Received: from eggs.gnu.org ([2001:4830:134:3::10]:38542) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZYZTy-0004Gn-Nd for qemu-devel@nongnu.org; Sun, 06 Sep 2015 08:55:44 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZYZTt-0004AM-Lb for qemu-devel@nongnu.org; Sun, 06 Sep 2015 08:55:42 -0400 Received: from zose-mta02.web4all.fr ([185.49.20.43]:56048) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZYZTt-00049B-1B for qemu-devel@nongnu.org; Sun, 06 Sep 2015 08:55:37 -0400 References: From: Jean-Christophe DUBOIS Message-ID: <55EC37C3.7020204@tribudubois.net> Date: Sun, 6 Sep 2015 14:55:31 +0200 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v1 1/3] i.MX: Add GPIO device List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Crosthwaite Cc: "qemu-devel@nongnu.org Developers" Le 05/09/2015 11:03, Peter Crosthwaite a =C3=A9crit : > On Sat, Sep 5, 2015 at 1:17 AM, Jean-Christophe Dubois > wrote: >> Signed-off-by: Jean-Christophe Dubois >> --- >> hw/gpio/Makefile.objs | 1 + >> hw/gpio/imx_gpio.c | 358 +++++++++++++++++++++++++++++++++++= ++++++++++ >> include/hw/gpio/imx_gpio.h | 60 ++++++++ >> 3 files changed, 419 insertions(+) >> create mode 100644 hw/gpio/imx_gpio.c >> create mode 100644 include/hw/gpio/imx_gpio.h >> >> diff --git a/hw/gpio/Makefile.objs b/hw/gpio/Makefile.objs >> index 1abcf17..52233f7 100644 >> --- a/hw/gpio/Makefile.objs >> +++ b/hw/gpio/Makefile.objs >> @@ -5,3 +5,4 @@ common-obj-$(CONFIG_ZAURUS) +=3D zaurus.o >> common-obj-$(CONFIG_E500) +=3D mpc8xxx.o >> >> obj-$(CONFIG_OMAP) +=3D omap_gpio.o >> +obj-$(CONFIG_IMX) +=3D imx_gpio.o >> diff --git a/hw/gpio/imx_gpio.c b/hw/gpio/imx_gpio.c >> new file mode 100644 >> index 0000000..8ec1d4c >> --- /dev/null >> +++ b/hw/gpio/imx_gpio.c >> @@ -0,0 +1,358 @@ >> +/* >> + * i.MX processors GPIO emulation. >> + * >> + * Copyright (C) 2015 Jean-Christophe Dubois >> + * >> + * This program is free software; you can redistribute it and/or >> + * modify it under the terms of the GNU General Public License as >> + * published by the Free Software Foundation; either version 2 or >> + * (at your option) version 3 of the License. >> + * >> + * This program is distributed in the hope that it will be useful, >> + * but WITHOUT ANY WARRANTY; without even the implied warranty of >> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the >> + * GNU General Public License for more details. >> + * >> + * You should have received a copy of the GNU General Public License = along >> + * with this program; if not, see . >> + */ >> + >> +#include "hw/gpio/imx_gpio.h" >> + >> +#ifndef IMX_GPIO_DEBUG >> +#define IMX_GPIO_DEBUG 0 >> +#endif >> + >> +#if IMX_GPIO_DEBUG >> +# define DPRINTF(fmt, args...) \ >> + do { fprintf(stder, "%s: " fmt , __func__, ##args); } while= (0) >> + > Use a regular if for debug conditional. This is so the debug code is > always compiled. Not sure what you mean here ... Most files in qemu have DEBUG functions=20 compiled out in the normal case ... Is there a change in "politic" on this topic? Could you point to an implementation doing things right? > >> +static char const *imx_gpio_reg_name(uint32_t