pci: introduce get_info_quirk callback.
PATCH 1/1 pci: introduce get_info_quirk callback.
This patch fixes 525e05147d5a3bdc08caa422d108c1ef71b584b5
by introducing device specific get_info_quirk callback.
It wrongly assumes that pci host bridge class device has
header type of pci-pci bridge. But this isn't always true.
In fact i440fx pci host bridge has header type of normal device, hence it breaks i440fx and other pci host bridges.
The right fix is that header type should be checked, instead of device class.
The change set's purpose is to show PBM pci host bridge info which doesn't conform to PCI specification. So introduce get_info_quirk callback and use it for PBM.
--- hw/apb_pci.c | 1 + hw/pci.c | 79 ++++++++++++++++++++++++++++++++------------------------- hw/pci.h | 3 ++ 3 files changed, 48 insertions(+), 35 deletions(-)
hw/apb_pci.c +1 -0
@@ -479,6 +479,7 @@ static PCIDeviceInfo pbm_pci_host_info = { .qdev.name = "pbm", .qdev.size = sizeof(PCIDevice), .init = pbm_pci_host_init, + .get_info_quirk = pci_bridge_get_info, .header_type = PCI_HEADER_TYPE_BRIDGE, };
hw/pci.c +44 -35
99 lines not on this page · raw message
hw/pci.h +3 -0
@@ -80,6 +80,7 @@ typedef uint32_t PCIConfigReadFunc(PCIDevice *pci_dev,typedef void PCIMapIORegionFunc (PCIDevice *pci_dev,int region_num,pcibus_t addr,pcibus_t size,int type);typedef int PCIUnregisterFunc (PCIDevice *pci_dev); +typedef void PCIGetInfoQuirkFunc (PCIDevice *pci_dev,PCIBus *bus,QDict *qdict); typedef struct PCIIORegion { pcibus_t addr; /* current PCI mapping address. -1 means not mapped */
@@ -240,6 +241,7 @@ void do_pci_info(Monitor *mon, QObject **ret_data);PCIBus *pci_bridge_init (PCIBus *bus,int devfn,uint16_t vid,uint16_t did,pci_map_irq_fn map_irq,const char *name);PCIDevice *pci_bridge_get_device (PCIBus *bus); +void pci_bridge_get_info (PCIDevice *dev,PCIBus *bus,QDict *qdict);static inline void pci_set_byte (uint8_t *config,uint8_t val)
@@ -314,6 +316,7 @@ typedef struct {PCIUnregisterFunc *exit;PCIConfigReadFunc *config_read;PCIConfigWriteFunc *config_write; +PCIGetInfoQuirkFunc *get_info_quirk;/* pci config header type */ uint8_t header_type;
Signature · 1 line
1.6.6.1
Quoted 10 lines · Isaku Yamahata
This patch fixes
525e05147d5a3bdc08caa422d108c1ef71b584b5
by introducing device specificget_info_quirkcallback.
It wrongly assumes that pci host bridge class device has
header type of pci-pci bridge. But this isn't always true.
In fact i440fx pci host bridge has header type of normal device, hence it breaks i440fx and other pci host bridges.
The right fix is that header type should be checked, instead of device class.The change set's purpose is to show PBM pci host bridge info which doesn't conform to PCI specification.
So, PBM has header type PCI_HEADER_TYPE_NORMAL
but all config space is in bridge format?
