Message ID | 1291978340-15933-1-git-send-email-b25806@freescale.com (mailing list archive) |
---|---|
State | Not Applicable |
Delegated to: | Grant Likely |
Headers | show |
On Fri, Dec 10, 2010 at 06:52:20PM +0800, Lan Chunhe wrote: > Currently, of_platform_bus_probe() completely skips nodes which do not > explicitly match the 'matches' table passed in. Or, if the root node > matches, then it registers all the children unconditionally. However, > there are situations, such as registering devices from the root node, > when it is desirable to register child nodes, but only if they actually > represent devices. For example, the root node may contain both a local > bus and a PCI device, but it also contains the chosen, aliases and cpus > nodes which don't represent real devices. > > This patch changes of_platform_bus_probe() to register all nodes at the > top level if they either match the matches table (the current behaviour), > or if they have a 'compatible' value (indicating it represents a device). > > Signed-off-by: Lan Chunhe <b25806@freescale.com> I believe this is my patch I wrote and pushed out to my tree back in October. Was this cherry-picked out of the test-devicetree branch? > --- > drivers/of/platform.c | 28 +++++++++++++++++++++++----- > 1 files changed, 23 insertions(+), 5 deletions(-) > > diff --git a/drivers/of/platform.c b/drivers/of/platform.c > index 5b4a07f..02755ab 100644 > --- a/drivers/of/platform.c > +++ b/drivers/of/platform.c > @@ -714,6 +714,8 @@ int of_platform_bus_probe(struct device_node *root, > struct device_node *child; > struct platform_device *dev; > int rc = 0; > + const void *compat; > + const struct of_device_id *match; > > if (WARN_ON(!matches || matches == OF_NO_DEEP_PROBE)) > return -EINVAL; > @@ -741,16 +743,32 @@ int of_platform_bus_probe(struct device_node *root, > rc = of_platform_bus_create(root, matches, &dev->dev); > goto bail; > } > + > + /* > + * Register each child node if either: > + * a) it has a 'compatible' value indicating they are a device, or > + * b) it is specified by the 'matches' table (by name or device_type) > + * If a node is specified in the matches table, then all its children > + * also get registered. > + */ > for_each_child_of_node(root, child) { > - if (!of_match_node(matches, child)) > + compat = of_get_property(child, "compatible", NULL); > + match = of_match_node(matches, child); > + if (!compat && !match) > continue; > > - pr_debug(" match: %s\n", child->full_name); > + pr_debug(" register device: %s\n", child->full_name); > dev = of_platform_device_create(child, NULL, parent); > - if (dev == NULL) > + if (!dev) { > rc = -ENOMEM; > - else > - rc = of_platform_bus_create(child, matches, &dev->dev); > + of_node_put(child); > + break; > + } > + if (!match) > + continue; > + > + pr_debug(" register children of: %s\n", child->full_name); > + rc = of_platform_bus_create(child, matches, &dev->dev); > if (rc) { > of_node_put(child); > break; > -- > 1.5.4.5 > >
On Thu, 30 Dec 2010 14:50:45 +0800, Grant Likely <grant.likely@secretlab.ca> wrote: > On Fri, Dec 10, 2010 at 06:52:20PM +0800, Lan Chunhe wrote: >> Currently, of_platform_bus_probe() completely skips nodes which do not >> explicitly match the 'matches' table passed in. Or, if the root node >> matches, then it registers all the children unconditionally. However, >> there are situations, such as registering devices from the root node, >> when it is desirable to register child nodes, but only if they actually >> represent devices. For example, the root node may contain both a local >> bus and a PCI device, but it also contains the chosen, aliases and cpus >> nodes which don't represent real devices. >> >> This patch changes of_platform_bus_probe() to register all nodes at the >> top level if they either match the matches table (the current >> behaviour), >> or if they have a 'compatible' value (indicating it represents a >> device). >> >> Signed-off-by: Lan Chunhe <b25806@freescale.com> > > I believe this is my patch I wrote