| Message ID | 1602828151-24784-2-git-send-email-yilun.xu@intel.com (mailing list archive) |
|---|---|
| State | Superseded, archived |
| Headers | show |
| Series | UIO support for dfl devices | expand |
On 10/15/20 11:02 PM, Xu Yilun wrote: > Add support for overriding the default matching of a dfl device to a dfl > driver. It follows the same way that can be used for PCI and platform > devices. This patch adds the 'driver_override' sysfs file. > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > --- > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- > include/linux/dfl.h | 2 ++ > 3 files changed, 79 insertions(+), 5 deletions(-) > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl > index 23543be..db7e8d3 100644 > --- a/Documentation/ABI/testing/sysfs-bus-dfl > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > @@ -1,15 +1,35 @@ > What: /sys/bus/dfl/devices/dfl_dev.X/type > -Date: Aug 2020 > -KernelVersion: 5.10 > +Date: Oct 2020 > +KernelVersion: 5.11 > Contact: Xu Yilun <yilun.xu@intel.com> > Description: Read-only. It returns type of DFL FIU of the device. Now DFL > supports 2 FIU types, 0 for FME, 1 for PORT. > Format: 0x%x > > What: /sys/bus/dfl/devices/dfl_dev.X/feature_id > -Date: Aug 2020 > -KernelVersion: 5.10 > +Date: Oct 2020 > +KernelVersion: 5.11 > Contact: Xu Yilun <yilun.xu@intel.com> > Description: Read-only. It returns feature identifier local to its DFL FIU > type. > Format: 0x%x These updates, do not match the comment. Consider splitting this out. > + > +What: /sys/bus/dfl/devices/.../driver_override > +Date: Oct 2020 > +KernelVersion: 5.11 > +Contact: Xu Yilun <yilun.xu@intel.com> I am looking at description and trying to make it consistent with sysfs-bus-pci > +Description: This file allows the driver for a device to be specified. 'to be specified which will override the standard dfl bus feature id to driver mapping.' > When > + specified, only a driver with a name matching the value written > + to driver_override will have an opportunity to bind to the > + device. The override is specified by writing a string to the > + driver_override file (echo dfl-uio-pdev > driver_override) and > + may be cleared with an empty string (echo > driver_override). > + This returns the device to standard matching rules binding. > + Writing to driver_override does not automatically unbind the > + device from its current driver or make any attempt to > + automatically load the specified driver. If no driver with a > + matching name is currently loaded in the kernel, the device > + will not bind to any driver. This also allows devices to > + opt-out of driver binding using a driver_override name such as > + "none". Only a single driver may be specified in the override, > + there is no support for parsing delimiters. > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > index 511b20f..bc35750 100644 > --- a/drivers/fpga/dfl.c > +++ b/drivers/fpga/dfl.c > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) > struct dfl_driver *ddrv = to_dfl_drv(drv); > const struct dfl_device_id *id_entry; > > + /* When driver_override is set, only bind to the matching driver */ > + if (ddev->driver_override) > + return !strcmp(ddev->driver_override, drv->name); > + > id_entry = ddrv->id_table; > if (id_entry) { > while (id_entry->feature_id) { > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) > ddev->type, ddev->feature_id); > } > I am looking at other implementations of driver_override* and looking for consistency. > +static ssize_t driver_override_show(struct device *dev, > + struct device_attribute *attr, char *buf) > +{ > + struct dfl_device *ddev = to_dfl_dev(dev); > + ssize_t len; > + > + device_lock(dev); > + len = sprintf(buf, "%s\n", ddev->driver_override); len = snprintf(buf, PAGE_SIZE ... > + device_unlock(dev); > + return len; > +} > + > +static ssize_t driver_override_store(struct device *dev, > + struct device_attribute *attr, > + const char *buf, size_t count) > +{ > + struct dfl_device *ddev = to_dfl_dev(dev); > + char *driver_override, *old, *cp; > + > + /* We need to keep extra room for a newline */ > + if (count >= (PAGE_SIZE - 1)) > + return -EINVAL; > + > + driver_override = kstrndup(buf, count, GFP_KERNEL); > + if (!driver_override) > + return -ENOMEM; > + > + cp = strchr(driver_override, '\n'); > + if (cp) > + *cp = '\0'; > + > + device_lock(dev); > + old = ddev->driver_override; > + if (strlen(driver_override)) { > + ddev->driver_override = driver_override; > + } else { > + kfree(driver_override); > + ddev->driver_override = NULL; > + } > + device_unlock(dev); > + > + kfree(old); > + > + return count; > +} > +static DEVICE_ATTR_RW(driver_override); > + > static ssize_t > type_show(struct device *dev, struct device_attribute *attr, char *buf) > { > @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); > static struct attribute *dfl_dev_attrs[] = { > &dev_attr_type.attr, > &dev_attr_feature_id.attr, > + &dev_attr_driver_override.attr, > NULL, > }; > ATTRIBUTE_GROUPS(dfl_dev); > @@ -469,7 +521,7 @@ static int dfl_devs_add(struct dfl_feature_platform_data *pdata) > > int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) > { > - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) > + if (!dfl_drv || !dfl_drv->probe) id_table is still needed for the normal case. Instead of removing this check, could you add something like || (!dfl_drv->is_override && !dfl_drv->id_table) Tom > return -EINVAL; > > dfl_drv->drv.owner = owner; > diff --git a/include/linux/dfl.h b/include/linux/dfl.h > index 7affba2f..e1b2471 100644 > --- a/include/linux/dfl.h > +++ b/include/linux/dfl.h > @@ -32,6 +32,7 @@ enum dfl_id_type { > * @num_irqs: number of IRQs supported by this dfl device. > * @cdev: pointer to DFL FPGA container device this dfl device belongs to. > * @id_entry: matched id entry in dfl driver's id table. > + * @driver_override: driver name to force a match > */ > struct dfl_device { > struct device dev; > @@ -43,6 +44,7 @@ struct dfl_device { > unsigned int num_irqs; > struct dfl_fpga_cdev *cdev; > const struct dfl_device_id *id_entry; > + char *driver_override; > }; > > /**
On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > Add support for overriding the default matching of a dfl device to a dfl > > driver. It follows the same way that can be used for PCI and platform > > devices. This patch adds the 'driver_override' sysfs file. > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > --- > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- > > include/linux/dfl.h | 2 ++ > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl > > index 23543be..db7e8d3 100644 > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > @@ -1,15 +1,35 @@ > > What: /sys/bus/dfl/devices/dfl_dev.X/type > > -Date: Aug 2020 > > -KernelVersion: 5.10 > > +Date: Oct 2020 > > +KernelVersion: 5.11 > > Contact: Xu Yilun <yilun.xu@intel.com> > > Description: Read-only. It returns type of DFL FIU of the device. Now DFL > > supports 2 FIU types, 0 for FME, 1 for PORT. > > Format: 0x%x > > > > What: /sys/bus/dfl/devices/dfl_dev.X/feature_id > > -Date: Aug 2020 > > -KernelVersion: 5.10 > > +Date: Oct 2020 > > +KernelVersion: 5.11 > > Contact: Xu Yilun <yilun.xu@intel.com> > > Description: Read-only. It returns feature identifier local to its DFL FIU > > type. > > Format: 0x%x > > These updates, do not match the comment. > > Consider splitting this out. I'm sorry it's a typo. The above code should not be changed. > > > + > > +What: /sys/bus/dfl/devices/.../driver_override > > +Date: Oct 2020 > > +KernelVersion: 5.11 > > +Contact: Xu Yilun <yilun.xu@intel.com> > I am looking at description and trying to make it consistent with sysfs-bus-pci > > +Description: This file allows the driver for a device to be specified. > > 'to be specified which will override the standard dfl bus feature id to driver mapping.' Yes, it could be improved. Actually now it is the "type" and "feature id" matching, the 2 fields are defined for dfl_driver.id_table. In future for dfl v1, it may be GUID matching, which will be added to id_table. So how about we make it more generic: 'to be specified which will override the standard ID table matching.' > > > > When > > + specified, only a driver with a name matching the value written > > + to driver_override will have an opportunity to bind to the > > + device. The override is specified by writing a string to the > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > + may be cleared with an empty string (echo > driver_override). > > + This returns the device to standard matching rules binding. > > + Writing to driver_override does not automatically unbind the > > + device from its current driver or make any attempt to > > + automatically load the specified driver. If no driver with a > > + matching name is currently loaded in the kernel, the device > > + will not bind to any driver. This also allows devices to > > + opt-out of driver binding using a driver_override name such as > > + "none". Only a single driver may be specified in the override, > > + there is no support for parsing delimiters. > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > index 511b20f..bc35750 100644 > > --- a/drivers/fpga/dfl.c > > +++ b/drivers/fpga/dfl.c > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > const struct dfl_device_id *id_entry; > > > > + /* When driver_override is set, only bind to the matching driver */ > > + if (ddev->driver_override) > > + return !strcmp(ddev->driver_override, drv->name); > > + > > id_entry = ddrv->id_table; > > if (id_entry) { > > while (id_entry->feature_id) { > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) > > ddev->type, ddev->feature_id); > > } > > > > I am looking at other implementations of driver_override* and looking for consistency. > > > +static ssize_t driver_override_show(struct device *dev, > > + struct device_attribute *attr, char *buf) > > +{ > > + struct dfl_device *ddev = to_dfl_dev(dev); > > + ssize_t