* [RFC 0/3] of: Refactor device tree linked lists
@ 2014-06-05 15:43 Grant Likely
[not found] ` <1401983021-13829-1-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
0 siblings, 1 reply; 10+ messages in thread
From: Grant Likely @ 2014-06-05 15:43 UTC (permalink / raw)
To: devicetree-u79uwXL29TY76Z2rM5mHXA; +Cc: rob.herring-QSEj5FYQhm4dnm+yROfE0A
I threw these patches together as part of getting the device_node
lifecycle issues sorted out so that the overlay patches can be merged.
It makes it possible to add and remove nodes without jumping through
crazy hoops.
I haven't yet decided if it is a good idea. It does increase the size of
deviec_node by 2 pointers which increases the footprint, but it
simplifies a lot of the code. Although, if I was really concerned about
footprint I would be trying to get rid of the ->full_name string and
just use the path component instead. The primary user of full_name is
various printk format strings around the kernel and we could get the
same behaviour using a %p variant.
--
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
^ permalink raw reply [flat|nested] 10+ messages in thread
* [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <1401983021-13829-1-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
@ 2014-06-05 15:43 ` Grant Likely
[not found] ` <1401983021-13829-2-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2014-06-05 15:43 ` [RFC 2/3] of: Replace custom linked list with list_head Grant Likely
2014-06-05 15:43 ` [RFC 3/3] of: Refactor device_node child list to use list_head Grant Likely
2 siblings, 1 reply; 10+ messages in thread
From: Grant Likely @ 2014-06-05 15:43 UTC (permalink / raw)
To: devicetree-u79uwXL29TY76Z2rM5mHXA
Cc: rob.herring-QSEj5FYQhm4dnm+yROfE0A, Grant Likely
There are a bunch of users open coding the for_each_node_by_name() by
calling of_find_node_by_name() directly instead of using the macro. This
is getting in the way of some cleanups, and the possibility of removing
of_find_node_by_name() entirely. Clean it up so that all the users are
consistent.
Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
arch/powerpc/platforms/powermac/pci.c | 2 +-
arch/powerpc/platforms/powermac/smp.c | 2 +-
arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
arch/powerpc/platforms/pseries/setup.c | 3 +--
drivers/cpufreq/pmac64-cpufreq.c | 3 +--
drivers/edac/cell_edac.c | 3 +--
drivers/pci/hotplug/rpaphp_core.c | 4 ++--
drivers/tty/serial/pmac_zilog.c | 9 +++------
sound/ppc/pmac.c | 6 +++---
10 files changed, 19 insertions(+), 35 deletions(-)
diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
index 63d82bbc05e9..39e1d163c427 100644
--- a/arch/powerpc/platforms/powermac/feature.c
+++ b/arch/powerpc/platforms/powermac/feature.c
@@ -2805,25 +2805,20 @@ set_initial_features(void)
/* Enable GMAC for now for PCI probing. It will be disabled
* later on after PCI probe
*/
- np = of_find_node_by_name(NULL, "ethernet");
- while(np) {
+ for_each_node_by_name(np, "ethernet")
if (of_device_is_compatible(np, "K2-GMAC"))
g5_gmac_enable(np, 0, 1);
- np = of_find_node_by_name(np, "ethernet");
- }
/* Enable FW before PCI probe. Will be disabled later on
* Note: We should have a batter way to check that we are
* dealing with uninorth internal cell and not a PCI cell
* on the external PCI. The code below works though.
*/
- np = of_find_node_by_name(NULL, "firewire");
- while(np) {
+ for_each_node_by_name(np, "firewire") {
if (of_device_is_compatible(np, "pci106b,5811")) {
macio_chips[0].flags |= MACIO_FLAG_FW_SUPPORTED;
g5_fw_enable(np, 0, 1);
}
- np = of_find_node_by_name(np, "firewire");
}
}
#else /* CONFIG_POWER4 */
@@ -2834,13 +2829,11 @@ set_initial_features(void)
/* Enable GMAC for now for PCI probing. It will be disabled
* later on after PCI probe
*/
- np = of_find_node_by_name(NULL, "ethernet");
- while(np) {
+ for_each_node_by_name(np, "ethernet") {
if (np->parent
&& of_device_is_compatible(np->parent, "uni-north")
&& of_device_is_compatible(np, "gmac"))
core99_gmac_enable(np, 0, 1);
- np = of_find_node_by_name(np, "ethernet");
}
/* Enable FW before PCI probe. Will be disabled later on
@@ -2848,8 +2841,7 @@ set_initial_features(void)
* dealing with uninorth internal cell and not a PCI cell
* on the external PCI. The code below works though.
*/
- np = of_find_node_by_name(NULL, "firewire");
- while(np) {
+ for_each_node_by_name(np, "firewire") {
if (np->parent
&& of_device_is_compatible(np->parent, "uni-north")
&& (of_device_is_compatible(np, "pci106b,18") ||
@@ -2858,18 +2850,16 @@ set_initial_features(void)
macio_chips[0].flags |= MACIO_FLAG_FW_SUPPORTED;
core99_firewire_enable(np, 0, 1);
}
- np = of_find_node_by_name(np, "firewire");
}
/* Enable ATA-100 before PCI probe. */
np = of_find_node_by_name(NULL, "ata-6");
- while(np) {
+ for_each_node_by_name(np, "ata-6") {
if (np->parent
&& of_device_is_compatible(np->parent, "uni-north")
&& of_device_is_compatible(np, "kauai-ata")) {
core99_ata100_enable(np, 1);
}
- np = of_find_node_by_name(np, "ata-6");
}
/* Switch airport off */
diff --git a/arch/powerpc/platforms/powermac/pci.c b/arch/powerpc/platforms/powermac/pci.c
index cf7009b8c7b6..7e868ccf3b0d 100644
--- a/arch/powerpc/platforms/powermac/pci.c
+++ b/arch/powerpc/platforms/powermac/pci.c
@@ -698,7 +698,7 @@ static void __init fixup_nec_usb2(void)
{
struct device_node *nec;
- for (nec = NULL; (nec = of_find_node_by_name(nec, "usb")) != NULL;) {
+ for_each_node_by_name(nec, "usb") {
struct pci_controller *hose;
u32 data;
const u32 *prop;
diff --git a/arch/powerpc/platforms/powermac/smp.c b/arch/powerpc/platforms/powermac/smp.c
index 5cbd4d67d5c4..af094ae03dbb 100644
--- a/arch/powerpc/platforms/powermac/smp.c
+++ b/arch/powerpc/platforms/powermac/smp.c
@@ -577,7 +577,7 @@ static void __init smp_core99_setup_i2c_hwsync(int ncpus)
int ok;
/* Look for the clock chip */
- while ((cc = of_find_node_by_name(cc, "i2c-hwclock")) != NULL) {
+ for_each_node_by_name(cc, "i2c-hwclock") {
p = of_get_parent(cc);
ok = p && of_device_is_compatible(p, "uni-n-i2c");
of_node_put(p);
diff --git a/arch/powerpc/platforms/powermac/udbg_adb.c b/arch/powerpc/platforms/powermac/udbg_adb.c
index 44e0b55a2a02..366bd221edec 100644
--- a/arch/powerpc/platforms/powermac/udbg_adb.c
+++ b/arch/powerpc/platforms/powermac/udbg_adb.c
@@ -191,7 +191,7 @@ int __init udbg_adb_init(int force_btext)
* of type "adb". If not, we return a failure, but we keep the
* bext output set for now
*/
- for (np = NULL; (np = of_find_node_by_name(np, "keyboard")) != NULL;) {
+ for_each_node_by_name(np, "keyboard") {
struct device_node *parent = of_get_parent(np);
int found = (parent && strcmp(parent->type, "adb") == 0);
of_node_put(parent);
diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c
index 099d2df976a2..4131f6644fd6 100644
