Wire up USB controllers in i.MX6 emulations
PATCH v3 4/5 hw/arm/fsl-imx6ul: Wire up USB controllers
In reply to Guenter RoeckIMX6UL USB controllers are quite similar to IMX7 USB controllers. Wire them up the same way.
The only real difference is that wiring up phy devices is necessary to avoid phy reset timeouts in the Linux kernel.
---
v3: Wire up USB and USB PHY controllers in separate loops
While the number of USB controllers matches the number of USB PHYs, they are logically different.
v2: Use USB PHY emulation
hw/arm/fsl-imx6ul.c | 35 +++++++++++++++++++++++++++++++++++ include/hw/arm/fsl-imx6ul.h | 10 ++++++++++ 2 files changed, 45 insertions(+)
hw/arm/fsl-imx6ul.c +35 -0
@@ -20,6 +20,7 @@#include "qapi/error.h" #include "hw/arm/fsl-imx6ul.h" #include "hw/misc/unimp.h" +#include "hw/usb/imx-usb-phy.h" #include "hw/boards.h" #include "sysemu/sysemu.h" #include "qemu/error-report.h"
@@ -133,6 +134,18 @@ static void fsl_imx6ul_init(Object *obj) TYPE_IMX_ENET); } +/* USB */ +for (i =0 ; i <FSL_IMX6UL_NUM_USB_PHYS ; i++) { +snprintf (name,NAME_SIZE ,"usbphy%d" , i); +sysbus_init_child_obj (obj, name, &s->usbphy[i],sizeof (s->usbphy[i]), +TYPE_IMX_USBPHY ); + } +for (i =0 ; i <FSL_IMX6UL_NUM_USBS ; i++) { +snprintf (name,NAME_SIZE ,"usb%d" , i); +sysbus_init_child_obj (obj, name, &s->usb[i],sizeof (s->usb[i]), +TYPE_CHIPIDEA ); + } +/* * SDHCI */
@@ -456,6 +469,28 @@ static void fsl_imx6ul_realize(DeviceState *dev, Error **errp) FSL_IMX6UL_ENETn_TIMER_IRQ[i])); } +/* USB */ +for (i =0 ; i <FSL_IMX6UL_NUM_USB_PHYS ; i++) { +object_property_set_bool (OBJECT (&s->usbphy[i]), true,"realized" , + &error_abort); +sysbus_mmio_map (SYS_BUS_DEVICE (&s->usbphy[i]),0 , +FSL_IMX6UL_USBPHY1_ADDR + i *0x1000 ); + } + +for (i =0 ; i <FSL_IMX6UL_NUM_USBS ; i++) { +static const int FSL_IMX6UL_USBn_IRQ[] = { +FSL_IMX6UL_USB1_IRQ , +FSL_IMX6UL_USB2_IRQ , + }; +object_property_set_bool (OBJECT (&s->usb[i]), true,"realized" , + &error_abort); +sysbus_mmio_map (SYS_BUS_DEVICE (&s->usb[i]),0 , +FSL_IMX6UL_USBO2_USB_ADDR + i *0x200 ); +sysbus_connect_irq (SYS_BUS_DEVICE (&s->usb[i]),0 , +qdev_get_gpio_in (DEVICE (&s->a7mpcore), + FSL_IMX6UL_USBn_IRQ[i])); + } +/* * USDHC */
include/hw/arm/fsl-imx6ul.h +10 -0
@@ -34,6 +34,8 @@#include "hw/sd/sdhci.h" #include "hw/ssi/imx_spi.h" #include "hw/net/imx_fec.h" +#include "hw/usb/chipidea.h" +#include "hw/usb/imx-usb-phy.h" #include "exec/memory.h" #include "cpu.h"
@@ -54,6 +56,8 @@ enum FslIMX6ULConfiguration {FSL_IMX6UL_NUM_I2CS =4 ,FSL_IMX6UL_NUM_ECSPIS =4 ,FSL_IMX6UL_NUM_ADCS =2 , +FSL_IMX6UL_NUM_USB_PHYS =2 , + FSL_IMX6UL_NUM_USBS = 2, }; typedef struct FslIMX6ULState {
@@ -77,6 +81,8 @@ typedef struct FslIMX6ULState {IMXFECState eth[FSL_IMX6UL_NUM_ETHS ];SDHCIState usdhc[FSL_IMX6UL_NUM_USDHCS ];IMX2WdtState wdt[FSL_IMX6UL_NUM_WDTS ]; +IMXUSBPHYState usbphy[FSL_IMX6UL_NUM_USB_PHYS ]; +ChipideaState usb[FSL_IMX6UL_NUM_USBS ];MemoryRegion rom;MemoryRegion caam;MemoryRegion ocram;