len; > > + > > + device_lock(dev); > > + len = sprintf(buf, "%s\n", ddev->driver_override); > len = snprintf(buf, PAGE_SIZE ... It is good to me. Some bus drivers use snprintf, some use sprintf. I think it is reasonable snprintf is used here, unlike %d, %u ... it is uncertain for the output size of %s. > > + device_unlock(dev); > > + return len; > > +} > > + > > +static ssize_t driver_override_store(struct device *dev, > > + struct device_attribute *attr, > > + const char *buf, size_t count) > > +{ > > + struct dfl_device *ddev = to_dfl_dev(dev); > > + char *driver_override, *old, *cp; > > + > > + /* We need to keep extra room for a newline */ > > + if (count >= (PAGE_SIZE - 1)) > > + return -EINVAL; > > + > > + driver_override = kstrndup(buf, count, GFP_KERNEL); > > + if (!driver_override) > > + return -ENOMEM; > > + > > + cp = strchr(driver_override, '\n'); > > + if (cp) > > + *cp = '\0'; > > + > > + device_lock(dev); > > + old = ddev->driver_override; > > + if (strlen(driver_override)) { > > + ddev->driver_override = driver_override; > > + } else { > > + kfree(driver_override); > > + ddev->driver_override = NULL; > > + } > > + device_unlock(dev); > > + > > + kfree(old); > > + > > + return count; > > +} > > +static DEVICE_ATTR_RW(driver_override); > > + > > static ssize_t > > type_show(struct device *dev, struct device_attribute *attr, char *buf) > > { > > @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); > > static struct attribute *dfl_dev_attrs[] = { > > &dev_attr_type.attr, > > &dev_attr_feature_id.attr, > > + &dev_attr_driver_override.attr, > > NULL, > > }; > > ATTRIBUTE_GROUPS(dfl_dev); > > @@ -469,7 +521,7 @@ static int dfl_devs_add(struct dfl_feature_platform_data *pdata) > > > > int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) > > { > > - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) > > + if (!dfl_drv || !dfl_drv->probe) > > id_table is still needed for the normal case. > > Instead of removing this check, could you add something like > > || (!dfl_drv->is_override && !dfl_drv->id_table) I don't think it is needed. Seems is_override and !id_table are duplicated conditions for this implementation. And it may make confusing, e.g. could a driver been force matched when is_override is not set? I think we could make it simple, if the dfl driver didn't provide the id_table, normally it could not match any device. I think it could be easily understood by dfl driver developers. Thanks, Yilun > > Tom > > > return -EINVAL; > > > > dfl_drv->drv.owner = owner; > > diff --git a/include/linux/dfl.h b/include/linux/dfl.h > > index 7affba2f..e1b2471 100644 > > --- a/include/linux/dfl.h > > +++ b/include/linux/dfl.h > > @@ -32,6 +32,7 @@ enum dfl_id_type { > > * @num_irqs: number of IRQs supported by this dfl device. > > * @cdev: pointer to DFL FPGA container device this dfl device belongs to. > > * @id_entry: matched id entry in dfl driver's id table. > > + * @driver_override: driver name to force a match > > */ > > struct dfl_device { > > struct device dev; > > @@ -43,6 +44,7 @@ struct dfl_device { > > unsigned int num_irqs; > > struct dfl_fpga_cdev *cdev; > > const struct dfl_device_id *id_entry; > > + char *driver_override; > > }; > > > > /**
> On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > > Add support for overriding the default matching of a dfl device to a dfl > > > driver. It follows the same way that can be used for PCI and platform > > > devices. This patch adds the 'driver_override' sysfs file. > > > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > > --- > > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > > drivers/fpga/dfl.c | 54 > ++++++++++++++++++++++++++++++++- > > > include/linux/dfl.h | 2 ++ > > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl > b/Documentation/ABI/testing/sysfs-bus-dfl > > > index 23543be..db7e8d3 100644 > > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > > @@ -1,15 +1,35 @@ > > > What: /sys/bus/dfl/devices/dfl_dev.X/type > > > -Date: Aug 2020 > > > -KernelVersion: 5.10 > > > +Date: Oct 2020 > > > +KernelVersion: 5.11 > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > Description: Read-only. It returns type of DFL FIU of the device. > Now DFL > > > supports 2 FIU types, 0 for FME, 1 for PORT. > > > Format: 0x%x > > > > > > What: /sys/bus/dfl/devices/dfl_dev.X/feature_id > > > -Date: Aug 2020 > > > -KernelVersion: 5.10 > > > +Date: Oct 2020 > > > +KernelVersion: 5.11 > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > Description: Read-only. It returns feature identifier local to its DFL > FIU > > > type. > > > Format: 0x%x > > > > These updates, do not match the comment. > > > > Consider splitting this out. > > I'm sorry it's a typo. The above code should not be changed. > > > > > > + > > > +What: /sys/bus/dfl/devices/.../driver_override > > > +Date: Oct 2020 > > > +KernelVersion: 5.11 > > > +Contact: Xu Yilun <yilun.xu@intel.com> > > I am looking at description and trying to make it consistent with sysfs-bus- > pci > > > +Description: This file allows the driver for a device to be specified. > > > > 'to be specified which will override the standard dfl bus feature id to driver > mapping.' > > Yes, it could be improved. > > Actually now it is the "type" and "feature id" matching, the 2 fields > are defined for dfl_driver.id_table. In future for dfl v1, it may be > GUID matching, which will be added to id_table. So how about we make it > more generic: > > 'to be specified which will override the standard ID table matching.' > > > > > > > > When > > > + specified, only a driver with a name matching the value written > > > + to driver_override will have an opportunity to bind to the > > > + device. The override is specified by writing a string to the > > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > > + may be cleared with an empty string (echo > driver_override). > > > + This returns the device to standard matching rules binding. > > > + Writing to driver_override does not automatically unbind the > > > + device from its current driver or make any attempt to > > > + automatically load the specified driver. If no driver with a > > > + matching name is currently loaded in the kernel, the device > > > + will not bind to any driver. This also allows devices to > > > + opt-out of driver binding using a driver_override name such as > > > + "none". Only a single driver may be specified in the override, > > > + there is no support for parsing delimiters. > > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > > index 511b20f..bc35750 100644 > > > --- a/drivers/fpga/dfl.c > > > +++ b/drivers/fpga/dfl.c > > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, > struct device_driver *drv) > > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > > const struct dfl_device_id *id_entry; > > > > > > + /* When driver_override is set, only bind to the matching driver */ > > > + if (ddev->driver_override) > > > + return !strcmp(ddev->driver_override, drv->name); > > > + > > > id_entry = ddrv->id_table; > > > if (id_entry) { > > > while (id_entry->feature_id) { > > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, > struct kobj_uevent_env *env) > > > ddev->type, ddev->feature_id); > > > } > > > > > > > I am looking at other implementations of driver_override* and looking for > consistency. > > > > > +static ssize_t driver_override_show(struct device *dev, > > > + struct device_attribute *attr, char *buf) > > > +{ > > > + struct dfl_device *ddev = to_dfl_dev(dev); > > > + ssize_t len; > > > + > > > + device_lock(dev); > > > + len = sprintf(buf, "%s\n", ddev->driver_override); > > len = snprintf(buf, PAGE_SIZE ... > > It is good to me. > > Some bus drivers use snprintf, some use sprintf. > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > uncertain for the output size of %s. you limited the size < a page in store function for driver_override? Hao > > > > + device_unlock(dev); > > > + return len; > > > +} > > > + > > > +static ssize_t driver_override_store(struct device *dev, > > > + struct device_attribute *attr, > > > + const char *buf, size_t count) > > > +{ > > > + struct dfl_device *ddev = to_dfl_dev(dev); > > > + char *driver_override, *old, *cp; > > > + > > > + /* We need to keep extra room for a newline */ > > > + if (count >= (PAGE_SIZE - 1)) > > > + return -EINVAL; > > > + > > > + driver_override = kstrndup(buf, count, GFP_KERNEL); > > > + if (!driver_override) > > > + return -ENOMEM; > > > + > > > + cp = strchr(driver_override, '\n'); > > > + if (cp) > > > + *cp = '\0'; > > > + > > > + device_lock(dev); > > > + old = ddev->driver_override; > > > + if (strlen(driver_override)) { > > > + ddev->driver_override = driver_override; > > > + } else { > > > + kfree(driver_override); > > > + ddev->driver_override = NULL; > > > + } > > > + device_unlock(dev); > > > + > > > + kfree(old); > > > + > > > + return count; > > > +} > > > +static DEVICE_ATTR_RW(driver_override); > > > + > > > static ssize_t > > > type_show(struct device *dev, struct device_attribute *attr, char *buf) > > > { > > > @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); > > > static struct attribute *dfl_dev_attrs[] = { > > > &dev_attr_type.attr, > > > &dev_attr_feature_id.attr, > > > + &dev_attr_driver_override.attr, > > > NULL, > > > }; > > > ATTRIBUTE_GROUPS(dfl_dev); > > > @@ -469,7 +521,7 @@ static int dfl_devs_add(struct > dfl_feature_platform_data *pdata) > > > > > > int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) > > > { > > > - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) > > > + if (!dfl_drv || !dfl_drv->probe) > > > > id_table is still needed for the normal case. > > > > Instead of removing this check, could you add something like > > > > || (!dfl_drv->is_override && !dfl_drv->id_table) > > I don't think it is needed. Seems is_override and !id_table are duplicated > conditions for this implementation. And it may make confusing, e.g. could > a driver been force matched when is_override is not set? > > I think we could make it simple, if the dfl driver didn't provide the > id_table, normally it could not match any device. I think it could be > easily understood by dfl driver developers. > > > Thanks, > Yilun > > > > > Tom > > > > > return -EINVAL; > > > > > > dfl_drv->drv.owner = owner; > > > diff --git a/include/linux/dfl.h b/include/linux/dfl.h > > > index 7affba2f..e1b2471 100644 > > > --- a/include/linux/dfl.h > > > +++ b/include/linux/dfl.h > > > @@ -32,6 +32,7 @@ enum dfl_id_type { > > > * @num_irqs: number of IRQs supported by this dfl device. > > > * @cdev: pointer to DFL FPGA container device this dfl device belongs to. > > > * @id_entry: matched id entry in dfl driver's id table. > > > + * @driver_override: driver name to force a match > > > */ > > > struct dfl_device { > > > struct device dev; > > > @@ -43,6 +44,7 @@ struct dfl_device { > > > unsigned int num_irqs; > > > struct dfl_fpga_cdev *cdev; > > > const struct dfl_device_id *id_entry; > > > + char *driver_override; > > > }; > > > > > > /**