--- a/arch/powerpc/platforms/pseries/setup.c
+++ b/arch/powerpc/platforms/pseries/setup.c
@@ -232,8 +232,7 @@ static void __init pseries_discover_pic(void)
struct device_node *np;
const char *typep;
- for (np = NULL; (np = of_find_node_by_name(np,
- "interrupt-controller"));) {
+ for_each_node_by_name(np, "interrupt-controller") {
typep = of_get_property(np, "compatible", NULL);
if (strstr(typep, "open-pic")) {
pSeries_mpic_node = of_node_get(np);
diff --git a/drivers/cpufreq/pmac64-cpufreq.c b/drivers/cpufreq/pmac64-cpufreq.c
index 8bc422977b5b..4ff86878727f 100644
--- a/drivers/cpufreq/pmac64-cpufreq.c
+++ b/drivers/cpufreq/pmac64-cpufreq.c
@@ -499,8 +499,7 @@ static int __init g5_pm72_cpufreq_init(struct device_node *cpunode)
}
/* Lookup the i2c hwclock */
- for (hwclock = NULL;
- (hwclock = of_find_node_by_name(hwclock, "i2c-hwclock")) != NULL;){
+ for_each_node_by_name(hwclock, "i2c-hwclock") {
const char *loc = of_get_property(hwclock,
"hwctrl-location", NULL);
if (loc == NULL)
diff --git a/drivers/edac/cell_edac.c b/drivers/edac/cell_edac.c
index 374b57fc596d..a12c8552f6a6 100644
--- a/drivers/edac/cell_edac.c
+++ b/drivers/edac/cell_edac.c
@@ -134,8 +134,7 @@ static void cell_edac_init_csrows(struct mem_ctl_info *mci)
int j;
u32 nr_pages;
- for (np = NULL;
- (np = of_find_node_by_name(np, "memory")) != NULL;) {
+ for_each_node_by_name(np, "memory") {
struct resource r;
/* We "know" that the Cell firmware only creates one entry
diff --git a/drivers/pci/hotplug/rpaphp_core.c b/drivers/pci/hotplug/rpaphp_core.c
index 4796c15fba94..f276ddcb27cc 100644
--- a/drivers/pci/hotplug/rpaphp_core.c
+++ b/drivers/pci/hotplug/rpaphp_core.c
@@ -369,11 +369,11 @@ static void __exit cleanup_slots(void)
static int __init rpaphp_init(void)
{
- struct device_node *dn = NULL;
+ struct device_node *dn;
info(DRIVER_DESC " version: " DRIVER_VERSION "\n");
- while ((dn = of_find_node_by_name(dn, "pci")))
+ for_each_node_by_name(dn, "pci")
rpaphp_add_slot(dn);
return 0;
diff --git a/drivers/tty/serial/pmac_zilog.c b/drivers/tty/serial/pmac_zilog.c
index e9d420ff3931..1d6d16b8f8b1 100644
--- a/drivers/tty/serial/pmac_zilog.c
+++ b/drivers/tty/serial/pmac_zilog.c
@@ -1650,8 +1650,7 @@ static int __init pmz_probe(void)
/*
* Find all escc chips in the system
*/
- node_p = of_find_node_by_name(NULL, "escc");
- while (node_p) {
+ for_each_node_by_name(node_p, "escc") {
/*
* First get channel A/B node pointers
*
@@ -1669,7 +1668,7 @@ static int __init pmz_probe(void)
of_node_put(node_b);
printk(KERN_ERR "pmac_zilog: missing node %c for escc %s\n",
(!node_a) ? 'a' : 'b', node_p->full_name);
- goto next;
+ continue;
}
/*
@@ -1696,11 +1695,9 @@ static int __init pmz_probe(void)
of_node_put(node_b);
memset(&pmz_ports[count], 0, sizeof(struct uart_pmac_port));
memset(&pmz_ports[count+1], 0, sizeof(struct uart_pmac_port));
- goto next;
+ continue;
}
count += 2;
-next:
- node_p = of_find_node_by_name(node_p, "escc");
}
pmz_ports_count = count;
diff --git a/sound/ppc/pmac.c b/sound/ppc/pmac.c
index 7a43c0c38316..8a431bcb056c 100644
--- a/sound/ppc/pmac.c
+++ b/sound/ppc/pmac.c
@@ -992,9 +992,9 @@ static int snd_pmac_detect(struct snd_pmac *chip)
return -ENODEV;
if (!sound) {
- sound = of_find_node_by_name(NULL, "sound");
- while (sound && sound->parent != chip->node)
- sound = of_find_node_by_name(sound, "sound");
+ for_each_node_by_name(sound, "sound")
+ if (sound->parent == chip->node)
+ break;
}
if (! sound) {
of_node_put(chip->node);
--
1.9.1
--
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
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [RFC 2/3] of: Replace custom linked list with list_head
[not found] ` <1401983021-13829-1-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2014-06-05 15:43 ` [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name() Grant Likely
@ 2014-06-05 15:43 ` Grant Likely
2014-06-05 15:43 ` [RFC 3/3] of: Refactor device_node child list to use list_head Grant Likely
2 siblings, 0 replies; 10+ messages in thread
From: Grant Likely @ 2014-06-05 15:43 UTC (permalink / raw)
To: devicetree-u79uwXL29TY76Z2rM5mHXA
Cc: rob.herring-QSEj5FYQhm4dnm+yROfE0A, Grant Likely
The device tree data structure uses a custom linked list implementation
which is baroque and prone to bugs. Replace the allnodes list with a
list_head and the common list_head accessor functions.
At the same time, move the private for_each_of_allnodes() macro into
drivers/of/base.c. It must not be used outside of the core code because
the macro doesn't take into account device tree locking.
Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
Cc: Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
arch/arm/mach-vexpress/v2m.c | 2 +-
drivers/of/base.c | 126 +++++++++++++++++++-----------------
drivers/of/fdt.c | 27 ++++----
drivers/of/pdt.c | 26 +++-----
drivers/pci/hotplug/rpadlpar_core.c | 2 +-
include/linux/of.h | 9 ++-
include/linux/of_fdt.h | 3 +-
include/linux/of_pdt.h | 3 +-
8 files changed, 96 insertions(+), 102 deletions(-)
diff --git a/arch/arm/mach-vexpress/v2m.c b/arch/arm/mach-vexpress/v2m.c
index 4f8b8cb17ff5..57151ba5edee 100644
--- a/arch/arm/mach-vexpress/v2m.c
+++ b/arch/arm/mach-vexpress/v2m.c
@@ -412,7 +412,7 @@ void __init v2m_dt_init_early(void)
vexpress_sysreg_of_early_init();
/* Confirm board type against DT property, if available */
- if (of_property_read_u32(of_allnodes, "arm,hbi", &dt_hbi) == 0) {
+ if (of_property_read_u32(of_root, "arm,hbi", &dt_hbi) == 0) {
u32 hbi = vexpress_get_hbi(VEXPRESS_SITE_MASTER);
if (WARN_ON(dt_hbi != hbi))
diff --git a/drivers/of/base.c b/drivers/of/base.c
index 567e6e1b7921..6c0bd86b802f 100644
--- a/drivers/of/base.c
+++ b/drivers/of/base.c
@@ -31,14 +31,22 @@
LIST_HEAD(aliases_lookup);
-struct device_node *of_allnodes;
+LIST_HEAD(of_allnodes);
EXPORT_SYMBOL(of_allnodes);
-struct device_node *of_chosen;
-struct device_node *of_aliases;
+struct device_node *of_root = NULL;
+struct device_node *of_chosen = NULL;
+struct device_node *of_aliases = NULL;
static struct device_node *of_stdout;
static struct kset *of_kset;
+#define of_allnodes_prepare(dn) \
+ list_prepare_entry(dn, &of_allnodes, allnext)
+#define for_each_of_allnodes(dn) \
+ list_for_each_entry(dn, &of_allnodes, allnext)
+#define for_each_of_allnodes_continue(dn) \
+ list_for_each_entry_continue(dn, &of_allnodes, allnext)
+
/*
* Used to protect the of_aliases; but also overloaded to hold off addition of
* nodes to sysfs
@@ -299,7 +307,7 @@ static int __init of_init(void)
mutex_unlock(&of_aliases_mutex);
/* Symlink in /proc as required by userspace ABI */
- if (of_allnodes)
+ if (of_root)
proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");
return 0;
@@ -348,18 +356,18 @@ EXPORT_SYMBOL(of_find_property);
* Returns a node pointer with refcount incremented, use
* of_node_put() on it when done.
