Re: [PATCH] target/s390x: Add a CONFIG switch to disable legacy CPUs

[Thomas Huth <thuth@redhat.com>]

On 13/06/2024 19.17, Philippe Mathieu-Daudé wrote:
> Hi Thomas,
> 
> On 13/6/24 19:07, Thomas Huth wrote:
>> Old CPU models are not officially supported anymore by IBM, and for
>> downstream builds of QEMU, we would like to be able to disable these
>> CPUs in the build. Thus add a CONFIG switch that can be used to
>> disable these CPUs (and old machine types that use them by default).
>>
>> Signed-off-by: Thomas Huth <thuth@redhat.com>
>> ---
>>   If you're interested, the PDF that can be downloaded from
>>   https://www.ibm.com/support/pages/ibm-mainframe-life-cycle-history
>>   shows the supported CPUs in a nice diagram
> 
> I'd add this link ...
> 
>>   hw/s390x/s390-virtio-ccw.c | 9 +++++++++
>>   target/s390x/cpu_models.c  | 3 +++
>>   target/s390x/Kconfig       | 5 +++++
>>   3 files changed, 17 insertions(+)
>>
>> diff --git a/target/s390x/Kconfig b/target/s390x/Kconfig
>> index d886be48b4..8a95f2bc3f 100644
>> --- a/target/s390x/Kconfig
>> +++ b/target/s390x/Kconfig
>> @@ -2,3 +2,8 @@ config S390X
>>       bool
>>       select PCI
>>       select S390_FLIC
>> +
>> +config S390X_LEGACY_CPUS
>> +    bool
>> +    default y
>> +    depends on S390X
>> diff --git a/target/s390x/cpu_models.c b/target/s390x/cpu_models.c
>> index efb508cd2e..ffae95dcb3 100644
>> --- a/target/s390x/cpu_models.c
>> +++ b/target/s390x/cpu_models.c
>> @@ -22,6 +22,7 @@
>>   #include "qemu/module.h"
>>   #include "qemu/hw-version.h"
>>   #include "qemu/qemu-print.h"
>> +#include CONFIG_DEVICES
>>   #ifndef CONFIG_USER_ONLY
>>   #include "sysemu/sysemu.h"
>>   #include "target/s390x/kvm/pv.h"
>> @@ -47,6 +48,7 @@
>>    * generation 15 one base feature and one optional feature have been 
>> deprecated.
>>    */
>>   static S390CPUDef s390_cpu_defs[] = {
>> +#ifdef CONFIG_S390X_LEGACY_CPUS
> 
> ... here :)

Can do ... let's just hope that the link is stable in the course of time!

>>       CPUDEF_INIT(0x2064, 7, 1, 38, 0x00000000U, "z900", "IBM zSeries 900 
>> GA1"),
>>       CPUDEF_INIT(0x2064, 7, 2, 38, 0x00000000U, "z900.2", "IBM zSeries 
>> 900 GA2"),
>>       CPUDEF_INIT(0x2064, 7, 3, 38, 0x00000000U, "z900.3", "IBM zSeries 
>> 900 GA3"),
>> @@ -78,6 +80,7 @@ static S390CPUDef s390_cpu_defs[] = {
>>       CPUDEF_INIT(0x2964, 13, 1, 47, 0x08000000U, "z13", "IBM z13 GA1"),
>>       CPUDEF_INIT(0x2964, 13, 2, 47, 0x08000000U, "z13.2", "IBM z13 GA2"),
>>       CPUDEF_INIT(0x2965, 13, 2, 47, 0x08000000U, "z13s", "IBM z13s GA1"),
>> +#endif
>>       CPUDEF_INIT(0x3906, 14, 1, 47, 0x08000000U, "z14", "IBM z14 GA1"),
>>       CPUDEF_INIT(0x3906, 14, 2, 47, 0x08000000U, "z14.2", "IBM z14 GA2"),
>>       CPUDEF_INIT(0x3907, 14, 1, 47, 0x08000000U, "z14ZR1", "IBM z14 Model 
>> ZR1 GA1"),
>> diff --git a/hw/s390x/s390-virtio-ccw.c b/hw/s390x/s390-virtio-ccw.c
>> index 3d0bc3e7f2..7529d2fba8 100644
>> --- a/hw/s390x/s390-virtio-ccw.c
>> +++ b/hw/s390x/s390-virtio-ccw.c
>> @@ -47,6 +47,7 @@
>>   #include "migration/blocker.h"
>>   #include "qapi/visitor.h"
>>   #include "hw/s390x/cpu-topology.h"
>> +#include CONFIG_DEVICES
>>   static Error *pv_mig_blocker;
>> @@ -603,6 +604,8 @@ static void s390_nmi(NMIState *n, int cpu_index, Error 
>> **errp)
>>       s390_cpu_restart(S390_CPU(cs));
>>   }
>> +#ifdef CONFIG_S390X_LEGACY_CPUS
>> +
>>   static ram_addr_t s390_fixup_ram_size(ram_addr_t sz)
>>   {
>>       /* same logic as in sclp.c */
>> @@ -623,6 +626,8 @@ static ram_addr_t s390_fixup_ram_size(ram_addr_t sz)
>>       return newsz;
>>   }
>> +#endif
>> +
>>   static inline bool machine_get_aes_key_wrap(Object *obj, Error **errp)
>>   {
>>       S390CcwMachineState *ms = S390_CCW_MACHINE(obj);
>> @@ -989,6 +994,8 @@ static void ccw_machine_6_1_class_options(MachineClass 
>> *mc)
>>   }
>>   DEFINE_CCW_MACHINE(6_1, "6.1", false);
>> +#ifdef CONFIG_S390X_LEGACY_CPUS
>> +
>>   static void ccw_machine_6_0_instance_options(MachineState *machine)
>>   {
>>       static const S390FeatInit qemu_cpu_feat = { S390_FEAT_LIST_QEMU_V6_0 };
> 
> Should we deprecate machines up to v6.0?

I'm still hoping that Daniel will be able to get his auto-deprecation 
patches merged in this cycle - then we shouldn't derive from that, I think.

By the way, what's up with your i440fx removal series? ... it would be good 
to get this finally merged now...?

  Thomas