On Mon, Oct 19, 2020 at 03:46:23PM +0800, Wu, Hao wrote: > > On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > > > > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > > > Add support for overriding the default matching of a dfl device to a dfl > > > > driver. It follows the same way that can be used for PCI and platform > > > > devices. This patch adds the 'driver_override' sysfs file. > > > > > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > > > --- > > > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > > > drivers/fpga/dfl.c | 54 > > ++++++++++++++++++++++++++++++++- > > > > include/linux/dfl.h | 2 ++ > > > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl > > b/Documentation/ABI/testing/sysfs-bus-dfl > > > > index 23543be..db7e8d3 100644 > > > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > > > @@ -1,15 +1,35 @@ > > > > What:/sys/bus/dfl/devices/dfl_dev.X/type > > > > -Date:Aug 2020 > > > > -KernelVersion:5.10 > > > > +Date:Oct 2020 > > > > +KernelVersion:5.11 > > > > Contact:Xu Yilun <yilun.xu@intel.com> > > > > Description:Read-only. It returns type of DFL FIU of the device. > > Now DFL > > > > supports 2 FIU types, 0 for FME, 1 for PORT. > > > > Format: 0x%x > > > > > > > > What:/sys/bus/dfl/devices/dfl_dev.X/feature_id > > > > -Date:Aug 2020 > > > > -KernelVersion:5.10 > > > > +Date:Oct 2020 > > > > +KernelVersion:5.11 > > > > Contact:Xu Yilun <yilun.xu@intel.com> > > > > Description:Read-only. It returns feature identifier local to its DFL > > FIU > > > > type. > > > > Format: 0x%x > > > > > > These updates, do not match the comment. > > > > > > Consider splitting this out. > > > > I'm sorry it's a typo. The above code should not be changed. > > > > > > > > > + > > > > +What: /sys/bus/dfl/devices/.../driver_override > > > > +Date: Oct 2020 > > > > +KernelVersion: 5.11 > > > > +Contact: Xu Yilun <yilun.xu@intel.com> > > > I am looking at description and trying to make it consistent with sysfs-bus- > > pci > > > > +Description: This file allows the driver for a device to be specified. > > > > > > 'to be specified which will override the standard dfl bus feature id to driver > > mapping.' > > > > Yes, it could be improved. > > > > Actually now it is the "type" and "feature id" matching, the 2 fields > > are defined for dfl_driver.id_table. In future for dfl v1, it may be > > GUID matching, which will be added to id_table. So how about we make it > > more generic: > > > > 'to be specified which will override the standard ID table matching.' > > > > > > > > > > > > When > > > > + specified, only a driver with a name matching the value written > > > > + to driver_override will have an opportunity to bind to the > > > > + device. The override is specified by writing a string to the > > > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > > > + may be cleared with an empty string (echo > driver_override). > > > > + This returns the device to standard matching rules binding. > > > > + Writing to driver_override does not automatically unbind the > > > > + device from its current driver or make any attempt to > > > > + automatically load the specified driver. If no driver with a > > > > + matching name is currently loaded in the kernel, the device > > > > + will not bind to any driver. This also allows devices to > > > > + opt-out of driver binding using a driver_override name such as > > > > + "none". Only a single driver may be specified in the override, > > > > + there is no support for parsing delimiters. > > > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > > > index 511b20f..bc35750 100644 > > > > --- a/drivers/fpga/dfl.c > > > > +++ b/drivers/fpga/dfl.c > > > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, > > struct device_driver *drv) > > > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > > > const struct dfl_device_id *id_entry; > > > > > > > > +/* When driver_override is set, only bind to the matching driver */ > > > > +if (ddev->driver_override) > > > > +return !strcmp(ddev->driver_override, drv->name); > > > > + > > > > id_entry = ddrv->id_table; > > > > if (id_entry) { > > > > while (id_entry->feature_id) { > > > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, > > struct kobj_uevent_env *env) > > > > ddev->type, ddev->feature_id); > > > > } > > > > > > > > > > I am looking at other implementations of driver_override* and looking for > > consistency. > > > > > > > +static ssize_t driver_override_show(struct device *dev, > > > > + struct device_attribute *attr, char *buf) > > > > +{ > > > > +struct dfl_device *ddev = to_dfl_dev(dev); > > > > +ssize_t len; > > > > + > > > > +device_lock(dev); > > > > +len = sprintf(buf, "%s\n", ddev->driver_override); > > > len = snprintf(buf, PAGE_SIZE ... > > > > It is good to me. > > > > Some bus drivers use snprintf, some use sprintf. > > > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > > uncertain for the output size of %s. > > you limited the size < a page in store function for driver_override? Yes. So normally the sprintf should be OK. But I think it may be safer if the driver_override pointer is corrupted in some unexpected cases. Thanks, Yilun > > Hao > > > > > > > +device_unlock(dev); > > > > +return len; > > > > +} > > > > + > > > > +static ssize_t driver_override_store(struct device *dev, > > > > + struct device_attribute *attr, > > > > + const char *buf, size_t count) > > > > +{ > > > > +struct dfl_device *ddev = to_dfl_dev(dev); > > > > +char *driver_override, *old, *cp; > > > > + > > > > +/* We need to keep extra room for a newline */ > > > > +if (count >= (PAGE_SIZE - 1)) > > > > +return -EINVAL; > > > > + > > > > +driver_override = kstrndup(buf, count, GFP_KERNEL); > > > > +if (!driver_override) > > > > +return -ENOMEM; > > > > + > > > > +cp = strchr(driver_override, '\n'); > > > > +if (cp) > > > > +*cp = '\0'; > > > > + > > > > +device_lock(dev); > > > > +old = ddev->driver_override; > > > > +if (strlen(driver_override)) { > > > > +ddev->driver_override = driver_override; > > > > +} else { > > > > +kfree(driver_override); > > > > +ddev->driver_override = NULL; > > > > +} > > > > +device_unlock(dev); > > > > + > > > > +kfree(old); > > > > + > > > > +return count; > > > > +} > > > > +static DEVICE_ATTR_RW(driver_override); > > > > + > > > > static ssize_t > > > > type_show(struct device *dev, struct device_attribute *attr, char *buf) > > > > { > > > > @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); > > > > static struct attribute *dfl_dev_attrs[] = { > > > > &dev_attr_type.attr, > > > > &dev_attr_feature_id.attr, > > > > +&dev_attr_driver_override.attr, > > > > NULL, > > > > }; > > > > ATTRIBUTE_GROUPS(dfl_dev); > > > > @@ -469,7 +521,7 @@ static int dfl_devs_add(struct > > dfl_feature_platform_data *pdata) > > > > > > > > int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) > > > > { > > > > -if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) > > > > +if (!dfl_drv || !dfl_drv->probe) > > > > > > id_table is still needed for the normal case. > > > > > > Instead of removing this check, could you add something like > > > > > > || (!dfl_drv->is_override && !dfl_drv->id_table) > > > > I don't think it is needed. Seems is_override and !id_table are duplicated > > conditions for this implementation. And it may make confusing, e.g. could > > a driver been force matched when is_override is not set? > > > > I think we could make it simple, if the dfl driver didn't provide the > > id_table, normally it could not match any device. I think it could be > > easily understood by dfl driver developers. > > > > > > Thanks, > > Yilun > > > > > > > > Tom > > > > > > > return -EINVAL; > > > > > > > > dfl_drv->drv.owner = owner; > > > > diff --git a/include/linux/dfl.h b/include/linux/dfl.h > > > > index 7affba2f..e1b2471 100644 > > > > --- a/include/linux/dfl.h > > > > +++ b/include/linux/dfl.h > > > > @@ -32,6 +32,7 @@ enum dfl_id_type { > > > > * @num_irqs: number of IRQs supported by this dfl device. > > > > * @cdev: pointer to DFL FPGA container device this dfl device belongs to. > > > > * @id_entry: matched id entry in dfl driver's id table. > > > > + * @driver_override: driver name to force a match > > > > */ > > > > struct dfl_device { > > > > struct device dev; > > > > @@ -43,6 +44,7 @@ struct dfl_device { > > > > unsigned int num_irqs; > > > > struct dfl_fpga_cdev *cdev; > > > > const struct dfl_device_id *id_entry; > > > > +char *driver_override; > > > > }; > > > > > > > > /**
