tracing: Fix polling on trace_pipe_raw
authorSteven Rostedt <srostedt@redhat.com>
Thu, 28 Feb 2013 14:17:16 +0000 (09:17 -0500)
committerSteven Rostedt <rostedt@goodmis.org>
Fri, 15 Mar 2013 04:34:49 +0000 (00:34 -0400)
The trace_pipe_raw never implemented polling and this was casing
issues for several utilities. This is now implemented.

Blocked reads still are on the TODO list.

Reported-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Tested-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
kernel/trace/trace.c

index 598a7aa7d0ae9f81d74205b11c0728629aee6f94..4a6e461273a9b7baa5188a70fdb92ae637cb7d36 100644 (file)
@@ -3555,10 +3555,8 @@ static int tracing_release_pipe(struct inode *inode, struct file *file)
 }
 
 static unsigned int
-tracing_poll_pipe(struct file *filp, poll_table *poll_table)
+trace_poll(struct trace_iterator *iter, struct file *filp, poll_table *poll_table)
 {
-       struct trace_iterator *iter = filp->private_data;
-
        if (trace_flags & TRACE_ITER_BLOCK) {
                /*
                 * Always select as readable when in blocking mode
@@ -3567,6 +3565,7 @@ tracing_poll_pipe(struct file *filp, poll_table *poll_table)
        } else {
                if (!trace_empty(iter))
                        return POLLIN | POLLRDNORM;
+               trace_wakeup_needed = true;
                poll_wait(filp, &trace_wait, poll_table);
                if (!trace_empty(iter))
                        return POLLIN | POLLRDNORM;
@@ -3575,6 +3574,14 @@ tracing_poll_pipe(struct file *filp, poll_table *poll_table)
        }
 }
 
+static unsigned int
+tracing_poll_pipe(struct file *filp, poll_table *poll_table)
+{
+       struct trace_iterator *iter = filp->private_data;
+
+       return trace_poll(iter, filp, poll_table);
+}
+
 /*
  * This is a make-shift waitqueue.
  * A tracer might use this callback on some rare cases:
@@ -4362,9 +4369,8 @@ static const struct file_operations snapshot_fops = {
 #endif /* CONFIG_TRACER_SNAPSHOT */
 
 struct ftrace_buffer_info {
-       struct trace_array      *tr;
+       struct trace_iterator   iter;
        void                    *spare;
-       int                     cpu;
        unsigned int            read;
 };
 
@@ -4381,22 +4387,32 @@ static int tracing_buffers_open(struct inode *inode, struct file *filp)
        if (!info)
                return -ENOMEM;
 
-       info->tr        = tr;
-       info->cpu       = tc->cpu;
-       info->spare     = NULL;
+       info->iter.tr           = tr;
+       info->iter.cpu_file     = tc->cpu;
+       info->spare             = NULL;
        /* Force reading ring buffer for first read */
-       info->read      = (unsigned int)-1;
+       info->read              = (unsigned int)-1;
 
        filp->private_data = info;
 
        return nonseekable_open(inode, filp);
 }
 
+static unsigned int
+tracing_buffers_poll(struct file *filp, poll_table *poll_table)
+{
+       struct ftrace_buffer_info *info = filp->private_data;
+       struct trace_iterator *iter = &info->iter;
+
+       return trace_poll(iter, filp, poll_table);
+}
+
 static ssize_t
 tracing_buffers_read(struct file *filp, char __user *ubuf,
                     size_t count, loff_t *ppos)
 {
        struct ftrace_buffer_info *info = filp->private_data;
+       struct trace_iterator *iter = &info->iter;
        ssize_t ret;
        size_t size;
 
@@ -4404,7 +4420,7 @@ tracing_buffers_read(struct file *filp, char __user *ubuf,
                return 0;
 
        if (!info->spare)
-               info->spare = ring_buffer_alloc_read_page(info->tr->buffer, info->cpu);
+               info->spare = ring_buffer_alloc_read_page(iter->tr->buffer, iter->cpu_file);
        if (!info->spare)
                return -ENOMEM;
 
