On Mon, Mar 6, 2017 at 9:41 PM, David Gibson
[off-list ref] wrote:

On Mon, Mar 06, 2017 at 04:48:18AM -0600, Rob Herring wrote:

quoted

On Thu, Mar 2, 2017 at 8:12 PM, David Gibson
[off-list ref] wrote:

quoted

On Tue, Feb 28, 2017 at 04:43:09PM -0600, Rob Herring wrote:

quoted

Add checks to identify simple-bus bus types and checks for child
devices. Simple-bus type is generally identified by "simple-bus"
compatible string. We also treat the root as a simple-bus, but only for
child nodes with reg property.

Signed-off-by: Rob Herring <redacted>
---
v2:
- new patch

 checks.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)

diff --git a/checks.c b/checks.c
index 5ed91ac50a10..c4865b4c8da0 100644
--- a/checks.c
+++ b/checks.c

@@ -817,6 +817,72 @@ static void check_pci_device_reg(struct check *c, struct dt_info *dti, struct no
 }
 WARNING(pci_device_reg, check_pci_device_reg, NULL, &reg_format);

+static const struct bus_type simple_bus = {
+     .name = "simple-bus",
+};
+
+static bool node_is_compatible(struct node *node, const char *compat)
+{
+     struct property *prop;
+     const char *str;
+
+     prop = get_property(node, "compatible");
+     if (!prop)
+             return false;
+
+     for (str = prop->val.val; str < prop->val.val + prop->val.len; str += strlen(str) + 1) {

This isn't safe if the compatible property is filled with garbage (not
'\0' terminated) - the strlen() could access beyond the end of the
property value.

Okay, I guess I can check that prop->val.val[prop->val.len - 1] == 0 up front.

Sure.  Or use strnlen.

Duh...

quoted

quoted

quoted

+             if (streq(str, compat))
+                     return true;
+     }
+     return false;
+}
+
+static void check_simple_bus_bridge(struct check *c, struct dt_info *dti, struct node *node)
+{
+     if (node_is_compatible(node, "simple-bus") || !node->parent)
+             node->bus = &simple_bus;

I don't think it's correct to assume the root bus is always a
simple-bus.  If it is, it really should be listed explicitly in the
root node's compatible property.

It is in the sense that Linux treats the root the same and creates
devices for top level children and then descends for nodes with
"simple-bus".

Hmm.. where in Linux is that?  I think that's a bug, technically
speaking, traversing the root node's children without regard to the
type of the root node.

drivers/of/platform.c:of_platform_populate() which is called on the
root node with "simple-bus" in the match table.

Plus I know we have some DT's like Tegra that didn't put all their
devices under a bus (but should have). Maybe I should warn on that
(i.e. warn on having unit-addresses without a bus type set on root).

Rob
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html