@@ -145,6 +151,10 @@ enum FslIMX6ULMemoryMap {FSL_IMX6UL_EPIT2_ADDR =0x020D4000 ,FSL_IMX6UL_EPIT1_ADDR =0x020D0000 ,FSL_IMX6UL_SNVS_HP_ADDR =0x020CC000 , +FSL_IMX6UL_USBPHY2_ADDR =0x020CA000 , +FSL_IMX6UL_USBPHY2_SIZE = (4 *1024 ), +FSL_IMX6UL_USBPHY1_ADDR =0x020C9000 , +FSL_IMX6UL_USBPHY1_SIZE = (4 *1024 ),FSL_IMX6UL_ANALOG_ADDR =0x020C8000 ,FSL_IMX6UL_CCM_ADDR =0x020C4000 , FSL_IMX6UL_WDOG2_ADDR = 0x020C0000,
Signature · 1 line
2.17.1
v3 cover Wire up USB controllers in i.MX6 emulations
This patch series wires up the USB controllers on fsl-imx6 and fsl-imx6ul emulations.
The first patch provides a basic implementation of the USB PHY controller used in i.MX28 and later chips. Only reset bit handling in the control register is actually implemented. Basic USB PHY support is needed to make the USB ports operational in Linux.
The second patch fixes USB and USB PHY interrupt numbers for i.MX6UL.
The third patch instantiates unimplemented pwm and can devices. This patch is necessary to avoid crashes in Linux when it tries to access those devices. The crashes are observed when trying to boot Linux v4.21 or later.
The final two patches instantiate the USB controllers for i.mMX6 and i.MX6UL.
v3:
- Minor cleanup in patch 1/5 (see details in patch)
- Added patch to fix USB and USB PHY interrupt numbers for fsl-imx6ul.
- Added patch to instantiate unimplemented pwm and CAN devices.
- Instantiate USB and USB PHY separately. They are logically different, and the number of instances is not always the same.
v2:
- Implement and instantiate basic USB PHY implementation instead of emulating a single USB PHY register
----------------------------------------------------------------
Guenter Roeck (5):
hw/usb: Add basic i.MX USB Phy support hw/arm/fsl-imx6ul: Fix USB interrupt numbers hw/arm/fsl-imx6ul: Instantiate unimplemented pwm and can devices hw/arm/fsl-imx6ul: Wire up USB controllers hw/arm/fsl-imx6: Wire up USB controllers
MAINTAINERS | 2 + hw/arm/Kconfig | 1 + hw/arm/fsl-imx6.c | 36 +++++++ hw/arm/fsl-imx6ul.c | 49 ++++++++++ hw/usb/Kconfig | 5 + hw/usb/Makefile.objs | 2 + hw/usb/imx-usb-phy.c | 225 +++++++++++++++++++++++++++++++++++++++++++ include/hw/arm/fsl-imx6.h | 6 ++ include/hw/arm/fsl-imx6ul.h | 16 ++- include/hw/usb/imx-usb-phy.h | 53 ++++++++++ 10 files changed, 392 insertions(+), 3 deletions(-) create mode 100644 hw/usb/imx-usb-phy.c create mode 100644 include/hw/usb/imx-usb-phy.h
PATCH v3 2/5 hw/arm/fsl-imx6ul: Fix USB interrupt numbers
USB1 and USB2 interrupt numbers were swapped. USB_PHY2 interrupt number is 45. That didn't really matter up to now since the interrupts were not used, but it needs to be fixed to be able to wire up the USB controllers.
