[Qemu-devel] [PATCH 0/8] misc cleanups for x86_64 disabling config settings

[Qemu-devel] [PATCH 0/8] misc cleanups for x86_64 disabling config settings

[David Ahern]

David Ahern (8):
  fix 'no such file' error from make_device_config.sh
  x86_64 config: fix compile for CONFIG_VMWARE_VGA=n
  x86_64 config: fix compile for CONFIG_PARALLEL=n
  remove unused parallel_* variables from pc.c
  x86_64 config: fix compile for CONFIG_NE2000_ISA=n
  x86_64 config: fix compile for CONFIG_VGA_ISA=n
  x86_64 config: fix compile for CONFIG_IDE_ISA=n
  x86_64 config: fix compile for CONFIG_FDC=n

 hw/pc.c               |   21 ++++++++++++++++++---
 hw/pc_piix.c          |   11 +++++++++++
 make_device_config.sh |    2 +-
 3 files changed, 30 insertions(+), 4 deletions(-)

-- 
1.7.3.4

[Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[David Ahern]

make_device_config currently emits an error:

make config-all-devices.mak
  GEN   x86_64-softmmu/config-devices.mak
/home/dsa/kvm/releases/qemu.git/make_device_config.sh: line 21: /home/dsa/kvm/releases/qemu.git/default-configs/pci.mak : No such file or directory

/bin/sh does not like the file redirection for tr so change it to cat.

Signed-off-by: David Ahern <daahern@cisco.com>
---
 make_device_config.sh |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/make_device_config.sh b/make_device_config.sh
index 596fc5b..1ce4088 100644
--- a/make_device_config.sh
+++ b/make_device_config.sh
@@ -18,7 +18,7 @@ process_includes () {
 
 f=$src
 while [ -n "$f" ] ; do
-  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
+  f=`cat $f | tr -d '\r' | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
   [ $? = 0 ] || exit 1
   all_includes="$all_includes $f"
 done
-- 
1.7.3.4

Re: [Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[Markus Armbruster]

David Ahern <daahern@cisco.com> writes:

> make_device_config currently emits an error:
>
> make config-all-devices.mak
>   GEN   x86_64-softmmu/config-devices.mak
> /home/dsa/kvm/releases/qemu.git/make_device_config.sh: line 21: /home/dsa/kvm/releases/qemu.git/default-configs/pci.mak : No such file or directory
>
> /bin/sh does not like the file redirection for tr so change it to cat.
>
> Signed-off-by: David Ahern <daahern@cisco.com>
> ---
>  make_device_config.sh |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/make_device_config.sh b/make_device_config.sh
> index 596fc5b..1ce4088 100644
> --- a/make_device_config.sh
> +++ b/make_device_config.sh
> @@ -18,7 +18,7 @@ process_includes () {
>  
>  f=$src
>  while [ -n "$f" ] ; do
> -  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
> +  f=`cat $f | tr -d '\r' | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>    [ $? = 0 ] || exit 1
>    all_includes="$all_includes $f"
>  done

This sweeps the real error under the carpet: $f has a trailing space.

What about:

diff --git a/make_device_config.sh b/make_device_config.sh
index 596fc5b..711829c 100644
--- a/make_device_config.sh
+++ b/make_device_config.sh
@@ -18,7 +18,7 @@ process_includes () {
 
 f=$src
 while [ -n "$f" ] ; do
-  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
+  f=`tr -d '\r' <"$f" | awk '/^include / {print "'$src_dir'/" $2}'`
   [ $? = 0 ] || exit 1
   all_includes="$all_includes $f"
 done

Re: [Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[David Ahern]

