Skip to contentSkip to conversation

Fix ohci_die() and move PCI code to separate file

PATCH In review Thomas Huth 3 participants 9 messages Apr 19, 2019 – May 2, 2019

lore
Share & tools
lore
#0 lore raw

v2 cover Fix ohci_die() and move PCI code to separate file

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Thomas Huth (2):
hw/usb/hcd-ohci: Do not use PCI functions with sysbus devices in

    ohci_die()

hw/usb/hcd-ohci: Move PCI-related code into a separate file

 hw/sh4/Kconfig        |   2 +-
 hw/usb/Kconfig        |   6 +-
 hw/usb/Makefile.objs  |   1 +
 hw/usb/hcd-ohci-pci.c | 163 +++++++++++++++++++++++++++++++
 hw/usb/hcd-ohci.c     | 219 ++++--------------------------------------
 hw/usb/hcd-ohci.h     | 104 ++++++++++++++++++++
 6 files changed, 293 insertions(+), 202 deletions(-)
 create mode 100644 hw/usb/hcd-ohci-pci.c
 create mode 100644 hw/usb/hcd-ohci.h
Signature · 1 line
2.21.0
#1 lore raw

PATCH v2 1/2 hw/usb/hcd-ohci: Do not use PCI functions with sysbus devices in ohci_die()

The ohci_die() function always assumes to be running with a PCI OHCI controller and calls the PCI-specific functions pci_set_word(). However, this function might also get called for the sysbus OHCI devices, so it likely fails in that case. To fix this issue, change the code now, so that there are two implementations now, one for sysbus and one for PCI, and use the right function via a function pointer in the OHCIState structure.

---
 hw/usb/hcd-ohci.c | 39 +++++++++++++++++++++++++++------------
 1 file changed, 27 insertions(+), 12 deletions(-)
hw/usb/hcd-ohci.c +27 -12
index 81cf5ab7a5..6d3f556989 100644
@@ -52,7 +52,7 @@ typedef struct OHCIPort {
     uint32_t ctrl;
 } OHCIPort;
 
-typedef struct {
+typedef struct OHCIState {
     USBBus bus;
     qemu_irq irq;
     MemoryRegion mem;
@@ -108,6 +108,7 @@ typedef struct {
     uint32_t async_td;
     bool async_complete;
 
+    void (*ohci_die)(struct OHCIState *ohci);
 } OHCIState;
 
 /* Host Controller Communications Area */
@@ -302,7 +303,10 @@ struct ohci_iso_td {
 
 #define OHCI_HRESET_FSBIR       (1 << 0)
 
-static void ohci_die(OHCIState *ohci);
+static void ohci_die(OHCIState *ohci)
+{
+    ohci->ohci_die(ohci);
+}
 
 /* Update IRQ levels */
 static inline void ohci_intr_update(OHCIState *ohci)
@@ -1854,13 +1858,14 @@ static USBBusOps ohci_bus_ops = {
 
 static void usb_ohci_init(OHCIState *ohci, DeviceState *dev,
                           uint32_t num_ports, dma_addr_t localmem_base,
-                          char *masterbus, uint32_t firstport,
-                          AddressSpace *as, Error **errp)
+                          char *masterbus, uint32_t firstport, AddressSpace *as,
+                          void (*ohci_die_fn)(struct OHCIState *), Error **errp)
 {
     Error *err = NULL;
     int i;
 
     ohci->as = as;
+    ohci->ohci_die = ohci_die_fn;
 
     if (num_ports > OHCI_MAX_PORTS) {
         error_setg(errp, "OHCI num-ports=%u is too big (limit is %u ports)",
@@ -1933,18 +1938,28 @@ typedef struct {
     uint32_t firstport;
 } OHCIPCIState;
 
-/** A typical O/EHCI will stop operating, set itself into error state
- * (which can be queried by MMIO) and will set PERR in its config
- * space to signal that it got an error
+/**
+ * A typical OHCI will stop operating and set itself into error state
+ * (which can be queried by MMIO) to signal that it got an error.
  */
-static void ohci_die(OHCIState *ohci)
+static void ohci_sysbus_die(struct OHCIState *ohci)
 {
-    OHCIPCIState *dev = container_of(ohci, OHCIPCIState, state);
-
     trace_usb_ohci_die();
 
     ohci_set_interrupt(ohci, OHCI_INTR_UE);
     ohci_bus_stop(ohci);
+}
+
+/**
+ * A typical PCI OHCI will additionally set PERR in its configspace to
+ * signal that it got an error.
+ */
+static void ohci_pci_die(struct OHCIState *ohci)
+{
+    OHCIPCIState *dev = container_of(ohci, OHCIPCIState, state);
+
+    ohci_sysbus_die(ohci);
+
     pci_set_word(dev->parent_obj.config + PCI_STATUS,
                  PCI_STATUS_DETECTED_PARITY);
 }
@@ -1959,7 +1974,7 @@ static void usb_ohci_realize_pci(PCIDevice *dev, Error **errp)
 