---
v3: New patch
include/hw/arm/fsl-imx6ul.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
include/hw/arm/fsl-imx6ul.h +3 -3
@@ -241,10 +241,10 @@ enum FslIMX6ULIRQs {FSL_IMX6UL_UART7_IRQ =39 ,FSL_IMX6UL_UART8_IRQ =40 , -FSL_IMX6UL_USB1_IRQ =42 , -FSL_IMX6UL_USB2_IRQ =43 , +FSL_IMX6UL_USB1_IRQ =43 , +FSL_IMX6UL_USB2_IRQ =42 ,FSL_IMX6UL_USB_PHY1_IRQ =44 , -FSL_IMX6UL_USB_PHY2_IRQ =44 , +FSL_IMX6UL_USB_PHY2_IRQ =45 ,FSL_IMX6UL_CAAM_JQ2_IRQ =46 , FSL_IMX6UL_CAAM_ERR_IRQ = 47,
Signature · 1 line
2.17.1
PATCH v3 1/5 hw/usb: Add basic i.MX USB Phy support
In reply to Guenter RoeckAdd basic USB PHY support as implemented in i.MX23, i.MX28, i.MX6, and i.MX7 SoCs.
The only support really needed - at least to boot Linux - is support for soft reset, which needs to reset various registers to their initial value. Otherwise, just record register values.
---
v3: Added Reviewed-by:;
dropped duplicate "the" in comments;
added new files to MAINTAINERS
v2: New patch, replacing dummy STMP register support with basic USB PHY
emulation.
MAINTAINERS | 2 + hw/arm/Kconfig | 1 + hw/usb/Kconfig | 5 + hw/usb/Makefile.objs | 2 + hw/usb/imx-usb-phy.c | 225 +++++++++++++++++++++++++++++++++++ include/hw/usb/imx-usb-phy.h | 53 +++++++++ 6 files changed, 288 insertions(+) create mode 100644 hw/usb/imx-usb-phy.c create mode 100644 include/hw/usb/imx-usb-phy.h
MAINTAINERS +2 -0
@@ -748,6 +748,8 @@ F: hw/arm/sabrelite.c F: hw/arm/fsl-imx6.c F: hw/misc/imx6_*.c F: hw/ssi/imx_spi.c +F: hw/usb/imx-usb-phy.c +F: include/hw/usb/imx-usb-phy.h F: include/hw/arm/fsl-imx6.h F: include/hw/misc/imx6_*.h F: include/hw/ssi/imx_spi.h
hw/arm/Kconfig +1 -0
@@ -361,6 +361,7 @@ config FSL_IMX6 select IMX select IMX_FEC select IMX_I2C + select IMX_USBPHY select SDHCI config ASPEED_SOC
hw/usb/Kconfig +5 -0
@@ -91,3 +91,8 @@ config USB_STORAGE_MTP bool default y depends on USB + +config IMX_USBPHY + bool + default y + depends on USB
hw/usb/Makefile.objs +2 -0
@@ -61,3 +61,5 @@ common-obj-$(CONFIG_XEN) += xen-usb.o xen-usb.o-cflags := $(LIBUSB_CFLAGS) xen-usb.o-libs := $(LIBUSB_LIBS) endif + +common-obj-$(CONFIG_IMX_USBPHY) += imx-usb-phy.o
hw/usb/imx-usb-phy.c new file +225 -0
225 lines not on this page · raw message
include/hw/usb/imx-usb-phy.h new file +53 -0
53 lines not on this page · raw message
Signature · 1 line
2.17.1
PATCH v3 3/5 hw/arm/fsl-imx6ul: Instantiate unimplemented pwm and can devices
In reply to Guenter RoeckRecent Linux kernels (post v4.20) crash due to accesses to flexcan and pwm controllers. Instantiate as unimplemented devices to work around the problem.