*/
-struct device_node *of_find_all_nodes(struct device_node *prev)
+struct device_node *of_find_all_nodes(struct device_node *np)
{
- struct device_node *np;
unsigned long flags;
raw_spin_lock_irqsave(&devtree_lock, flags);
- np = prev ? prev->allnext : of_allnodes;
- for (; np != NULL; np = np->allnext)
- if (of_node_get(np))
- break;
- of_node_put(prev);
+ of_node_put(np);
+ np = list_next_entry(of_allnodes_prepare(np), allnext);
+ if (&np->allnext == &of_allnodes)
+ np = NULL;
+ of_node_get(np);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
+
return np;
}
EXPORT_SYMBOL(of_find_all_nodes);
@@ -831,7 +839,7 @@ struct device_node *of_find_node_by_path(const char *path)
unsigned long flags;
if (strcmp(path, "/") == 0)
- return of_node_get(of_allnodes);
+ return of_node_get(of_root);
/* The path could begin with an alias */
if (*path != '/') {
@@ -856,7 +864,7 @@ struct device_node *of_find_node_by_path(const char *path)
/* Step down the tree matching path components */
raw_spin_lock_irqsave(&devtree_lock, flags);
if (!np)
- np = of_node_get(of_allnodes);
+ np = of_node_get(of_root);
while (np && *path == '/') {
path++; /* Increment past '/' delimiter */
np = __of_find_node_by_path(np, path);
@@ -881,16 +889,18 @@ EXPORT_SYMBOL(of_find_node_by_path);
struct device_node *of_find_node_by_name(struct device_node *from,
const char *name)
{
- struct device_node *np;
+ struct device_node *np = NULL;
unsigned long flags;
raw_spin_lock_irqsave(&devtree_lock, flags);
- np = from ? from->allnext : of_allnodes;
- for (; np; np = np->allnext)
- if (np->name && (of_node_cmp(np->name, name) == 0)
- && of_node_get(np))
- break;
of_node_put(from);
+ from = of_allnodes_prepare(from);
+ for_each_of_allnodes_continue(from) {
+ if (from->name && (of_node_cmp(from->name, name) == 0)) {
+ np = of_node_get(from);
+ break;
+ }
+ }
raw_spin_unlock_irqrestore(&devtree_lock, flags);
return np;
}
@@ -911,16 +921,18 @@ EXPORT_SYMBOL(of_find_node_by_name);
struct device_node *of_find_node_by_type(struct device_node *from,
const char *type)
{
- struct device_node *np;
+ struct device_node *np = NULL;
unsigned long flags;
raw_spin_lock_irqsave(&devtree_lock, flags);
- np = from ? from->allnext : of_allnodes;
- for (; np; np = np->allnext)
- if (np->type && (of_node_cmp(np->type, type) == 0)
- && of_node_get(np))
- break;
of_node_put(from);
+ from = of_allnodes_prepare(from);
+ for_each_of_allnodes_continue(from) {
+ if (from->type && (of_node_cmp(from->type, type) == 0)) {
+ np = of_node_get(from);
+ break;
+ }
+ }
raw_spin_unlock_irqrestore(&devtree_lock, flags);
return np;
}
@@ -943,18 +955,20 @@ EXPORT_SYMBOL(of_find_node_by_type);
struct device_node *of_find_compatible_node(struct device_node *from,
const char *type, const char *compatible)
{
- struct device_node *np;
+ struct device_node *np = NULL;
unsigned long flags;
raw_spin_lock_irqsave(&devtree_lock, flags);
- np = from ? from->allnext : of_allnodes;
- for (; np; np = np->allnext) {
- if (__of_device_is_compatible(np, compatible, type, NULL) &&
- of_node_get(np))
+ of_node_put(from);
+ from = of_allnodes_prepare(from);
+ for_each_of_allnodes_continue(from) {
+ if (__of_device_is_compatible(from, compatible, type, NULL)) {
+ np = of_node_get(from);
break;
+ }
}
- of_node_put(from);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
+
return np;
}
EXPORT_SYMBOL(of_find_compatible_node);
@@ -974,22 +988,22 @@ EXPORT_SYMBOL(of_find_compatible_node);
struct device_node *of_find_node_with_property(struct device_node *from,
const char *prop_name)
{
- struct device_node *np;
+ struct device_node *np = NULL;
struct property *pp;
unsigned long flags;
raw_spin_lock_irqsave(&devtree_lock, flags);
- np = from ? from->allnext : of_allnodes;
- for (; np; np = np->allnext) {
- for (pp = np->properties; pp; pp = pp->next) {
+ of_node_put(from);
+ from = of_allnodes_prepare(from);
+ for_each_of_allnodes_continue(from) {
+ for (pp = from->properties; pp; pp = pp->next) {
if (of_prop_cmp(pp->name, prop_name) == 0) {
- of_node_get(np);
+ np = of_node_get(from);
goto out;
}
}
}
-out:
- of_node_put(from);
+ out:
raw_spin_unlock_irqrestore(&devtree_lock, flags);
return np;
}
@@ -1054,7 +1068,7 @@ struct device_node *of_find_matching_node_and_match(struct device_node *from,
const struct of_device_id *matches,
const struct of_device_id **match)
{
- struct device_node *np;
+ struct device_node *np = NULL;
const struct of_device_id *m;
unsigned long flags;
@@ -1062,16 +1076,16 @@ struct device_node *of_find_matching_node_and_match(struct device_node *from,
*match = NULL;
raw_spin_lock_irqsave(&devtree_lock, flags);
- np = from ? from->allnext : of_allnodes;
- for (; np; np = np->allnext) {
- m = __of_match_node(matches, np);
- if (m && of_node_get(np)) {
+ of_node_put(from);
+ from = of_allnodes_prepare(from);
+ for_each_of_allnodes_continue(from) {
+ if ((m = __of_match_node(matches, from))) {
+ np = of_node_get(from);
if (match)
*match = m;
break;
}
}
- of_node_put(from);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
return np;
}
@@ -1113,14 +1127,16 @@ EXPORT_SYMBOL_GPL(of_modalias_node);
*/
struct device_node *of_find_node_by_phandle(phandle handle)
{
- struct device_node *np;
+ struct device_node *tmp, *np = NULL;
unsigned long flags;
raw_spin_lock_irqsave(&devtree_lock, flags);
- for (np = of_allnodes; np; np = np->allnext)
- if (np->phandle == handle)
+ for_each_of_allnodes(tmp) {
+ if (tmp->phandle == handle) {
+ np = of_node_get(tmp);
break;
- of_node_get(np);
+ }
+ }
raw_spin_unlock_irqrestore(&devtree_lock, flags);
return np;
}
@@ -1959,10 +1975,9 @@ int of_attach_node(struct device_node *np)
return rc;
raw_spin_lock_irqsave(&devtree_lock, flags);
+ list_add_tail(&np->allnext, &of_allnodes);
np->sibling = np->parent->child;
- np->allnext = of_allnodes;
np->parent->child = np;
- of_allnodes = np;
of_node_clear_flag(np, OF_DETACHED);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
@@ -2000,16 +2015,7 @@ int of_detach_node(struct device_node *np)
return rc;
}
- if (of_allnodes == np)
- of_allnodes = np->allnext;
- else {
- struct device_node *prev;
- for (prev = of_allnodes;
- prev->allnext != np;
- prev = prev->allnext)
- ;
- prev->allnext = np->allnext;
- }
+ list_del(&np->allnext);
if (parent->child == np)
parent->child = np->sibling;
diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
index 17be90f5445f..6876ac3d4b30 100644
--- a/drivers/of/fdt.c
+++ b/drivers/of/fdt.c
@@ -96,14 +96,14 @@ static void *unflatten_dt_alloc(void **mem, unsigned long size,
* @mem: Memory chunk to use for allocating device nodes and properties
* @p: pointer to node in flat tree
* @dad: Parent struct device_node
- * @allnextpp: pointer to ->allnext from last allocated device_node
+ * @allnext: list of ->allnext from this tree
* @fpsize: Size of the node path up at the current depth.