     usb_ohci_init(&ohci->state, DEVICE(dev), ohci->num_ports, 0,
                   ohci->masterbus, ohci->firstport,
-                  pci_get_address_space(dev), &err);
+                  pci_get_address_space(dev), ohci_pci_die, &err);
     if (err) {
         error_propagate(errp, err);
         return;
@@ -2023,7 +2038,7 @@ static void ohci_realize_pxa(DeviceState *dev, Error **errp)
 
     usb_ohci_init(&s->ohci, dev, s->num_ports, s->dma_offset,
                   s->masterbus, s->firstport,
-                  &address_space_memory, &err);
+                  &address_space_memory, ohci_sysbus_die, &err);
     if (err) {
         error_propagate(errp, err);
         return;
Signature · 1 line
2.21.0
#2 lore raw

PATCH v2 2/2 hw/usb/hcd-ohci: Move PCI-related code into a separate file

In reply to Thomas Huth

Some machines (like the pxa2xx-based ARM machines) only have a sysbus OHCI controller, but no PCI. With the new Kconfig-style build system, it will soon be possible to create QEMU binaries that only contain such PCI-less machines. However, the two OHCI controllers, for sysbus and for PCI, are currently both located in one file, so the PCI code is still required for linking here. Move the OHCI-PCI device code into a separate file, so that it is possible to use the sysbus OHCI device also without the PCI dependency.

---
 hw/sh4/Kconfig        |   2 +-
 hw/usb/Kconfig        |   6 +-
 hw/usb/Makefile.objs  |   1 +
 hw/usb/hcd-ohci-pci.c | 163 +++++++++++++++++++++++++++++++
 hw/usb/hcd-ohci.c     | 216 ++----------------------------------------
 hw/usb/hcd-ohci.h     | 104 ++++++++++++++++++++
 6 files changed, 284 insertions(+), 208 deletions(-)
 create mode 100644 hw/usb/hcd-ohci-pci.c
 create mode 100644 hw/usb/hcd-ohci.h
hw/sh4/Kconfig +1 -1
index 593662d28a..4cbce3a0ed 100644
@@ -6,7 +6,7 @@ config R2D
     select I82378 if TEST_DEVICES
     select IDE_MMIO
     select PFLASH_CFI02
-    select USB_OHCI
+    select USB_OHCI_PCI
     select PCI
     select SM501
     select SH4
hw/usb/Kconfig +5 -1
index a1b7acb12a..564305e283 100644
@@ -8,10 +8,14 @@ config USB_UHCI
     select USB
 
 config USB_OHCI
+    bool
+    select USB
+
+config USB_OHCI_PCI
     bool
     default y if PCI_DEVICES
     depends on PCI
-    select USB
+    select USB_OHCI
 
 config USB_EHCI
     bool
hw/usb/Makefile.objs +1 -0
index 2b929649ac..81688f6e70 100644
@@ -5,6 +5,7 @@ common-obj-$(CONFIG_USB) += desc.o desc-msos.o
 # usb host adapters
 common-obj-$(CONFIG_USB_UHCI) += hcd-uhci.o
 common-obj-$(CONFIG_USB_OHCI) += hcd-ohci.o
+common-obj-$(CONFIG_USB_OHCI_PCI) += hcd-ohci-pci.o
 common-obj-$(CONFIG_USB_EHCI) += hcd-ehci.o hcd-ehci-pci.o
 common-obj-$(CONFIG_USB_EHCI_SYSBUS) += hcd-ehci.o hcd-ehci-sysbus.o
 common-obj-$(CONFIG_USB_XHCI) += hcd-xhci.o
hw/usb/hcd-ohci-pci.c new file +163 -0

163 lines not on this page · raw message

hw/usb/hcd-ohci.c +10 -206

304 lines not on this page · raw message

hw/usb/hcd-ohci.h new file +104 -0

104 lines not on this page · raw message

Signature · 1 line
2.21.0
Reviewed-by Tested-by on the v2 series
#3 lore raw
In reply to Thomas Huth

Hi Thomas,

Quoted 6 lines · Thomas Huth

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Way cleaner. I wonder why you don't use a typedef for the void (*ohci_die_fn)(struct OHCIState *) prototype.
Anyway to this series:

Quoted 16 lines

Thomas Huth (2):
hw/usb/hcd-ohci: Do not use PCI functions with sysbus devices in

    ohci_die()

hw/usb/hcd-ohci: Move PCI-related code into a separate file

 hw/sh4/Kconfig        |   2 +-
 hw/usb/Kconfig        |   6 +-
 hw/usb/Makefile.objs  |   1 +
 hw/usb/hcd-ohci-pci.c | 163 +++++++++++++++++++++++++++++++
 hw/usb/hcd-ohci.c     | 219 ++++--------------------------------------
 hw/usb/hcd-ohci.h     | 104 ++++++++++++++++++++
 6 files changed, 293 insertions(+), 202 deletions(-)
 create mode 100644 hw/usb/hcd-ohci-pci.c
 create mode 100644 hw/usb/hcd-ohci.h
#4 lore raw
Quoted 7 lines · Philippe Mathieu-Daudé

Hi Thomas,

Quoted 4 lines · Thomas Huth

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

Quoted 5 lines
Quoted 2 lines

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Way cleaner. I wonder why you don't use a typedef for the void (*ohci_die_fn)(struct OHCIState *) prototype.