---
v3: New patch
hw/arm/fsl-imx6ul.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
hw/arm/fsl-imx6ul.c +14 -0
@@ -516,6 +516,20 @@ static void fsl_imx6ul_realize(DeviceState *dev, Error **errp)*/ create_unimplemented_device ("sdma" ,FSL_IMX6UL_SDMA_ADDR ,0x4000 ); +/* +* PWM +*/ +create_unimplemented_device ("pwm1" ,FSL_IMX6UL_PWM1_ADDR ,0x4000 ); +create_unimplemented_device ("pwm2" ,FSL_IMX6UL_PWM2_ADDR ,0x4000 ); +create_unimplemented_device ("pwm3" ,FSL_IMX6UL_PWM3_ADDR ,0x4000 ); +create_unimplemented_device ("pwm4" ,FSL_IMX6UL_PWM4_ADDR ,0x4000 ); + +/* +* CAN +*/ +create_unimplemented_device ("can1" ,FSL_IMX6UL_CAN1_ADDR ,0x4000 ); +create_unimplemented_device ("can2" ,FSL_IMX6UL_CAN2_ADDR ,0x4000 ); +/* * APHB_DMA */
Signature · 1 line
2.17.1
PATCH v3 5/5 hw/arm/fsl-imx6: Wire up USB controllers
In reply to Guenter RoeckWith this patch, the USB controllers on 'sabrelite' are detected and can be used to boot the system.
---
v3: Wire up USB and USB PHY controllers separately.
The number of USB controllers does not match the number of USB PHYs, and they are logically different.
v2: Use USB PHY emulation
hw/arm/fsl-imx6.c | 36 ++++++++++++++++++++++++++++++++++++ include/hw/arm/fsl-imx6.h | 6 ++++++ 2 files changed, 42 insertions(+)
hw/arm/fsl-imx6.c +36 -0
@@ -22,6 +22,7 @@#include "qemu/osdep.h" #include "qapi/error.h" #include "hw/arm/fsl-imx6.h" +#include "hw/usb/imx-usb-phy.h" #include "hw/boards.h" #include "hw/qdev-properties.h" #include "sysemu/sysemu.h"
@@ -86,6 +87,17 @@ static void fsl_imx6_init(Object *obj) TYPE_IMX_USDHC); } +for (i =0 ; i <FSL_IMX6_NUM_USB_PHYS ; i++) { +snprintf (name,NAME_SIZE ,"usbphy%d" , i); +sysbus_init_child_obj (obj, name, &s->usbphy[i],sizeof (s->usbphy[i]), +TYPE_IMX_USBPHY ); + } +for (i =0 ; i <FSL_IMX6_NUM_USBS ; i++) { +snprintf (name,NAME_SIZE ,"usb%d" , i); +sysbus_init_child_obj (obj, name, &s->usb[i],sizeof (s->usb[i]), +TYPE_CHIPIDEA ); + } +for (i =0 ; i <FSL_IMX6_NUM_ECSPIS ; i++) {snprintf (name,NAME_SIZE ,"spi%d" , i +1 ); sysbus_init_child_obj(obj, name, &s->spi[i], sizeof(s->spi[i]),
@@ -349,6 +361,30 @@ static void fsl_imx6_realize(DeviceState *dev, Error **errp) esdhc_table[i].irq)); } +/* USB */ +for (i =0 ; i <FSL_IMX6_NUM_USB_PHYS ; i++) { +object_property_set_bool (OBJECT (&s->usbphy[i]), true,"realized" , + &error_abort); +sysbus_mmio_map (SYS_BUS_DEVICE (&s->usbphy[i]),0 , +FSL_IMX6_USBPHY1_ADDR + i *0x1000 ); + } +for (i =0 ; i <FSL_IMX6_NUM_USBS ; i++) { +static const int FSL_IMX6_USBn_IRQ[] = { +FSL_IMX6_USB_OTG_IRQ , +FSL_IMX6_USB_HOST1_IRQ , +FSL_IMX6_USB_HOST2_IRQ , +FSL_IMX6_USB_HOST3_IRQ , + }; + +object_property_set_bool (OBJECT (&s->usb[i]), true,"realized" , + &error_abort); +sysbus_mmio_map (SYS_BUS_DEVICE (&s->usb[i]),0 , +FSL_IMX6_USBOH3_USB_ADDR + i *0x200 ); +sysbus_connect_irq (SYS_BUS_DEVICE (&s->usb[i]),0 , +qdev_get_gpio_in (DEVICE (&s->a9mpcore), + FSL_IMX6_USBn_IRQ[i])); + } +/* Initialize all ECSPI */ for (i =0 ; i <FSL_IMX6_NUM_ECSPIS ; i++) { static const struct {