*/
static void * unflatten_dt_node(void *blob,
void *mem,
int *poffset,
struct device_node *dad,
- struct device_node ***allnextpp,
+ struct list_head *allnext,
unsigned long fpsize)
{
const __be32 *p;
@@ -151,7 +151,7 @@ static void * unflatten_dt_node(void *blob,
np = unflatten_dt_alloc(&mem, sizeof(struct device_node) + allocl,
__alignof__(struct device_node));
- if (allnextpp) {
+ if (allnext) {
char *fn;
of_node_init(np);
np->full_name = fn = ((char *)np) + sizeof(*np);
@@ -173,8 +173,7 @@ static void * unflatten_dt_node(void *blob,
memcpy(fn, pathp, l);
prev_pp = &np->properties;
- **allnextpp = np;
- *allnextpp = &np->allnext;
+ list_add_tail(&np->allnext, allnext);
if (dad != NULL) {
np->parent = dad;
/* we temporarily use the next field as `last_child'*/
@@ -205,7 +204,7 @@ static void * unflatten_dt_node(void *blob,
has_name = 1;
pp = unflatten_dt_alloc(&mem, sizeof(struct property),
__alignof__(struct property));
- if (allnextpp) {
+ if (allnext) {
/* We accept flattened tree phandles either in
* ePAPR-style "phandle" properties, or the
* legacy "linux,phandle" properties. If both
@@ -247,7 +246,7 @@ static void * unflatten_dt_node(void *blob,
sz = (pa - ps) + 1;
pp = unflatten_dt_alloc(&mem, sizeof(struct property) + sz,
__alignof__(struct property));
- if (allnextpp) {
+ if (allnext) {
pp->name = "name";
pp->length = sz;
pp->value = pp + 1;
@@ -259,7 +258,7 @@ static void * unflatten_dt_node(void *blob,
(char *)pp->value);
}
}
- if (allnextpp) {
+ if (allnext) {
*prev_pp = NULL;
np->name = of_get_property(np, "name", NULL);
np->type = of_get_property(np, "device_type", NULL);
@@ -275,7 +274,7 @@ static void * unflatten_dt_node(void *blob,
if (depth < 0)
depth = 0;
while (*poffset > 0 && depth > old_depth)
- mem = unflatten_dt_node(blob, mem, poffset, np, allnextpp,
+ mem = unflatten_dt_node(blob, mem, poffset, np, allnext,
fpsize);
if (*poffset < 0 && *poffset != -FDT_ERR_NOTFOUND)
@@ -297,13 +296,12 @@ static void * unflatten_dt_node(void *blob,
* for the resulting tree
*/
static void __unflatten_device_tree(void *blob,
- struct device_node **mynodes,
+ struct list_head *allnext,
void * (*dt_alloc)(u64 size, u64 align))
{
unsigned long size;
int start;
void *mem;
- struct device_node **allnextp = mynodes;
pr_debug(" -> unflatten_device_tree()\n");
@@ -339,11 +337,10 @@ static void __unflatten_device_tree(void *blob,
/* Second pass, do actual unflattening */
start = 0;
- unflatten_dt_node(blob, mem, &start, NULL, &allnextp, 0);
+ unflatten_dt_node(blob, mem, &start, NULL, allnext, 0);
if (be32_to_cpup(mem + size) != 0xdeadbeef)
pr_warning("End of tree marker overwritten: %08x\n",
be32_to_cpup(mem + size));
- *allnextp = NULL;
pr_debug(" <- unflatten_device_tree()\n");
}
@@ -361,8 +358,7 @@ static void *kernel_tree_alloc(u64 size, u64 align)
* pointers of the nodes so the normal device-tree walking functions
* can be used.
*/
-void of_fdt_unflatten_tree(unsigned long *blob,
- struct device_node **mynodes)
+void of_fdt_unflatten_tree(unsigned long *blob, struct list_head *mynodes)
{
__unflatten_device_tree(blob, mynodes, &kernel_tree_alloc);
}
@@ -904,6 +900,7 @@ void __init unflatten_device_tree(void)
{
__unflatten_device_tree(initial_boot_params, &of_allnodes,
early_init_dt_alloc_memory_arch);
+ of_root = list_first_entry_or_null(&of_allnodes, struct device_node, allnext);
/* Get pointer to "/chosen" and "/aliases" nodes for use everywhere */
of_alias_scan(early_init_dt_alloc_memory_arch);
diff --git a/drivers/of/pdt.c b/drivers/of/pdt.c
index 36b4035881b0..38dd814870b4 100644
--- a/drivers/of/pdt.c
+++ b/drivers/of/pdt.c
@@ -25,8 +25,7 @@
static struct of_pdt_ops *of_pdt_prom_ops __initdata;
-void __initdata (*of_pdt_build_more)(struct device_node *dp,
- struct device_node ***nextp);
+void __initdata (*of_pdt_build_more)(struct device_node *dp);
#if defined(CONFIG_SPARC)
unsigned int of_pdt_unique_id __initdata;
@@ -192,8 +191,7 @@ static struct device_node * __init of_pdt_create_node(phandle node,
}
static struct device_node * __init of_pdt_build_tree(struct device_node *parent,
- phandle node,
- struct device_node ***nextp)
+ phandle node)
{
struct device_node *ret = NULL, *prev_sibling = NULL;
struct device_node *dp;
@@ -210,16 +208,15 @@ static struct device_node * __init of_pdt_build_tree(struct device_node *parent,
ret = dp;
prev_sibling = dp;
- *(*nextp) = dp;
- *nextp = &dp->allnext;
+ list_add_tail(&of_allnodes, &np->allnext);
dp->full_name = of_pdt_build_full_name(dp);
dp->child = of_pdt_build_tree(dp,
- of_pdt_prom_ops->getchild(node), nextp);
+ of_pdt_prom_ops->getchild(node));
if (of_pdt_build_more)
- of_pdt_build_more(dp, nextp);
+ of_pdt_build_more(dp);
node = of_pdt_prom_ops->getsibling(node);
}
@@ -234,20 +231,17 @@ static void * __init kernel_tree_alloc(u64 size, u64 align)
void __init of_pdt_build_devicetree(phandle root_node, struct of_pdt_ops *ops)
{
- struct device_node **nextp;
-
BUG_ON(!ops);
of_pdt_prom_ops = ops;
- of_allnodes = of_pdt_create_node(root_node, NULL);
+ of_root = of_pdt_create_node(root_node, NULL);
#if defined(CONFIG_SPARC)
- of_allnodes->path_component_name = "";
+ of_root->path_component_name = "";
#endif
- of_allnodes->full_name = "/";
+ of_root->full_name = "/";
- nextp = &of_allnodes->allnext;
- of_allnodes->child = of_pdt_build_tree(of_allnodes,
- of_pdt_prom_ops->getchild(of_allnodes->phandle), &nextp);
+ list_add_tail(&of_allnodes, &of_root->allnext);
+ of_root->child = of_pdt_build_tree(of_root, of_pdt_prom_ops->getchild(of_root->phandle));
/* Get pointer to "/chosen" and "/aliases" nodes for use everywhere */
of_alias_scan(kernel_tree_alloc);
diff --git a/drivers/pci/hotplug/rpadlpar_core.c b/drivers/pci/hotplug/rpadlpar_core.c
index 4fcdeedda31b..e25e4d1ec241 100644
--- a/drivers/pci/hotplug/rpadlpar_core.c
+++ b/drivers/pci/hotplug/rpadlpar_core.c
@@ -68,7 +68,7 @@ static struct device_node *find_php_slot_pci_node(char *drc_name,
char *type;
int rc;
- while ((np = of_find_node_by_name(np, "pci"))) {
+ for_each_node_by_name(np, "pci") {
rc = rpaphp_get_drc_props(np, NULL, &name, &type, NULL);
if (rc == 0)
if (!strcmp(drc_name, name) && !strcmp(drc_type, type))
diff --git a/include/linux/of.h b/include/linux/of.h
index 3bad8d106e0e..e082db3c284f 100644
--- a/include/linux/of.h
+++ b/include/linux/of.h
@@ -56,7 +56,7 @@ struct device_node {
struct device_node *child;
struct device_node *sibling;
struct device_node *next; /* next device of same type */
- struct device_node *allnext; /* next in list of all nodes */
+ struct list_head allnext;
struct kobject kobj;
unsigned long _flags;
