Currently, user-space extracts data from the ring-buffer via splice,
which is handy for storage or network sharing. However, due to splice
limitations, it is imposible to do real-time analysis without a copy.
A solution for that problem is to let the user-space map the ring-buffer
directly.
The mapping is exposed via the per-CPU file trace_pipe_raw. The first
element of the mapping is the meta-page. It is followed by each
subbuffer constituting the ring-buffer, ordered by their unique page ID:
* Meta-page -- include/uapi/linux/trace_mmap.h for a description
* Subbuf ID 0
* Subbuf ID 1
...
It is therefore easy to translate a subbuf ID into an offset in the
mapping:
reader_id = meta->reader->id;
reader_offset = meta->meta_page_size + reader_id * meta->subbuf_size;
When new data is available, the mapper must call a newly introduced ioctl:
TRACE_MMAP_IOCTL_GET_READER. This will update the Meta-page reader ID to
point to the next reader containing unread data.
Mapping will prevent snapshot and buffer size modifications.
CC: <[email protected]>
Signed-off-by: Vincent Donnefort <[email protected]>
diff --git a/include/uapi/linux/trace_mmap.h b/include/uapi/linux/trace_mmap.h
index ffcd8dfcaa4f..d25b9d504a7c 100644
--- a/include/uapi/linux/trace_mmap.h
+++ b/include/uapi/linux/trace_mmap.h
@@ -43,4 +43,6 @@ struct trace_buffer_meta {
__u64 Reserved2;
};
+#define TRACE_MMAP_IOCTL_GET_READER _IO('T', 0x1)
+
#endif /* _TRACE_MMAP_H_ */
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 233d1af39fff..0f37aa9860fd 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -1191,6 +1191,12 @@ static void tracing_snapshot_instance_cond(struct trace_array *tr,
return;
}
+ if (tr->mapped) {
+ trace_array_puts(tr, "*** BUFFER MEMORY MAPPED ***\n");
+ trace_array_puts(tr, "*** Can not use snapshot (sorry) ***\n");
+ return;
+ }
+
local_irq_save(flags);
update_max_tr(tr, current, smp_processor_id(), cond_data);
local_irq_restore(flags);
@@ -1323,7 +1329,7 @@ static int tracing_arm_snapshot_locked(struct trace_array *tr)
lockdep_assert_held(&trace_types_lock);
spin_lock(&tr->snapshot_trigger_lock);
- if (tr->snapshot == UINT_MAX) {
+ if (tr->snapshot == UINT_MAX || tr->mapped) {
spin_unlock(&tr->snapshot_trigger_lock);
return -EBUSY;
}
@@ -6068,7 +6074,7 @@ static void tracing_set_nop(struct trace_array *tr)
{
if (tr->current_trace == &nop_trace)
return;
-
+
tr->current_trace->enabled--;
if (tr->current_trace->reset)
@@ -8194,15 +8200,32 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
return ret;
}
-/* An ioctl call with cmd 0 to the ring buffer file will wake up all waiters */
static long tracing_buffers_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
struct ftrace_buffer_info *info = file->private_data;
struct trace_iterator *iter = &info->iter;
+ int err;
+
+ if (cmd == TRACE_MMAP_IOCTL_GET_READER) {
+ if (!(file->f_flags & O_NONBLOCK)) {
+ err = ring_buffer_wait(iter->array_buffer->buffer,
+ iter->cpu_file,
+ iter->tr->buffer_percent,
+ NULL, NULL);
+ if (err)
+ return err;
+ }
- if (cmd)
- return -ENOIOCTLCMD;
+ return ring_buffer_map_get_reader(iter->array_buffer->buffer,
+ iter->cpu_file);
+ } else if (cmd) {
+ return -ENOTTY;
+ }
+ /*
+ * An ioctl call with cmd 0 to the ring buffer file will wake up all
+ * waiters
+ */
mutex_lock(&trace_types_lock);
/* Make sure the waiters see the new wait_index */
@@ -8214,6 +8237,94 @@ static long tracing_buffers_ioctl(struct file *file, unsigned int cmd, unsigned
return 0;
}
+static vm_fault_t tracing_buffers_mmap_fault(struct vm_fault *vmf)
+{
+ return VM_FAULT_SIGBUS;
+}
+
+#ifdef CONFIG_TRACER_MAX_TRACE
+static int get_snapshot_map(struct trace_array *tr)
+{
+ int err = 0;
+
+ /*
+ * Called with mmap_lock held. lockdep would be unhappy if we would now
+ * take trace_types_lock. Instead use the specific
+ * snapshot_trigger_lock.