@@ -4412,12 +4428,12 @@ tracing_buffers_read(struct file *filp, char __user *ubuf,
        if (info->read < PAGE_SIZE)
                goto read;
 
-       trace_access_lock(info->cpu);
-       ret = ring_buffer_read_page(info->tr->buffer,
+       trace_access_lock(iter->cpu_file);
+       ret = ring_buffer_read_page(iter->tr->buffer,
                                    &info->spare,
                                    count,
-                                   info->cpu, 0);
-       trace_access_unlock(info->cpu);
+                                   iter->cpu_file, 0);
+       trace_access_unlock(iter->cpu_file);
        if (ret < 0)
                return 0;
 
@@ -4442,9 +4458,10 @@ read:
 static int tracing_buffers_release(struct inode *inode, struct file *file)
 {
        struct ftrace_buffer_info *info = file->private_data;
+       struct trace_iterator *iter = &info->iter;
 
        if (info->spare)
-               ring_buffer_free_read_page(info->tr->buffer, info->spare);
+               ring_buffer_free_read_page(iter->tr->buffer, info->spare);
        kfree(info);
 
        return 0;
@@ -4511,6 +4528,7 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
                            unsigned int flags)
 {
        struct ftrace_buffer_info *info = file->private_data;
+       struct trace_iterator *iter = &info->iter;
        struct partial_page partial_def[PIPE_DEF_BUFFERS];
        struct page *pages_def[PIPE_DEF_BUFFERS];
        struct splice_pipe_desc spd = {
@@ -4541,8 +4559,9 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
                len &= PAGE_MASK;
        }
 
-       trace_access_lock(info->cpu);
-       entries = ring_buffer_entries_cpu(info->tr->buffer, info->cpu);
+ again:
+       trace_access_lock(iter->cpu_file);
+       entries = ring_buffer_entries_cpu(iter->tr->buffer, iter->cpu_file);
 
        for (i = 0; i < pipe->buffers && len && entries; i++, len -= PAGE_SIZE) {
                struct page *page;
@@ -4553,15 +4572,15 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
                        break;
 
                ref->ref = 1;
-               ref->buffer = info->tr->buffer;
-               ref->page = ring_buffer_alloc_read_page(ref->buffer, info->cpu);
+               ref->buffer = iter->tr->buffer;
+               ref->page = ring_buffer_alloc_read_page(ref->buffer, iter->cpu_file);
                if (!ref->page) {
                        kfree(ref);
                        break;
                }
 
                r = ring_buffer_read_page(ref->buffer, &ref->page,
-                                         len, info->cpu, 1);
+                                         len, iter->cpu_file, 1);
                if (r < 0) {
                        ring_buffer_free_read_page(ref->buffer, ref->page);
                        kfree(ref);
@@ -4585,20 +4604,24 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
                spd.nr_pages++;
                *ppos += PAGE_SIZE;
 
-               entries = ring_buffer_entries_cpu(info->tr->buffer, info->cpu);
+               entries = ring_buffer_entries_cpu(iter->tr->buffer, iter->cpu_file);
        }
 
-       trace_access_unlock(info->cpu);
+       trace_access_unlock(iter->cpu_file);
        spd.nr_pages = i;
 
        /* did we read anything? */
        if (!spd.nr_pages) {
-               if ((file->f_flags & O_NONBLOCK) || (flags & SPLICE_F_NONBLOCK))
+               if ((file->f_flags & O_NONBLOCK) || (flags & SPLICE_F_NONBLOCK)) {
                        ret = -EAGAIN;
-               else
-                       ret = 0;
-               /* TODO: block */
-               goto out;
+                       goto out;
+               }
+               default_wait_pipe(iter);
+               if (signal_pending(current)) {
+                       ret = -EINTR;
+                       goto out;
+               }
+               goto again;
        }
 
        ret = splice_to_pipe(pipe, &spd);
@@ -4610,6 +4633,7 @@ out:
 static const struct file_operations tracing_buffers_fops = {
        .open           = tracing_buffers_open,
        .read           = tracing_buffers_read,
+       .poll           = tracing_buffers_poll,
        .release        = tracing_buffers_release,
        .splice_read    = tracing_buffers_splice_read,
        .llseek         = no_llseek,