reg) >> +{ >> + switch (reg) { >> + case DR_ADDR: >> + return "DR"; >> + case GDIR_ADDR: >> + return "GDIR"; >> + case PSR_ADDR: >> + return "PSR"; >> + case ICR1_ADDR: >> + return "ICR1"; >> + case ICR2_ADDR: >> + return "ICR2"; >> + case IMR_ADDR: >> + return "IMR"; >> + case ISR_ADDR: >> + return "ISR"; >> + case EDGE_SEL_ADDR: >> + /* This register is not present in i.MX31 */ >> + return "EDGE_SEL"; >> + default: >> + return "[?]"; >> + } > But I'm guessing this will be an issue for the non debug case. Perhaps > return "" from here in non-debug or just leave this always compiled > (optimisiser should be smart enough to trim it). > >> +} >> +#else >> +# define DPRINTF(fmt, args...) do {} while (0) >> +#endif >> + >> +static void imx_gpio_update_int(IMXGPIOState *s) >> +{ >> + if (s->isr) { >> + qemu_irq_raise(s->irq); >> + } else { >> + qemu_irq_lower(s->irq); >> + } > qemu_set_irq. OK > >> +} >> + >> +static void imx_gpio_set_int_line(IMXGPIOState *s, int line, int leve= l) >> +{ >> + /* is this signal configured as an interrupt source */ >> + if (extract32(s->imr, line, 1)) { > Short return on this condition (negated) rather than indent entire logi= c. OK > >> + /* When set EDGE_SEL override the ICR config */ >> + if (extract32(s->edge_sel, line, 1)) { >> + /* we detect interrupt on rising and falling edge */ >> + if (extract32(s->psr, line, 1) !=3D level) { > This is dangerous, as level is boolean and this will promote to > integer for comparison. I think last time people talked about this on > list, we decided that all users or GPIO setters should use 0 vs 1 but > it may not be the in-tree case. !!level will resolve. OK, I'll work out something with an enum to be explicit. > >> + /* level changed */ >> + s->isr =3D deposit32(s->isr, line, 1, 1); >> + } >> + } else if (extract64(s->icr, 2*line + 1, 1)) { >> + /* interrupt is edge sensitive */ >> + if (extract32(s->psr, line, 1) !=3D level) { >> + /* level changed */ >> + int falling =3D extract64(s->icr, 2*line, 1); >> + >> + if ((falling && !level) || (!falling && level)) { > falling =3D=3D !level OK > >> + s->isr =3D deposit32(s->isr, line, 1, 1); >> + } >> + } >> + } else { >> + /* interrupt is level sensitive */ >> + int high =3D extract64(s->icr, 2*line, 1); >> + >> + if ((high && level) || (!high && !level)) { > =3D=3D OK > >> + s->isr =3D deposit32(s->isr, line, 1, 1); >> + } >> + } >> + } >> +} >> + >> +static void imx_gpio_set(void *opaque, int line, int level) >> +{ >> + IMXGPIOState *s =3D IMX_GPIO(opaque); >> + >> + /* if the line is configured as output nothing to do */ >> + if (extract32(s->gdir, line, 1)) { >> + /* actually we should not be called */ >> + qemu_log_mask(LOG_GUEST_ERROR, "%s[%s]: requesting to set lin= e %d" >> + " to %d but this is an output line\n", >> + TYPE_IMX_GPIO, __func__, line, level); > Short return. OK > >> + } else { >> + imx_gpio_set_int_line(s, line, level); >> + >> + /* this is an input signal, so set PSR */ >> + s->psr =3D deposit32(s->psr, line, 1, level); >> + >> + imx_gpio_update_int(s); >> + } >> +} >> + >> +static void imx_gpio_set_all_int_lines(IMXGPIOState *s) >> +{ >> + int i; >> + >> + for (i =3D 0; i < 32; i++) { > 32 should be macroified. OK > >> + imx_gpio_set_int_line(s, i, extract32(s->psr, i, 1)); >> + } >> + >> + imx_gpio_update_int(s); >> +} >> + >> +static uint64_t imx_gpio_read(void *opaque, hwaddr offset, unsigned s= ize) >> +{ >> + IMXGPIOState *s =3D IMX_GPIO(opaque); >> + uint32_t reg_value =3D 0; >> + int i; >> + >> + switch (offset) { >> + case DR_ADDR: /* DATA REGISTER */ >> + for (i =3D 0; i < 32; i++) { >> + uint32_t ref_value; >> + >> + /* >> + * depending on the "line" configuration, the bit values >> + * are comming either from DR ou PSR > "coming" "or" OK > >> + */ >> + if (extract32(s->gdir, i, 1)) { >> + ref_value =3D s->dr; >> + } else { >> + ref_value =3D s->psr; >> + } >> + >> + reg_value =3D deposit32(reg_value, i, 1, >> + extract32(ref_value, i, 1)); > This can be one-shot with some bitwise logicals to avoid the loop. > > reg_value =3D s->dr & s->gdir | s->psr & ~s->gdir > > I think. OK > >> + } >> + break; >> + >> + case GDIR_ADDR: /* DIRECTION REGISTER */ > These trailing comments are duped between read and write handlers. I > would move to header as a more global single reference. I was thinking it was more helpful here to help understand the code=20 without having to refer to a header file. The point is that the abreviated spec name is not always very helpful. But, OK, I'll move them to the header file ... > >> + reg_value =3D s->gdir; >> + break; >> + >> + case PSR_ADDR: /* PAD STATUS REGISTER */ >> + reg_value =3D s->psr; >> + break; >> + >> + case ICR1_ADDR: /* INTERRUPT CONFIGURATION REGISTER 1 */ >> + reg_value =3D (uint32_t) (s->icr & 0xffffffff); >> + break; >> + >> + case ICR2_ADDR: /* INTERRUPT CONFIGURATION REGISTER 2 */ >> + reg_value =3D (uint32_t) (s->icr >> 32); >> + break; >> + >> + case IMR_ADDR: /* INTERRUPT MASK REGISTER */ >> + reg_value =3D s->imr; >> + break; >> + >> + case ISR_ADDR: /* INTERRUPT STATUS REGISTER */ >> + reg_value =3D s->isr; >> + break; >> + >> + case EDGE_SEL_ADDR: /* EDGE SEL REGISTER */ >> + /* This register is not present in i.MX31 */ >> + reg_value =3D s->edge_sel; >> + break; >> + >> + default: >> + qemu_log_mask(LOG_GUEST_ERROR, "%s[%s]: Bad register at offse= t %d\n", >> + TYPE_IMX_GPIO, __func__, (int)offset); >> + break; >> + } >> + >> + DPRINTF("(%s) =3D 0x%08x\n", imx_gpio_reg_name(offset), reg_value= ); > PRIx32. OK > >> + >> + return reg_value; >> +} >> + >> +static inline void imx_gpio_set_output(IMXGPIOState *s) > set_all_outputs to be consistent with set_all_int_lines. Group the two > functions and any similars together (either here or above with > int_lines). OK > >> +{ >> + int i; >> + >> + for (i =3D 0; i < 32; i++) { >> + /* >> + * if the line is set as output, then forward the line >> + * level to its user. >> + */ >> + if (extract32(s->gdir, i, 1) && s->handler[i]) { >> + qemu_set_irq(s->handler[i], extract32(s->dr, i, 1)); >> + } >> + } >> +} >> + >> +static void imx_gpio_write(void *opaque, hwaddr offset, uint64_t valu= e, >> + unsigned size) >> +{ >> + IMXGPIOState *s =3D IMX_GPIO(opaque); >> + >> + DPRINTF("(%s, value =3D 