+ */
+ spin_lock(&tr->snapshot_trigger_lock);
+
+ if (tr->snapshot || tr->mapped == UINT_MAX)
+ err = -EBUSY;
+ else
+ tr->mapped++;
+
+ spin_unlock(&tr->snapshot_trigger_lock);
+
+ /* Wait for update_max_tr() to observe iter->tr->mapped */
+ if (tr->mapped == 1)
+ synchronize_rcu();
+
+ return err;
+
+}
+static void put_snapshot_map(struct trace_array *tr)
+{
+ spin_lock(&tr->snapshot_trigger_lock);
+ if (!WARN_ON(!tr->mapped))
+ tr->mapped--;
+ spin_unlock(&tr->snapshot_trigger_lock);
+}
+#else
+static inline int get_snapshot_map(struct trace_array *tr) { return 0; }
+static inline void put_snapshot_map(struct trace_array *tr) { }
+#endif
+
+static void tracing_buffers_mmap_close(struct vm_area_struct *vma)
+{
+ struct ftrace_buffer_info *info = vma->vm_file->private_data;
+ struct trace_iterator *iter = &info->iter;
+
+ WARN_ON(ring_buffer_unmap(iter->array_buffer->buffer, iter->cpu_file));
+ put_snapshot_map(iter->tr);
+}
+
+static void tracing_buffers_mmap_open(struct vm_area_struct *vma) { }
+
+static const struct vm_operations_struct tracing_buffers_vmops = {
+ .open = tracing_buffers_mmap_open,
+ .close = tracing_buffers_mmap_close,
+ .fault = tracing_buffers_mmap_fault,
+};
+
+static int tracing_buffers_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+ struct ftrace_buffer_info *info = filp->private_data;
+ struct trace_iterator *iter = &info->iter;
+ int ret = 0;
+
+ if (vma->vm_flags & VM_WRITE || vma->vm_flags & VM_EXEC ||
+ !(vma->vm_flags & VM_MAYSHARE))
+ return -EPERM;
+
+ vm_flags_mod(vma,
+ VM_MIXEDMAP | VM_PFNMAP |
+ VM_DONTCOPY | VM_DONTDUMP | VM_DONTEXPAND | VM_IO,
+ VM_MAYWRITE);
+
+ vma->vm_ops = &tracing_buffers_vmops;
+
+ ret = get_snapshot_map(iter->tr);
+ if (ret)
+ return ret;
+
+ ret = ring_buffer_map(iter->array_buffer->buffer, iter->cpu_file, vma);
+ if (ret)
+ put_snapshot_map(iter->tr);
+
+ return ret;
+}
+
static const struct file_operations tracing_buffers_fops = {
.open = tracing_buffers_open,
.read = tracing_buffers_read,
@@ -8223,6 +8334,7 @@ static const struct file_operations tracing_buffers_fops = {
.splice_read = tracing_buffers_splice_read,
.unlocked_ioctl = tracing_buffers_ioctl,
.llseek = no_llseek,
+ .mmap = tracing_buffers_mmap,
};
static ssize_t
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 64450615ca0c..749a182dab48 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -336,6 +336,7 @@ struct trace_array {
bool allocated_snapshot;
spinlock_t snapshot_trigger_lock;
unsigned int snapshot;
+ unsigned int mapped;
unsigned long max_latency;
#ifdef CONFIG_FSNOTIFY
struct dentry *d_max_latency;
--
2.44.0.396.g6e790dbe36-goog
On Tue, 26 Mar 2024 10:08:28 +0000
Vincent Donnefort <[email protected]> wrote:
> Currently, user-space extracts data from the ring-buffer via splice,
> which is handy for storage or network sharing. However, due to splice
> limitations, it is imposible to do real-time analysis without a copy.
>
> A solution for that problem is to let the user-space map the ring-buffer
> directly.
>
> The mapping is exposed via the per-CPU file trace_pipe_raw. The first
> element of the mapping is the meta-page. It is followed by each
> subbuffer constituting the ring-buffer, ordered by their unique page ID:
>
> * Meta-page -- include/uapi/linux/trace_mmap.h for a description
> * Subbuf ID 0
> * Subbuf ID 1
> ...
>
> It is therefore easy to translate a subbuf ID into an offset in the
> mapping:
>
> reader_id = meta->reader->id;
> reader_offset = meta->meta_page_size + reader_id * meta->subbuf_size;
>
> When new data is available, the mapper must call a newly introduced ioctl:
> TRACE_MMAP_IOCTL_GET_READER. This will update the Meta-page reader ID to
> point to the next reader containing unread data.
