[PATCH v3 08/15] acpi: serial: don't use _STA method

Gerd Hoffmann posted 15 patches 5 years, 9 months ago
Maintainers: John Snow <jsnow@redhat.com>, Marcel Apfelbaum <marcel.apfelbaum@gmail.com>, Kevin Wolf <kwolf@redhat.com>, Richard Henderson <rth@twiddle.net>, "Michael S. Tsirkin" <mst@redhat.com>, Laurent Vivier <lvivier@redhat.com>, Thomas Huth <thuth@redhat.com>, Eduardo Habkost <ehabkost@redhat.com>, Paolo Bonzini <pbonzini@redhat.com>, Igor Mammedov <imammedo@redhat.com>, "Marc-André Lureau" <marcandre.lureau@redhat.com>, Max Reitz <mreitz@redhat.com>
There is a newer version of this series
[PATCH v3 08/15] acpi: serial: don't use _STA method
Posted by Gerd Hoffmann 5 years, 9 months ago
The _STA method dates back to the days where we had a static DSDT.  The
device is listed in the DSDT table unconditionally and the _STA method
checks a bit in the isa bridge pci config space to figure whenever a
given is isa device is present or not, then evaluates to 0x0f (present)
or 0x00 (absent).

These days the DSDT is generated by qemu anyway, so if a device is not
present we can simply drop it from the DSDT instead.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
---
 hw/i386/acpi-build.c | 32 ++++++++------------------------
 1 file changed, 8 insertions(+), 24 deletions(-)

diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c
index f79e5eaf170d..a99e5bbd1156 100644
--- a/hw/i386/acpi-build.c
+++ b/hw/i386/acpi-build.c
@@ -1202,50 +1202,34 @@ static Aml *build_lpt_device_aml(void)
     return dev;
 }
 
-static Aml *build_com_device_aml(uint8_t uid)
+static void build_com_device_aml(Aml *scope, uint8_t uid)
 {
     Aml *dev;
     Aml *crs;
-    Aml *method;
-    Aml *if_ctx;
-    Aml *else_ctx;
-    Aml *zero = aml_int(0);
-    Aml *is_present = aml_local(0);
-    const char *enabled_field = "CAEN";
     uint8_t irq = 4;
     uint16_t io_port = 0x03F8;
 
     assert(uid == 1 || uid == 2);
     if (uid == 2) {
-        enabled_field = "CBEN";
         irq = 3;
         io_port = 0x02F8;
     }
+    if (!memory_region_present(get_system_io(), io_port)) {
+        return;
+    }
 
     dev = aml_device("COM%d", uid);
     aml_append(dev, aml_name_decl("_HID", aml_eisaid("PNP0501")));
     aml_append(dev, aml_name_decl("_UID", aml_int(uid)));
 
-    method = aml_method("_STA", 0, AML_NOTSERIALIZED);
-    aml_append(method, aml_store(aml_name("%s", enabled_field), is_present));
-    if_ctx = aml_if(aml_equal(is_present, zero));
-    {
-        aml_append(if_ctx, aml_return(aml_int(0x00)));
-    }
-    aml_append(method, if_ctx);
-    else_ctx = aml_else();
-    {
-        aml_append(else_ctx, aml_return(aml_int(0x0f)));
-    }
-    aml_append(method, else_ctx);
-    aml_append(dev, method);
+    aml_append(dev, aml_name_decl("_STA", aml_int(0xf)));
 
     crs = aml_resource_template();
     aml_append(crs, aml_io(AML_DECODE16, io_port, io_port, 0x00, 0x08));
     aml_append(crs, aml_irq_no_flags(irq));
     aml_append(dev, aml_name_decl("_CRS", crs));
 
-    return dev;
+    aml_append(scope, dev);
 }
 
 static void build_isa_devices_aml(Aml *table)
@@ -1262,8 +1246,8 @@ static void build_isa_devices_aml(Aml *table)
         aml_append(scope, build_fdc_device_aml(fdc));
     }
     aml_append(scope, build_lpt_device_aml());
-    aml_append(scope, build_com_device_aml(1));
-    aml_append(scope, build_com_device_aml(2));
+    build_com_device_aml(scope, 1);
+    build_com_device_aml(scope, 2);
 
     if (ambiguous) {
         error_report("Multiple ISA busses, unable to define IPMI ACPI data");
-- 
2.18.2


Re: [PATCH v3 08/15] acpi: serial: don't use _STA method
Posted by Philippe Mathieu-Daudé 5 years, 9 months ago
Hi Gerd,

On 4/29/20 3:59 PM, Gerd Hoffmann wrote:
> The _STA method dates back to the days where we had a static DSDT.  The
> device is listed in the DSDT table unconditionally and the _STA method
> checks a bit in the isa bridge pci config space to figure whenever a
> given is isa device is present or not, then evaluates to 0x0f (present)
> or 0x00 (absent).
> 
> These days the DSDT is generated by qemu anyway, so if a device is not
> present we can simply drop it from the DSDT instead.
> 
> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
> ---
>   hw/i386/acpi-build.c | 32 ++++++++------------------------
>   1 file changed, 8 insertions(+), 24 deletions(-)
> 
> diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c
> index f79e5eaf170d..a99e5bbd1156 100644
> --- a/hw/i386/acpi-build.c
> +++ b/hw/i386/acpi-build.c
> @@ -1202,50 +1202,34 @@ static Aml *build_lpt_device_aml(void)
>       return dev;
>   }
>   
> -static Aml *build_com_device_aml(uint8_t uid)
> +static void build_com_device_aml(Aml *scope, uint8_t uid)
>   {
>       Aml *dev;
>       Aml *crs;
> -    Aml *method;
> -    Aml *if_ctx;
> -    Aml *else_ctx;
> -    Aml *zero = aml_int(0);
> -    Aml *is_present = aml_local(0);
> -    const char *enabled_field = "CAEN";
>       uint8_t irq = 4;
>       uint16_t io_port = 0x03F8;
>   
>       assert(uid == 1 || uid == 2);
>       if (uid == 2) {
> -        enabled_field = "CBEN";
>           irq = 3;
>           io_port = 0x02F8;
>       }
> +    if (!memory_region_present(get_system_io(), io_port)) {
> +        return;
> +    }

The patch looks OK, but an you split this check into a separate 
(previous?) patch please?

