| Message ID | 20180412035145.21488-2-andrew@aj.id.au |
|---|---|
| State | Accepted, archived |
| Headers | show |
| Series | Miscellaneous BMC interfaces | expand |
On 12 April 2018 at 13:21, Andrew Jeffery <andrew@aj.id.au> wrote: > The ASPEED BMC SoCs have many knobs and switches that are sometimes > design-specific and often defy any approach to unify them under an > existing subsystem. > > Add a driver to translate a devicetree table into sysfs entries to > expose bits and fields for manipulation from userspace. This encompasses > concepts from scratch registers to boolean conditions to enable or > disable host interface features. > > Signed-off-by: Andrew Jeffery <andrew@aj.id.au> > --- > drivers/soc/Kconfig | 1 + > drivers/soc/Makefile | 1 + > drivers/soc/aspeed/Kconfig | 11 +++ > drivers/soc/aspeed/Makefile | 1 + > drivers/soc/aspeed/aspeed-bmc-misc.c | 187 +++++++++++++++++++++++++++++++++++ > 5 files changed, 201 insertions(+) > create mode 100644 drivers/soc/aspeed/Kconfig > create mode 100644 drivers/soc/aspeed/Makefile > create mode 100644 drivers/soc/aspeed/aspeed-bmc-misc.c > > diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig > index 07fc0ac51c52..776fd5978f70 100644 > --- a/drivers/soc/Kconfig > +++ b/drivers/soc/Kconfig > @@ -1,6 +1,7 @@ > menu "SOC (System On Chip) specific Drivers" > > source "drivers/soc/actions/Kconfig" > +source "drivers/soc/aspeed/Kconfig" > source "drivers/soc/atmel/Kconfig" > source "drivers/soc/bcm/Kconfig" > source "drivers/soc/fsl/Kconfig" > diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile > index 9241125416ba..94a5b97c4466 100644 > --- a/drivers/soc/Makefile > +++ b/drivers/soc/Makefile > @@ -3,6 +3,7 @@ > # > > obj-$(CONFIG_ARCH_ACTIONS) += actions/ > +obj-$(CONFIG_ARCH_ASPEED) += aspeed/ > obj-$(CONFIG_ARCH_AT91) += atmel/ > obj-y += bcm/ > obj-$(CONFIG_ARCH_DOVE) += dove/ > diff --git a/drivers/soc/aspeed/Kconfig b/drivers/soc/aspeed/Kconfig > new file mode 100644 > index 000000000000..704a4efe180f > --- /dev/null > +++ b/drivers/soc/aspeed/Kconfig > @@ -0,0 +1,11 @@ > +menu "ASPEED SoC drivers" > + > +config ASPEED_BMC_MISC > + bool "Miscellaneous ASPEED BMC interfaces" Do you mean bool or tristate? if you mean bool, drop the module-y bits below. > + depends on ARCH_ASPEED || COMPILE_TEST > + default ARCH_ASPEED > + help > + Say yes to expose VGA and LPC scratch registers, and other > + miscellaneous control interfaces specific to the ASPEED BMC SoCs > + > +endmenu > diff --git a/drivers/soc/aspeed/Makefile b/drivers/soc/aspeed/Makefile > new file mode 100644 > index 000000000000..d1a80f9a584f > --- /dev/null > +++ b/drivers/soc/aspeed/Makefile > @@ -0,0 +1 @@ > +obj-$(CONFIG_ASPEED_BMC_MISC) += aspeed-bmc-misc.o > diff --git a/drivers/soc/aspeed/aspeed-bmc-misc.c b/drivers/soc/aspeed/aspeed-bmc-misc.c > new file mode 100644 > index 000000000000..48522a736a05 > --- /dev/null > +++ b/drivers/soc/aspeed/aspeed-bmc-misc.c copyright headahs > @@ -0,0 +1,187 @@ > +#include <linux/kobject.h> > +#include <linux/module.h> > +#include <linux/of.h> > +#include <linux/platform_device.h> > +#include <linux/regmap.h> > +#include <linux/mfd/syscon.h> > + > +#define DEVICE_NAME "aspeed-bmc-misc" This is the name of our chardev? No, it appears to just be the driver name. You can probably just use it directly. An observation: It's the only thing inside this file that is aspeed specific. If not for the name, it could be generic code. lgtm. I suggest writing up some bindings and putting on the asbestos pants. I'll merge this into openbmc with the IBM copyright added to the top. With this adding userspace ABI, anyone who writes code against it will need to make themselves available for rework as it goes through upstream review. Cheers, Joel