>
Thanks for the update Vincent!
> Mapping will prevent snapshot and buffer size modifications.
>
> CC: <[email protected]>
> Signed-off-by: Vincent Donnefort <[email protected]>
>
> diff --git a/include/uapi/linux/trace_mmap.h b/include/uapi/linux/trace_mmap.h
> index ffcd8dfcaa4f..d25b9d504a7c 100644
> --- a/include/uapi/linux/trace_mmap.h
> +++ b/include/uapi/linux/trace_mmap.h
> @@ -43,4 +43,6 @@ struct trace_buffer_meta {
> __u64 Reserved2;
> };
>
> +#define TRACE_MMAP_IOCTL_GET_READER _IO('T', 0x1)
> +
> #endif /* _TRACE_MMAP_H_ */
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index 233d1af39fff..0f37aa9860fd 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -1191,6 +1191,12 @@ static void tracing_snapshot_instance_cond(struct trace_array *tr,
> return;
> }
>
> + if (tr->mapped) {
> + trace_array_puts(tr, "*** BUFFER MEMORY MAPPED ***\n");
> + trace_array_puts(tr, "*** Can not use snapshot (sorry) ***\n");
> + return;
> + }
> +
> local_irq_save(flags);
> update_max_tr(tr, current, smp_processor_id(), cond_data);
> local_irq_restore(flags);
> @@ -1323,7 +1329,7 @@ static int tracing_arm_snapshot_locked(struct trace_array *tr)
> lockdep_assert_held(&trace_types_lock);
>
> spin_lock(&tr->snapshot_trigger_lock);
> - if (tr->snapshot == UINT_MAX) {
> + if (tr->snapshot == UINT_MAX || tr->mapped) {
> spin_unlock(&tr->snapshot_trigger_lock);
> return -EBUSY;
> }
> @@ -6068,7 +6074,7 @@ static void tracing_set_nop(struct trace_array *tr)
> {
> if (tr->current_trace == &nop_trace)
> return;
> -
> +
> tr->current_trace->enabled--;
>
> if (tr->current_trace->reset)
> @@ -8194,15 +8200,32 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos,
> return ret;
> }
>
> -/* An ioctl call with cmd 0 to the ring buffer file will wake up all waiters */
> static long tracing_buffers_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
> {
> struct ftrace_buffer_info *info = file->private_data;
> struct trace_iterator *iter = &info->iter;
> + int err;
> +
> + if (cmd == TRACE_MMAP_IOCTL_GET_READER) {
> + if (!(file->f_flags & O_NONBLOCK)) {
> + err = ring_buffer_wait(iter->array_buffer->buffer,
> + iter->cpu_file,
> + iter->tr->buffer_percent,
> + NULL, NULL);
> + if (err)
> + return err;
> + }
>
> - if (cmd)
> - return -ENOIOCTLCMD;
> + return ring_buffer_map_get_reader(iter->array_buffer->buffer,
> + iter->cpu_file);
> + } else if (cmd) {
> + return -ENOTTY;
> + }
>
> + /*
> + * An ioctl call with cmd 0 to the ring buffer file will wake up all
> + * waiters
> + */
> mutex_lock(&trace_types_lock);
>
> /* Make sure the waiters see the new wait_index */
> @@ -8214,6 +8237,94 @@ static long tracing_buffers_ioctl(struct file *file, unsigned int cmd, unsigned
> return 0;
> }
>
> +static vm_fault_t tracing_buffers_mmap_fault(struct vm_fault *vmf)
> +{
> + return VM_FAULT_SIGBUS;
> +}
If this is all it does, I don't believe it's needed.
> +
> +#ifdef CONFIG_TRACER_MAX_TRACE
> +static int get_snapshot_map(struct trace_array *tr)
> +{
> + int err = 0;
> +
> + /*
> + * Called with mmap_lock held. lockdep would be unhappy if we would now
> + * take trace_types_lock. Instead use the specific
> + * snapshot_trigger_lock.
