| Message ID | 20220219064210.3145381-4-shorne@gmail.com |
|---|---|
| State | New |
| Headers | show |
| Series | OpenRISC Device Tree Generation | expand |
On Sat, 19 Feb 2022 at 06:42, Stafford Horne <shorne@gmail.com> wrote: > > Currently the OpenRISC SMP configuration only supports 2 cores due to > the UART IRQ routing being limited to 2 cores. As was done in commit > 1eeffbeb11 ("hw/openrisc/openrisc_sim: Use IRQ splitter when connecting > IRQ to multiple CPUs") we can use a splitter to wire more than 2 CPUs. > > This patch moves serial initialization out to it's own function and > uses a splitter to connect multiple CPU irq lines to the UART. > > Signed-off-by: Stafford Horne <shorne@gmail.com> > --- Reviewed-by: Peter Maydell <peter.maydell@linaro.org> thanks -- PMM
On 19/2/22 07:42, Stafford Horne wrote: > Currently the OpenRISC SMP configuration only supports 2 cores due to > the UART IRQ routing being limited to 2 cores. As was done in commit > 1eeffbeb11 ("hw/openrisc/openrisc_sim: Use IRQ splitter when connecting > IRQ to multiple CPUs") we can use a splitter to wire more than 2 CPUs. > > This patch moves serial initialization out to it's own function and > uses a splitter to connect multiple CPU irq lines to the UART. > > Signed-off-by: Stafford Horne <shorne@gmail.com> > --- > hw/openrisc/openrisc_sim.c | 32 ++++++++++++++++++++++++-------- > 1 file changed, 24 insertions(+), 8 deletions(-) > > diff --git a/hw/openrisc/openrisc_sim.c b/hw/openrisc/openrisc_sim.c > index d12b3e0c5e..5bfbac00f8 100644 > --- a/hw/openrisc/openrisc_sim.c > +++ b/hw/openrisc/openrisc_sim.c > @@ -137,6 +137,28 @@ static void openrisc_sim_ompic_init(hwaddr base, int num_cpus, > sysbus_mmio_map(s, 0, base); > } > > +static void openrisc_sim_serial_init(hwaddr base, int num_cpus, > + OpenRISCCPU *cpus[], int irq_pin) > +{ > + qemu_irq serial_irq; > + int i; > + > + if (num_cpus > 1) { > + DeviceState *splitter = qdev_new(TYPE_SPLIT_IRQ); > + qdev_prop_set_uint32(splitter, "num-lines", num_cpus); > + qdev_realize_and_unref(splitter, NULL, &error_fatal); > + for (i = 0; i < num_cpus; i++) { > + qdev_connect_gpio_out(splitter, i, get_cpu_irq(cpus, i, irq_pin)); > + } > + serial_irq = qdev_get_gpio_in(splitter, 0); > + } else { > + serial_irq = get_cpu_irq(cpus, 0, irq_pin); > + } Up to here the code seems a generic helper: or1k_cpus_connect_device(OpenRISCCPU *cpus[], unsigned num_cpus, unsigned irq_pin); > + serial_mm_init(get_system_memory(), base, 0, serial_irq, 115200, > + serial_hd(0), DEVICE_NATIVE_ENDIAN); This part specific to UART. > +} Anyhow, Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
On Sun, Feb 20, 2022 at 09:06:41PM +0100, Philippe Mathieu-Daudé wrote: > On 19/2/22 07:42, Stafford Horne wrote: > > Currently the OpenRISC SMP configuration only supports 2 cores due to > > the UART IRQ routing being limited to 2 cores. As was done in commit > > 1eeffbeb11 ("hw/openrisc/openrisc_sim: Use IRQ splitter when connecting > > IRQ to multiple CPUs") we can use a splitter to wire more than 2 CPUs. > > > > This patch moves serial initialization out to it's own function and > > uses a splitter to connect multiple CPU irq lines to the UART. > > > > Signed-off-by: Stafford Horne <shorne@gmail.com> > > --- > > hw/openrisc/openrisc_sim.c | 32 ++++++++++++++++++++++++-------- > > 1 file changed, 24 insertions(+), 8 deletions(-) > > > > diff --git a/hw/openrisc/openrisc_sim.c b/hw/openrisc/openrisc_sim.c > > index d12b3e0c5e..5bfbac00f8 100644 > > --- a/hw/openrisc/openrisc_sim.c > > +++ b/hw/openrisc/openrisc_sim.c > > @@ -137,6 +137,28 @@ static void openrisc_sim_ompic_init(hwaddr base, int num_cpus, > > sysbus_mmio_map(s, 0, base); > > } > > +static void openrisc_sim_serial_init(hwaddr base, int num_cpus, > > + OpenRISCCPU *cpus[], int irq_pin) > > +{ > > + qemu_irq serial_irq; > > + int i; > > + > > + if (num_cpus > 1) { > > + DeviceState *splitter = qdev_new(TYPE_SPLIT_IRQ); > > + qdev_prop_set_uint32(splitter, "num-lines", num_cpus); > > + qdev_realize_and_unref(splitter, NULL, &error_fatal); > > + for (i = 0; i < num_cpus; i++) { > > + qdev_connect_gpio_out(splitter, i, get_cpu_irq(cpus, i, irq_pin)); > > + } > > + serial_irq = qdev_get_gpio_in(splitter, 0); > > + } else { > > + serial_irq = get_cpu_irq(cpus, 0, irq_pin); > > + } > > Up to here the code seems a generic helper: > > or1k_cpus_connect_device(OpenRISCCPU *cpus[], > unsigned num_cpus, > unsigned irq_pin); Right, this is similar to that used in openrisc_sim_net_init. I thought about sharing the code but I didn't think it worth adding helper. The main reason for me is that openrisc_sim_net_init doesn't expose the qemu_irq and just does sysbus_connect_irq. While openrisc_sim_serial_init exposes the qemu_irq. I think a generic function would have to look like: qemu_irq openrisc_cpus_irq_pin_init(OpenRISCCPU *cpus[], unsigned num_cpus, unsigned irq_pin); I would like to leave this as is for now as. > > + serial_mm_init(get_system_memory(), base, 0, serial_irq, 115200, > > + serial_hd(0), DEVICE_NATIVE_ENDIAN); > > This part specific to UART. Right. > > +} > > Anyhow, > Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Thank you, Adding this to the patch as is. -Stafford
diff --git a/hw/openrisc/openrisc_sim.c b/hw/openrisc/openrisc_sim.c index d12b3e0c5e..5bfbac00f8 100644 --- a/hw/openrisc/openrisc_sim.c +++ b/hw/openrisc/openrisc_sim.c @@ -137,6 +137,28 @@ static void openrisc_sim_ompic_init(hwaddr base, int num_cpus, sysbus_mmio_map(s, 0, base); } +static void openrisc_sim_serial_init(hwaddr base, int num_cpus, + OpenRISCCPU *cpus[], int irq_pin) +{ + qemu_irq serial_irq; + int i; + + if (num_cpus > 1) { + DeviceState *splitter = qdev_new(TYPE_SPLIT_IRQ); + qdev_prop_set_uint32(splitter, "num-lines", num_cpus); + qdev_realize_and_unref(splitter, NULL, &error_fatal); + for (i = 0; i < num_cpus; i++) { + qdev_connect_gpio_out(splitter, i, get_cpu_irq(cpus, i, irq_pin)); + } + serial_irq = qdev_get_gpio_in(splitter, 0); + } else { + serial_irq = get_cpu_irq(cpus, 0, irq_pin); + } + serial_mm_init(get_system_memory(), base, 0, serial_irq, 115200, + serial_hd(0), DEVICE_NATIVE_ENDIAN); +} + + static void openrisc_load_kernel(ram_addr_t ram_size, const char *kernel_filename) { @@ -177,7 +199,6 @@ static void openrisc_sim_init(MachineState *machine) const char *kernel_filename = machine->kernel_filename; OpenRISCCPU *cpus[2] = {}; MemoryRegion *ram; - qemu_irq serial_irq; int n; unsigned int smp_cpus = machine->smp.cpus; @@ -208,15 +229,10 @@ static void openrisc_sim_init(MachineState *machine) if (smp_cpus > 1) { openrisc_sim_ompic_init(or1ksim_memmap[OR1KSIM_OMPIC].base, smp_cpus, cpus, OR1KSIM_OMPIC_IRQ); - - serial_irq = qemu_irq_split(get_cpu_irq(cpus, 0, OR1KSIM_UART_IRQ), - get_cpu_irq(cpus, 1, OR1KSIM_UART_IRQ)); - } else { - serial_irq = get_cpu_irq(cpus, 0, OR1KSIM_UART_IRQ); } - serial_mm_init(get_system_memory(), or1ksim_memmap[OR1KSIM_UART].base, 0, - serial_irq, 115200, serial_hd(0), DEVICE_NATIVE_ENDIAN); + openrisc_sim_serial_init(or1ksim_memmap[OR1KSIM_UART].base, smp_cpus, cpus, + OR1KSIM_UART_IRQ); openrisc_load_kernel(ram_size, kernel_filename); }
Currently the OpenRISC SMP configuration only supports 2 cores due to the UART IRQ routing being limited to 2 cores. As was done in commit 1eeffbeb11 ("hw/openrisc/openrisc_sim: Use IRQ splitter when connecting IRQ to multiple CPUs") we can use a splitter to wire more than 2 CPUs. This patch moves serial initialization out to it's own function and uses a splitter to connect multiple CPU irq lines to the UART. Signed-off-by: Stafford Horne <shorne@gmail.com> --- hw/openrisc/openrisc_sim.c | 32 ++++++++++++++++++++++++-------- 1 file changed, 24 insertions(+), 8 deletions(-)