and pushed out to my tree back in > October. Was this cherry-picked out of the test-devicetree branch? Yes, I will add the signed off with you. Signed-off-by: Grant Likely <grant.likely@secretlab.ca> >> Signed-off-by: Lan Chunhe <b25806@freescale.com> At present I think that it is cherry-picked, and I have tested it which is OK. Do you have better method? Thanks. -Lan >> --- >> drivers/of/platform.c | 28 +++++++++++++++++++++++----- >> 1 files changed, 23 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/of/platform.c b/drivers/of/platform.c >> index 5b4a07f..02755ab 100644 >> --- a/drivers/of/platform.c >> +++ b/drivers/of/platform.c >> @@ -714,6 +714,8 @@ int of_platform_bus_probe(struct device_node *root, >> struct device_node *child; >> struct platform_device *dev; >> int rc = 0; >> + const void *compat; >> + const struct of_device_id *match; >> >> if (WARN_ON(!matches || matches == OF_NO_DEEP_PROBE)) >> return -EINVAL; >> @@ -741,16 +743,32 @@ int of_platform_bus_probe(struct device_node >> *root, >> rc = of_platform_bus_create(root, matches, &dev->dev); >> goto bail; >> } >> + >> + /* >> + * Register each child node if either: >> + * a) it has a 'compatible' value indicating they are a device, or >> + * b) it is specified by the 'matches' table (by name or device_type) >> + * If a node is specified in the matches table, then all its children >> + * also get registered. >> + */ >> for_each_child_of_node(root, child) { >> - if (!of_match_node(matches, child)) >> + compat = of_get_property(child, "compatible", NULL); >> + match = of_match_node(matches, child); >> + if (!compat && !match) >> continue; >> >> - pr_debug(" match: %s\n", child->full_name); >> + pr_debug(" register device: %s\n", child->full_name); >> dev = of_platform_device_create(child, NULL, parent); >> - if (dev == NULL) >> + if (!dev) { >> rc = -ENOMEM; >> - else >> - rc = of_platform_bus_create(child, matches, &dev->dev); >> + of_node_put(child); >> + break; >> + } >> + if (!match) >> + continue; >> + >> + pr_debug(" register children of: %s\n", child->full_name); >> + rc = of_platform_bus_create(child, matches, &dev->dev); >> if (rc) { >> of_node_put(child); >> break; >> -- >> 1.5.4.5 >> >> >
On Thu, Dec 30, 2010 at 03:23:11PM +0800, Lan Chunhe wrote: > On Thu, 30 Dec 2010 14:50:45 +0800, Grant Likely > <grant.likely@secretlab.ca> wrote: > > >On Fri, Dec 10, 2010 at 06:52:20PM +0800, Lan Chunhe wrote: > >>Currently, of_platform_bus_probe() completely skips nodes which do not > >>explicitly match the 'matches' table passed in. Or, if the root node > >>matches, then it registers all the children unconditionally. However, > >>there are situations, such as registering devices from the root node, > >>when it is desirable to register child nodes, but only if they actually > >>represent devices. For example, the root node may contain both a local > >>bus and a PCI device, but it also contains the chosen, aliases and cpus > >>nodes which don't represent real devices. > >> > >>This patch changes of_platform_bus_probe() to register all nodes at the > >>top level if they either match the matches table (the current > >>behaviour), > >>or if they have a 'compatible' value (indicating it represents a > >>device). > >> > >>Signed-off-by: Lan Chunhe <b25806@freescale.com> > > > >I believe this is my patch I wrote and pushed out to my tree back in > >October. Was this cherry-picked out of the test-devicetree branch? > > Yes, I will add the signed off with you. > > Signed-off-by: Grant Likely <grant.likely@secretlab.ca> > >>Signed-off-by: Lan Chunhe <b25806@freescale.com> > > At present I think that it is cherry-picked, and I have tested it which > is OK. > Do you have better method? You need to be very careful about stuff like this. Posting a patch without attribution of the original author and has the original signed-off-by lines stripped off is an absolute no-no. In literary circles this would be called plagiarism. I don't think you did it intentionally, but proper attribution is a