0x%08x)\n", imx_gpio_reg_name(offset), >> + (uint32_t)value); > PRIx32. OK > >> + >> + switch (offset) { >> + case DR_ADDR: /* DATA REGISTER */ >> + s->dr =3D (uint32_t)value; > cast un-needed. OK > >> + >> + imx_gpio_set_output(s); >> + >> + break; >> + >> + case GDIR_ADDR: /* DIRECTION REGISTER */ >> + s->gdir =3D (uint32_t)value; > cast un-needed (theres a few more below). OK > >> + >> + imx_gpio_set_output(s); >> + imx_gpio_set_all_int_lines(s); >> + >> + break; >> + >> + case ICR1_ADDR: /* INTERRUPT CONFIGURATION REGISTER 1 */ >> + s->icr =3D (s->icr | 0xffffffff) & (uint32_t)value; > deposit64. OK > >> + >> + imx_gpio_set_all_int_lines(s); >> + >> + break; >> + >> + case ICR2_ADDR: /* INTERRUPT CONFIGURATION REGISTER 2 */ >> + s->icr =3D (s->icr | 0xffffffff00000000) & (value << 32); > deposit64. OK > >> + >> + imx_gpio_set_all_int_lines(s); >> + >> + break; >> + >> + case IMR_ADDR: /* INTERRUPT MASK REGISTER */ >> + s->imr =3D (uint32_t)value; >> + >> + imx_gpio_set_all_int_lines(s); >> + >> + break; >> + >> + case ISR_ADDR: /* INTERRUPT STATUS REGISTER */ >> + s->isr |=3D ~(uint32_t)value; >> + >> + imx_gpio_set_all_int_lines(s); >> + >> + break; >> + >> + case EDGE_SEL_ADDR: /* EDGE SEL REGISTER */ >> + /* This register is not present in i.MX31 */ >> + s->edge_sel =3D (uint32_t)value; >> + >> + imx_gpio_set_all_int_lines(s); >> + >> + break; >> + > I don't think you need the extra white. For short side effects it is > usual to just format as: > > case ADDR: > register =3D value; > side_effects(); > break; OK > >> + default: >> + qemu_log_mask(LOG_GUEST_ERROR, "%s[%s]: Bad register at offse= t %d\n", >> + TYPE_IMX_GPIO, __func__, (int)offset); >> + break; >> + } >> + >> + return; >> +} >> + >> +static void imx_gpio_reset(DeviceState *dev) >> +{ >> + IMXGPIOState *s =3D IMX_GPIO(dev); >> + >> + s->dr =3D 0; >> + s->gdir =3D 0; >> + s->psr =3D 0; >> + s->icr =3D 0; >> + s->imr =3D 0; >> + s->isr =3D 0; >> + s->edge_sel =3D 0; >> + >> + imx_gpio_set_output(s); >> + imx_gpio_update_int(s); >> +} >> + >> +static const MemoryRegionOps imx_gpio_ops =3D { >> + .read =3D imx_gpio_read, >> + .write =3D imx_gpio_write, >> + .valid.min_access_size =3D 4, >> + .valid.max_access_size =3D 4, >> + .endianness =3D DEVICE_NATIVE_ENDIAN, >> +}; >> + >> +static const VMStateDescription vmstate_imx_gpio =3D { >> + .name =3D TYPE_IMX_GPIO, >> + .version_id =3D 1, >> + .minimum_version_id =3D 1, >> + .minimum_version_id_old =3D 1, >> + .fields =3D (VMStateField[]) { >> + VMSTATE_UINT32(dr, IMXGPIOState), >> + VMSTATE_UINT32(gdir, IMXGPIOState), >> + VMSTATE_UINT32(psr, IMXGPIOState), >> + VMSTATE_UINT64(icr, IMXGPIOState), >> + VMSTATE_UINT32(imr, IMXGPIOState), >> + VMSTATE_UINT32(isr, IMXGPIOState), >> + /* This register is not present in i.MX31 */ >> + VMSTATE_UINT32(edge_sel, IMXGPIOState), > You can make a boolean property of the device and turn its presence on > and off to get the functionality. OK > >> + VMSTATE_END_OF_LIST() >> + } >> +}; >> + >> +static void imx_gpio_realize(DeviceState *dev, Error **errp) >> +{ >> + IMXGPIOState *s =3D