On 01/13/11 06:43, Markus Armbruster wrote:
> David Ahern <daahern@cisco.com> writes:
> 
>> make_device_config currently emits an error:
>>
>> make config-all-devices.mak
>>   GEN   x86_64-softmmu/config-devices.mak
>> /home/dsa/kvm/releases/qemu.git/make_device_config.sh: line 21: /home/dsa/kvm/releases/qemu.git/default-configs/pci.mak : No such file or directory
>>
>> /bin/sh does not like the file redirection for tr so change it to cat.
>>
>> Signed-off-by: David Ahern <daahern@cisco.com>
>> ---
>>  make_device_config.sh |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/make_device_config.sh b/make_device_config.sh
>> index 596fc5b..1ce4088 100644
>> --- a/make_device_config.sh
>> +++ b/make_device_config.sh
>> @@ -18,7 +18,7 @@ process_includes () {
>>  
>>  f=$src
>>  while [ -n "$f" ] ; do
>> -  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>> +  f=`cat $f | tr -d '\r' | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>>    [ $? = 0 ] || exit 1
>>    all_includes="$all_includes $f"
>>  done
> 
> This sweeps the real error under the carpet: $f has a trailing space.
> 
> What about:
> 
> diff --git a/make_device_config.sh b/make_device_config.sh
> index 596fc5b..711829c 100644
> --- a/make_device_config.sh
> +++ b/make_device_config.sh
> @@ -18,7 +18,7 @@ process_includes () {
>  
>  f=$src
>  while [ -n "$f" ] ; do
> -  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
> +  f=`tr -d '\r' <"$f" | awk '/^include / {print "'$src_dir'/" $2}'`
>    [ $? = 0 ] || exit 1
>    all_includes="$all_includes $f"
>  done

That works too.

Any idea why the '%/config-devices.mak' rule does not work for a build
directory != src directory?

e.g., src=/my/qemu.git, build=/tmp/build-qemu

x86_64-softmmu/config-devices.mak is not created, so
config-all-devices.mak is empty and the build blows up.

David

Re: [Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[Markus Armbruster]

David Ahern <daahern@cisco.com> writes:

> On 01/13/11 06:43, Markus Armbruster wrote:
>> David Ahern <daahern@cisco.com> writes:
>> 
>>> make_device_config currently emits an error:
>>>
>>> make config-all-devices.mak
>>>   GEN   x86_64-softmmu/config-devices.mak
>>> /home/dsa/kvm/releases/qemu.git/make_device_config.sh: line 21: /home/dsa/kvm/releases/qemu.git/default-configs/pci.mak : No such file or directory
>>>
>>> /bin/sh does not like the file redirection for tr so change it to cat.
>>>
>>> Signed-off-by: David Ahern <daahern@cisco.com>
>>> ---
>>>  make_device_config.sh |    2 +-
>>>  1 files changed, 1 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/make_device_config.sh b/make_device_config.sh
>>> index 596fc5b..1ce4088 100644
>>> --- a/make_device_config.sh
>>> +++ b/make_device_config.sh
>>> @@ -18,7 +18,7 @@ process_includes () {
>>>  
>>>  f=$src
>>>  while [ -n "$f" ] ; do
>>> -  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>>> +  f=`cat $f | tr -d '\r' | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>>>    [ $? = 0 ] || exit 1
>>>    all_includes="$all_includes $f"
>>>  done
>> 
>> This sweeps the real error under the carpet: $f has a trailing space.
>> 
>> What about:
>> 
>> diff --git a/make_device_config.sh b/make_device_config.sh
>> index 596fc5b..711829c 100644
>> --- a/make_device_config.sh
>> +++ b/make_device_config.sh
>> @@ -18,7 +18,7 @@ process_includes () {
>>  
>>  f=$src
>>  while [ -n "$f" ] ; do
>> -  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>> +  f=`tr -d '\r' <"$f" | awk '/^include / {print "'$src_dir'/" $2}'`
>>    [ $? = 0 ] || exit 1
>>    all_includes="$all_includes $f"
>>  done
>
> That works too.

Care to respin your patch?