include/hw/arm/fsl-imx6.h +6 -0
@@ -30,6 +30,8 @@#include "hw/sd/sdhci.h" #include "hw/ssi/imx_spi.h" #include "hw/net/imx_fec.h" +#include "hw/usb/chipidea.h" +#include "hw/usb/imx-usb-phy.h" #include "exec/memory.h" #include "cpu.h"
@@ -44,6 +46,8 @@#define FSL_IMX6_NUM_ESDHCS 4#define FSL_IMX6_NUM_ECSPIS 5#define FSL_IMX6_NUM_WDTS 2 +#define FSL_IMX6_NUM_USB_PHYS 2 +#define FSL_IMX6_NUM_USBS 4 typedef struct FslIMX6State { /*< private >*/
@@ -62,6 +66,8 @@ typedef struct FslIMX6State {SDHCIState esdhc[FSL_IMX6_NUM_ESDHCS ];IMXSPIState spi[FSL_IMX6_NUM_ECSPIS ];IMX2WdtState wdt[FSL_IMX6_NUM_WDTS ]; +IMXUSBPHYState usbphy[FSL_IMX6_NUM_USB_PHYS ]; +ChipideaState usb[FSL_IMX6_NUM_USBS ];IMXFECState eth;MemoryRegion rom;MemoryRegion caam;
Signature · 1 line
2.17.1
Peter Maydell
AppliedQuoted 25 lines · Guenter Roeck
This patch series wires up the USB controllers on fsl-imx6 and fsl-imx6ul emulations.
The first patch provides a basic implementation of the USB PHY controller used in i.MX28 and later chips. Only reset bit handling in the control register is actually implemented. Basic USB PHY support is needed to make the USB ports operational in Linux.
The second patch fixes USB and USB PHY interrupt numbers for i.MX6UL.
The third patch instantiates unimplemented pwm and can devices. This patch is necessary to avoid crashes in Linux when it tries to access those devices. The crashes are observed when trying to boot Linux v4.21 or later.
The final two patches instantiate the USB controllers for i.mMX6 and i.MX6UL.
v3:
- Minor cleanup in patch 1/5 (see details in patch)
- Added patch to fix USB and USB PHY interrupt numbers for fsl-imx6ul.
- Added patch to instantiate unimplemented pwm and CAN devices.
- Instantiate USB and USB PHY separately. They are logically different, and the number of instances is not always the same.
Quoted 3 lines
v2:
- Implement and instantiate basic USB PHY implementation instead of emulating a single USB PHY register
Applied to target-arm.next, thanks.
-- PMM
Re: [PATCH v3 1/5] hw/usb: Add basic i.MX USB Phy support
In reply to Guenter RoeckQuoted 31 lines · Guenter Roeck
Add basic USB PHY support as implemented in i.MX23, i.MX28, i.MX6, and i.MX7 SoCs.
The only support really needed - at least to boot Linux - is support for soft reset, which needs to reset various registers to their initial value. Otherwise, just record register values.
