summaryrefslogtreecommitdiff
path: root/drivers/staging/ozwpan/ozevent.c
diff options
context:
space:
mode:
authorRupesh Gujare <rgujare@ozmodevices.com>2012-03-29 12:11:50 (GMT)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>2012-04-10 16:40:52 (GMT)
commit68a75f3f1aeae099388f63e4cec3ad08aff8e7da (patch)
tree442ca9c3e25df79a5913d5796588448d9ece6101 /drivers/staging/ozwpan/ozevent.c
parent13c398aa5c9a7eae7792726f6229d961e2915b12 (diff)
downloadlinux-fsl-qoriq-68a75f3f1aeae099388f63e4cec3ad08aff8e7da.tar.xz
staging: ozwpan: Replace existing event logging mechanism
This patch replaces existing event logging mechanism from ioctl to debugfs. This patch replaces previous patch submitted by Chris Kelly. Previous patch can be found at :- http://article.gmane.org/gmane.linux.usb.general/60026/ Signed-off-by: Rupesh Gujare <rgujare@ozmodevices.com> Signed-off-by: Chris Kelly <ckelly@ozmodevices.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Diffstat (limited to 'drivers/staging/ozwpan/ozevent.c')
-rw-r--r--drivers/staging/ozwpan/ozevent.c195
1 files changed, 137 insertions, 58 deletions
diff --git a/drivers/staging/ozwpan/ozevent.c b/drivers/staging/ozwpan/ozevent.c
index 73703d3..7f66b4f 100644
--- a/drivers/staging/ozwpan/ozevent.c
+++ b/drivers/staging/ozwpan/ozevent.c
@@ -5,29 +5,46 @@
*/
#include "ozconfig.h"
#ifdef WANT_EVENT_TRACE
+#include <linux/module.h>
+#include <linux/debugfs.h>
#include <linux/jiffies.h>
#include <linux/uaccess.h>
#include "oztrace.h"
#include "ozevent.h"
+#include "ozappif.h"
/*------------------------------------------------------------------------------
+ * Although the event mask is logically part of the oz_evtdev structure, it is
+ * needed outside of this file so define it seperately to avoid the need to
+ * export definition of struct oz_evtdev.
*/
-unsigned long g_evt_mask = 0xffffffff;
+u32 g_evt_mask;
/*------------------------------------------------------------------------------
*/
#define OZ_MAX_EVTS 2048 /* Must be power of 2 */
-DEFINE_SPINLOCK(g_eventlock);
-static int g_evt_in;
-static int g_evt_out;
-static int g_missed_events;
-static struct oz_event g_events[OZ_MAX_EVTS];
+struct oz_evtdev {
+ struct dentry *root_dir;
+ int evt_in;
+ int evt_out;
+ int missed_events;
+ int present;
+ atomic_t users;
+ spinlock_t lock;
+ struct oz_event evts[OZ_MAX_EVTS];
+};
+
+static struct oz_evtdev g_evtdev;
+
/*------------------------------------------------------------------------------
* Context: process
*/
void oz_event_init(void)
{
+ /* Because g_evtdev is static external all fields initally zero so no
+ * need to reinitialised those.
+ */
oz_trace("Event tracing initialized\n");
- g_evt_in = g_evt_out = 0;
- g_missed_events = 0;
+ spin_lock_init(&g_evtdev.lock);
+ atomic_set(&g_evtdev.users, 0);
}
/*------------------------------------------------------------------------------
* Context: process
@@ -43,74 +60,136 @@ void oz_event_log2(u8 evt, u8 ctx1, u16 ctx2, void *ctx3, unsigned ctx4)
{
unsigned long irqstate;
int ix;
- spin_lock_irqsave(&g_eventlock, irqstate);
- ix = (g_evt_in + 1) & (OZ_MAX_EVTS - 1);
- if (ix != g_evt_out) {
- struct oz_event *e = &g_events[g_evt_in];
+ spin_lock_irqsave(&g_evtdev.lock, irqstate);
+ ix = (g_evtdev.evt_in + 1) & (OZ_MAX_EVTS - 1);
+ if (ix != g_evtdev.evt_out) {
+ struct oz_event *e = &g_evtdev.evts[g_evtdev.evt_in];
e->jiffies = jiffies;
e->evt = evt;
e->ctx1 = ctx1;
e->ctx2 = ctx2;
- e->ctx3 = ctx3;
+ e->ctx3 = (__u32)(unsigned long)ctx3;
e->ctx4 = ctx4;
- g_evt_in = ix;
+ g_evtdev.evt_in = ix;
} else {
- g_missed_events++;
+ g_evtdev.missed_events++;
}
- spin_unlock_irqrestore(&g_eventlock, irqstate);
+ spin_unlock_irqrestore(&g_evtdev.lock, irqstate);
}
/*------------------------------------------------------------------------------
* Context: process
*/
-int oz_events_copy(struct oz_evtlist __user *lst)
+static void oz_events_clear(struct oz_evtdev *dev)
{
- int first;
- int ix;
- struct hdr {
- int count;
- int missed;
- } hdr;
- ix = g_evt_out;