> Any idea why the '%/config-devices.mak' rule does not work for a build
> directory != src directory?
>
> e.g., src=/my/qemu.git, build=/tmp/build-qemu
>
> x86_64-softmmu/config-devices.mak is not created, so
> config-all-devices.mak is empty and the build blows up.

Should be debuggable :)

"mkdir bld && cd bld && ./configure --disable-werror && make" works for
me.  config-all-devices.mak has some 80 entries.

Re: [Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[David Ahern]

On 01/14/11 03:34, Markus Armbruster wrote:

>>> What about:
>>>
>>> diff --git a/make_device_config.sh b/make_device_config.sh
>>> index 596fc5b..711829c 100644
>>> --- a/make_device_config.sh
>>> +++ b/make_device_config.sh
>>> @@ -18,7 +18,7 @@ process_includes () {
>>>  
>>>  f=$src
>>>  while [ -n "$f" ] ; do
>>> -  f=`tr -d '\r' < $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>>> +  f=`tr -d '\r' <"$f" | awk '/^include / {print "'$src_dir'/" $2}'`
>>>    [ $? = 0 ] || exit 1
>>>    all_includes="$all_includes $f"
>>>  done
>>
>> That works too.
> 
> Care to respin your patch?

done.

> 
>> Any idea why the '%/config-devices.mak' rule does not work for a build
>> directory != src directory?
>>
>> e.g., src=/my/qemu.git, build=/tmp/build-qemu
>>
>> x86_64-softmmu/config-devices.mak is not created, so
>> config-all-devices.mak is empty and the build blows up.
> 
> Should be debuggable :)
> 
> "mkdir bld && cd bld && ./configure --disable-werror && make" works for
> me.  config-all-devices.mak has some 80 entries.

And it works for me too. Must have been an 11pm phenomenon.

David

Re: [Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[Stefan Weil]

Am 13.01.2011 07:34, schrieb David Ahern:
> make_device_config currently emits an error:
>
> make config-all-devices.mak
>    GEN   x86_64-softmmu/config-devices.mak
> /home/dsa/kvm/releases/qemu.git/make_device_config.sh: line 21: /home/dsa/kvm/releases/qemu.git/default-configs/pci.mak : No such file or directory
>
> /bin/sh does not like the file redirection for tr so change it to cat.
>
> Signed-off-by: David Ahern<daahern@cisco.com>
> ---
>   make_device_config.sh |    2 +-
>   1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/make_device_config.sh b/make_device_config.sh
> index 596fc5b..1ce4088 100644
> --- a/make_device_config.sh
> +++ b/make_device_config.sh
> @@ -18,7 +18,7 @@ process_includes () {
>
>   f=$src
>   while [ -n "$f" ] ; do
> -  f=`tr -d '\r'<  $f | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
> +  f=`cat $f | tr -d '\r' | awk '/^include / {ORS=" "; print "'$src_dir'/" $2}'`
>     [ $? = 0 ] || exit 1
>     all_includes="$all_includes $f"
>   done
>    

Hello Blue,

a patch from last year which fixes the same problem is still waiting
for being committed to qemu master. See
http://patchwork.ozlabs.org/patch/77021/

Thanks,
Stefan

Re: [Qemu-devel] [PATCH 1/8] fix 'no such file' error from make_device_config.sh

[Blue Swirl]

