Re: [Qemu-devel] [PATCH 1/2] prep_pci: set isa_mem_base in the PCI host bridge

[Hervé Poussineau]

Andreas Färber a écrit :
Am 23.07.2013 23:16, schrieb Hervé Poussineau:
Currently, it is done by i82378 device, which shouldn't care of it.
Signed-off-by: Hervé Poussineau <hpous...@reactos.org>
---
 hw/isa/i82378.c    |    3 ---
 hw/pci-host/prep.c |    2 ++
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/hw/isa/i82378.c b/hw/isa/i82378.c
index b25ed04..de71d81 100644
--- a/hw/isa/i82378.c
+++ b/hw/isa/i82378.c
@@ -45,7 +45,6 @@ typedef struct I82378State {
 typedef struct PCIi82378State {
     PCIDevice pci_dev;
     uint32_t isa_io_base;
-    uint32_t isa_mem_base;
     I82378State state;
 } PCIi82378State;
 
@@ -234,7 +233,6 @@ static int pci_i82378_init(PCIDevice *dev)
     pci_set_long(dev->wmask + PCI_BASE_ADDRESS_0, 0);
     pci_set_long(pci_conf + PCI_BASE_ADDRESS_0, pci->isa_io_base);
 
-    isa_mem_base = pci->isa_mem_base;
     isa_bus_new(&dev->qdev, pci_address_space_io(dev));
 
     i82378_init(&dev->qdev, s);
@@ -244,7 +242,6 @@ static int pci_i82378_init(PCIDevice *dev)
 
 static Property i82378_properties[] = {
     DEFINE_PROP_HEX32("iobase", PCIi82378State, isa_io_base, 0x80000000),
-    DEFINE_PROP_HEX32("membase", PCIi82378State, isa_mem_base, 0xc0000000),
     DEFINE_PROP_END_OF_LIST()
 };
 
diff --git a/hw/pci-host/prep.c b/hw/pci-host/prep.c
index b41d564..d6bcc63 100644
--- a/hw/pci-host/prep.c
+++ b/hw/pci-host/prep.c
@@ -119,6 +119,8 @@ static void raven_pcihost_realizefn(DeviceState *d, Error 
**errp)
     MemoryRegion *address_space_mem = get_system_memory();
     int i;
 
+    isa_mem_base = 0xc0000000;
+
     for (i = 0; i < 4; i++) {
         sysbus_init_irq(dev, &s->irq[i]);
     }

Patch is okay with me, but I wonder what we still need the global
isa_mem_base for? The only users seem to be VGA, adding offsets to it.

Indeed, isa_mem_base should be removed, and PCI bus regions (memory and 
I/O) should be adapted in PCI host.
However, while I have a patch pending, it is not ready yet, so I prefer 
to postpone it past QEMU 1.6.

Hervé