On Tue, 28 Oct 2025, Philippe Mathieu-Daudé wrote:
On 27/10/25 20:47, BALATON Zoltan wrote:
On Mon, 27 Oct 2025, Philippe Mathieu-Daudé wrote:
On 25/10/25 01:31, BALATON Zoltan wrote:
These memory windows are a result of the address decoding in the
Articia S north bridge so better model it there and not in board code.
Suggested-by: Philippe Mathieu-Daudé <[email protected]>
Signed-off-by: BALATON Zoltan <[email protected]>
---
hw/pci-host/articia.c | 15 ++++++++++++++-
hw/ppc/amigaone.c | 28 +++++-----------------------
hw/ppc/pegasos2.c | 13 -------------
3 files changed, 19 insertions(+), 37 deletions(-)
@@ -169,6 +174,7 @@ static void articia_realize(DeviceState *dev, Error
**errp)
{
ArticiaState *s = ARTICIA(dev);
PCIHostState *h = PCI_HOST_BRIDGE(dev);
+ MemoryRegion *mr;
PCIDevice *pdev;
bitbang_i2c_init(&s->smbus, i2c_init_bus(dev, "smbus"));
@@ -180,6 +186,14 @@ static void articia_realize(DeviceState *dev, Error
**errp)
memory_region_init_io(&s->reg, OBJECT(s), &articia_reg_ops, s,
TYPE_ARTICIA, 0x1000000);
memory_region_add_subregion_overlap(&s->reg, 0, &s->io, 1);
+ mr = g_new(MemoryRegion, 1);
Won't Coverity or other analysis tools complain about the leak?
(this is why we usually keep a reference in the device state, here
ArticiaState). Otherwise:
According to https://www.qemu.org/docs/master/devel/memory.html#region-
lifecycle
there should be no leak and keeping a reference should not be necessary as
the lifetime is managed by attaching it to the owner object so no need to
keep a reference when it's not needed otherwise. Not littering the state
struct with unneded references makes it easier to comprehend so I'd only
keep things there that are necessary.
IIUC this doc is about what happens within the allocated MemoryRegion,
regardless of where it is allocated.
That doc explicitely says:
"Destruction of a memory region happens automatically when the owner
object dies. When there are multiple memory regions under the same owner
object, the memory API will guarantee all memory regions will be properly
detached and finalized one by one. The order in which memory regions will
be finalized is not guaranteed."
(and these pci-host objects are created once at machine init and never die
so the question seems quite theoretical). I'd like to keep object state
simple and not keep around references in it that nothing uses and should
be managed automatically. I'd only add fields to the state struct that
other methods need.
Regards,
BALATON Zoltan