On Fri, Jan 14, 2011 at 7:57 PM, Stefan Weil <weil@mail.berlios.de> wrote:
> Am 13.01.2011 07:34, schrieb David Ahern:
>>
>> make_device_config currently emits an error:
>>
>> make config-all-devices.mak
>>   GEN   x86_64-softmmu/config-devices.mak
>> /home/dsa/kvm/releases/qemu.git/make_device_config.sh: line 21:
>> /home/dsa/kvm/releases/qemu.git/default-configs/pci.mak : No such file or
>> directory
>>
>> /bin/sh does not like the file redirection for tr so change it to cat.
>>
>> Signed-off-by: David Ahern<daahern@cisco.com>
>> ---
>>  make_device_config.sh |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/make_device_config.sh b/make_device_config.sh
>> index 596fc5b..1ce4088 100644
>> --- a/make_device_config.sh
>> +++ b/make_device_config.sh
>> @@ -18,7 +18,7 @@ process_includes () {
>>
>>  f=$src
>>  while [ -n "$f" ] ; do
>> -  f=`tr -d '\r'<  $f | awk '/^include / {ORS=" "; print "'$src_dir'/"
>> $2}'`
>> +  f=`cat $f | tr -d '\r' | awk '/^include / {ORS=" "; print "'$src_dir'/"
>> $2}'`
>>    [ $? = 0 ] || exit 1
>>    all_includes="$all_includes $f"
>>  done
>>
>
> Hello Blue,
>
> a patch from last year which fixes the same problem is still waiting
> for being committed to qemu master. See
> http://patchwork.ozlabs.org/patch/77021/

Thanks, applied.

[Qemu-devel] [PATCH 2/8] x86_64 config: fix compile for CONFIG_VMWARE_VGA=n

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/hw/pc.c b/hw/pc.c
index fface7d..50795e8 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -1069,11 +1069,13 @@ void pc_vga_init(PCIBus *pci_bus)
         } else {
             isa_cirrus_vga_init();
         }
+#ifdef CONFIG_VMWARE_VGA
     } else if (vmsvga_enabled) {
         if (pci_bus)
             pci_vmsvga_init(pci_bus);
         else
             fprintf(stderr, "%s: vmware_vga: no PCI bus\n", __FUNCTION__);
+#endif
 #ifdef CONFIG_SPICE
     } else if (qxl_enabled) {
         if (pci_bus)
-- 
1.7.3.4

[Qemu-devel] [PATCH 3/8] x86_64 config: fix compile for CONFIG_PARALLEL=n

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/hw/pc.c b/hw/pc.c
index 50795e8..18ec0df 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -1138,11 +1138,13 @@ void pc_basic_device_init(qemu_irq *isa_irq,
         }
     }
 
+#ifdef CONFIG_PARALLEL
     for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
         if (parallel_hds[i]) {
             parallel_init(i, parallel_hds[i]);
         }
     }
+#endif
 
     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
     i8042 = isa_create_simple("i8042");
-- 
1.7.3.4

[Qemu-devel] [PATCH 4/8] remove unused parallel_* variables from pc.c

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc.c |    3 ---
 1 files changed, 0 insertions(+), 3 deletions(-)

diff --git a/hw/pc.c b/hw/pc.c
index 18ec0df..6a55d2f 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -826,9 +826,6 @@ static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
                                               0x280, 0x380 };
 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
 
-static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
-static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
-
 void pc_audio_init (PCIBus *pci_bus, qemu_irq *pic)
 {
     struct soundhw *c;
-- 
1.7.3.4

[Qemu-devel] [PATCH 5/8] x86_64 config: fix compile for CONFIG_NE2000_ISA=n

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc.c      |    4 ++++
 hw/pc_piix.c |    2 ++
 2 files changed, 6 insertions(+), 0 deletions(-)

diff --git a/hw/pc.c b/hw/pc.c
index 6a55d2f..e7514fd 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -822,9 +822,11 @@ static void load_linux(void *fw_cfg,
 
 #define NE2000_NB_MAX 6
 
+#ifdef CONFIG_NE2000_ISA
 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
                                               0x280, 0x380 };
 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
+#endif
 
 void pc_audio_init (PCIBus *pci_bus, qemu_irq *pic)
 {
@@ -843,6 +845,7 @@ void pc_audio_init (PCIBus *pci_bus, qemu_irq *pic)
     }
 }
 