Reviewed-by Peter Maydell <[email protected]>Signed-off-by Guenter Roeck <[email protected]>---
v3: Added Reviewed-by:;
dropped duplicate "the" in comments;
added new files toMAINTAINERS
v2: New patch, replacing dummy STMP register support with basic USB PHY
emulation.MAINTAINERS | 2 + hw/arm/Kconfig | 1 + hw/usb/Kconfig | 5 + hw/usb/Makefile.objs | 2 + hw/usb/imx-usb-phy.c | 225 +++++++++++++++++++++++++++++++++++ include/hw/usb/imx-usb-phy.h | 53 +++++++++ 6 files changed, 288 insertions(+) create mode 100644 hw/usb/imx-usb-phy.c create mode 100644 include/hw/usb/imx-usb-phy.hhw/arm/Kconfig +0
-0
hw/arm/Kconfig config FSL_IMX6 · line 365
Quoted 5 lines
hw/arm/Kconfig +1
-0@@ -361,6 +361,7 @@ config FSL_IMX6 select IMX select IMX_FEC select IMX_I2C + select IMX_USBPHY select SDHCI
I know it is merged, but FYI this change belongs to patch 5 of this series "hw/arm/fsl-imx6: Wire up USB controllers" where you add the dependency to the machine.
Re: [PATCH v3 1/5] hw/usb: Add basic i.MX USB Phy support
In reply to Guenter RoeckQuoted 10 lines · Guenter Roeck
Add basic USB PHY support as implemented in i.MX23, i.MX28, i.MX6, and i.MX7 SoCs.
The only support really needed - at least to boot Linux - is support for soft reset, which needs to reset various registers to their initial value. Otherwise, just record register values.
Reviewed-by Peter Maydell <[email protected]>Signed-off-by Guenter Roeck <[email protected]>
Hi Guenter; we've had a fuzzer report that this device model accesses off the end of the usbphy[] array:
https://gitlab.com/qemu-project/qemu/-/issues/1408
hw/usb/imx-usb-phy.c line 94
Quoted 9 lines
hw/usb/imx-usb-phy.c new file +7
-0+static uint64_t imx_usbphy_read (void *opaque,hwaddr offset,unsigned size) +{ +IMXUSBPHYState *s = (IMXUSBPHYState *)opaque; +uint32_t index = offset >>2 ; +uint32_t value; +default : + value = s->usbphy[index];
No bounds check in the default case (or ditto in the write function)...
hw/usb/imx-usb-phy.c line 199
Quoted 11 lines
hw/usb/imx-usb-phy.c new file +10
-0+break ; + } +return (uint64_t )value; +} +static void imx_usbphy_realize (DeviceState *dev,Error **errp) +{ +IMXUSBPHYState *s =IMX_USBPHY (dev); + +memory_region_init_io (&s->iomem,OBJECT (s), &imx_usbphy_ops, s, +"imx-usbphy" ,0x1000 );
...and the memory region is created at size 0x1000 so the read/write fns can be called with offsets up to that size...
Quoted 37 lines
include/hw/usb/imx-usb-phy.h new file +36
-0+sysbus_init_mmio (SYS_BUS_DEVICE (s), &s->iomem); +} +enum IMXUsbPhyRegisters { +USBPHY_PWD , +USBPHY_PWD_SET , +USBPHY_PWD_CLR , +USBPHY_PWD_TOG , +USBPHY_TX , +USBPHY_TX_SET , +USBPHY_TX_CLR , +USBPHY_TX_TOG , +USBPHY_RX , +USBPHY_RX_SET , +USBPHY_RX_CLR , +USBPHY_RX_TOG , +USBPHY_CTRL , +USBPHY_CTRL_SET , +USBPHY_CTRL_CLR , +USBPHY_CTRL_TOG , +USBPHY_STATUS , +USBPHY_DEBUG =0x14 , +USBPHY_DEBUG_SET , +USBPHY_DEBUG_CLR , +USBPHY_DEBUG_TOG , +USBPHY_DEBUG0_STATUS , +USBPHY_DEBUG1 =0x1c , +USBPHY_DEBUG1_SET , +USBPHY_DEBUG1_CLR , +USBPHY_DEBUG1_TOG , +USBPHY_VERSION , +USBPHY_MAX +}; + +#define USBPHY_CTRL_SFTRST BIT(31) + +#define TYPE_IMX_USBPHY "imx.usbphy"
include/hw/usb/imx-usb-phy.h line 50
Quoted 10 lines
include/hw/usb/imx-usb-phy.h new file +10
-0+#define IMX_USBPHY (obj) OBJECT_CHECK(IMXUSBPHYState, (obj), TYPE_IMX_USBPHY) + +typedef struct IMXUSBPHYState { + /* <private> */ + SysBusDevice parent_obj; + + /* <public> */ + MemoryRegion iomem; + + uint32_t usbphy[USBPHY_MAX];
...but the array is only created with USBPHY_MAX entries.