On Mon, Oct 19, 2020 at 12:06:13PM +0800, Xu Yilun wrote: > On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > > Add support for overriding the default matching of a dfl device to a dfl > > > driver. It follows the same way that can be used for PCI and platform > > > devices. This patch adds the 'driver_override' sysfs file. > > > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > > --- > > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > > drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- > > > include/linux/dfl.h | 2 ++ > > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl > > > index 23543be..db7e8d3 100644 > > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > > @@ -1,15 +1,35 @@ > > > What: /sys/bus/dfl/devices/dfl_dev.X/type > > > -Date: Aug 2020 > > > -KernelVersion: 5.10 > > > +Date: Oct 2020 > > > +KernelVersion: 5.11 > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > Description: Read-only. It returns type of DFL FIU of the device. Now DFL > > > supports 2 FIU types, 0 for FME, 1 for PORT. > > > Format: 0x%x > > > > > > What: /sys/bus/dfl/devices/dfl_dev.X/feature_id > > > -Date: Aug 2020 > > > -KernelVersion: 5.10 > > > +Date: Oct 2020 > > > +KernelVersion: 5.11 > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > Description: Read-only. It returns feature identifier local to its DFL FIU > > > type. > > > Format: 0x%x > > > > These updates, do not match the comment. > > > > Consider splitting this out. > > I'm sorry it's a typo. The above code should not be changed. > > > > > > + > > > +What: /sys/bus/dfl/devices/.../driver_override > > > +Date: Oct 2020 > > > +KernelVersion: 5.11 > > > +Contact: Xu Yilun <yilun.xu@intel.com> > > I am looking at description and trying to make it consistent with sysfs-bus-pci > > > +Description: This file allows the driver for a device to be specified. > > > > 'to be specified which will override the standard dfl bus feature id to driver mapping.' > > Yes, it could be improved. > > Actually now it is the "type" and "feature id" matching, the 2 fields > are defined for dfl_driver.id_table. In future for dfl v1, it may be > GUID matching, which will be added to id_table. So how about we make it > more generic: > > 'to be specified which will override the standard ID table matching.' > > > > > > > > When > > > + specified, only a driver with a name matching the value written > > > + to driver_override will have an opportunity to bind to the > > > + device. The override is specified by writing a string to the > > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > > + may be cleared with an empty string (echo > driver_override). > > > + This returns the device to standard matching rules binding. > > > + Writing to driver_override does not automatically unbind the > > > + device from its current driver or make any attempt to > > > + automatically load the specified driver. If no driver with a > > > + matching name is currently loaded in the kernel, the device > > > + will not bind to any driver. This also allows devices to > > > + opt-out of driver binding using a driver_override name such as > > > + "none". Only a single driver may be specified in the override, > > > + there is no support for parsing delimiters. > > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > > index 511b20f..bc35750 100644 > > > --- a/drivers/fpga/dfl.c > > > +++ b/drivers/fpga/dfl.c > > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) > > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > > const struct dfl_device_id *id_entry; > > > > > > + /* When driver_override is set, only bind to the matching driver */ > > > + if (ddev->driver_override) > > > + return !strcmp(ddev->driver_override, drv->name); > > > + > > > id_entry = ddrv->id_table; > > > if (id_entry) { > > > while (id_entry->feature_id) { > > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) > > > ddev->type, ddev->feature_id); > > > } > > > > > > > I am looking at other implementations of driver_override* and looking for consistency. > > > > > +static ssize_t driver_override_show(struct device *dev, > > > + struct device_attribute *attr, char *buf) > > > +{ > > > + struct dfl_device *ddev = to_dfl_dev(dev); > > > + ssize_t len; > > > + > > > + device_lock(dev); > > > + len = sprintf(buf, "%s\n", ddev->driver_override); > > len = snprintf(buf, PAGE_SIZE ... > > It is good to me. > > Some bus drivers use snprintf, some use sprintf. > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > uncertain for the output size of %s. Sorry, I checked the Documentation/filesystems/sysfs.rst again and found I didn't remember it correctly. The snprintf is must not be used, sprintf() or scnprintf() should be used. So I prefer sprintf() here, following other implementations. sprintf() should be safe here, as we already limited the input of driver_override string to less then PAGE_SIZE on store(). Thanks, Yilun > > > > + device_unlock(dev); > > > + return len; > > > +} > > > + > > > +static ssize_t driver_override_store(struct device *dev, > > > + struct device_attribute *attr, > > > + const char *buf, size_t count) > > > +{ > > > + struct dfl_device *ddev = to_dfl_dev(dev); > > > + char *driver_override, *old, *cp; > > > + > > > + /* We need to keep extra room for a newline */ > > > + if (count >= (PAGE_SIZE - 1)) > > > + return -EINVAL; > > > + > > > + driver_override = kstrndup(buf, count, GFP_KERNEL); > > > + if (!driver_override) > > > + return -ENOMEM; > > > + > > > + cp = strchr(driver_override, '\n'); > > > + if (cp) > > > + *cp = '\0'; > > > + > > > + device_lock(dev); > > > + old = ddev->driver_override; > > > + if (strlen(driver_override)) { > > > + ddev->driver_override = driver_override; > > > + } else { > > > + kfree(driver_override); > > > + ddev->driver_override = NULL; > > > + } > > > + device_unlock(dev); > > > + > > > + kfree(old); > > > + > > > + return count; > > > +} > > > +static DEVICE_ATTR_RW(driver_override); > > > + > > > static ssize_t > > > type_show(struct device *dev, struct device_attribute *attr, char *buf) > > > { > > > @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); > > > static struct attribute *dfl_dev_attrs[] = { > > > &dev_attr_type.attr, > > > &dev_attr_feature_id.attr, > > > + &dev_attr_driver_override.attr, > > > NULL, > > > }; > > > ATTRIBUTE_GROUPS(dfl_dev); > > > @@ -469,7 +521,7 @@ static int dfl_devs_add(struct dfl_feature_platform_data *pdata) > > > > > > int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) > > > { > > > - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) > > > + if (!dfl_drv || !dfl_drv->probe) > > > > id_table is still needed for the normal case. > > > > Instead of removing this check, could you add something like > > > > || (!dfl_drv->is_override && !dfl_drv->id_table) > > I don't think it is needed. Seems is_override and !id_table are duplicated > conditions for this implementation. And it may make confusing, e.g. could > a driver been force matched when is_override is not set? > > I think we could make it simple, if the dfl driver didn't provide the > id_table, normally it could not match any device. I think it could be > easily understood by dfl driver developers. > > > Thanks, > Yilun > > > > > Tom > > > > > return -EINVAL; > > > > > > dfl_drv->drv.owner = owner; > > > diff --git a/include/linux/dfl.h b/include/linux/dfl.h > > > index 7affba2f..e1b2471 100644 > > > --- a/include/linux/dfl.h > > > +++ b/include/linux/dfl.h > > > @@ -32,6 +32,7 @@ enum dfl_id_type { > > > * @num_irqs: number of IRQs supported by this dfl device. > > > * @cdev: pointer to DFL FPGA container device this dfl device belongs to. > > > * @id_entry: matched id entry in dfl driver's id table. > > > + * @driver_override: driver name to force a match > > > */ > > > struct dfl_device { > > > struct device dev; > > > @@ -43,6 +44,7 @@ struct dfl_device { > > > unsigned int num_irqs; > > > struct dfl_fpga_cdev *cdev; > > > const struct dfl_device_id *id_entry; > > > + char *driver_override; > > > }; > > > > > > /**
On Mon, Oct 19, 2020 at 03:50:32PM +0800, Xu Yilun wrote: > On Mon, Oct 19, 2020 at 03:46:23PM +0800, Wu, Hao wrote: > > > On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > > > > > > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > > > > Add support for overriding the default matching of a dfl device to a dfl > > > > > driver. It follows the same way that can be used for PCI and platform > > > > > devices. This patch adds the 'driver_override' sysfs file. > > > > > > > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > > > > --- > > > > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > > > > drivers/fpga/dfl.c | 54 > > > ++++++++++++++++++++++++++++++++- > > > > > include/linux/dfl.h | 2 ++ > > > > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > > > > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl > > > b/Documentation/ABI/testing/sysfs-bus-dfl > > > > > index 23543be..db7e8d3 100644 > > > > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > > > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > > > > @@ -1,15 +1,35 @@ > > > > > What:/sys/bus/dfl/devices/dfl_dev.X/type > > > > > -Date:Aug 2020 > > > > > -KernelVersion:5.10 > > > > > +Date:Oct 2020 > > > > > +KernelVersion:5.11 > > > > > Contact:Xu Yilun <yilun.xu@intel.com> > > > > > Description:Read-only. It returns type of DFL FIU of the device. > > > Now DFL > > > > > supports 2 FIU types, 0 for FME, 1 for PORT. > > > > > Format: 0x%x > > > > > > > > > > What:/sys/bus/dfl/devices/dfl_dev.X/feature_id > > > > > -Date:Aug 2020 > > > > > -KernelVersion:5.10 > > > > > +Date:Oct 2020 > > > > > +KernelVersion:5.11 > > > > > Contact:Xu Yilun <yilun.xu@intel.com> > > > > > Description:Read-only. It returns feature identifier local to its DFL > > > FIU > > > > > type. > > > > > Format: 0x%x > > > > > > > > These updates, do not match the comment. > > > > > > > > Consider splitting this out. > > > > > > I'm sorry it's a typo. The above code should not be changed. > > > > > > > > > > > > + > > > > > +What: /sys/bus/dfl/devices/.../driver_override > > > > > +Date: Oct 2020 > > > > > +KernelVersion: 5.11 > > > > > +Contact: Xu Yilun <yilun.xu@intel.com> > > > > I am looking at description and trying to make it consistent with sysfs-bus- > > > pci > > > > > +Description: This file allows the driver for a device to be specified. > > > > > > > > 'to be specified which will override the standard dfl bus feature id to