void *data;
@@ -110,14 +110,15 @@ static inline void of_node_put(struct device_node *node) { }
#ifdef CONFIG_OF
/* Pointer for first entry in chain of all nodes. */
-extern struct device_node *of_allnodes;
+extern struct list_head of_allnodes;
+extern struct device_node *of_root;
extern struct device_node *of_chosen;
extern struct device_node *of_aliases;
extern raw_spinlock_t devtree_lock;
static inline bool of_have_populated_dt(void)
{
- return of_allnodes != NULL;
+ return !list_empty(&of_allnodes);
}
static inline bool of_node_is_root(const struct device_node *node)
@@ -208,8 +209,6 @@ static inline const char *of_node_full_name(const struct device_node *np)
return np ? np->full_name : "<no-node>";
}
-#define for_each_of_allnodes(dn) \
- for (dn = of_allnodes; dn; dn = dn->allnext)
extern struct device_node *of_find_node_by_name(struct device_node *from,
const char *name);
extern struct device_node *of_find_node_by_type(struct device_node *from,
diff --git a/include/linux/of_fdt.h b/include/linux/of_fdt.h
index 5c0ab057eecf..efa7615a1ce9 100644
--- a/include/linux/of_fdt.h
+++ b/include/linux/of_fdt.h
@@ -35,8 +35,7 @@ extern int of_fdt_is_compatible(const void *blob,
const char *compat);
extern int of_fdt_match(const void *blob, unsigned long node,
const char *const *compat);
-extern void of_fdt_unflatten_tree(unsigned long *blob,
- struct device_node **mynodes);
+extern void of_fdt_unflatten_tree(unsigned long *blob, struct list_head *mynodes);
/* TBD: Temporary export of fdt globals - remove when code fully merged */
extern int __initdata dt_root_addr_cells;
diff --git a/include/linux/of_pdt.h b/include/linux/of_pdt.h
index c65a18a0cfdf..7e09244bb679 100644
--- a/include/linux/of_pdt.h
+++ b/include/linux/of_pdt.h
@@ -39,7 +39,6 @@ extern void *prom_early_alloc(unsigned long size);
/* for building the device tree */
extern void of_pdt_build_devicetree(phandle root_node, struct of_pdt_ops *ops);
-extern void (*of_pdt_build_more)(struct device_node *dp,
- struct device_node ***nextp);
+extern void (*of_pdt_build_more)(struct device_node *dp);
#endif /* _LINUX_OF_PDT_H */
--
1.9.1
--
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
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [RFC 3/3] of: Refactor device_node child list to use list_head
[not found] ` <1401983021-13829-1-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2014-06-05 15:43 ` [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name() Grant Likely
2014-06-05 15:43 ` [RFC 2/3] of: Replace custom linked list with list_head Grant Likely
@ 2014-06-05 15:43 ` Grant Likely
2 siblings, 0 replies; 10+ messages in thread
From: Grant Likely @ 2014-06-05 15:43 UTC (permalink / raw)
To: devicetree-u79uwXL29TY76Z2rM5mHXA
Cc: rob.herring-QSEj5FYQhm4dnm+yROfE0A, Grant Likely
The device tree data structure uses a custom linked list implemenation
which is baroque and prone to bugs. Replace the child node lists with a
list_head and the common list_head accessor functions.
Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
Cc: Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
drivers/of/base.c | 43 +++++++++++++++++++------------------------
drivers/of/fdt.c | 12 +++---------
include/linux/of.h | 8 +++++---
3 files changed, 27 insertions(+), 36 deletions(-)
diff --git a/drivers/of/base.c b/drivers/of/base.c
index 6c0bd86b802f..69f5648babd8 100644
--- a/drivers/of/base.c
+++ b/drivers/of/base.c
@@ -47,6 +47,12 @@ static struct kset *of_kset;
#define for_each_of_allnodes_continue(dn) \
list_for_each_entry_continue(dn, &of_allnodes, allnext)
+#define of_children_prepare(parent, child) \
+ list_prepare_entry(child, &parent->children, childnext)
+#define for_each_of_children(parent, child) \
+ list_for_each_entry(child, &parent->children, childnext)
+#define for_each_of_children_continue(parent, child) \
+ list_for_each_entry_continue(child, &parent->children, childnext)
/*
* Used to protect the of_aliases; but also overloaded to hold off addition of
* nodes to sysfs
@@ -711,11 +717,11 @@ static struct device_node *__of_get_next_child(const struct device_node *node,
if (!node)
return NULL;
- next = prev ? prev->sibling : node->child;
- for (; next; next = next->sibling)
- if (of_node_get(next))
- break;
of_node_put(prev);
+ next = list_next_entry(of_children_prepare(node, prev), childnext);
+ if (&next->childnext == &node->children)
+ next = NULL;
+ of_node_get(next);
return next;
}
#define __for_each_child_of_node(parent, child) \
@@ -754,21 +760,21 @@ EXPORT_SYMBOL(of_get_next_child);
struct device_node *of_get_next_available_child(const struct device_node *node,
struct device_node *prev)
{
- struct device_node *next;
+ struct device_node *next = NULL;
unsigned long flags;
if (!node)
return NULL;
raw_spin_lock_irqsave(&devtree_lock, flags);
- next = prev ? prev->sibling : node->child;
- for (; next; next = next->sibling) {
- if (!__of_device_is_available(next))
- continue;
- if (of_node_get(next))
+ of_node_put(prev);
+ prev = of_children_prepare(node, prev);
+ for_each_of_children_continue(node, prev) {
+ if (__of_device_is_available(prev)) {
+ next = of_node_get(prev);
break;
+ }
}
- of_node_put(prev);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
return next;
}
@@ -1976,8 +1982,7 @@ int of_attach_node(struct device_node *np)
raw_spin_lock_irqsave(&devtree_lock, flags);
list_add_tail(&np->allnext, &of_allnodes);
- np->sibling = np->parent->child;
- np->parent->child = np;
+ list_add_tail(&np->childnext, &np->parent->children);
of_node_clear_flag(np, OF_DETACHED);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
@@ -2016,17 +2021,7 @@ int of_detach_node(struct device_node *np)
}
list_del(&np->allnext);
-
- if (parent->child == np)
- parent->child = np->sibling;
- else {
- struct device_node *prevsib;
- for (prevsib = np->parent->child;
- prevsib->sibling != np;
- prevsib = prevsib->sibling)
- ;
- prevsib->sibling = np->sibling;
- }
+ list_del(&np->childnext);
of_node_set_flag(np, OF_DETACHED);
raw_spin_unlock_irqrestore(&devtree_lock, flags);
diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
index 6876ac3d4b30..2ff96cac1496 100644
--- a/drivers/of/fdt.c
+++ b/drivers/of/fdt.c
@@ -155,6 +155,7 @@ static void * unflatten_dt_node(void *blob,
char *fn;
of_node_init(np);
np->full_name = fn = ((char *)np) + sizeof(*np);
+ np->parent = dad;
if (new_format) {
/* rebuild full path for new format */
if (dad && dad->parent) {
@@ -174,15 +175,8 @@ static void * unflatten_dt_node(void *blob,
prev_pp = &np->properties;
list_add_tail(&np->allnext, allnext);
- if (dad != NULL) {