>   
>       dev = aml_device("COM%d", uid);
>       aml_append(dev, aml_name_decl("_HID", aml_eisaid("PNP0501")));
>       aml_append(dev, aml_name_decl("_UID", aml_int(uid)));
>   
> -    method = aml_method("_STA", 0, AML_NOTSERIALIZED);
> -    aml_append(method, aml_store(aml_name("%s", enabled_field), is_present));
> -    if_ctx = aml_if(aml_equal(is_present, zero));
> -    {
> -        aml_append(if_ctx, aml_return(aml_int(0x00)));
> -    }
> -    aml_append(method, if_ctx);
> -    else_ctx = aml_else();
> -    {
> -        aml_append(else_ctx, aml_return(aml_int(0x0f)));
> -    }
> -    aml_append(method, else_ctx);
> -    aml_append(dev, method);
> +    aml_append(dev, aml_name_decl("_STA", aml_int(0xf)));
>   
>       crs = aml_resource_template();
>       aml_append(crs, aml_io(AML_DECODE16, io_port, io_port, 0x00, 0x08));
>       aml_append(crs, aml_irq_no_flags(irq));
>       aml_append(dev, aml_name_decl("_CRS", crs));
>   
> -    return dev;
> +    aml_append(scope, dev);
>   }
>   
>   static void build_isa_devices_aml(Aml *table)
> @@ -1262,8 +1246,8 @@ static void build_isa_devices_aml(Aml *table)
>           aml_append(scope, build_fdc_device_aml(fdc));
>       }
>       aml_append(scope, build_lpt_device_aml());
> -    aml_append(scope, build_com_device_aml(1));
> -    aml_append(scope, build_com_device_aml(2));
> +    build_com_device_aml(scope, 1);
> +    build_com_device_aml(scope, 2);
>   
>       if (ambiguous) {
>           error_report("Multiple ISA busses, unable to define IPMI ACPI data");
> 


Re: [PATCH v3 08/15] acpi: serial: don't use _STA method
Posted by Gerd Hoffmann 5 years, 9 months ago
> > -static Aml *build_com_device_aml(uint8_t uid)
> > +static void build_com_device_aml(Aml *scope, uint8_t uid)
> >   {
> >       Aml *dev;
> >       Aml *crs;
> > -    Aml *method;
> > -    Aml *if_ctx;
> > -    Aml *else_ctx;
> > -    Aml *zero = aml_int(0);
> > -    Aml *is_present = aml_local(0);
> > -    const char *enabled_field = "CAEN";
> >       uint8_t irq = 4;
> >       uint16_t io_port = 0x03F8;
> >       assert(uid == 1 || uid == 2);
> >       if (uid == 2) {
> > -        enabled_field = "CBEN";
> >           irq = 3;
> >           io_port = 0x02F8;
> >       }
> > +    if (!memory_region_present(get_system_io(), io_port)) {
> > +        return;
> > +    }
> 
> The patch looks OK, but an you split this check into a separate (previous?)
> patch please?

I don't think this belongs to a separate patch.  It is basically the
same check the lpc bridge is doing when filling the pci config space
(see hw/isa/lpc_ich9.c).  So this effectively maintains the existing
logic, only that we now check directly instead of letting the guest
check the pci config space bit via _STA method.

Also note this is only temporary for bisecting, the next patch in the
series moves the code to a device callback so this kind of "device
exists" check is not needed any more.

take care,
  Gerd


Re: [PATCH v3 08/15] acpi: serial: don't use _STA method
Posted by Philippe Mathieu-Daudé 5 years, 9 months ago
On 5/4/20 3:19 PM, Gerd Hoffmann wrote:
>>> -static Aml *build_com_device_aml(uint8_t uid)
>>> +static void build_com_device_aml(Aml *scope, uint8_t uid)
>>>    {
>>>        Aml *dev;
>>>        Aml *crs;
>>> -    Aml *method;
>>> -    Aml *if_ctx;
>>> -    Aml *else_ctx;
>>> -    Aml *zero = aml_int(0);
>>> -    Aml *is_present = aml_local(0);
>>> -    const char *enabled_field = "CAEN";
>>>        uint8_t irq = 4;
>>>        uint16_t io_port = 0x03F8;
>>>        assert(uid == 1 || uid == 2);
>>>        if (uid == 2) {
>>> -        enabled_field = "CBEN";
>>>            irq = 3;
>>>            io_port = 0x02F8;
>>>        }
>>> +    if (!memory_region_present(get_system_io(), io_port)) {
>>> +        return;
>>> +    }
>>
>> The patch looks OK, but an you split this check into a separate (previous?)
>> patch please?
> 
> I don't think this belongs to a separate patch.  It is basically the
> same check the lpc bridge is doing when filling the pci config space
> (see hw/isa/lpc_ich9.c).  So this effectively maintains the existing
> logic, only that we now check directly instead of letting the guest
> check the pci config space bit via _STA method.
> 
> Also note this is only temporary for bisecting, the next patch in the
> series moves the code to a device callback so this kind of "device
> exists" check is not needed any more.

Oh correct.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>

> 
> take care,
>    Gerd
>