Quoted 116 lines
So introduce
get_info_quirkcallback and use it for PBM.Cc Blue Swirl <[email protected]>Cc "Michael S. Tsirkin" <[email protected]>Signed-off-by Isaku Yamahata <[email protected]>--- hw/apb_pci.c | 1 + hw/pci.c | 79 ++++++++++++++++++++++++++++++++------------------------- hw/pci.h | 3 ++ 3 files changed, 48 insertions(+), 35 deletions(-)hw/apb_pci.c +1
-0@@ -479,6 +479,7 @@ static PCIDeviceInfo pbm_pci_host_info = { .qdev.name = "pbm", .qdev.size = sizeof(PCIDevice), .init = pbm_pci_host_init, + .get_info_quirk = pci_bridge_get_info, .header_type = PCI_HEADER_TYPE_BRIDGE, };hw/pci.c +36
-34@@ -1271,10 +1271,44 @@ static QObject *pci_get_regions_list(const PCIDevice *dev)static QObject *pci_get_devices_list (PCIBus *bus,int bus_num); +void pci_bridge_get_info (PCIDevice *dev,PCIBus *bus,QDict *qdict) +{ +QObject *pci_bridge; + + pci_bridge =qobject_from_jsonf ("{ 'bus': " +"{ 'number': %d, 'secondary': %d, 'subordinate': %d }, " +"'io_range': { 'base': %" PRId64", 'limit': %" PRId64"}, " +"'memory_range': { 'base': %" PRId64", 'limit': %" PRId64"}, " +"'prefetchable_range': { 'base': %" PRId64", 'limit': %" PRId64"} }" , + dev->config[PCI_PRIMARY_BUS ], dev->config[PCI_SECONDARY_BUS ], + dev->config[PCI_SUBORDINATE_BUS ], +pci_bridge_get_base (dev,PCI_BASE_ADDRESS_SPACE_IO ), +pci_bridge_get_limit (dev,PCI_BASE_ADDRESS_SPACE_IO ), +pci_bridge_get_base (dev,PCI_BASE_ADDRESS_SPACE_MEMORY ), +pci_bridge_get_limit (dev,PCI_BASE_ADDRESS_SPACE_MEMORY ), +pci_bridge_get_base (dev,PCI_BASE_ADDRESS_SPACE_MEMORY | +PCI_BASE_ADDRESS_MEM_PREFETCH ), +pci_bridge_get_limit (dev,PCI_BASE_ADDRESS_SPACE_MEMORY | +PCI_BASE_ADDRESS_MEM_PREFETCH )); + +if (dev->config[PCI_SECONDARY_BUS ] !=0 ) { +PCIBus *child_bus =pci_find_bus (bus, dev->config[PCI_SECONDARY_BUS ]); +if (child_bus) { +qdict_put_obj (qobject_to_qdict (pci_bridge),"devices" , +pci_get_devices_list (child_bus, + dev->config[PCI_SECONDARY_BUS ])); + } + } + +qdict_put_obj (qdict,"pci_bridge" , pci_bridge); +} +static QObject *pci_get_dev_dict (PCIDevice *dev,PCIBus *bus,int bus_num) { -int class; +PCIDeviceInfo *info =DO_UPCAST (PCIDeviceInfo, qdev, dev->qdev.info);QObject *obj; +QDict *qdict; +uint8_t type; obj = qobject_from_jsonf("{ 'bus': %d, 'slot': %d, 'function': %d," "'class_info': %p, 'id': %p, 'regions': %p," " 'qdev_id': %s }",@@ -1283,45 +1317,20 @@ static QObject *pci_get_dev_dict(PCIDevice *dev, PCIBus *bus, int bus_num)pci_get_dev_class (dev),pci_get_dev_id (dev),pci_get_regions_list (dev), dev->qdev.id ? dev->qdev.id : ""); + qdict =qobject_to_qdict (obj);if (dev->config[PCI_INTERRUPT_PIN ] !