- np->parent = dad;
- /* we temporarily use the next field as `last_child'*/
- if (dad->next == NULL)
- dad->child = np;
- else
- dad->next->sibling = np;
- dad->next = np;
- }
+ if (np->parent)
+ list_add_tail(&np->childnext, &np->parent->children);
}
/* process properties */
for (offset = fdt_first_property_offset(blob, *poffset);
diff --git a/include/linux/of.h b/include/linux/of.h
index e082db3c284f..11bef100bc43 100644
--- a/include/linux/of.h
+++ b/include/linux/of.h
@@ -52,10 +52,11 @@ struct device_node {
struct property *properties;
struct property *deadprops; /* removed properties */
+
+ struct list_head children; /* next device of same type */
struct device_node *parent;
- struct device_node *child;
- struct device_node *sibling;
- struct device_node *next; /* next device of same type */
+
+ struct list_head childnext;
struct list_head allnext;
struct kobject kobj;
unsigned long _flags;
@@ -80,6 +81,7 @@ extern int of_node_add(struct device_node *node);
extern struct kobj_type of_node_ktype;
static inline void of_node_init(struct device_node *node)
{
+ INIT_LIST_HEAD(&node->children);
kobject_init(&node->kobj, &of_node_ktype);
}
--
1.9.1
--
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
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <1401983021-13829-2-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
@ 2014-06-05 17:48 ` Rob Herring
[not found] ` <CABGGisx=fhp1P68M+9aQUTzi6Bp+PnXK7Xq_TQ+=87AkGCZvwQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 10+ messages in thread
From: Rob Herring @ 2014-06-05 17:48 UTC (permalink / raw)
To: Grant Likely; +Cc: devicetree-u79uwXL29TY76Z2rM5mHXA
On Thu, Jun 5, 2014 at 10:43 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> There are a bunch of users open coding the for_each_node_by_name() by
> calling of_find_node_by_name() directly instead of using the macro. This
> is getting in the way of some cleanups, and the possibility of removing
> of_find_node_by_name() entirely. Clean it up so that all the users are
> consistent.
>
> Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> ---
> arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
> arch/powerpc/platforms/powermac/pci.c | 2 +-
> arch/powerpc/platforms/powermac/smp.c | 2 +-
> arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
> arch/powerpc/platforms/pseries/setup.c | 3 +--
> drivers/cpufreq/pmac64-cpufreq.c | 3 +--
> drivers/edac/cell_edac.c | 3 +--
> drivers/pci/hotplug/rpaphp_core.c | 4 ++--
> drivers/tty/serial/pmac_zilog.c | 9 +++------
> sound/ppc/pmac.c | 6 +++---
> 10 files changed, 19 insertions(+), 35 deletions(-)
>
> diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
> index 63d82bbc05e9..39e1d163c427 100644
> --- a/arch/powerpc/platforms/powermac/feature.c
> +++ b/arch/powerpc/platforms/powermac/feature.c
> @@ -2805,25 +2805,20 @@ set_initial_features(void)
> /* Enable GMAC for now for PCI probing. It will be disabled
> * later on after PCI probe
> */
> - np = of_find_node_by_name(NULL, "ethernet");
> - while(np) {
> + for_each_node_by_name(np, "ethernet")
> if (of_device_is_compatible(np, "K2-GMAC"))
Can't for_each_compatible_node be used here instead?
> g5_gmac_enable(np, 0, 1);
> - np = of_find_node_by_name(np, "ethernet");
> - }
>
> /* Enable FW before PCI probe. Will be disabled later on
> * Note: We should have a batter way to check that we are
> * dealing with uninorth internal cell and not a PCI cell
> * on the external PCI. The code below works though.
> */
> - np = of_find_node_by_name(NULL, "firewire");
> - while(np) {
> + for_each_node_by_name(np, "firewire") {
> if (of_device_is_compatible(np, "pci106b,5811")) {
and here...
> macio_chips[0].flags |= MACIO_FLAG_FW_SUPPORTED;
> g5_fw_enable(np, 0, 1);
> }
> - np = of_find_node_by_name(np, "firewire");
> }
> }
> #else /* CONFIG_POWER4 */
> @@ -2834,13 +2829,11 @@ set_initial_features(void)
> /* Enable GMAC for now for PCI probing. It will be disabled
> * later on after PCI probe
> */
> - np = of_find_node_by_name(NULL, "ethernet");
> - while(np) {
> + for_each_node_by_name(np, "ethernet") {
> if (np->parent
> && of_device_is_compatible(np->parent, "uni-north")
> && of_device_is_compatible(np, "gmac"))
This one is a bit more complicated, but should still work with
for_each_compatible_node.
There's a few more that could be converted. I guess it depends if you
think dropping looking at node names is okay.
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
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <CABGGisx=fhp1P68M+9aQUTzi6Bp+PnXK7Xq_TQ+=87AkGCZvwQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2014-06-26 16:03 ` Grant Likely
[not found] ` <20140626160345.6613CC41060-WNowdnHR2B42iJbIjFUEsiwD8/FfD2ys@public.gmane.org>
0 siblings, 1 reply; 10+ messages in thread
From: Grant Likely @ 2014-06-26 16:03 UTC (permalink / raw)
To: Rob Herring; +Cc: devicetree-u79uwXL29TY76Z2rM5mHXA
On Thu, 5 Jun 2014 12:48:54 -0500, Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> On Thu, Jun 5, 2014 at 10:43 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> > There are a bunch of users open coding the for_each_node_by_name() by
> > calling of_find_node_by_name() directly instead of using the macro. This
> > is getting in the way of some cleanups, and the possibility of removing
> > of_find_node_by_name() entirely. Clean it up so that all the users are
> > consistent.
> >
> > Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> > ---
> > arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
> > arch/powerpc/platforms/powermac/pci.c | 2 +-
> > arch/powerpc/platforms/powermac/smp.c | 2 +-
> > arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
> > arch/powerpc/platforms/pseries/setup.c | 3 +--
> > drivers/cpufreq/pmac64-cpufreq.c | 3 +--
> > drivers/edac/cell_edac.c | 3 +--
> > drivers/pci/hotplug/rpaphp_core.c | 4 ++--
> > drivers/tty/serial/pmac_zilog.c | 9 +++------
> > sound/ppc/pmac.c | 6 +++---
> > 10 files changed, 19 insertions(+), 35 deletions(-)
> >
> > diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
> > index 63d82bbc05e9..39e1d163c427 100644
> > --- a/arch/powerpc/platforms/powermac/feature.c
> > +++ b/arch/powerpc/platforms/powermac/feature.c
> > @@ -2805,25 +2805,20 @@ set_initial_features(void)
> > /* Enable GMAC for now for PCI probing. It will be disabled
> > * later on after PCI probe
> > */
> > - np = of_find_node_by_name(NULL, "ethernet");
> > - while(np) {
> > + for_each_node_by_name(np, "ethernet")
> > if (of_device_is_compatible(np, "K2-GMAC"))
>
> Can't for_each_compatible_node be used here instead?