driver > > > mapping.' > > > > > > Yes, it could be improved. > > > > > > Actually now it is the "type" and "feature id" matching, the 2 fields > > > are defined for dfl_driver.id_table. In future for dfl v1, it may be > > > GUID matching, which will be added to id_table. So how about we make it > > > more generic: > > > > > > 'to be specified which will override the standard ID table matching.' > > > > > > > > > > > > > > > > When > > > > > + specified, only a driver with a name matching the value written > > > > > + to driver_override will have an opportunity to bind to the > > > > > + device. The override is specified by writing a string to the > > > > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > > > > + may be cleared with an empty string (echo > driver_override). > > > > > + This returns the device to standard matching rules binding. > > > > > + Writing to driver_override does not automatically unbind the > > > > > + device from its current driver or make any attempt to > > > > > + automatically load the specified driver. If no driver with a > > > > > + matching name is currently loaded in the kernel, the device > > > > > + will not bind to any driver. This also allows devices to > > > > > + opt-out of driver binding using a driver_override name such as > > > > > + "none". Only a single driver may be specified in the override, > > > > > + there is no support for parsing delimiters. > > > > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > > > > index 511b20f..bc35750 100644 > > > > > --- a/drivers/fpga/dfl.c > > > > > +++ b/drivers/fpga/dfl.c > > > > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, > > > struct device_driver *drv) > > > > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > > > > const struct dfl_device_id *id_entry; > > > > > > > > > > +/* When driver_override is set, only bind to the matching driver */ > > > > > +if (ddev->driver_override) > > > > > +return !strcmp(ddev->driver_override, drv->name); > > > > > + > > > > > id_entry = ddrv->id_table; > > > > > if (id_entry) { > > > > > while (id_entry->feature_id) { > > > > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, > > > struct kobj_uevent_env *env) > > > > > ddev->type, ddev->feature_id); > > > > > } > > > > > > > > > > > > > I am looking at other implementations of driver_override* and looking for > > > consistency. > > > > > > > > > +static ssize_t driver_override_show(struct device *dev, > > > > > + struct device_attribute *attr, char *buf) > > > > > +{ > > > > > +struct dfl_device *ddev = to_dfl_dev(dev); > > > > > +ssize_t len; > > > > > + > > > > > +device_lock(dev); > > > > > +len = sprintf(buf, "%s\n", ddev->driver_override); > > > > len = snprintf(buf, PAGE_SIZE ... > > > > > > It is good to me. > > > > > > Some bus drivers use snprintf, some use sprintf. > > > > > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > > > uncertain for the output size of %s. No, no one should care at all about this. If you are caring, then you are using sysfs wrong. Just use sprintf(), you never care about the size of the buffer for a show/store function. Or better yet, use the new sysfs_emit() call, that is what everything will be converted over to anyway. > > you limited the size < a page in store function for driver_override? > > Yes. So normally the sprintf should be OK. But I think it may be safer > if the driver_override pointer is corrupted in some unexpected cases. How can that happen? greg k-h
On Mon, Oct 19, 2020 at 04:52:33PM +0800, Xu Yilun wrote: > On Mon, Oct 19, 2020 at 12:06:13PM +0800, Xu Yilun wrote: > > On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > > > > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > > > Add support for overriding the default matching of a dfl device to a dfl > > > > driver. It follows the same way that can be used for PCI and platform > > > > devices. This patch adds the 'driver_override' sysfs file. > > > > > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > > > --- > > > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > > > drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- > > > > include/linux/dfl.h | 2 ++ > > > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl > > > > index 23543be..db7e8d3 100644 > > > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > > > @@ -1,15 +1,35 @@ > > > > What: /sys/bus/dfl/devices/dfl_dev.X/type > > > > -Date: Aug 2020 > > > > -KernelVersion: 5.10 > > > > +Date: Oct 2020 > > > > +KernelVersion: 5.11 > > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > > Description: Read-only. It returns type of DFL FIU of the device. Now DFL > > > > supports 2 FIU types, 0 for FME, 1 for PORT. > > > > Format: 0x%x > > > > > > > > What: /sys/bus/dfl/devices/dfl_dev.X/feature_id > > > > -Date: Aug 2020 > > > > -KernelVersion: 5.10 > > > > +Date: Oct 2020 > > > > +KernelVersion: 5.11 > > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > > Description: Read-only. It returns feature identifier local to its DFL FIU > > > > type. > > > > Format: 0x%x > > > > > > These updates, do not match the comment. > > > > > > Consider splitting this out. > > > > I'm sorry it's a typo. The above code should not be changed. > > > > > > > > > + > > > > +What: /sys/bus/dfl/devices/.../driver_override > > > > +Date: Oct 2020 > > > > +KernelVersion: 5.11 > > > > +Contact: Xu Yilun <yilun.xu@intel.com> > > > I am looking at description and trying to make it consistent with sysfs-bus-pci > > > > +Description: This file allows the driver for a device to be specified. > > > > > > 'to be specified which will override the standard dfl bus feature id to driver mapping.' > > > > Yes, it could be improved. > > > > Actually now it is the "type" and "feature id" matching, the 2 fields > > are defined for dfl_driver.id_table. In future for dfl v1, it may be > > GUID matching, which will be added to id_table. So how about we make it > > more generic: > > > > 'to be specified which will override the standard ID table matching.' > > > > > > > > > > > > When > > > > + specified, only a driver with a name matching the value written > > > > + to driver_override will have an opportunity to bind to the > > > > + device. The override is specified by writing a string to the > > > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > > > + may be cleared with an empty string (echo > driver_override). > > > > + This returns the device to standard matching rules binding. > > > > + Writing to driver_override does not automatically unbind the > > > > + device from its current driver or make any attempt to > > > > + automatically load the specified driver. If no driver with a > > > > + matching name is currently loaded in the kernel, the device > > > > + will not bind to any driver. This also allows devices to > > > > + opt-out of driver binding using a driver_override name such as > > > > + "none". Only a single driver may be specified in the override, > > > > + there is no support for parsing delimiters. > > > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > > > index 511b20f..bc35750 100644 > > > > --- a/drivers/fpga/dfl.c > > > > +++ b/drivers/fpga/dfl.c > > > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) > > > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > > > const struct dfl_device_id *id_entry; > > > > > > > > + /* When driver_override is set, only bind to the matching driver */ > > > > + if (ddev->driver_override) > > > > + return !strcmp(ddev->driver_override, drv->name); > > > > + > > > > id_entry = ddrv->id_table; > > > > if (id_entry) { > > > > while (id_entry->feature_id) { > > > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) > > > > ddev->type, ddev->feature_id); > > > > } > > > > > > > > > > I am looking at other implementations of driver_override* and looking for consistency. > > > > > > > +static ssize_t driver_override_show(struct device *dev, > > > > + struct device_attribute *attr, char *buf) > > > > +{ > > > > + struct dfl_device *ddev = to_dfl_dev(dev); > > > > + ssize_t len; > > > > + > > > > + device_lock(dev); > > > > + len = sprintf(buf, "%s\n", ddev->driver_override); > > > len = snprintf(buf, PAGE_SIZE ... > > > > It is good to me. > > > > Some bus drivers use snprintf, some use sprintf. > > > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > > uncertain for the output size of %s. > > Sorry, I checked the Documentation/filesystems/sysfs.rst again and found > I didn't remember it correctly. > > The snprintf is must not be used, sprintf() or scnprintf() should be > used. So I prefer sprintf() here, following other implementations. > > sprintf() should be safe here, as we already limited the input of > driver_override string to less then PAGE_SIZE on store(). As already mentioned, please use sysfs_emit() instead. thanks, greg k-h
