Re: [PATCH] hw/i386/isapc: Remove dead Xen code

["Alex Bennée" <alex.bennee@linaro.org>]

Philippe Mathieu-Daudé <philmd@linaro.org> writes:

> On 24/11/25 11:27, Mark Cave-Ayland wrote:
>> On 23/11/2025 18:58, Philippe Mathieu-Daudé wrote:
>> 
>>> We don't use Xen on the isapc machine: let's remove
>>> pointless code.
>>>
>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>> ---
>>>   hw/i386/isapc.c | 27 ++++-----------------------
>>>   1 file changed, 4 insertions(+), 23 deletions(-)
>>>
>>> diff --git a/hw/i386/isapc.c b/hw/i386/isapc.c
>>> index 1ba9ae22cc3..1e13a30c505 100644
>>> --- a/hw/i386/isapc.c
>>> +++ b/hw/i386/isapc.c
>>> @@ -17,8 +17,6 @@
>>>   #include "hw/ide/ide-bus.h"
>>>   #include "system/kvm.h"
>>>   #include "hw/i386/kvm/clock.h"
>>> -#include "hw/xen/xen-x86.h"
>>> -#include "system/xen.h"
>>>   #include "hw/rtc/mc146818rtc.h"
>>>   #include "target/i386/cpu.h"
>>> @@ -37,7 +35,6 @@ static void pc_init_isa(MachineState *machine)
>>>       ISABus *isa_bus;
>>>       uint32_t irq;
>>>       GSIState *gsi_state;
>>> -    MemoryRegion *ram_memory;
>>>       DriveInfo *hd[MAX_IDE_BUS * MAX_IDE_DEVS];
>>>       int i;
>>> @@ -77,15 +74,9 @@ static void pc_init_isa(MachineState *machine)
>>>       /*
>>>        * There is no RAM split for the isapc machine
>>>        */
>>> -    if (xen_enabled()) {
>>> -        xen_hvm_init_pc(pcms, &ram_memory);
>>> -    } else {
>>> -        ram_memory = machine->ram;
>>> -
>>> -        pcms->max_ram_below_4g = 3.5 * GiB;
>>> -        x86ms->above_4g_mem_size = 0;
>>> -        x86ms->below_4g_mem_size = machine->ram_size;
>>> -    }
>>> +    pcms->max_ram_below_4g = 3.5 * GiB;
>>> +    x86ms->above_4g_mem_size = 0;
>>> +    x86ms->below_4g_mem_size = machine->ram_size;
>>>       x86_cpus_init(x86ms, pcmc->default_cpu_version);
>>> @@ -94,17 +85,7 @@ static void pc_init_isa(MachineState *machine)
>>>       }
>>>       /* allocate ram and load rom/bios */
>>> -    if (!xen_enabled()) {
>>> -        pc_memory_init(pcms, system_memory, system_memory, 0);
>>> -    } else {
>>> -        assert(machine->ram_size == x86ms->below_4g_mem_size +
>>> -                                    x86ms->above_4g_mem_size);
>>> -
>>> -        if (machine->kernel_filename != NULL) {
>>> -            /* For xen HVM direct kernel boot, load linux here */
>>> -            xen_load_linux(pcms);
>>> -        }
>>> -    }
>>> +    pc_memory_init(pcms, system_memory, system_memory, 0);
>>>       gsi_state = pc_gsi_create(&x86ms->gsi, false);
>> FWIW I did contact upstream about this (see https://patchew.org/
>> QEMU/20250618112828.235087-1-
>> mark.caveayland@nutanix.com/20250618112828.235087-13-
>> mark.caveayland@nutanix.com/#91b96b24-af60-4a89-91ec-
>> d0e660a76a0a@nutanix.com) and they mentioned that it should work,
>> which is why I left it in.
>> Unfortunately I don't currently have any way to test it here :/
>
> OK :| IMHO upstream Xen maintainers should provide us a test, otherwise
> we can not make our code base evolve.

Ok how about this:

  As of the xen commit 9e8672f1c36 (tools: remove xend and associated
  python modules) there is no longer the ability to emulate an ISA only
  system. As a result we can drop the code in isapc. We still need to
  support the ISA bus for the -m pc type as Xen doesn't support the q35
  machine model for HVM.

I think the actual support got removed earlier but this was when the
last bits of "Simulate an ISA only system" where removed in 2013.

-- 
Alex Bennée
Virtualisation Tech Lead @ Linaro