must, particularly because there are copyright issues involved. As for this particular patch, I'm not particularly happy with it and I was planning to rewrite it. It is an inelegant solution and I think the registration of devices from the device tree can be made simpler instead of more complex like in this patch. g. > > Thanks. > -Lan > > >>--- > >> drivers/of/platform.c | 28 +++++++++++++++++++++++----- > >> 1 files changed, 23 insertions(+), 5 deletions(-) > >> > >>diff --git a/drivers/of/platform.c b/drivers/of/platform.c > >>index 5b4a07f..02755ab 100644 > >>--- a/drivers/of/platform.c > >>+++ b/drivers/of/platform.c > >>@@ -714,6 +714,8 @@ int of_platform_bus_probe(struct device_node *root, > >> struct device_node *child; > >> struct platform_device *dev; > >> int rc = 0; > >>+ const void *compat; > >>+ const struct of_device_id *match; > >> > >> if (WARN_ON(!matches || matches == OF_NO_DEEP_PROBE)) > >> return -EINVAL; > >>@@ -741,16 +743,32 @@ int of_platform_bus_probe(struct > >>device_node *root, > >> rc = of_platform_bus_create(root, matches, &dev->dev); > >> goto bail; > >> } > >>+ > >>+ /* > >>+ * Register each child node if either: > >>+ * a) it has a 'compatible' value indicating they are a device, or > >>+ * b) it is specified by the 'matches' table (by name or device_type) > >>+ * If a node is specified in the matches table, then all its children > >>+ * also get registered. > >>+ */ > >> for_each_child_of_node(root, child) { > >>- if (!of_match_node(matches, child)) > >>+ compat = of_get_property(child, "compatible", NULL); > >>+ match = of_match_node(matches, child); > >>+ if (!compat && !match) > >> continue; > >> > >>- pr_debug(" match: %s\n", child->full_name); > >>+ pr_debug(" register device: %s\n", child->full_name); > >> dev = of_platform_device_create(child, NULL, parent); > >>- if (dev == NULL) > >>+ if (!dev) { > >> rc = -ENOMEM; > >>- else > >>- rc = of_platform_bus_create(child, matches, &dev->dev); > >>+ of_node_put(child); > >>+ break; > >>+ } > >>+ if (!match) > >>+ continue; > >>+ > >>+ pr_debug(" register children of: %s\n", child->full_name); > >>+ rc = of_platform_bus_create(child, matches, &dev->dev); > >> if (rc) { > >> of_node_put(child); > >> break; > >>-- > >>1.5.4.5 > >> > >> > > > >
On Thu, 30 Dec 2010 15:31:57 +0800, Grant Likely <grant.likely@secretlab.ca> wrote: > On Thu, Dec 30, 2010 at 03:23:11PM +0800, Lan Chunhe wrote: >> On Thu, 30 Dec 2010 14:50:45 +0800, Grant Likely >> <grant.likely@secretlab.ca> wrote: >> >> >On Fri, Dec 10, 2010 at 06:52:20PM +0800, Lan Chunhe wrote: >> >>Currently, of_platform_bus_probe() completely skips nodes which do not >> >>explicitly match the 'matches' table passed in. Or, if the root node >> >>matches, then it registers all the children unconditionally. However, >> >>there are situations, such as registering devices from the root node, >> >>when it is desirable to register child nodes, but only if they >> actually >> >>represent devices. For example, the root node may contain both a >> local >> >>bus and a PCI device, but it also contains the chosen, aliases and >> cpus >> >>nodes which don't represent real devices. >> >> >> >>This patch changes of_platform_bus_probe() to register all nodes at >> the >> >>top level if they either match the matches table (the current >> >>behaviour), >> >>or if they have a 'compatible' value (indicating it represents a >> >>device). >> >> >> >>Signed-off-by: Lan Chunhe <b25806@freescale.com> >> > >> >I believe this is my patch I wrote and pushed out to my tree back in >> >October. Was this cherry-picked out of the test-devicetree branch? >> >> Yes, I will add the signed off with you. >> >> Signed-off-by: Grant Likely <grant.likely@secretlab.ca> >> >>Signed-off-by: Lan Chunhe <b25806@freescale.com> >> >> At present I think that it is cherry-picked, and I have tested it >> which >> is OK. >> Do you have better method? > > You need to be very careful about stuff like this. Posting a patch > without attribution of the original author and has