Not easily without changing the behaviour. It would need to then check
the name inside the block.
>
> > g5_gmac_enable(np, 0, 1);
> > - np = of_find_node_by_name(np, "ethernet");
> > - }
> >
> > /* Enable FW before PCI probe. Will be disabled later on
> > * Note: We should have a batter way to check that we are
> > * dealing with uninorth internal cell and not a PCI cell
> > * on the external PCI. The code below works though.
> > */
> > - np = of_find_node_by_name(NULL, "firewire");
> > - while(np) {
> > + for_each_node_by_name(np, "firewire") {
> > if (of_device_is_compatible(np, "pci106b,5811")) {
>
> and here...
same.
>
> > macio_chips[0].flags |= MACIO_FLAG_FW_SUPPORTED;
> > g5_fw_enable(np, 0, 1);
> > }
> > - np = of_find_node_by_name(np, "firewire");
> > }
> > }
> > #else /* CONFIG_POWER4 */
> > @@ -2834,13 +2829,11 @@ set_initial_features(void)
> > /* Enable GMAC for now for PCI probing. It will be disabled
> > * later on after PCI probe
> > */
> > - np = of_find_node_by_name(NULL, "ethernet");
> > - while(np) {
> > + for_each_node_by_name(np, "ethernet") {
> > if (np->parent
> > && of_device_is_compatible(np->parent, "uni-north")
> > && of_device_is_compatible(np, "gmac"))
>
> This one is a bit more complicated, but should still work with
> for_each_compatible_node.
>
> There's a few more that could be converted. I guess it depends if you
> think dropping looking at node names is okay.
I'll take another look through and see if anything can be simplified.
g.
--
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
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <20140626160345.6613CC41060-WNowdnHR2B42iJbIjFUEsiwD8/FfD2ys@public.gmane.org>
@ 2014-06-26 18:22 ` Rob Herring
[not found] ` <CAL_JsqKDsdqd3PJ0gwvESMER=Nxi1us55pKWDhTuZtwgjo_MGA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 10+ messages in thread
From: Rob Herring @ 2014-06-26 18:22 UTC (permalink / raw)
To: Grant Likely
Cc: Rob Herring, devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
On Thu, Jun 26, 2014 at 11:03 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> On Thu, 5 Jun 2014 12:48:54 -0500, Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>> On Thu, Jun 5, 2014 at 10:43 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>> > There are a bunch of users open coding the for_each_node_by_name() by
>> > calling of_find_node_by_name() directly instead of using the macro. This
>> > is getting in the way of some cleanups, and the possibility of removing
>> > of_find_node_by_name() entirely. Clean it up so that all the users are
>> > consistent.
>> >
>> > Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>> > ---
>> > arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
>> > arch/powerpc/platforms/powermac/pci.c | 2 +-
>> > arch/powerpc/platforms/powermac/smp.c | 2 +-
>> > arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
>> > arch/powerpc/platforms/pseries/setup.c | 3 +--
>> > drivers/cpufreq/pmac64-cpufreq.c | 3 +--
>> > drivers/edac/cell_edac.c | 3 +--
>> > drivers/pci/hotplug/rpaphp_core.c | 4 ++--
>> > drivers/tty/serial/pmac_zilog.c | 9 +++------
>> > sound/ppc/pmac.c | 6 +++---
>> > 10 files changed, 19 insertions(+), 35 deletions(-)
>> >
>> > diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
>> > index 63d82bbc05e9..39e1d163c427 100644
>> > --- a/arch/powerpc/platforms/powermac/feature.c
>> > +++ b/arch/powerpc/platforms/powermac/feature.c
>> > @@ -2805,25 +2805,20 @@ set_initial_features(void)
>> > /* Enable GMAC for now for PCI probing. It will be disabled
>> > * later on after PCI probe
>> > */
>> > - np = of_find_node_by_name(NULL, "ethernet");
>> > - while(np) {
>> > + for_each_node_by_name(np, "ethernet")
>> > if (of_device_is_compatible(np, "K2-GMAC"))
>>
>> Can't for_each_compatible_node be used here instead?
>
> Not easily without changing the behaviour. It would need to then check
> the name inside the block.
Why would it change behavior? If the compatible string matches, do you
really have cases where the node name is not "ethernet"? I don't
believe it's the kernel's job to validate DT bindings.
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
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <CAL_JsqKDsdqd3PJ0gwvESMER=Nxi1us55pKWDhTuZtwgjo_MGA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2014-06-26 18:56 ` Grant Likely
[not found] ` <CACxGe6tiiNvz9VCq=enTWMGDBUS_0gsDZh2DvzpVP5itQtiRpA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 10+ messages in thread
From: Grant Likely @ 2014-06-26 18:56 UTC (permalink / raw)
To: Rob Herring
Cc: Rob Herring, devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
On Thu, Jun 26, 2014 at 7:22 PM, Rob Herring <robherring2-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> On Thu, Jun 26, 2014 at 11:03 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>> On Thu, 5 Jun 2014 12:48:54 -0500, Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>> On Thu, Jun 5, 2014 at 10:43 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>> > There are a bunch of users open coding the for_each_node_by_name() by
>>> > calling of_find_node_by_name() directly instead of using the macro. This
>>> > is getting in the way of some cleanups, and the possibility of removing
>>> > of_find_node_by_name() entirely. Clean it up so that all the users are
>>> > consistent.
>>> >
>>> > Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>>> > ---
>>> > arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
>>> > arch/powerpc/platforms/powermac/pci.c | 2 +-
>>> > arch/powerpc/platforms/powermac/smp.c | 2 +-
>>> > arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
>>> > arch/powerpc/platforms/pseries/setup.c | 3 +--
>>> > drivers/cpufreq/pmac64-cpufreq.c | 3 +--
>>> > drivers/edac/cell_edac.c | 3 +--
>>> > drivers/pci/hotplug/rpaphp_core.c | 4 ++--
>>> > drivers/tty/serial/pmac_zilog.c | 9 +++------
>>> > sound/ppc/pmac.c | 6 +++---
>>> > 10 files changed, 19 insertions(+), 35 deletions(-)
>>> >
>>> > diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
>>> > index 63d82bbc05e9..39e1d163c427 100644
>>> > --- a/arch/powerpc/platforms/powermac/feature.c
>>> > +++ b/arch/powerpc/platforms/powermac/feature.c
>>> > @@ -2805,25 +2805,20 @@ set_initial_features(void)
>>> > /* Enable GMAC for now for PCI probing. It will be disabled
>>> > * later on after PCI probe
>>> > */
>>> > - np = of_find_node_by_name(NULL, "ethernet");
>>> > - while(np) {
>>> > + for_each_node_by_name(np, "ethernet")
>>> > if (of_device_is_compatible(np, "K2-GMAC"))
>>>
>>> Can't for_each_compatible_node be used here instead?
>>
>> Not easily without changing the behaviour. It would need to then check
>> the name inside the block.
>
> Why would it change behavior? If the compatible string matches, do you
> really have cases where the node name is not "ethernet"? I don't
> believe it's the kernel's job to validate DT bindings.
Yes, there are actually some bindings that have the same compatible
property but behaviour changes based on node name! I don't want to do
the legwork to figure out if these are in that group. Someone else can
do that job.
g.