+#ifdef CONFIG_NE2000_ISA
 void pc_init_ne2k_isa(NICInfo *nd)
 {
     static int nb_ne2k = 0;
@@ -853,6 +856,7 @@ void pc_init_ne2k_isa(NICInfo *nd)
                     ne2000_irq[nb_ne2k], nd);
     nb_ne2k++;
 }
+#endif
 
 int cpu_is_bsp(CPUState *env)
 {
diff --git a/hw/pc_piix.c b/hw/pc_piix.c
index f82508d..5046947 100644
--- a/hw/pc_piix.c
+++ b/hw/pc_piix.c
@@ -119,9 +119,11 @@ static void pc_init1(ram_addr_t ram_size,
     for(i = 0; i < nb_nics; i++) {
         NICInfo *nd = &nd_table[i];
 
+#ifdef CONFIG_NE2000_ISA
         if (!pci_enabled || (nd->model && strcmp(nd->model, "ne2k_isa") == 0))
             pc_init_ne2k_isa(nd);
         else
+#endif
             pci_nic_init_nofail(nd, "e1000", NULL);
     }
 
-- 
1.7.3.4

[Qemu-devel] [PATCH 6/8] x86_64 config: fix compile for CONFIG_VGA_ISA=n

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/hw/pc.c b/hw/pc.c
index e7514fd..11b570f 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -1087,8 +1087,10 @@ void pc_vga_init(PCIBus *pci_bus)
     } else if (std_vga_enabled) {
         if (pci_bus) {
             pci_vga_init(pci_bus);
+#ifdef CONFIG_VGA_ISA
         } else {
             isa_vga_init();
+#endif
         }
     }
 }
-- 
1.7.3.4

[Qemu-devel] Re: [PATCH 6/8] x86_64 config: fix compile for CONFIG_VGA_ISA=n

[Paolo Bonzini]

On 01/13/2011 07:34 AM, David Ahern wrote:
> Signed-off-by: David Ahern<daahern@cisco.com>
> ---
>   hw/pc.c |    2 ++
>   1 files changed, 2 insertions(+), 0 deletions(-)
>
> diff --git a/hw/pc.c b/hw/pc.c
> index e7514fd..11b570f 100644
> --- a/hw/pc.c
> +++ b/hw/pc.c
> @@ -1087,8 +1087,10 @@ void pc_vga_init(PCIBus *pci_bus)
>       } else if (std_vga_enabled) {
>           if (pci_bus) {
>               pci_vga_init(pci_bus);
> +#ifdef CONFIG_VGA_ISA
>           } else {
>               isa_vga_init();
> +#endif

Should this be an abort for #ifndef CONFIG_VGA_ISA?  And maybe the isapc 
machine should be disabled altogether?

Paolo

[Qemu-devel] Re: [PATCH 6/8] x86_64 config: fix compile for CONFIG_VGA_ISA=n

[David Ahern]

On 01/13/11 10:02, Paolo Bonzini wrote:
> On 01/13/2011 07:34 AM, David Ahern wrote:
>> Signed-off-by: David Ahern<daahern@cisco.com>
>> ---
>>   hw/pc.c |    2 ++
>>   1 files changed, 2 insertions(+), 0 deletions(-)
>>
>> diff --git a/hw/pc.c b/hw/pc.c
>> index e7514fd..11b570f 100644
>> --- a/hw/pc.c
>> +++ b/hw/pc.c
>> @@ -1087,8 +1087,10 @@ void pc_vga_init(PCIBus *pci_bus)
>>       } else if (std_vga_enabled) {
>>           if (pci_bus) {
>>               pci_vga_init(pci_bus);
>> +#ifdef CONFIG_VGA_ISA
>>           } else {
>>               isa_vga_init();
>> +#endif
> 
> Should this be an abort for #ifndef CONFIG_VGA_ISA?  And maybe the isapc
> machine should be disabled altogether?
> 
> Paolo

If that's desired I can change that to:

#ifdef CONFIG_VGA_ISA
         } else {
             isa_vga_init();
#else
             abort();
#endif


David

[Qemu-devel] Re: [PATCH 6/8] x86_64 config: fix compile for CONFIG_VGA_ISA=n

[Paolo Bonzini]

On 01/13/2011 06:28 PM, David Ahern wrote:
>>> @@ -1087,8 +1087,10 @@ void pc_vga_init(PCIBus *pci_bus)
>>>        } else if (std_vga_enabled) {
>>>            if (pci_bus) {
>>>                pci_vga_init(pci_bus);
>>> +#ifdef CONFIG_VGA_ISA
>>>            } else {
>>>                isa_vga_init();
>>> +#endif
>>
>> Should this be an abort for #ifndef CONFIG_VGA_ISA?  And maybe the isapc
>> machine should be disabled altogether?
>>
>> Paolo
>
> If that's desired I can change that to:
>
> #ifdef CONFIG_VGA_ISA
>           } else {
>               isa_vga_init();
> #else
>               abort();
> #endif

Please check if you can trigger the abort via -M isapc (I think so, but 
I'm not sure).  If I am correct, the better solution is to disable isapc.

Paolo

[Qemu-devel] Re: [PATCH 6/8] x86_64 config: fix compile for CONFIG_VGA_ISA=n

[David Ahern]

On 01/14/11 00:31, Paolo Bonzini wrote:
> On 01/13/2011 06:28 PM, David Ahern wrote:
>>>> @@ -1087,8 +1087,10 @@ void pc_vga_init(PCIBus *pci_bus)
>>>>        } else if (std_vga_enabled) {
>>>>            if (pci_bus) {
>>>>                pci_vga_init(pci_bus);
>>>> +#ifdef CONFIG_VGA_ISA
>>>>            } else {
>>>>                isa_vga_init();
>>>> +#endif
>>>
>>> Should this be an abort for #ifndef CONFIG_VGA_ISA?  And maybe the isapc
>>> machine should be disabled altogether?
>>>
>>> Paolo
>>
>> If that's desired I can change that to:
>>
>> #ifdef CONFIG_VGA_ISA
>>           } else {
>>               isa_vga_init();
>> #else
>>               abort();
>> #endif
> 
> Please check if you can trigger the abort via -M isapc (I think so, but
> I'm not sure).  If I am correct, the better solution is to disable isapc.
> 
> Paolo

Yes, it would trigger the abort.

And, that test case exposed a larger issue: config-devices.mak and
config-all-devices.mak are not added to the config-host.h and
x86_64-softmmu/config-target.h config files which means the addition of
the ifdef makes it disabled all the time -- regardless of the y/n
setting in the config*.mak file. ie., a patch is needed to get the
config*device.mak options added to the config*.h file.

David

[Qemu-devel] [PATCH 7/8] x86_64 config: fix compile for CONFIG_IDE_ISA=n

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc_piix.c |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/hw/pc_piix.c b/hw/pc_piix.c
index 5046947..e794a82 100644
--- a/hw/pc_piix.c
+++ b/hw/pc_piix.c
@@ -38,9 +38,11 @@
 
 #define MAX_IDE_BUS 2
 
+#ifdef CONFIG_IDE_ISA
 static const int ide_iobase[MAX_IDE_BUS] = { 0x1f0, 0x170 };
 static const int ide_iobase2[MAX_IDE_BUS] = { 0x3f6, 0x376 };
 static const int ide_irq[MAX_IDE_BUS] = { 14, 15 };
+#endif
 