the original > signed-off-by lines stripped off is an absolute no-no. In literary > circles this would be called plagiarism. > > I don't think you did it intentionally, but proper attribution is a > must, particularly because there are copyright issues involved. Sorry, thank you that you remind me. > As for this particular patch, I'm not particularly happy with it and I > was planning to rewrite it. It is an inelegant solution and I think > the registration of devices from the device tree can be made simpler > instead of more complex like in this patch. Now this patch registers devices from the device tree. Then which function of which file do you will modify? Thanks. -Lan > g. > >> >> Thanks. >> -Lan >> >> >>--- >> >> drivers/of/platform.c | 28 +++++++++++++++++++++++----- >> >> 1 files changed, 23 insertions(+), 5 deletions(-) >> >> >> >>diff --git a/drivers/of/platform.c b/drivers/of/platform.c >> >>index 5b4a07f..02755ab 100644 >> >>--- a/drivers/of/platform.c >> >>+++ b/drivers/of/platform.c >> >>@@ -714,6 +714,8 @@ int of_platform_bus_probe(struct device_node >> *root, >> >> struct device_node *child; >> >> struct platform_device *dev; >> >> int rc = 0; >> >>+ const void *compat; >> >>+ const struct of_device_id *match; >> >> >> >> if (WARN_ON(!matches || matches == OF_NO_DEEP_PROBE)) >> >> return -EINVAL; >> >>@@ -741,16 +743,32 @@ int of_platform_bus_probe(struct >> >>device_node *root, >> >> rc = of_platform_bus_create(root, matches, &dev->dev); >> >> goto bail; >> >> } >> >>+ >> >>+ /* >> >>+ * Register each child node if either: >> >>+ * a) it has a 'compatible' value indicating they are a device, or >> >>+ * b) it is specified by the 'matches' table (by name or >> device_type) >> >>+ * If a node is specified in the matches table, then all its >> children >> >>+ * also get registered. >> >>+ */ >> >> for_each_child_of_node(root, child) { >> >>- if (!of_match_node(matches, child)) >> >>+ compat = of_get_property(child, "compatible", NULL); >> >>+ match = of_match_node(matches, child); >> >>+ if (!compat && !match) >> >> continue; >> >> >> >>- pr_debug(" match: %s\n", child->full_name); >> >>+ pr_debug(" register device: %s\n", child->full_name); >> >> dev = of_platform_device_create(child, NULL, parent); >> >>- if (dev == NULL) >> >>+ if (!dev) { >> >> rc = -ENOMEM; >> >>- else >> >>- rc = of_platform_bus_create(child, matches, &dev->dev); >> >>+ of_node_put(child); >> >>+ break; >> >>+ } >> >>+ if (!match) >> >>+ continue; >> >>+ >> >>+ pr_debug(" register children of: %s\n", child->full_name); >> >>+ rc = of_platform_bus_create(child, matches, &dev->dev); >> >> if (rc) { >> >> of_node_put(child); >> >> break; >> >>-- >> >>1.5.4.5 >> >> >> >> >> > >> >> >
On Thu, Dec 30, 2010 at 04:04:10PM +0800, Lan Chunhe wrote: > On Thu, 30 Dec 2010 15:31:57 +0800, Grant Likely > <grant.likely@secretlab.ca> wrote: > > >On Thu, Dec 30, 2010 at 03:23:11PM +0800, Lan Chunhe wrote: > >>On Thu, 30 Dec 2010 14:50:45 +0800, Grant Likely > >><grant.likely@secretlab.ca> wrote: > >> > >>>On Fri, Dec 10, 2010 at 06:52:20PM +0800, Lan Chunhe wrote: > >>>>Currently, of_platform_bus_probe() completely skips nodes which do not > >>>>explicitly match the 'matches' table passed in. Or, if the root node > >>>>matches, then it registers all the children unconditionally. However, > >>>>there are situations, such as registering devices from the root node, > >>>>when it is desirable to register child nodes, but only if they > >>actually > >>>>represent devices. For example, the root node may contain > >>both a local > >>>>bus and a PCI device, but it also contains the chosen, aliases > >>and cpus > >>>>nodes which don't represent real devices. > >>>> > >>>>This patch changes of_platform_bus_probe() to register all > >>nodes at the > >>>>top level if they either match the matches table (the current > >>>>behaviour), > >>>>or if they have a 'compatible' value (indicating it represents a > >>>>device). > >>>> > >>>>Signed-off-by: Lan Chunhe <b25806@freescale.com> > >>> > >>>I