It does not work in that case:

typedef struct OHCIState { // <-- struct OHCIState definition

    [...]
    uint32_t async_td;
    bool async_complete;

    void (*ohci_die)(struct OHCIState *ohci); // <-- ohci_die definition
} OHCIState;                         // <-- typedef OHCIState definition

The typedef is defined after the ohci_die entry.

Quoted 3 lines

Anyway to this series:

Thanks!
Thomas

#5 lore raw
Quoted 17 lines · Thomas Huth
Quoted 7 lines · Philippe Mathieu-Daudé

Hi Thomas,

Quoted 4 lines · Thomas Huth

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

Quoted 5 lines
Quoted 2 lines

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Way cleaner. I wonder why you don't use a typedef for the void (*ohci_die_fn)(struct OHCIState *) prototype.

It does not work in that case:

typedef struct OHCIState { // <-- struct OHCIState definition

Quoted 8 lines
    [...]
    uint32_t async_td;
    bool async_complete;

    void (*ohci_die)(struct OHCIState *ohci); // <-- ohci_die definition
} OHCIState;                         // <-- typedef OHCIState definition

The typedef is defined after the ohci_die entry.

I was thinking of forward declaration:

typedef struct OHCIState OHCIState;

typedef void (ohci_die_fn)(OHCIState *ohci);

struct OHCIState {
    [...]
    uint32_t async_td;
    bool async_complete;

    ohci_die_fn *ohci_die;
} OHCIState;

static void usb_ohci_init(OHCIState *ohci, DeviceState *dev,
uint32_t num_ports, dma_addr_t localmem_base, char *masterbus, uint32_t firstport, AddressSpace *as, ohci_die_fn *ohci_die, Error **errp)
{ ...

Quoted 8 lines
Quoted 4 lines

Anyway to this series:

Thanks!
Thomas

#6 lore raw
Quoted 25 lines · Philippe Mathieu-Daudé
Quoted 24 lines · Thomas Huth
Quoted 7 lines · Philippe Mathieu-Daudé

Hi Thomas,

Quoted 4 lines · Thomas Huth

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

Quoted 5 lines
Quoted 2 lines

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Way cleaner. I wonder why you don't use a typedef for the void (*ohci_die_fn)(struct OHCIState *) prototype.

It does not work in that case:

typedef struct OHCIState { // <-- struct OHCIState definition

    [...]
    uint32_t async_td;
    bool async_complete;

    void (*ohci_die)(struct OHCIState *ohci); // <-- ohci_die definition
} OHCIState;                         // <-- typedef OHCIState definition
Quoted 7 lines
Quoted 1 line

The typedef is defined after the ohci_die entry.

I was thinking of forward declaration:

typedef struct OHCIState OHCIState;

typedef void (ohci_die_fn)(OHCIState *ohci);

Could work, too, but I don't like typedeferities... so unless Gerd forces me to use that here, I'd prefer to keep the patch in its current shape.

Thomas

#7 lore raw
Quoted 30 lines · Thomas Huth
Quoted 29 lines · Philippe Mathieu-Daudé
Quoted 17 lines · Thomas Huth
Quoted 7 lines · Philippe Mathieu-Daudé

Hi Thomas,

Quoted 4 lines · Thomas Huth

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

Quoted 5 lines
Quoted 2 lines

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Way cleaner. I wonder why you don't use a typedef for the void (*ohci_die_fn)(struct OHCIState *) prototype.

It does not work in that case:

typedef struct OHCIState { // <-- struct OHCIState definition

Quoted 8 lines
    [...]
    uint32_t async_td;
    bool async_complete;

    void (*ohci_die)(struct OHCIState *ohci); // <-- ohci_die definition
} OHCIState;                         // <-- typedef OHCIState definition

The typedef is defined after the ohci_die entry.

I was thinking of forward declaration:

Quoted 7 lines
Quoted 3 lines

typedef struct OHCIState OHCIState;

typedef void (ohci_die_fn)(OHCIState *ohci);

Could work, too, but I don't like typedeferities... so unless Gerd forces me to use that here, I'd prefer to keep the patch in its current shape.

Fine with me ;)

#8 lore raw
In reply to Thomas Huth
Quoted 11 lines · Thomas Huth

First patch fixes a problem with ohci_die(), second patch moves PCI code into a separate file, so that the sysbus OHCI device can also be used without the dependency on the PCI code.

v2: Split the patch into two patches, one for the ohci_die() fix and one
for the PCI code movement.

Thomas Huth (2):
hw/usb/hcd-ohci: Do not use PCI functions with sysbus devices in

    ohci_die()

hw/usb/hcd-ohci: Move PCI-related code into a separate file

Added to usb queue.

thanks,
Gerd