--
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
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <CACxGe6tiiNvz9VCq=enTWMGDBUS_0gsDZh2DvzpVP5itQtiRpA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2014-06-26 19:00 ` Rob Herring
[not found] ` <CAL_JsqLUjcxHrHctfVe7M79=8MpUCYq1cBhfsoXVzrBoxECwvw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 10+ messages in thread
From: Rob Herring @ 2014-06-26 19:00 UTC (permalink / raw)
To: Grant Likely
Cc: Rob Herring, devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
On Thu, Jun 26, 2014 at 1:56 PM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> On Thu, Jun 26, 2014 at 7:22 PM, Rob Herring <robherring2-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>> On Thu, Jun 26, 2014 at 11:03 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>> On Thu, 5 Jun 2014 12:48:54 -0500, Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>>> On Thu, Jun 5, 2014 at 10:43 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>>> > There are a bunch of users open coding the for_each_node_by_name() by
>>>> > calling of_find_node_by_name() directly instead of using the macro. This
>>>> > is getting in the way of some cleanups, and the possibility of removing
>>>> > of_find_node_by_name() entirely. Clean it up so that all the users are
>>>> > consistent.
>>>> >
>>>> > Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>>>> > ---
>>>> > arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
>>>> > arch/powerpc/platforms/powermac/pci.c | 2 +-
>>>> > arch/powerpc/platforms/powermac/smp.c | 2 +-
>>>> > arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
>>>> > arch/powerpc/platforms/pseries/setup.c | 3 +--
>>>> > drivers/cpufreq/pmac64-cpufreq.c | 3 +--
>>>> > drivers/edac/cell_edac.c | 3 +--
>>>> > drivers/pci/hotplug/rpaphp_core.c | 4 ++--
>>>> > drivers/tty/serial/pmac_zilog.c | 9 +++------
>>>> > sound/ppc/pmac.c | 6 +++---
>>>> > 10 files changed, 19 insertions(+), 35 deletions(-)
>>>> >
>>>> > diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
>>>> > index 63d82bbc05e9..39e1d163c427 100644
>>>> > --- a/arch/powerpc/platforms/powermac/feature.c
>>>> > +++ b/arch/powerpc/platforms/powermac/feature.c
>>>> > @@ -2805,25 +2805,20 @@ set_initial_features(void)
>>>> > /* Enable GMAC for now for PCI probing. It will be disabled
>>>> > * later on after PCI probe
>>>> > */
>>>> > - np = of_find_node_by_name(NULL, "ethernet");
>>>> > - while(np) {
>>>> > + for_each_node_by_name(np, "ethernet")
>>>> > if (of_device_is_compatible(np, "K2-GMAC"))
>>>>
>>>> Can't for_each_compatible_node be used here instead?
>>>
>>> Not easily without changing the behaviour. It would need to then check
>>> the name inside the block.
>>
>> Why would it change behavior? If the compatible string matches, do you
>> really have cases where the node name is not "ethernet"? I don't
>> believe it's the kernel's job to validate DT bindings.
>
> Yes, there are actually some bindings that have the same compatible
> property but behaviour changes based on node name! I don't want to do
> the legwork to figure out if these are in that group. Someone else can
> do that job.
Well, that's just wrong.
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
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name()
[not found] ` <CAL_JsqLUjcxHrHctfVe7M79=8MpUCYq1cBhfsoXVzrBoxECwvw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2014-06-26 19:02 ` Grant Likely
0 siblings, 0 replies; 10+ messages in thread
From: Grant Likely @ 2014-06-26 19:02 UTC (permalink / raw)
To: Rob Herring
Cc: Rob Herring, devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
On Thu, Jun 26, 2014 at 8:00 PM, Rob Herring <robherring2-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> On Thu, Jun 26, 2014 at 1:56 PM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>> On Thu, Jun 26, 2014 at 7:22 PM, Rob Herring <robherring2-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>>> On Thu, Jun 26, 2014 at 11:03 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>>> On Thu, 5 Jun 2014 12:48:54 -0500, Rob Herring <rob.herring-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>>>> On Thu, Jun 5, 2014 at 10:43 AM, Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>>>>> > There are a bunch of users open coding the for_each_node_by_name() by
>>>>> > calling of_find_node_by_name() directly instead of using the macro. This
>>>>> > is getting in the way of some cleanups, and the possibility of removing
>>>>> > of_find_node_by_name() entirely. Clean it up so that all the users are
>>>>> > consistent.
>>>>> >
>>>>> > Signed-off-by: Grant Likely <grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>>>>> > ---
>>>>> > arch/powerpc/platforms/powermac/feature.c | 20 +++++---------------
>>>>> > arch/powerpc/platforms/powermac/pci.c | 2 +-
>>>>> > arch/powerpc/platforms/powermac/smp.c | 2 +-
>>>>> > arch/powerpc/platforms/powermac/udbg_adb.c | 2 +-
>>>>> > arch/powerpc/platforms/pseries/setup.c | 3 +--
>>>>> > drivers/cpufreq/pmac64-cpufreq.c | 3 +--
>>>>> > drivers/edac/cell_edac.c | 3 +--
>>>>> > drivers/pci/hotplug/rpaphp_core.c | 4 ++--
>>>>> > drivers/tty/serial/pmac_zilog.c | 9 +++------
>>>>> > sound/ppc/pmac.c | 6 +++---
>>>>> > 10 files changed, 19 insertions(+), 35 deletions(-)
>>>>> >
>>>>> > diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c
>>>>> > index 63d82bbc05e9..39e1d163c427 100644
>>>>> > --- a/arch/powerpc/platforms/powermac/feature.c
>>>>> > +++ b/arch/powerpc/platforms/powermac/feature.c
>>>>> > @@ -2805,25 +2805,20 @@ set_initial_features(void)
>>>>> > /* Enable GMAC for now for PCI probing. It will be disabled
>>>>> > * later on after PCI probe
>>>>> > */
>>>>> > - np = of_find_node_by_name(NULL, "ethernet");
>>>>> > - while(np) {
>>>>> > + for_each_node_by_name(np, "ethernet")
>>>>> > if (of_device_is_compatible(np, "K2-GMAC"))
>>>>>
>>>>> Can't for_each_compatible_node be used here instead?
>>>>
>>>> Not easily without changing the behaviour. It would need to then check
>>>> the name inside the block.
>>>
>>> Why would it change behavior? If the compatible string matches, do you
>>> really have cases where the node name is not "ethernet"? I don't
>>> believe it's the kernel's job to validate DT bindings.
>>
>> Yes, there are actually some bindings that have the same compatible
>> property but behaviour changes based on node name! I don't want to do
>> the legwork to figure out if these are in that group. Someone else can
>> do that job.
>
> Well, that's just wrong.
true.
--
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
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2014-06-26 19:02 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-06-05 15:43 [RFC 0/3] of: Refactor device tree linked lists Grant Likely
[not found] ` <1401983021-13829-1-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2014-06-05 15:43 ` [RFC 1/3] of: Migrate of_find_node_by_name() users to for_each_node_by_name() Grant Likely
[not found] ` <1401983021-13829-2-git-send-email-grant.likely-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
2014-06-05 17:48 ` Rob Herring
[not found] ` <CABGGisx=fhp1P68M+9aQUTzi6Bp+PnXK7Xq_TQ+=87AkGCZvwQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2014-06-26 16:03 ` Grant Likely
[not found] ` <20140626160345.6613CC41060-WNowdnHR2B42iJbIjFUEsiwD8/FfD2ys@public.gmane.org>
2014-06-26 18:22 ` Rob Herring
[not found] ` <CAL_JsqKDsdqd3PJ0gwvESMER=Nxi1us55pKWDhTuZtwgjo_MGA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2014-06-26 18:56 ` Grant Likely
[not found] ` <CACxGe6tiiNvz9VCq=enTWMGDBUS_0gsDZh2DvzpVP5itQtiRpA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2014-06-26 19:00 ` Rob Herring
[not found] ` <CAL_JsqLUjcxHrHctfVe7M79=8MpUCYq1cBhfsoXVzrBoxECwvw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2014-06-26 19:02 ` Grant Likely
2014-06-05 15:43 ` [RFC 2/3] of: Replace custom linked list with list_head Grant Likely
2014-06-05 15:43 ` [RFC 3/3] of: Refactor device_node child list to use list_head Grant Likely
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).