Do you know what the device is supposed to do with these off-the-end acceses? We could either reduce the memory region size or bounds check and RAZ/WI the out-of-range accesses.
thanks
-- PMM
Re: [PATCH v3 1/5] hw/usb: Add basic i.MX USB Phy support
Hi Peter,
Quoted 9 lines · Peter Maydell
Quoted 8 lines · Guenter Roeck
Add basic USB PHY support as implemented in i.MX23, i.MX28, i.MX6, and i.MX7 SoCs.
The only support really needed - at least to boot Linux - is support for soft reset, which needs to reset various registers to their initial value. Otherwise, just record register values.
Quoted 7 lines
Quoted 2 lines
Reviewed-by Peter Maydell <[email protected]>Signed-off-by Guenter Roeck <[email protected]>Hi Guenter; we've had a fuzzer report that this device model accesses off the end of the usbphy[] array:
https://gitlab.com/qemu-project/qemu/-/issues/1408
Good catch. And an obvious bug, sorry.
Quoted 77 lines
hw/usb/imx-usb-phy.c line 94
Quoted 9 lines
hw/usb/imx-usb-phy.c new file +7
-0+static uint64_t imx_usbphy_read (void *opaque,hwaddr offset,unsigned size) +{ +IMXUSBPHYState *s = (IMXUSBPHYState *)opaque; +uint32_t index = offset >>2 ; +uint32_t value; +default : + value = s->usbphy[index];No bounds check in the default case (or ditto in the write function)...
hw/usb/imx-usb-phy.c line 199
Quoted 11 lines
hw/usb/imx-usb-phy.c new file +10
-0+break ; + } +return (uint64_t )value; +} +static void imx_usbphy_realize (DeviceState *dev,Error **errp) +{ +IMXUSBPHYState *s =IMX_USBPHY (dev); + +memory_region_init_io (&s->iomem,OBJECT (s), &imx_usbphy_ops, s, +"imx-usbphy" ,0x1000 );...and the memory region is created at size
0x1000so the read/write fns can be called with offsets up to that size...Quoted 37 lines
include/hw/usb/imx-usb-phy.h new file +36
-0+sysbus_init_mmio (SYS_BUS_DEVICE (s), &s->iomem); +} +enum IMXUsbPhyRegisters { +USBPHY_PWD , +USBPHY_PWD_SET , +USBPHY_PWD_CLR , +USBPHY_PWD_TOG , +USBPHY_TX , +USBPHY_TX_SET , +USBPHY_TX_CLR , +USBPHY_TX_TOG , +USBPHY_RX , +USBPHY_RX_SET , +USBPHY_RX_CLR , +USBPHY_RX_TOG , +USBPHY_CTRL , +USBPHY_CTRL_SET , +USBPHY_CTRL_CLR , +USBPHY_CTRL_TOG , +USBPHY_STATUS , +USBPHY_DEBUG =0x14 , +USBPHY_DEBUG_SET , +USBPHY_DEBUG_CLR , +USBPHY_DEBUG_TOG , +USBPHY_DEBUG0_STATUS , +USBPHY_DEBUG1 =0x1c , +USBPHY_DEBUG1_SET , +USBPHY_DEBUG1_CLR , +USBPHY_DEBUG1_TOG , +USBPHY_VERSION , +USBPHY_MAX +}; + +#define USBPHY_CTRL_SFTRST BIT(31) + +#define TYPE_IMX_USBPHY "imx.usbphy"include/hw/usb/imx-usb-phy.h line 50
Quoted 10 lines
include/hw/usb/imx-usb-phy.h new file +10
-0+#define IMX_USBPHY (obj) OBJECT_CHECK(IMXUSBPHYState, (obj), TYPE_IMX_USBPHY) + +typedef struct IMXUSBPHYState { + /* <private> */ + SysBusDevice parent_obj; + + /* <public> */ + MemoryRegion iomem; + + uint32_t usbphy[USBPHY_MAX];...but the array is only created with
USBPHY_MAXentries.