On 10/18/20 9:06 PM, Xu Yilun wrote: > On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: >> On 10/15/20 11:02 PM, Xu Yilun wrote: >>> Add support for overriding the default matching of a dfl device to a dfl >>> driver. It follows the same way that can be used for PCI and platform >>> devices. This patch adds the 'driver_override' sysfs file. >>> >>> Signed-off-by: Xu Yilun <yilun.xu@intel.com> >>> --- >>> Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- >>> drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- >>> include/linux/dfl.h | 2 ++ >>> 3 files changed, 79 insertions(+), 5 deletions(-) >>> >>> diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl >>> index 23543be..db7e8d3 100644 >>> --- a/Documentation/ABI/testing/sysfs-bus-dfl >>> +++ b/Documentation/ABI/testing/sysfs-bus-dfl >>> @@ -1,15 +1,35 @@ >>> What: /sys/bus/dfl/devices/dfl_dev.X/type >>> -Date: Aug 2020 >>> -KernelVersion: 5.10 >>> +Date: Oct 2020 >>> +KernelVersion: 5.11 >>> Contact: Xu Yilun <yilun.xu@intel.com> >>> Description: Read-only. It returns type of DFL FIU of the device. Now DFL >>> supports 2 FIU types, 0 for FME, 1 for PORT. >>> Format: 0x%x >>> >>> What: /sys/bus/dfl/devices/dfl_dev.X/feature_id >>> -Date: Aug 2020 >>> -KernelVersion: 5.10 >>> +Date: Oct 2020 >>> +KernelVersion: 5.11 >>> Contact: Xu Yilun <yilun.xu@intel.com> >>> Description: Read-only. It returns feature identifier local to its DFL FIU >>> type. >>> Format: 0x%x >> These updates, do not match the comment. >> >> Consider splitting this out. > I'm sorry it's a typo. The above code should not be changed. > >>> + >>> +What: /sys/bus/dfl/devices/.../driver_override >>> +Date: Oct 2020 >>> +KernelVersion: 5.11 >>> +Contact: Xu Yilun <yilun.xu@intel.com> >> I am looking at description and trying to make it consistent with sysfs-bus-pci >>> +Description: This file allows the driver for a device to be specified. >> 'to be specified which will override the standard dfl bus feature id to driver mapping.' > Yes, it could be improved. > > Actually now it is the "type" and "feature id" matching, the 2 fields > are defined for dfl_driver.id_table. In future for dfl v1, it may be > GUID matching, which will be added to id_table. So how about we make it > more generic: > > 'to be specified which will override the standard ID table matching.' yes, this is good. > >> >>> When >>> + specified, only a driver with a name matching the value written >>> + to driver_override will have an opportunity to bind to the >>> + device. The override is specified by writing a string to the >>> + driver_override file (echo dfl-uio-pdev > driver_override) and >>> + may be cleared with an empty string (echo > driver_override). >>> + This returns the device to standard matching rules binding. >>> + Writing to driver_override does not automatically unbind the >>> + device from its current driver or make any attempt to >>> + automatically load the specified driver. If no driver with a >>> + matching name is currently loaded in the kernel, the device >>> + will not bind to any driver. This also allows devices to >>> + opt-out of driver binding using a driver_override name such as >>> + "none". Only a single driver may be specified in the override, >>> + there is no support for parsing delimiters. >>> diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c >>> index 511b20f..bc35750 100644 >>> --- a/drivers/fpga/dfl.c >>> +++ b/drivers/fpga/dfl.c >>> @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) >>> struct dfl_driver *ddrv = to_dfl_drv(drv); >>> const struct dfl_device_id *id_entry; >>> >>> + /* When driver_override is set, only bind to the matching driver */ >>> + if (ddev->driver_override) >>> + return !strcmp(ddev->driver_override, drv->name); >>> + >>> id_entry = ddrv->id_table; >>> if (id_entry) { >>> while (id_entry->feature_id) { >>> @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) >>> ddev->type, ddev->feature_id); >>> } >>> >> I am looking at other implementations of driver_override* and looking for consistency. >> >>> +static ssize_t driver_override_show(struct device *dev, >>> + struct device_attribute *attr, char *buf) >>> +{ >>> + struct dfl_device *ddev = to_dfl_dev(dev); >>> + ssize_t len; >>> + >>> + device_lock(dev); >>> + len = sprintf(buf, "%s\n", ddev->driver_override); >> len = snprintf(buf, PAGE_SIZE ... > It is good to me. > > Some bus drivers use snprintf, some use sprintf. > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > uncertain for the output size of %s. I am not sure if you are agreeing to do this change. I was looking at the other driver_override_show() functions. Generally the 'n' variants of string functions should be used. > >>> + device_unlock(dev); >>> + return len; >>> +} >>> + >>> +static ssize_t driver_override_store(struct device *dev, >>> + struct device_attribute *attr, >>> + const char *buf, size_t count) >>> +{ >>> + struct dfl_device *ddev = to_dfl_dev(dev); >>> + char *driver_override, *old, *cp; >>> + >>> + /* We need to keep extra room for a newline */ >>> + if (count >= (PAGE_SIZE - 1)) >>> + return -EINVAL; >>> + >>> + driver_override = kstrndup(buf, count, GFP_KERNEL); >>> + if (!driver_override) >>> + return -ENOMEM; >>> + >>> + cp = strchr(driver_override, '\n'); >>> + if (cp) >>> + *cp = '\0'; >>> + >>> + device_lock(dev); >>> + old = ddev->driver_override; >>> + if (strlen(driver_override)) { >>> + ddev->driver_override = driver_override; >>> + } else { >>> + kfree(driver_override); >>> + ddev->driver_override = NULL; >>> + } >>> + device_unlock(dev); >>> + >>> + kfree(old); >>> + >>> + return count; >>> +} >>> +static DEVICE_ATTR_RW(driver_override); >>> + >>> static ssize_t >>> type_show(struct device *dev, struct device_attribute *attr, char *buf) >>> { >>> @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); >>> static struct attribute *dfl_dev_attrs[] = { >>> &dev_attr_type.attr, >>> &dev_attr_feature_id.attr, >>> + &dev_attr_driver_override.attr, >>> NULL, >>> }; >>> ATTRIBUTE_GROUPS(dfl_dev); >>> @@ -469,7 +521,7 @@ static int dfl_devs_add(struct dfl_feature_platform_data *pdata) >>> >>> int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) >>> { >>> - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) >>> + if (!dfl_drv || !dfl_drv->probe) >> id_table is still needed for the normal case. >> >> Instead of removing this check, could you add something like >> >> || (!dfl_drv->is_override && !dfl_drv->id_table) > I don't think it is needed. Seems is_override and !id_table are duplicated > conditions for this implementation. And it may make confusing, e.g. could > a driver been force matched when is_override is not set? > > I think we could make it simple, if the dfl driver didn't provide the > id_table, normally it could not match any device. I think it could be > easily understood by dfl driver developers. > Then an ASSERT should be added in dfl_bus_match() for id_entry. Tom > Thanks, > Yilun > >> Tom >> >>> return -EINVAL; >>> >>> dfl_drv->drv.owner = owner; >>> diff --git a/include/linux/dfl.h b/include/linux/dfl.h >>> index 7affba2f..e1b2471 100644 >>> --- a/include/linux/dfl.h >>> +++ b/include/linux/dfl.h >>> @@ -32,6 +32,7 @@ enum dfl_id_type { >>> * @num_irqs: number of IRQs supported by this dfl device. >>> * @cdev: pointer to DFL FPGA container device this dfl device belongs to. >>> * @id_entry: matched id entry in dfl driver's id table. >>> + * @driver_override: driver name to force a match >>> */ >>> struct dfl_device { >>> struct device dev; >>> @@ -43,6 +44,7 @@ struct dfl_device { >>> unsigned int num_irqs; >>> struct dfl_fpga_cdev *cdev; >>> const struct dfl_device_id *id_entry; >>> + char *driver_override; >>> }; >>> >>> /**
On Mon, Oct 19, 2020 at 11:03:16AM +0200, Greg KH wrote: > On Mon, Oct 19, 2020 at 04:52:33PM +0800, Xu Yilun wrote: > > On Mon, Oct 19, 2020 at 12:06:13PM +0800, Xu Yilun wrote: > > > On Fri, Oct 16, 2020 at 09:21:50AM -0700, Tom Rix wrote: > > > > > > > > On 10/15/20 11:02 PM, Xu Yilun wrote: > > > > > Add support for overriding the default matching of a dfl device to a dfl > > > > > driver. It follows the same way that can be used for PCI and platform > > > > > devices. This patch adds the 'driver_override' sysfs file. > > > > > > > > > > Signed-off-by: Xu Yilun <yilun.xu@intel.com> > > > > > --- > > > > > Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- > > > > > drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- > > > > > include/linux/dfl.h | 2 ++ > > > > > 3 files changed, 79 insertions(+), 5 deletions(-) > > > > > > > > > > diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl > > > > > index 23543be..db7e8d3 100644 > > > > > --- a/Documentation/ABI/testing/sysfs-bus-dfl > > > > > +++ b/Documentation/ABI/testing/sysfs-bus-dfl > > > > > @@ -1,15 +1,35 @@ > > > > > What: /sys/bus/dfl/devices/dfl_dev.X/type > > > > > -Date: Aug 2020 > > > > > -KernelVersion: 5.10 > > > > > +Date: Oct 2020 > > > > > +KernelVersion: 5.11 > > > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > > > Description: Read-only. It returns type of DFL FIU of the device. Now DFL > > > > > supports 2 FIU types, 0 for FME, 1 for PORT. > > > > > Format: 0x%x > > > > > > > > > > What: /sys/bus/dfl/devices/dfl_dev.X/feature_id > > > > > -Date: Aug 2020 > > > > > -KernelVersion: 5.10 > > > > > +Date: Oct 2020 > > > > > +KernelVersion: 5.11 > > > > > Contact: Xu Yilun <yilun.xu@intel.com> > > > > > Description: Read-only. It returns feature identifier local to its DFL FIU > > > > > type. > > > > > Format: 0x%x > > > > > > > > These updates, do not match the comment. > > > > > > > > Consider splitting this out. > > > > > > I'm sorry it's a typo. The above code should not be changed. > > > > > > > > > > > > + > > > > > +What: /sys/bus/dfl/devices/.../driver_override > > > > > +Date: Oct 2020 > > > > > +KernelVersion: 5.11 > > > > > +Contact: Xu Yilun <yilun.xu@intel.com> > > > > I am looking at description and trying to make it consistent with sysfs-bus-pci > > > > > +Description: This file allows the driver for a device to be specified. > > > > > > > > 'to be specified which will override the standard dfl bus feature id to driver mapping.' > > > > > > Yes, it could be improved. > > > > > > Actually now it is the "type" and "feature id" matching, the 2 fields > > > are defined for dfl_driver.id_table. In future for dfl v1, it may be > > > GUID matching, which will be added to id_table. So how about we make it > > > more generic: > > > > > > 'to be specified which will override the standard ID table matching.' > > > > > > > > > > > > > > > > When > > > > > + specified, only a driver with a name matching the value written > > > > > + to driver_override will have an opportunity to bind to the > > > > > + device. The override is specified by writing a string to the > > > > > + driver_override file (echo dfl-uio-pdev > driver_override) and > > > > > + may be cleared with an empty string (echo > driver_override). > > > > > + This returns the device to standard matching rules binding. > > > > > + Writing to driver_override does not automatically unbind the > > > > > + device from its current driver or make