On Thu, 19 Apr 2018, at 12:37, Joel Stanley wrote: > On 12 April 2018 at 13:21, Andrew Jeffery <andrew@aj.id.au> wrote: > > The ASPEED BMC SoCs have many knobs and switches that are sometimes > > design-specific and often defy any approach to unify them under an > > existing subsystem. > > > > Add a driver to translate a devicetree table into sysfs entries to > > expose bits and fields for manipulation from userspace. This encompasses > > concepts from scratch registers to boolean conditions to enable or > > disable host interface features. > > > > Signed-off-by: Andrew Jeffery <andrew@aj.id.au> > > --- > > drivers/soc/Kconfig | 1 + > > drivers/soc/Makefile | 1 + > > drivers/soc/aspeed/Kconfig | 11 +++ > > drivers/soc/aspeed/Makefile | 1 + > > drivers/soc/aspeed/aspeed-bmc-misc.c | 187 +++++++++++++++++++++++++++++++++++ > > 5 files changed, 201 insertions(+) > > create mode 100644 drivers/soc/aspeed/Kconfig > > create mode 100644 drivers/soc/aspeed/Makefile > > create mode 100644 drivers/soc/aspeed/aspeed-bmc-misc.c > > > > diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig > > index 07fc0ac51c52..776fd5978f70 100644 > > --- a/drivers/soc/Kconfig > > +++ b/drivers/soc/Kconfig > > @@ -1,6 +1,7 @@ > > menu "SOC (System On Chip) specific Drivers" > > > > source "drivers/soc/actions/Kconfig" > > +source "drivers/soc/aspeed/Kconfig" > > source "drivers/soc/atmel/Kconfig" > > source "drivers/soc/bcm/Kconfig" > > source "drivers/soc/fsl/Kconfig" > > diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile > > index 9241125416ba..94a5b97c4466 100644 > > --- a/drivers/soc/Makefile > > +++ b/drivers/soc/Makefile > > @@ -3,6 +3,7 @@ > > # > > > > obj-$(CONFIG_ARCH_ACTIONS) += actions/ > > +obj-$(CONFIG_ARCH_ASPEED) += aspeed/ > > obj-$(CONFIG_ARCH_AT91) += atmel/ > > obj-y += bcm/ > > obj-$(CONFIG_ARCH_DOVE) += dove/ > > diff --git a/drivers/soc/aspeed/Kconfig b/drivers/soc/aspeed/Kconfig > > new file mode 100644 > > index 000000000000..704a4efe180f > > --- /dev/null > > +++ b/drivers/soc/aspeed/Kconfig > > @@ -0,0 +1,11 @@ > > +menu "ASPEED SoC drivers" > > + > > +config ASPEED_BMC_MISC > > + bool "Miscellaneous ASPEED BMC interfaces" > > Do you mean bool or tristate? if you mean bool, drop the module-y bits below. Should be tristate, was just slapping bits together :) > > > + depends on ARCH_ASPEED || COMPILE_TEST > > + default ARCH_ASPEED > > + help > > + Say yes to expose VGA and LPC scratch registers, and other > > + miscellaneous control interfaces specific to the ASPEED BMC SoCs > > + > > +endmenu > > diff --git a/drivers/soc/aspeed/Makefile b/drivers/soc/aspeed/Makefile > > new file mode 100644 > > index 000000000000..d1a80f9a584f > > --- /dev/null > > +++ b/drivers/soc/aspeed/Makefile > > @@ -0,0 +1 @@ > > +obj-$(CONFIG_ASPEED_BMC_MISC) += aspeed-bmc-misc.o > > diff --git a/drivers/soc/aspeed/aspeed-bmc-misc.c b/drivers/soc/aspeed/aspeed-bmc-misc.c > > new file mode 100644 > > index 000000000000..48522a736a05 > > --- /dev/null > > +++ b/drivers/soc/aspeed/aspeed-bmc-misc.c > > copyright headahs Good catch. > > > @@ -0,0 +1,187 @@ > > +#include <linux/kobject.h> > > +#include <linux/module.h> > > +#include <linux/of.h> > > +#include <linux/platform_device.h> > > +#include <linux/regmap.h> > > +#include <linux/mfd/syscon.h> > > + > > +#define DEVICE_NAME "aspeed-bmc-misc" > > This is the name of our chardev? No, it appears to just be the driver > name. You can probably just use it directly. Yep > > An observation: It's the only thing inside this file that is aspeed > specific. If not for the name, it could be generic code. I agree, hence some of the bullet points in the cover letter. But it's name and location was also partly driven by Arnd's suggestion that we stick the junk in drivers/soc, and it felt a bit weird not to use aspeed in the name given the convention there. > > lgtm. I suggest writing up some bindings and putting on the asbestos pants. Wheee... > > I'll merge this into openbmc with the IBM copyright added to the top. Sounds good. > With this adding userspace ABI, anyone who writes code against it will > need to make themselves available for rework as it goes through > upstream review. Yeah. Will try to make sure people are aware of this. Thanks for the feedback. Andrew
>> An observation: It's the only thing inside this file that is aspeed >> specific. If not for the name, it could be generic code. >I agree, hence some of the bullet points in the cover letter. But it's name and location was also partly driven by Arnd's suggestion that we stick the junk in drivers/soc, and it felt a bit weird not to use aspeed in the name >given the convention there. Just FYI - We've got the same exact problem on Nuvoton (just odds and ends in misc registers). Does this imply we will need a soc/npcm750 which also does the exact same thing? I'm guessing making it too generic - will make it harder to get upstream?
Hi Eugene, On Wed, 2 May 2018, at 02:51, Eugene.Cho@dell.com wrote: > > >> An observation: It's the only thing inside this file that is aspeed > >> specific. If not for the name, it could be generic code. > > >I agree, hence some of the bullet points in the cover letter. But it's name and location was also partly driven by Arnd's suggestion that we stick the junk in drivers/soc, and it felt a bit weird not to use aspeed in the name >given the convention there. > > Just FYI - We've got the same exact problem on Nuvoton (just odds and > ends in misc registers). Does this imply we will need a soc/npcm750 > which also does the exact same thing? I don't think so, if anything it's an argument for moving it out of soc/aspeed > > I'm guessing making it too generic - will make it harder to get upstream? > Probably, but the more evidence we have for requiring it the harder it is to argue against, and so the fact that this could help the NPCM as well might even make it easier to upstream. Have you played with the implementation at all? Would be good to get some consensus on it being a decent approach here before we go upstream with it. I also need to add devicetree bindings and ABI documentation... Cheers, Andrew
Hey Andrew -
> Have you played with the implementation at all? Would be good to get some consensus on it being a decent approach here before we go upstream with it. I also need to add devicetree bindings and ABI documentation...
Yea we have a "Dell" driver which almost matches exactly aspeed-bmc-misc.c (sysfs node per register field... label,reg-offset, and bit-mask configured via dt).
I think what you got is good, only thing I would say is maybe remove the references to aspeed?