 static void ioapic_init(IsaIrqState *isa_irq_state)
 {
@@ -141,6 +143,7 @@ static void pc_init1(ram_addr_t ram_size,
         dev = pci_piix3_ide_init(pci_bus, hd, piix3_devfn + 1);
         idebus[0] = qdev_get_child_bus(&dev->qdev, "ide.0");
         idebus[1] = qdev_get_child_bus(&dev->qdev, "ide.1");
+#ifdef CONFIG_IDE_ISA
     } else {
         for(i = 0; i < MAX_IDE_BUS; i++) {
             ISADevice *dev;
@@ -149,6 +152,12 @@ static void pc_init1(ram_addr_t ram_size,
             idebus[i] = qdev_get_child_bus(&dev->qdev, "ide.0");
         }
     }
+#else
+    } else {
+        idebus[0] = NULL;
+        idebus[1] = NULL;
+    }
+#endif
 
     pc_audio_init(pci_enabled ? pci_bus : NULL, isa_irq);
 
-- 
1.7.3.4

[Qemu-devel] [PATCH 8/8] x86_64 config: fix compile for CONFIG_FDC=n

[David Ahern]

Signed-off-by: David Ahern <daahern@cisco.com>
---
 hw/pc.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/hw/pc.c b/hw/pc.c
index 11b570f..71750f5 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -189,6 +189,7 @@ static void pic_irq_request(void *opaque, int irq, int level)
 
 /* PC cmos mappings */
 
+#ifdef CONFIG_FDC
 #define REG_EQUIPMENT_BYTE          0x14
 
 static int cmos_get_fd_drive_type(int fd0)
@@ -214,6 +215,7 @@ static int cmos_get_fd_drive_type(int fd0)
     }
     return val;
 }
+#endif
 
 static void cmos_init_hd(int type_ofs, int info_ofs, BlockDriverState *hd,
                          ISADevice *s)
@@ -335,7 +337,9 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
                   FDCtrl *floppy_controller, ISADevice *s)
 {
     int val;
+#ifdef CONFIG_FDC
     int fd0, fd1, nb;
+#endif
     static pc_cmos_init_late_arg arg;
 
     /* various important CMOS locations needed by PC/Bochs bios */
@@ -376,6 +380,7 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
         exit(1);
     }
 
+#ifdef CONFIG_FDC
     /* floppy type */
 
     fd0 = fdctrl_get_drive_type(floppy_controller, 0);
@@ -403,6 +408,7 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
     val |= 0x02; /* FPU is there */
     val |= 0x04; /* PS/2 mouse installed */
     rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
+#endif
 
     /* hard drives */
     arg.rtc_state = s;
@@ -1162,7 +1168,9 @@ void pc_basic_device_init(qemu_irq *isa_irq,
     for(i = 0; i < MAX_FD; i++) {
         fd[i] = drive_get(IF_FLOPPY, 0, i);
     }
+#ifdef CONFIG_FDC
     *floppy_controller = fdctrl_init_isa(fd);
+#endif
 }
 
 void pc_pci_device_init(PCIBus *pci_bus)
-- 
1.7.3.4

[Qemu-devel] Re: [PATCH 8/8] x86_64 config: fix compile for CONFIG_FDC=n

[Paolo Bonzini]

On 01/13/2011 07:34 AM, David Ahern wrote:
>
> +#ifdef CONFIG_FDC
>       /* floppy type */
>
>       fd0 = fdctrl_get_drive_type(floppy_controller, 0);
> @@ -403,6 +408,7 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
>       val |= 0x02; /* FPU is there */
>       val |= 0x04; /* PS/2 mouse installed */
>       rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
> +#endif

Isn't it a bit too eager?  The equipment byte should still be set, or 
have you checked with a real machine?

Paolo

[Qemu-devel] Re: [PATCH 8/8] x86_64 config: fix compile for CONFIG_FDC=n

[David Ahern]

On 01/13/11 10:04, Paolo Bonzini wrote:
> On 01/13/2011 07:34 AM, David Ahern wrote:
>>
>> +#ifdef CONFIG_FDC
>>       /* floppy type */
>>
>>       fd0 = fdctrl_get_drive_type(floppy_controller, 0);
>> @@ -403,6 +408,7 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t
>> above_4g_mem_size,
>>       val |= 0x02; /* FPU is there */
>>       val |= 0x04; /* PS/2 mouse installed */
>>       rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
>> +#endif
> 
> Isn't it a bit too eager?  The equipment byte should still be set, or
> have you checked with a real machine?
> 
> Paolo

A bit to eager.  I'll fix.