=0 ) { -QDict *qdict =qobject_to_qdict (obj);qdict_put (qdict,"irq" ,qint_from_int (dev->config[PCI_INTERRUPT_LINE ])); } - class =pci_get_word (dev->config +PCI_CLASS_DEVICE ); -if (class ==PCI_CLASS_BRIDGE_HOST || class ==PCI_CLASS_BRIDGE_PCI ) { -QDict *qdict; -QObject *pci_bridge; - - pci_bridge =qobject_from_jsonf ("{ 'bus': " -"{ 'number': %d, 'secondary': %d, 'subordinate': %d }, " -"'io_range': { 'base': %" PRId64", 'limit': %" PRId64"}, " -"'memory_range': { 'base': %" PRId64", 'limit': %" PRId64"}, " -"'prefetchable_range': { 'base': %" PRId64", 'limit': %" PRId64"} }" , - dev->config[PCI_PRIMARY_BUS ], dev->config[PCI_SECONDARY_BUS ], - dev->config[PCI_SUBORDINATE_BUS ], -pci_bridge_get_base (dev,PCI_BASE_ADDRESS_SPACE_IO ), -pci_bridge_get_limit (dev,PCI_BASE_ADDRESS_SPACE_IO ), -pci_bridge_get_base (dev,PCI_BASE_ADDRESS_SPACE_MEMORY ), -pci_bridge_get_limit (dev,PCI_BASE_ADDRESS_SPACE_MEMORY ), -pci_bridge_get_base (dev,PCI_BASE_ADDRESS_SPACE_MEMORY | -PCI_BASE_ADDRESS_MEM_PREFETCH ), -pci_bridge_get_limit (dev,PCI_BASE_ADDRESS_SPACE_MEMORY | -PCI_BASE_ADDRESS_MEM_PREFETCH )); - -if (dev->config[PCI_SECONDARY_BUS ] !=0 ) { -PCIBus *child_bus =pci_find_bus (bus, dev->config[PCI_SECONDARY_BUS ]); - -if (child_bus) { - qdict =qobject_to_qdict (pci_bridge); -qdict_put_obj (qdict,"devices" , -pci_get_devices_list (child_bus, - dev->config[PCI_SECONDARY_BUS ])); - } - } - qdict =qobject_to_qdict (obj);
hw/pci.c static QObject *pci_get_dev_dict(PCIDevice *dev, PCIBus *bus, int bus_num) · line 1334
Quoted 10 lines
hw/pci.c +8
-1-qdict_put_obj (qdict,"pci_bridge" , pci_bridge); + type = dev->config[PCI_HEADER_TYPE ] & ~PCI_HEADER_TYPE_MULTI_FUNCTION ; +if (type ==PCI_HEADER_TYPE_BRIDGE ) { +pci_bridge_get_info (dev, bus, qdict); + } + +if (info/* not all pci device aren't converted qdev yet */ && + info->get_info_quirk) { + info->get_info_quirk (dev, bus, qdict); }
this should be:
Quoted 5 lines
hw/pci.c +0
-0if (info/* not all pci device aren't converted qdev yet */ && info->get_info_quirk) {info->get_info_quirk(dev, bus, qdict); } else if type == PCI_HEADER_TYPE_BRIDGE) { pci_bridge_get_info(dev, bus, qdict);
}
Quoted 32 lines
return obj;hw/pci.h +3
-0@@ -80,6 +80,7 @@ typedef uint32_t PCIConfigReadFunc(PCIDevice *pci_dev,typedef void PCIMapIORegionFunc (PCIDevice *pci_dev,int region_num,pcibus_t addr,pcibus_t size,int type);typedef int PCIUnregisterFunc (PCIDevice *pci_dev); +typedef void PCIGetInfoQuirkFunc (PCIDevice *pci_dev,PCIBus *bus,QDict *qdict); typedef struct PCIIORegion { pcibus_t addr; /* current PCI mapping address. -1 means not mapped */@@ -240,6 +241,7 @@ void do_pci_info(Monitor *mon, QObject **ret_data);PCIBus *pci_bridge_init (PCIBus *bus,int devfn,uint16_t vid,uint16_t did,pci_map_irq_fn map_irq,const char *name);PCIDevice *pci_bridge_get_device (PCIBus *bus); +void pci_bridge_get_info (PCIDevice *dev,PCIBus *bus,QDict *qdict);static inline void pci_set_byte (uint8_t *config,uint8_t val)@@ -314,6 +316,7 @@ typedef struct {PCIUnregisterFunc *exit;PCIConfigReadFunc *config_read;PCIConfigWriteFunc *config_write; +PCIGetInfoQuirkFunc *get_info_quirk;/* pci config header type */ uint8_t header_type;Signature · 1 line
1.6.6.1
Quoted 9 lines · Michael S. Tsirkin
Quoted 8 lines · Isaku Yamahata
This patch fixes
525e05147d5a3bdc08caa422d108c1ef71b584b5
by introducing device specificget_info_quirkcallback.