On Fri, 11 May 2018, at 04:09, Eugene.Cho@dell.com wrote: > Hey Andrew - > > > Have you played with the implementation at all? Would be good to get some consensus on it being a decent approach here before we go upstream with it. I also need to add devicetree bindings and ABI documentation... > > Yea we have a "Dell" driver which almost matches exactly aspeed-bmc- > misc.c (sysfs node per register field... label,reg-offset, and bit-mask > configured via dt). > I think what you got is good, only thing I would say is maybe remove the > references to aspeed? > Yeah, I'll rework the patches to file-off the ASPEED parts and find a more appropriate home for the driver. Cheers for the feedback. Andrew
diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig index 07fc0ac51c52..776fd5978f70 100644 --- a/drivers/soc/Kconfig +++ b/drivers/soc/Kconfig @@ -1,6 +1,7 @@ menu "SOC (System On Chip) specific Drivers" source "drivers/soc/actions/Kconfig" +source "drivers/soc/aspeed/Kconfig" source "drivers/soc/atmel/Kconfig" source "drivers/soc/bcm/Kconfig" source "drivers/soc/fsl/Kconfig" diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile index 9241125416ba..94a5b97c4466 100644 --- a/drivers/soc/Makefile +++ b/drivers/soc/Makefile @@ -3,6 +3,7 @@ # obj-$(CONFIG_ARCH_ACTIONS) += actions/ +obj-$(CONFIG_ARCH_ASPEED) += aspeed/ obj-$(CONFIG_ARCH_AT91) += atmel/ obj-y += bcm/ obj-$(CONFIG_ARCH_DOVE) += dove/ diff --git a/drivers/soc/aspeed/Kconfig b/drivers/soc/aspeed/Kconfig new file mode 100644 index 000000000000..704a4efe180f --- /dev/null +++ b/drivers/soc/aspeed/Kconfig @@ -0,0 +1,11 @@ +menu "ASPEED SoC drivers" + +config ASPEED_BMC_MISC + bool "Miscellaneous ASPEED BMC interfaces" + depends on ARCH_ASPEED || COMPILE_TEST + default ARCH_ASPEED + help + Say yes to expose VGA and LPC scratch registers, and other + miscellaneous control interfaces specific to the ASPEED BMC SoCs + +endmenu diff --git a/drivers/soc/aspeed/Makefile b/drivers/soc/aspeed/Makefile new file mode 100644 index 000000000000..d1a80f9a584f --- /dev/null +++ b/drivers/soc/aspeed/Makefile @@ -0,0 +1 @@ +obj-$(CONFIG_ASPEED_BMC_MISC) += aspeed-bmc-misc.o diff --git a/drivers/soc/aspeed/aspeed-bmc-misc.c b/drivers/soc/aspeed/aspeed-bmc-misc.c new file mode 100644 index 000000000000..48522a736a05 --- /dev/null +++ b/drivers/soc/aspeed/aspeed-bmc-misc.c @@ -0,0 +1,187 @@ +#include <linux/kobject.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/platform_device.h> +#include <linux/regmap.h> +#include <linux/mfd/syscon.h> + +#define DEVICE_NAME "aspeed-bmc-misc" + +struct aspeed_bmc_ctrl { + const char *name; + u32 offset; + u32 mask; + u32 shift; + struct regmap *map; + struct kobj_attribute attr; +}; + +struct aspeed_bmc_misc { + struct device *dev; + struct regmap *map; + struct aspeed_bmc_ctrl *ctrls; + int nr_ctrls; +}; + +static int aspeed_bmc_misc_parse_dt_child(struct device_node *child, + struct aspeed_bmc_ctrl *ctrl) +{ + int rc; + + /* Example child: + * + * ilpc2ahb { + * offset = <0x80>; + * bit-mask = <0x1>; + * bit-shift = <6>; + * label = "foo"; + * } + */ + if (of_property_read_string(child, "label", &ctrl->name)) + ctrl->name = child->name; + + rc = of_property_read_u32(child, "offset", &ctrl->offset); + if (rc < 0) + return rc; + + rc = of_property_read_u32(child, "bit-mask", &ctrl->mask); + if (rc < 0) + return rc; + + rc = of_property_read_u32(child, "bit-shift", &ctrl->shift); + if (rc < 0) + return rc; + + ctrl->mask <<= ctrl->shift; + + return 0; +} + +static int aspeed_bmc_misc_parse_dt(struct aspeed_bmc_misc *bmc, + struct device_node *parent) +{ + struct aspeed_bmc_ctrl *ctrl; + struct device_node *child; + int rc; + + bmc->nr_ctrls = of_get_child_count(parent); + bmc->ctrls = devm_kcalloc(bmc->dev, bmc->nr_ctrls, sizeof(*bmc->ctrls), + GFP_KERNEL); + if (!bmc->ctrls) + return -ENOMEM; + + ctrl = bmc->ctrls; + for_each_child_of_node(parent, child) { + rc = aspeed_bmc_misc_parse_dt_child(child, ctrl++); + if (rc < 0) { + of_node_put(child); + return rc; + } + } + + return 0; +} + +static ssize_t aspeed_bmc_misc_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + struct aspeed_bmc_ctrl *ctrl; + unsigned int val; + int rc; + + ctrl = container_of(attr, struct aspeed_bmc_ctrl, attr); + rc = regmap_read(ctrl->map, ctrl->offset, &val); + if (rc) + return rc; + + val &= ctrl->mask; + val >>= ctrl->shift; + + return sprintf(buf, "%u\n", val); +} + +static ssize_t aspeed_bmc_misc_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t count) +{ + struct aspeed_bmc_ctrl *ctrl; + long val; + int rc; + + rc = kstrtol(buf, 0, &val); + if (rc) + return rc; + + ctrl = container_of(attr, struct aspeed_bmc_ctrl, attr); + val <<= ctrl->shift; + rc = regmap_update_bits(ctrl->map, ctrl->offset, ctrl->mask, val); + + return rc < 0 ? rc : count; +} + +static int aspeed_bmc_misc_add_sysfs_attr(struct aspeed_bmc_misc *bmc, + struct aspeed_bmc_ctrl *ctrl) +{ + ctrl->map = bmc->map; + + sysfs_attr_init(&ctrl->attr.attr); + ctrl->attr.attr.name = ctrl->name; + ctrl->attr.attr.mode = 0664; + ctrl->attr.show = aspeed_bmc_misc_show; + ctrl->attr.store = aspeed_bmc_misc_store; + + return sysfs_create_file(&bmc->dev->kobj, &ctrl->attr.attr); +} + +static int aspeed_bmc_misc_populate_sysfs(struct aspeed_bmc_misc *bmc) +{ + int rc; + int i; + + for (i = 0; i < bmc->nr_ctrls; i++) { + rc = aspeed_bmc_misc_add_sysfs_attr(bmc, &bmc->ctrls[i]); + if (rc < 0) + return rc; + } + + return 0; +} + +static int aspeed_bmc_misc_probe(struct platform_device *pdev) +{ + struct aspeed_bmc_misc *bmc; + int rc; + + bmc = devm_kzalloc(&pdev->dev, sizeof(*bmc), GFP_KERNEL); + if (!bmc) + return -ENOMEM; + + bmc->dev = &pdev->dev; + bmc->map = syscon_node_to_regmap(pdev->dev.parent->of_node); + if (IS_ERR(bmc->map)) + return PTR_ERR(bmc->map); + + rc = aspeed_bmc_misc_parse_dt(bmc, pdev->dev.of_node); + if (rc < 0) + return rc; + + return aspeed_bmc_misc_populate_sysfs(bmc); +} + +static const struct of_device_id aspeed_bmc_misc_match[] = { + { .compatible = "aspeed,bmc-misc" }, + { }, +}; + +static struct platform_driver aspeed_bmc_misc = { + .driver = { + .name = DEVICE_NAME, + .of_match_table = aspeed_bmc_misc_match, + }, + .probe = aspeed_bmc_misc_probe, +}; + +module_platform_driver(aspeed_bmc_misc); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Andrew Jeffery <andrew@aj.id.au>");
The ASPEED BMC SoCs have many knobs and switches that are sometimes design-specific and often defy any approach to unify them under an existing subsystem. Add a driver to translate a devicetree table into sysfs entries to expose bits and fields for manipulation from userspace. This encompasses concepts from scratch registers to boolean conditions to enable or disable host interface features. Signed-off-by: Andrew Jeffery <andrew@aj.id.au> --- drivers/soc/Kconfig | 1 + drivers/soc/Makefile | 1 + drivers/soc/aspeed/Kconfig | 11 +++ drivers/soc/aspeed/Makefile | 1 + drivers/soc/aspeed/aspeed-bmc-misc.c | 187 +++++++++++++++++++++++++++++++++++ 5 files changed, 201 insertions(+) create mode 100644 drivers/soc/aspeed/Kconfig create mode 100644 drivers/soc/aspeed/Makefile create mode 100644 drivers/soc/aspeed/aspeed-bmc-misc.c