David

Re: [Qemu-devel] [PATCH 0/8] misc cleanups for x86_64 disabling config settings

[Isaku Yamahata]

How about creating stub functions instead of #ifdefs?
something like

pc-stubs.c
#ifndef CONFIG_VMWARE_VGA
pci_vmsvga_init()
{
	error ...
}
#endif
#ifndef CONFIG_...
...


On Wed, Jan 12, 2011 at 11:34:29PM -0700, David Ahern wrote:
> David Ahern (8):
>   fix 'no such file' error from make_device_config.sh
>   x86_64 config: fix compile for CONFIG_VMWARE_VGA=n
>   x86_64 config: fix compile for CONFIG_PARALLEL=n
>   remove unused parallel_* variables from pc.c
>   x86_64 config: fix compile for CONFIG_NE2000_ISA=n
>   x86_64 config: fix compile for CONFIG_VGA_ISA=n
>   x86_64 config: fix compile for CONFIG_IDE_ISA=n
>   x86_64 config: fix compile for CONFIG_FDC=n
> 
>  hw/pc.c               |   21 ++++++++++++++++++---
>  hw/pc_piix.c          |   11 +++++++++++
>  make_device_config.sh |    2 +-
>  3 files changed, 30 insertions(+), 4 deletions(-)
> 
> -- 
> 1.7.3.4
> 
> 

-- 
yamahata

Re: [Qemu-devel] [PATCH 0/8] misc cleanups for x86_64 disabling config settings

[David Ahern]

On 01/13/11 02:57, Isaku Yamahata wrote:
> How about creating stub functions instead of #ifdefs?
> something like
> 
> pc-stubs.c
> #ifndef CONFIG_VMWARE_VGA
> pci_vmsvga_init()
> {
> 	error ...
> }
> #endif
> #ifndef CONFIG_...
> ...

I wasn't trying to re-architect how the CONFIG_XXXX is implemented; I
only wanted the existing design to actually work. There are a number of
CONFIG settings which exist in name, but cannot in fact be disabled.

David


> 
> 
> On Wed, Jan 12, 2011 at 11:34:29PM -0700, David Ahern wrote:
>> David Ahern (8):
>>   fix 'no such file' error from make_device_config.sh
>>   x86_64 config: fix compile for CONFIG_VMWARE_VGA=n
>>   x86_64 config: fix compile for CONFIG_PARALLEL=n
>>   remove unused parallel_* variables from pc.c
>>   x86_64 config: fix compile for CONFIG_NE2000_ISA=n
>>   x86_64 config: fix compile for CONFIG_VGA_ISA=n
>>   x86_64 config: fix compile for CONFIG_IDE_ISA=n
>>   x86_64 config: fix compile for CONFIG_FDC=n
>>
>>  hw/pc.c               |   21 ++++++++++++++++++---
>>  hw/pc_piix.c          |   11 +++++++++++
>>  make_device_config.sh |    2 +-
>>  3 files changed, 30 insertions(+), 4 deletions(-)
>>
>> -- 
>> 1.7.3.4
>>
>>
> 

[Qemu-devel] Re: [PATCH 0/8] misc cleanups for x86_64 disabling config settings

[Paolo Bonzini]

On 01/13/2011 10:57 AM, Isaku Yamahata wrote:
> How about creating stub functions instead of #ifdefs?

I think David's #ifdef's are okay.  Ultimately what you'd want is to 
drive all this from configuration files, and if this were done you 
wouldn't need stub functions.

Paolo