It wrongly assumes that pci host bridge class device has
header type of pci-pci bridge. But this isn't always true.
In fact i440fx pci host bridge has header type of normal device, hence it breaks i440fx and other pci host bridges.
The right fix is that header type should be checked, instead of device class.
Quoted 5 lines
Quoted 2 lines
The change set's purpose is to show PBM pci host bridge info which doesn't conform to PCI specification.
So, PBM has header type
PCI_HEADER_TYPE_NORMAL
but all config space is in bridge format?
Some of registers is in bridge format.
To be honest, I don't know whether all or some. Blue seems to want bus numbers registers at least.
http://lists.gnu.org/archive/html/qemu-devel/2010-02/msg00519.html
Signature · 1 line
yamahata
Quoted 13 lines · Isaku Yamahata
Quoted 12 lines · Michael S. Tsirkin
Quoted 10 lines · Isaku Yamahata
This patch fixes
525e05147d5a3bdc08caa422d108c1ef71b584b5
by introducing device specificget_info_quirkcallback.
It wrongly assumes that pci host bridge class device has
header type of pci-pci bridge. But this isn't always true.
In fact i440fx pci host bridge has header type of normal device, hence it breaks i440fx and other pci host bridges.
The right fix is that header type should be checked, instead of device class.The change set's purpose is to show PBM pci host bridge info which doesn't conform to PCI specification.
Quoted 8 lines
Quoted 2 lines
So, PBM has header type
PCI_HEADER_TYPE_NORMAL
but all config space is in bridge format?Some of registers is in bridge format.
To be honest, I don't know whether all or some. Blue seems to want bus numbers registers at least.http://lists.gnu.org/archive/html/qemu-devel/2010-02/msg00519.html
I'm not so sure anymore, I'll have to check this in the weekend. Perhaps there was some other problem with bus numbering or OpenBIOS PCI programming which resulted in the need to program PRIMARY/ SECONDARY/SUBORDINATE bus values for host bridge. Does it make any sense to use anything other than bus 0 for PCI bus attached to host bridge?
Quoted 21 lines · Blue Swirl
Quoted 20 lines · Isaku Yamahata
Quoted 9 lines · Michael S. Tsirkin
Quoted 8 lines · Isaku Yamahata
This patch fixes
525e05147d5a3bdc08caa422d108c1ef71b584b5
by introducing device specificget_info_quirkcallback.
It wrongly assumes that pci host bridge class device has
header type of pci-pci bridge. But this isn't always true.
In fact i440fx pci host bridge has header type of normal device, hence it breaks i440fx and other pci host bridges.
The right fix is that header type should be checked, instead of device class.Quoted 5 lines
Quoted 2 lines
The change set's purpose is to show PBM pci host bridge info which doesn't conform to PCI specification.
So, PBM has header type
PCI_HEADER_TYPE_NORMAL
but all config space is in bridge format?Some of registers is in bridge format.
To be honest, I don't know whether all or some. Blue seems to want bus numbers registers at least.
Quoted 8 lines
I'm not so sure anymore, I'll have to check this in the weekend. Perhaps there was some other problem with bus numbering or OpenBIOS PCI programming which resulted in the need to program PRIMARY/ SECONDARY/SUBORDINATE bus values for host bridge. Does it make any sense to use anything other than bus 0 for PCI bus attached to host bridge?
I found the bug in my new PCI probing code, 525e05147d5a3bdc08caa422d108c1ef71b584b5 was not correct after all. There is no need to configure host brige bus values.
So I'll apply Isaku's earlier fix that removes the bogus host bridge data from 'info pci'.