IMX_GPIO(dev); >> + >> + memory_region_init_io(&s->iomem, OBJECT(s), &imx_gpio_ops, s, >> + TYPE_IMX_GPIO, IMX_GPIO_MEM_SIZE); >> + >> + qdev_init_gpio_in(DEVICE(s), imx_gpio_set, 32); >> + qdev_init_gpio_out(DEVICE(s), s->handler, 32); >> + >> + sysbus_init_irq(SYS_BUS_DEVICE(dev), &s->irq); >> + sysbus_init_mmio(SYS_BUS_DEVICE(dev), &s->iomem); >> +} >> + >> +static void imx_gpio_class_init(ObjectClass *klass, void *data) >> +{ >> + DeviceClass *dc =3D DEVICE_CLASS(klass); >> + >> + dc->realize =3D imx_gpio_realize; >> + dc->reset =3D imx_gpio_reset; >> + dc->vmsd =3D &vmstate_imx_gpio; >> + dc->desc =3D "i.MX GPIO controller"; > dc->props =3D ... > > for the "has-edge-sel" property. OK > >> +} >> + >> +static const TypeInfo imx_gpio_info =3D { >> + .name =3D TYPE_IMX_GPIO, >> + .parent =3D TYPE_SYS_BUS_DEVICE, >> + .instance_size =3D sizeof(IMXGPIOState), >> + .class_init =3D imx_gpio_class_init, >> +}; >> + >> +static void imx_gpio_register_types(void) >> +{ >> + type_register_static(&imx_gpio_info); >> +} >> + >> +type_init(imx_gpio_register_types) >> diff --git a/include/hw/gpio/imx_gpio.h b/include/hw/gpio/imx_gpio.h >> new file mode 100644 >> index 0000000..6d3f149 >> --- /dev/null >> +++ b/include/hw/gpio/imx_gpio.h >> @@ -0,0 +1,60 @@ >> +/* >> + * i.MX processors GPIO registers definition. >> + * >> + * Copyright (C) 2015 Jean-Christophe Dubois >> + * >> + * This program is free software; you can redistribute it and/or >> + * modify it under the terms of the GNU General Public License as >> + * published by the Free Software Foundation; either version 2 or >> + * (at your option) version 3 of the License. >> + * >> + * This program is distributed in the hope that it will be useful, >> + * but WITHOUT ANY WARRANTY; without even the implied warranty of >> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the >> + * GNU General Public License for more details. >> + * >> + * You should have received a copy of the GNU General Public License = along >> + * with this program; if not, see . >> + */ >> + >> +#ifndef __IMX_GPIO_H_ >> +#define __IMX_GPIO_H_ >> + >> +#include >> + >> +#define TYPE_IMX_GPIO "imx.gpio" >> +#define IMX_GPIO(obj) OBJECT_CHECK(IMXGPIOState, (obj), TYPE_IMX_GPIO= ) >> + >> +#define IMX_GPIO_MEM_SIZE 0x20 >> + >> +/* i.MX GPIO memory map */ >> +#define DR_ADDR 0x00 >> +#define GDIR_ADDR 0x04 >> +#define PSR_ADDR 0x08 >> +#define ICR1_ADDR 0x0c >> +#define ICR2_ADDR 0x10 >> +#define IMR_ADDR 0x14 >> +#define ISR_ADDR 0x18 >> +#define EDGE_SEL_ADDR 0x1c >> + >> +typedef struct IMXGPIOState { >> + /*< private >*/ >> + SysBusDevice parent_obj; >> + >> + /*< public >*/ >> + MemoryRegion iomem; >> + >> + uint32_t dr; >> + uint32_t gdir; >> + uint32_t psr; >> + uint64_t icr; >> + uint32_t imr; >> + uint32_t isr; >> + /* This register is not present in i.MX31 */ >> + uint32_t edge_sel; >> + >> + qemu_irq irq; >> + qemu_irq handler[32]; > Should this just be "outputs"? Im guessing this is a bidirectional pin > which QEMU has trouble modelleing and this is the output mode variant > of that (hence it probably has no name in the TRM to use). OK > > Regards, > Peter > >> +} IMXGPIOState; >> + >> +#endif /* __IMX_GPIO_H_ */ >> -- >> 2.1.4 >> >>