believe this is my patch I wrote and pushed out to my tree back in > >>>October. Was this cherry-picked out of the test-devicetree branch? > >> > >> Yes, I will add the signed off with you. > >> > >> Signed-off-by: Grant Likely <grant.likely@secretlab.ca> > >>>>Signed-off-by: Lan Chunhe <b25806@freescale.com> > >> > >> At present I think that it is cherry-picked, and I have > >>tested it which > >> is OK. > >> Do you have better method? > > > >You need to be very careful about stuff like this. Posting a patch > >without attribution of the original author and has the original > >signed-off-by lines stripped off is an absolute no-no. In literary > >circles this would be called plagiarism. > > > >I don't think you did it intentionally, but proper attribution is a > >must, particularly because there are copyright issues involved. > > Sorry, thank you that you remind me. > > >As for this particular patch, I'm not particularly happy with it and I > >was planning to rewrite it. It is an inelegant solution and I think > >the registration of devices from the device tree can be made simpler > >instead of more complex like in this patch. > > Now this patch registers devices from the device tree. > Then which function of which file do you will modify? I'm thinking about deprecating the whole recursive device registration entirely and instead making child device registration always the responsibility of a device driver. That would mean each bus node would bind to a simple device driver that registers the child platform devices in the .probe() hook. g.
diff --git a/drivers/of/platform.c b/drivers/of/platform.c index 5b4a07f..02755ab 100644 --- a/drivers/of/platform.c +++ b/drivers/of/platform.c @@ -714,6 +714,8 @@ int of_platform_bus_probe(struct device_node *root, struct device_node *child; struct platform_device *dev; int rc = 0; + const void *compat; + const struct of_device_id *match; if (WARN_ON(!matches || matches == OF_NO_DEEP_PROBE)) return -EINVAL; @@ -741,16 +743,32 @@ int of_platform_bus_probe(struct device_node *root, rc = of_platform_bus_create(root, matches, &dev->dev); goto bail; } + + /* + * Register each child node if either: + * a) it has a 'compatible' value indicating they are a device, or + * b) it is specified by the 'matches' table (by name or device_type) + * If a node is specified in the matches table, then all its children + * also get registered. + */ for_each_child_of_node(root, child) { - if (!of_match_node(matches, child)) + compat = of_get_property(child, "compatible", NULL); + match = of_match_node(matches, child); + if (!compat && !match) continue; - pr_debug(" match: %s\n", child->full_name); + pr_debug(" register device: %s\n", child->full_name); dev = of_platform_device_create(child, NULL, parent); - if (dev == NULL) + if (!dev) { rc = -ENOMEM; - else - rc = of_platform_bus_create(child, matches, &dev->dev); + of_node_put(child); + break; + } + if (!match) + continue; + + pr_debug(" register children of: %s\n", child->full_name); + rc = of_platform_bus_create(child, matches, &dev->dev); if (rc) { of_node_put(child); break;
Currently, of_platform_bus_probe() completely skips nodes which do not explicitly match the 'matches' table passed in. Or, if the root node matches, then it registers all the children unconditionally. However, there are situations, such as registering devices from the root node, when it is desirable to register child nodes, but only if they actually represent devices. For example, the root node may contain both a local bus and a PCI device, but it also contains the chosen, aliases and cpus nodes which don't represent real devices. This patch changes of_platform_bus_probe() to register all nodes at the top level if they either match the matches table (the current behaviour), or if they have a 'compatible' value (indicating it represents a device). Signed-off-by: Lan Chunhe <b25806@freescale.com> --- drivers/of/platform.c | 28 +++++++++++++++++++++++----- 1 files changed, 23 insertions(+), 5 deletions(-)