- hdr.count = g_evt_in - ix;
- if (hdr.count < 0)
- hdr.count += OZ_MAX_EVTS;
- if (hdr.count > OZ_EVT_LIST_SZ)
- hdr.count = OZ_EVT_LIST_SZ;
- hdr.missed = g_missed_events;
- g_missed_events = 0;
- if (copy_to_user((void __user *)lst, &hdr, sizeof(hdr)))
- return -EFAULT;
- first = OZ_MAX_EVTS - ix;
- if (first > hdr.count)
- first = hdr.count;
- if (first) {
- int sz = first*sizeof(struct oz_event);
- void __user *p = (void __user *)lst->evts;
- if (copy_to_user(p, &g_events[ix], sz))
- return -EFAULT;
- if (hdr.count > first) {
- p = (void __user *)&lst->evts[first];
- sz = (hdr.count-first)*sizeof(struct oz_event);
- if (copy_to_user(p, g_events, sz))
- return -EFAULT;
- }
+ unsigned long irqstate;
+ oz_trace("Clearing events\n");
+ spin_lock_irqsave(&dev->lock, irqstate);
+ dev->evt_in = dev->evt_out = 0;
+ dev->missed_events = 0;
+ spin_unlock_irqrestore(&dev->lock, irqstate);
+}
+#ifdef CONFIG_DEBUG_FS
+/*------------------------------------------------------------------------------
+ * Context: process
+ */
+int oz_events_open(struct inode *inode, struct file *filp)
+{
+ oz_trace("oz_evt_open()\n");
+ oz_trace("Open flags: 0x%x\n", filp->f_flags);
+ if (atomic_add_return(1, &g_evtdev.users) == 1) {
+ oz_events_clear(&g_evtdev);
+ return nonseekable_open(inode, filp);
+ } else {
+ atomic_dec(&g_evtdev.users);
+ return -EBUSY;
}
- ix += hdr.count;
- if (ix >= OZ_MAX_EVTS)
- ix -= OZ_MAX_EVTS;
- g_evt_out = ix;
+}
+/*------------------------------------------------------------------------------
+ * Context: process
+ */
+int oz_events_release(struct inode *inode, struct file *filp)
+{
+ oz_events_clear(&g_evtdev);
+ atomic_dec(&g_evtdev.users);
+ g_evt_mask = 0;
+ oz_trace("oz_evt_release()\n");
return 0;
}
/*------------------------------------------------------------------------------
* Context: process
*/
-void oz_events_clear(void)
+ssize_t oz_events_read(struct file *filp, char __user *buf, size_t count,
+ loff_t *fpos)
{
- unsigned long irqstate;
- spin_lock_irqsave(&g_eventlock, irqstate);
- g_evt_in = g_evt_out = 0;
- g_missed_events = 0;
- spin_unlock_irqrestore(&g_eventlock, irqstate);
+ struct oz_evtdev *dev = &g_evtdev;
+ int rc = 0;
+ int nb_evts = count / sizeof(struct oz_event);
+ int n;
+ int sz;
+
+ n = dev->evt_in - dev->evt_out;
+ if (n < 0)
+ n += OZ_MAX_EVTS;
+ if (nb_evts > n)
+ nb_evts = n;
+ if (nb_evts == 0)
+ goto out;
+ n = OZ_MAX_EVTS - dev->evt_out;
+ if (n > nb_evts)
+ n = nb_evts;
+ sz = n * sizeof(struct oz_event);
+ if (copy_to_user(buf, &dev->evts[dev->evt_out], sz)) {
+ rc = -EFAULT;
+ goto out;
+ }
+ if (n == nb_evts)
+ goto out2;
+ n = nb_evts - n;
+ if (copy_to_user(buf + sz, dev->evts, n * sizeof(struct oz_event))) {
+ rc = -EFAULT;
+ goto out;
+ }
+out2:
+ dev->evt_out = (dev->evt_out + nb_evts) & (OZ_MAX_EVTS - 1);
+ rc = nb_evts * sizeof(struct oz_event);
+out:
+ return rc;
}
-#endif /* WANT_EVENT_TRACE */
+/*------------------------------------------------------------------------------
+ */
+const struct file_operations oz_events_fops = {
+ .owner = THIS_MODULE,
+ .open = oz_events_open,
+ .release = oz_events_release,
+ .read = oz_events_read,
+};
+/*------------------------------------------------------------------------------
+ * Context: process
+ */
+void oz_debugfs_init(void)
+{
+ struct dentry *parent;
+ parent = debugfs_create_dir("ozwpan", NULL);
+ if (parent == NULL) {
+ oz_trace("Failed to create debugfs directory ozmo\n");
+ return;
+ } else {
+ g_evtdev.root_dir = parent;
+ if (debugfs_create_file("events", S_IRUSR, parent, NULL,
+ &oz_events_fops) == NULL)
+ oz_trace("Failed to create file ozmo/events\n");
+ if (debugfs_create_x32("event_mask", S_IRUSR | S_IWUSR, parent,
+ &g_evt_mask) == NULL)
+ oz_trace("Failed to create file ozmo/event_mask\n");
+ }
+}
+/*------------------------------------------------------------------------------
+ * Context: process
+ */
+void oz_debugfs_remove(void)
+{
+ debugfs_remove_recursive(g_evtdev.root_dir);
+}
+#endif /* CONFIG_DEBUG_FS */
+#endif /* WANT_EVENT_TRACE */