get_irqresource(struct resource *res, u32 gsi, · line 451

Quoted 9 lines

/*
* In IO-APIC mode, use overridden attribute. Two reasons:
* 1. BIOS bug in DSDT

drivers/acpi/resource.c +1 -1
@@ -401,7 +448,7 @@ static void acpi_dev_get_irqresource(struct resource *res, u32 gsi,
          * using extended IRQ descriptors we take the IRQ configuration
          * from _CRS directly.
          */
-       if (legacy && !acpi_get_override_irq(gsi, &t, &p)) {
+       if (override && !acpi_get_override_irq(gsi, &t, &p)) {

if (check_override &&

Quoted 2 lines
     acpi_dev_irq_override(gsi, triggering, polarity, shareable) &&
     !acpi_get_override_irq(gsi, &t, &p)) {

OK, got it. will change to it in the v3.

Thanks.

Quoted 6 lines
Quoted 6 lines
drivers/acpi/resource.c +0 -1
                 u8 trig = t ? ACPI_LEVEL_SENSITIVE : ACPI_EDGE_SENSITIVE;
                 u8 pol = p ? ACPI_ACTIVE_LOW : ACPI_ACTIVE_HIGH;
 
--

2.25.1