Quoted 4 lines
Do you know what the device is supposed to do with these off-the-end acceses? We could either reduce the memory region size or bounds check and RAZ/WI the out-of-range accesses.
I have no idea what the real hardware would do. The datasheets (at least the ones I checked) don't say, only that the region size is 4k. I would suggest a bounds check, ignore out-of-bounds writes (maybe with a log message), and return 0 for reads (which I think is what you suggest with RAZ/WI).
Want me to send a patch ?
Thanks,
Guenter
Re: [PATCH v3 1/5] hw/usb: Add basic i.MX USB Phy support
Quoted 30 lines · Guenter Roeck
Hi Peter,
Quoted 9 lines · Peter Maydell
Quoted 8 lines · Guenter Roeck
Add basic USB PHY support as implemented in i.MX23, i.MX28, i.MX6, and i.MX7 SoCs.
The only support really needed - at least to boot Linux - is support for soft reset, which needs to reset various registers to their initial value. Otherwise, just record register values.
Quoted 7 lines
Quoted 2 lines
Reviewed-by Peter Maydell <[email protected]>Signed-off-by Guenter Roeck <[email protected]>Hi Guenter; we've had a fuzzer report that this device model accesses off the end of the usbphy[] array:
https://gitlab.com/qemu-project/qemu/-/issues/1408Good catch. And an obvious bug, sorry.
Quoted 4 lines
Do you know what the device is supposed to do with these off-the-end acceses? We could either reduce the memory region size or bounds check and RAZ/WI the out-of-range accesses.
Quoted 7 lines
I have no idea what the real hardware would do. The datasheets (at least the ones I checked) don't say, only that the region size is 4k. I would suggest a bounds check, ignore out-of-bounds writes (maybe with a log message), and return 0 for reads (which I think is what you suggest with RAZ/WI).
Want me to send a patch ?
If you have the time, that would be great. I expect you're better set up to test it than I am...
thanks
-- PMM
Re: [PATCH v3 1/5] hw/usb: Add basic i.MX USB Phy support
Quoted 37 lines · Peter Maydell
Quoted 36 lines · Guenter Roeck
Hi Peter,
Quoted 9 lines · Peter Maydell
Quoted 8 lines · Guenter Roeck
Add basic USB PHY support as implemented in i.MX23, i.MX28, i.MX6, and i.MX7 SoCs.
The only support really needed - at least to boot Linux - is support for soft reset, which needs to reset various registers to their initial value. Otherwise, just record register values.
Quoted 7 lines
Quoted 2 lines
Reviewed-by Peter Maydell <[email protected]>Signed-off-by Guenter Roeck <[email protected]>Hi Guenter; we've had a fuzzer report that this device model accesses off the end of the usbphy[] array:
https://gitlab.com/qemu-project/qemu/-/issues/1408Good catch. And an obvious bug, sorry.
Quoted 4 lines
Do you know what the device is supposed to do with these off-the-end acceses? We could either reduce the memory region size or bounds check and RAZ/WI the out-of-range accesses.
I have no idea what the real hardware would do. The datasheets (at least the ones I checked) don't say, only that the region size is 4k. I would suggest a bounds check, ignore out-of-bounds writes (maybe with a log message), and return 0 for reads (which I think is what you suggest with RAZ/WI).
Quoted 5 lines
Quoted 1 line
Want me to send a patch ?
If you have the time, that would be great. I expect you're better set up to test it than I am...
I prepared a patch. Currently testing.
Guenter