any attempt to > > > > > + automatically load the specified driver. If no driver with a > > > > > + matching name is currently loaded in the kernel, the device > > > > > + will not bind to any driver. This also allows devices to > > > > > + opt-out of driver binding using a driver_override name such as > > > > > + "none". Only a single driver may be specified in the override, > > > > > + there is no support for parsing delimiters. > > > > > diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c > > > > > index 511b20f..bc35750 100644 > > > > > --- a/drivers/fpga/dfl.c > > > > > +++ b/drivers/fpga/dfl.c > > > > > @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) > > > > > struct dfl_driver *ddrv = to_dfl_drv(drv); > > > > > const struct dfl_device_id *id_entry; > > > > > > > > > > + /* When driver_override is set, only bind to the matching driver */ > > > > > + if (ddev->driver_override) > > > > > + return !strcmp(ddev->driver_override, drv->name); > > > > > + > > > > > id_entry = ddrv->id_table; > > > > > if (id_entry) { > > > > > while (id_entry->feature_id) { > > > > > @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) > > > > > ddev->type, ddev->feature_id); > > > > > } > > > > > > > > > > > > > I am looking at other implementations of driver_override* and looking for consistency. > > > > > > > > > +static ssize_t driver_override_show(struct device *dev, > > > > > + struct device_attribute *attr, char *buf) > > > > > +{ > > > > > + struct dfl_device *ddev = to_dfl_dev(dev); > > > > > + ssize_t len; > > > > > + > > > > > + device_lock(dev); > > > > > + len = sprintf(buf, "%s\n", ddev->driver_override); > > > > len = snprintf(buf, PAGE_SIZE ... > > > > > > It is good to me. > > > > > > Some bus drivers use snprintf, some use sprintf. > > > > > > I think it is reasonable snprintf is used here, unlike %d, %u ... it is > > > uncertain for the output size of %s. > > > > Sorry, I checked the Documentation/filesystems/sysfs.rst again and found > > I didn't remember it correctly. > > > > The snprintf is must not be used, sprintf() or scnprintf() should be > > used. So I prefer sprintf() here, following other implementations. > > > > sprintf() should be safe here, as we already limited the input of > > driver_override string to less then PAGE_SIZE on store(). > > As already mentioned, please use sysfs_emit() instead. Yes, I'll do it. Thanks, Yilun > > thanks, > > greg k-h
> >>> int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) > >>> { > >>> - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) > >>> + if (!dfl_drv || !dfl_drv->probe) > >> id_table is still needed for the normal case. > >> > >> Instead of removing this check, could you add something like > >> > >> || (!dfl_drv->is_override && !dfl_drv->id_table) > > I don't think it is needed. Seems is_override and !id_table are duplicated > > conditions for this implementation. And it may make confusing, e.g. could > > a driver been force matched when is_override is not set? > > > > I think we could make it simple, if the dfl driver didn't provide the > > id_table, normally it could not match any device. I think it could be > > easily understood by dfl driver developers. > > > Then an ASSERT should be added in dfl_bus_match() for id_entry. I didn't get your idea. What Assertion should be added for id_entry? BUG_ON(id_entry == NULL)? Then dfl-uio-pdev can't be inserted before driver_override is set. I think it is normal case that a driver is successfully registered but doesn't match any device because it provides no id_table. Thanks, Yilun.
On Tue, Oct 20, 2020 at 03:11:58PM +0800, Xu Yilun wrote: > I think it is normal case that a driver is successfully registered but > doesn't match any device because it provides no id_table. How is that "normal"? What would ever cause that driver to be bound to a device then? And you better not say userspace is responsible for it... thanks, greg k-h
On Tue, Oct 20, 2020 at 09:32:59AM +0200, Greg KH wrote: > On Tue, Oct 20, 2020 at 03:11:58PM +0800, Xu Yilun wrote: > > I think it is normal case that a driver is successfully registered but > > doesn't match any device because it provides no id_table. > > How is that "normal"? What would ever cause that driver to be bound to > a device then? This patchset adds the driver_override sysfs node, to let userspace specify which driver (by name matching) to be bound to a device. The driver "dfl-uio-pdev" doesn't provide an id_table, it could only be bound to a device whose driver_override is set to "dfl-uio-pdev". Sorry, this is actually not "normal". Usually dfl drivers with valid id_table should be used to drive the dfl devices. But we also want to give an option for userspace to take full control of the device, some IP blocks in FPGA are written for specific purposes by FPGA user so a userspace driver may serve them better. > > And you better not say userspace is responsible for it... Actually it is the userspace's decision which device they want to access directly ... I'm not sure if this idea is OK. I see similar implementation for pci/platform devices. Thanks, Yilun > > thanks, > > greg k-h
On Tue, Oct 20, 2020 at 04:57:23PM +0800, Xu Yilun wrote: > On Tue, Oct 20, 2020 at 09:32:59AM +0200, Greg KH wrote: > > On Tue, Oct 20, 2020 at 03:11:58PM +0800, Xu Yilun wrote: > > > I think it is normal case that a driver is successfully registered but > > > doesn't match any device because it provides no id_table. > > > > How is that "normal"? What would ever cause that driver to be bound to > > a device then? > > This patchset adds the driver_override sysfs node, to let userspace > specify which driver (by name matching) to be bound to a device. The > driver "dfl-uio-pdev" doesn't provide an id_table, it could only be > bound to a device whose driver_override is set to "dfl-uio-pdev". > > Sorry, this is actually not "normal". Usually dfl drivers with valid id_table > should be used to drive the dfl devices. But we also want to give an option > for userspace to take full control of the device, some IP blocks in FPGA are > written for specific purposes by FPGA user so a userspace driver may > serve them better. > > > > > And you better not say userspace is responsible for it... > > Actually it is the userspace's decision which device they want to > access directly ... > > I'm not sure if this idea is OK. I see similar implementation for > pci/platform devices. That is there for debugging and for being able to support things when the kernel is not updated with a new device id yet. The virtio people took this to a new extreme and use it to bind real devices through to virtual machines, but really, that's horrid. And they have problems with it as is seen in random patches at times. Don't make this the only way to support this, that's not an ok api. greg k-h
On 10/20/20 12:11 AM, Xu Yilun wrote: >>>>> int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) >>>>> { >>>>> - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) >>>>> + if (!dfl_drv || !dfl_drv->probe) >>>> id_table is still needed for the normal case. >>>> >>>> Instead of removing this check, could you add something like >>>> >>>> || (!dfl_drv->is_override && !dfl_drv->id_table) >>> I don't think it is needed. Seems is_override and !id_table are duplicated >>> conditions for this implementation. And it may make confusing, e.g. could >>> a driver been force matched when is_override is not set? >>> >>> I think we could make it simple, if the dfl driver didn't provide the >>> id_table, normally it could not match any device. I think it could be >>> easily understood by dfl driver developers. >>> >> Then an ASSERT should be added in dfl_bus_match() for id_entry. > I didn't get your idea. What Assertion should be added for id_entry? > BUG_ON(id_entry == NULL)? Then dfl-uio-pdev can't be inserted before > driver_override is set. > > I think it is normal case that a driver is successfully registered but > doesn't match any device because it provides no id_table. Ok, i see your point. Thanks for the explanation. Tom > > Thanks, > Yilun. >
Hi Greg, On Tue, Oct 20, 2020 at 11:21:55AM +0200, Greg KH wrote: > On Tue, Oct 20, 2020 at 04:57:23PM +0800, Xu Yilun wrote: > > On Tue, Oct 20, 2020 at 09:32:59AM +0200, Greg KH wrote: > > > On Tue, Oct 20, 2020 at 03:11:58PM +0800, Xu Yilun wrote: > > > > I think it is normal case that a driver is successfully registered but > > > > doesn't match any device because it provides no id_table. > > > > > > How is that "normal"? What would ever cause that driver to be bound to > > > a device then? > > > > This patchset adds the driver_override sysfs node, to let userspace > > specify which driver (by name matching) to be bound to a device. The > > driver "dfl-uio-pdev" doesn't provide an id_table, it could only be > > bound to a device whose driver_override is set to "dfl-uio-pdev". > > > > Sorry, this is actually not "normal". Usually dfl drivers with valid id_table > > should be used to drive the dfl devices. But we also want to give an option > > for userspace to take full control of the device, some IP blocks in FPGA are > > written for specific purposes by FPGA user so a userspace driver may > > serve them better. > > > > > > > > And you better not say userspace is responsible for it... > > > > Actually it is the userspace's decision which device they want to > > access directly ... > > > > I'm not sure if this idea is OK. I see similar implementation for > > pci/platform devices. > > That is there for debugging and for being able to support things when > the kernel is not updated with a new device id yet. We are trying to use the driver_override for similar purposes on FPGA. The FPGA card could be reprogramed with a newly developed Hardware component by users. Sometimes only the FPGA users know which dfl device they want to operate on. This is why we give the option to users for direct I/O access. Also FPGA users have the requirement for diagnostic of the newly developed components. It is possible we define an id_table for dfl_uio_pdev driver, but then users have to modify the driver code everytime they create & reprogram a new HW component. > > The virtio people took this to a new extreme and use it to bind real > devices through to virtual machines, but really, that's horrid. And > they have problems with it as is seen in random patches at times. Sorry I didn't have much knowledge on this. Does the sysfs 'driver_override' or 'new_id' have problems, or the userspace direct I/O access mechanism (UIO/VFIO) has problems? If it is the former, I could consider some other way to generate the UIO devices for DFL. But could you help specify more about the issues? Thanks very much, Yilun > > Don't make this the only way to support this, that's not an ok api. > > greg k-h
Hi Greg, We created a 'driver_override' api for dfl bus, and a driver for userspace I/O access for dfl devices. Some new comments inline. On Wed, Oct 21, 2020 at 03:25:32PM +0800, Xu Yilun wrote: > Hi Greg, > > On Tue, Oct 20, 2020 at 11:21:55AM +0200, Greg KH wrote: > > On Tue, Oct 20, 2020 at 04:57:23PM +0800, Xu Yilun wrote: > > > On Tue, Oct 20, 2020 at 09:32:59AM +0200, Greg KH wrote: > > > > On Tue, Oct 20, 2020 at 03:11:58PM +0800, Xu Yilun wrote: > > > > > I think it is normal case that a driver is successfully registered but > > > > > doesn't match any device because it provides no id_table. > > > > > > > > How is that "normal"? What would ever cause that driver to be bound to > > > > a device then? > > > > > > This patchset adds the driver_override sysfs node, to let userspace > > > specify which driver (by name matching) to be bound to a device. The > > > driver "dfl-uio-pdev" doesn't provide an id_table, it could only be > > > bound to a device whose driver_override is set to "dfl-uio-pdev". > > > > > > Sorry, this is actually not "normal". Usually dfl drivers with valid id_table > > > should be used to drive the dfl devices. But we also want to give an option > > > for userspace to take full control of the device, some IP blocks in FPGA are > > > written for specific purposes by FPGA user so a userspace driver may > > > serve them better. > > > > > > > > > > > And you better not say userspace is responsible for it... > > > > > > Actually it is the userspace's decision which device they want to > > > access directly ... > > > > > > I'm not sure if this idea is OK. I see similar implementation for > > > pci/platform devices. > > > > That is there for debugging and for being able to support things when > > the kernel is not updated with a new device id yet. > > We are trying to use the driver_override for similar purposes on FPGA. > The FPGA card could be reprogramed with a newly developed Hardware > component by users. Sometimes only the FPGA users know which dfl device > they want to operate on. This is why we give the option to users for > direct I/O access. > > Also FPGA users have the requirement for diagnostic of the newly > developed components. > > It is possible we define an id_table for dfl_uio_pdev driver, but then > users have to modify the driver code everytime they create & reprogram > a new HW component. > > > > > The virtio people took this to a new extreme and use it to bind real > > devices through to virtual machines, but really, that's horrid. And > > they have problems with it as is seen in random patches at times. I searched the previous patches, it seems the main issue is the user could unbind the vfio-pci for a device and bind it to another kernel driver, while its iommu group is still being used by vfio for other devices in the same group, then the integrity of the group is compromised. The conclusion is the user should be responsible for the correct operation of bind/unbind/driver_override/new_id ... In this patchset, we use the driver_override for UIO case, I think the issue doesn't impact it, is it? > > > > Don't make this the only way to support this, that's not an ok api. We are not going to match driver for every dfl device by user assignment. We are developing generic kernel drivers for them, e.g. dfl-n3000-nios, dfl-emif, as you can see in recent patches. They follows the generic id_table matching. We use the driver_override for UIO case. Similar to what you've mentioned, FPGA users develop their own hardware components on FPGA, so they need to run diagnostic tools for white-box test. They also need to support some hardware designed for special purposes and does not fit into one of the standard kernel subsystems. The FPGA users are the developer of some hardware component, they have the knowledge of the new dfl devices. So the driver_override give the option to the users for driver assignment. For these purposes, is it OK we use the driver_override? Thanks, Yilun > > > > greg k-h
diff --git a/Documentation/ABI/testing/sysfs-bus-dfl b/Documentation/ABI/testing/sysfs-bus-dfl index 23543be..db7e8d3 100644 --- a/Documentation/ABI/testing/sysfs-bus-dfl +++ b/Documentation/ABI/testing/sysfs-bus-dfl @@ -1,15 +1,35 @@ What: /sys/bus/dfl/devices/dfl_dev.X/type -Date: Aug 2020 -KernelVersion: 5.10 +Date: Oct 2020 +KernelVersion: 5.11 Contact: Xu Yilun <yilun.xu@intel.com> Description: Read-only. It returns type of DFL FIU of the device. Now DFL supports 2 FIU types, 0 for FME, 1 for PORT. Format: 0x%x What: /sys/bus/dfl/devices/dfl_dev.X/feature_id -Date: Aug 2020 -KernelVersion: 5.10 +Date: Oct 2020 +KernelVersion: 5.11 Contact: Xu Yilun <yilun.xu@intel.com> Description: Read-only. It returns feature identifier local to its DFL FIU type. Format: 0x%x + +What: /sys/bus/dfl/devices/.../driver_override +Date: Oct 2020 +KernelVersion: 5.11 +Contact: Xu Yilun <yilun.xu@intel.com> +Description: This file allows the driver for a device to be specified. When + specified, only a driver with a name matching the value written + to driver_override will have an opportunity to bind to the + device. The override is specified by writing a string to the + driver_override file (echo dfl-uio-pdev > driver_override) and + may be cleared with an empty string (echo > driver_override). + This returns the device to standard matching rules binding. + Writing to driver_override does not automatically unbind the + device from its current driver or make any attempt to + automatically load the specified driver. If no driver with a + matching name is currently loaded in the kernel, the device + will not bind to any driver. This also allows devices to + opt-out of driver binding using a driver_override name such as + "none". Only a single driver may be specified in the override, + there is no support for parsing delimiters. diff --git a/drivers/fpga/dfl.c b/drivers/fpga/dfl.c index 511b20f..bc35750 100644 --- a/drivers/fpga/dfl.c +++ b/drivers/fpga/dfl.c @@ -262,6 +262,10 @@ static int dfl_bus_match(struct device *dev, struct device_driver *drv) struct dfl_driver *ddrv = to_dfl_drv(drv); const struct dfl_device_id *id_entry; + /* When driver_override is set, only bind to the matching driver */ + if (ddev->driver_override) + return !strcmp(ddev->driver_override, drv->name); + id_entry = ddrv->id_table; if (id_entry) { while (id_entry->feature_id) { @@ -303,6 +307,53 @@ static int dfl_bus_uevent(struct device *dev, struct kobj_uevent_env *env) ddev->type, ddev->feature_id); } +static ssize_t driver_override_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct dfl_device *ddev = to_dfl_dev(dev); + ssize_t len; + + device_lock(dev); + len = sprintf(buf, "%s\n", ddev->driver_override); + device_unlock(dev); + return len; +} + +static ssize_t driver_override_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct dfl_device *ddev = to_dfl_dev(dev); + char *driver_override, *old, *cp; + + /* We need to keep extra room for a newline */ + if (count >= (PAGE_SIZE - 1)) + return -EINVAL; + + driver_override = kstrndup(buf, count, GFP_KERNEL); + if (!driver_override) + return -ENOMEM; + + cp = strchr(driver_override, '\n'); + if (cp) + *cp = '\0'; + + device_lock(dev); + old = ddev->driver_override; + if (strlen(driver_override)) { + ddev->driver_override = driver_override; + } else { + kfree(driver_override); + ddev->driver_override = NULL; + } + device_unlock(dev); + + kfree(old); + + return count; +} +static DEVICE_ATTR_RW(driver_override); + static ssize_t type_show(struct device *dev, struct device_attribute *attr, char *buf) { @@ -324,6 +375,7 @@ static DEVICE_ATTR_RO(feature_id); static struct attribute *dfl_dev_attrs[] = { &dev_attr_type.attr, &dev_attr_feature_id.attr, + &dev_attr_driver_override.attr, NULL, }; ATTRIBUTE_GROUPS(dfl_dev); @@ -469,7 +521,7 @@ static int dfl_devs_add(struct dfl_feature_platform_data *pdata) int __dfl_driver_register(struct dfl_driver *dfl_drv, struct module *owner) { - if (!dfl_drv || !dfl_drv->probe || !dfl_drv->id_table) + if (!dfl_drv || !dfl_drv->probe) return -EINVAL; dfl_drv->drv.owner = owner; diff --git a/include/linux/dfl.h b/include/linux/dfl.h index 7affba2f..e1b2471 100644 --- a/include/linux/dfl.h +++ b/include/linux/dfl.h @@ -32,6 +32,7 @@ enum dfl_id_type { * @num_irqs: number of IRQs supported by this dfl device. * @cdev: pointer to DFL FPGA container device this dfl device belongs to. * @id_entry: matched id entry in dfl driver's id table. + * @driver_override: driver name to force a match */ struct dfl_device { struct device dev; @@ -43,6 +44,7 @@ struct dfl_device { unsigned int num_irqs; struct dfl_fpga_cdev *cdev; const struct dfl_device_id *id_entry; + char *driver_override; }; /**
Add support for overriding the default matching of a dfl device to a dfl driver. It follows the same way that can be used for PCI and platform devices. This patch adds the 'driver_override' sysfs file. Signed-off-by: Xu Yilun <yilun.xu@intel.com> --- Documentation/ABI/testing/sysfs-bus-dfl | 28 ++++++++++++++--- drivers/fpga/dfl.c | 54 ++++++++++++++++++++++++++++++++- include/linux/dfl.h | 2 ++ 3 files changed, 79 insertions(+), 5 deletions(-)