> + */
> + spin_lock(&tr->snapshot_trigger_lock);
> +
> + if (tr->snapshot || tr->mapped == UINT_MAX)
> + err = -EBUSY;
> + else
> + tr->mapped++;
> +
> + spin_unlock(&tr->snapshot_trigger_lock);
> +
> + /* Wait for update_max_tr() to observe iter->tr->mapped */
> + if (tr->mapped == 1)
> + synchronize_rcu();
> +
> + return err;
> +
> +}
> +static void put_snapshot_map(struct trace_array *tr)
> +{
> + spin_lock(&tr->snapshot_trigger_lock);
> + if (!WARN_ON(!tr->mapped))
> + tr->mapped--;
> + spin_unlock(&tr->snapshot_trigger_lock);
> +}
> +#else
> +static inline int get_snapshot_map(struct trace_array *tr) { return 0; }
> +static inline void put_snapshot_map(struct trace_array *tr) { }
> +#endif
> +
> +static void tracing_buffers_mmap_close(struct vm_area_struct *vma)
> +{
> + struct ftrace_buffer_info *info = vma->vm_file->private_data;
> + struct trace_iterator *iter = &info->iter;
> +
> + WARN_ON(ring_buffer_unmap(iter->array_buffer->buffer, iter->cpu_file));
> + put_snapshot_map(iter->tr);
> +}
> +
> +static void tracing_buffers_mmap_open(struct vm_area_struct *vma) { }
Same for the open.
> +
> +static const struct vm_operations_struct tracing_buffers_vmops = {
> + .open = tracing_buffers_mmap_open,
> + .close = tracing_buffers_mmap_close,
> + .fault = tracing_buffers_mmap_fault,
> +};
I replaced this with:
static const struct vm_operations_struct tracing_buffers_vmops = {
.close = tracing_buffers_mmap_close,
};
And it appears to work just fine. The mm code handles the NULL cases for
open and .fault.
Is there any reason to do something different than the mm defaults?
-- Steve
> +
> +static int tracing_buffers_mmap(struct file *filp, struct vm_area_struct *vma)
> +{
> + struct ftrace_buffer_info *info = filp->private_data;
> + struct trace_iterator *iter = &info->iter;
> + int ret = 0;
> +
> + if (vma->vm_flags & VM_WRITE || vma->vm_flags & VM_EXEC ||
> + !(vma->vm_flags & VM_MAYSHARE))
> + return -EPERM;
> +
> + vm_flags_mod(vma,
> + VM_MIXEDMAP | VM_PFNMAP |
> + VM_DONTCOPY | VM_DONTDUMP | VM_DONTEXPAND | VM_IO,
> + VM_MAYWRITE);
> +
> + vma->vm_ops = &tracing_buffers_vmops;
> +
> + ret = get_snapshot_map(iter->tr);
> + if (ret)
> + return ret;
> +
> + ret = ring_buffer_map(iter->array_buffer->buffer, iter->cpu_file, vma);
> + if (ret)
> + put_snapshot_map(iter->tr);
> +
> + return ret;
> +}
> +
> static const struct file_operations tracing_buffers_fops = {
> .open = tracing_buffers_open,
> .read = tracing_buffers_read,
> @@ -8223,6 +8334,7 @@ static const struct file_operations tracing_buffers_fops = {
> .splice_read = tracing_buffers_splice_read,
> .unlocked_ioctl = tracing_buffers_ioctl,
> .llseek = no_llseek,
> + .mmap = tracing_buffers_mmap,
> };
>
> static ssize_t
> diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
> index 64450615ca0c..749a182dab48 100644
> --- a/kernel/trace/trace.h
> +++ b/kernel/trace/trace.h
> @@ -336,6 +336,7 @@ struct trace_array {
> bool allocated_snapshot;
> spinlock_t snapshot_trigger_lock;
> unsigned int snapshot;
> + unsigned int mapped;
> unsigned long max_latency;
> #ifdef CONFIG_FSNOTIFY
> struct dentry *d_max_latency;
On Wed, Apr 03, 2024 at 10:13:52AM -0400, Steven Rostedt wrote:
> On Fri, 29 Mar 2024 14:40:55 -0400
> Steven Rostedt <[email protected]> wrote:
>
> > > +static vm_fault_t tracing_buffers_mmap_fault(struct vm_fault *vmf)
> > > +{
> > > + return VM_FAULT_SIGBUS;
> > > +}
> >
> > If this is all it does, I don't believe it's needed.
> >
> > > +
> > > +#ifdef CONFIG_TRACER_MAX_TRACE
> > > +static int get_snapshot_map(struct trace_array *tr)
> > > +{
> > > + int err = 0;
> > > +
> > > + /*
> > > + * Called with mmap_lock held. lockdep would be unhappy if we would now
> > > + * take trace_types_lock. Instead use the specific
> > > + * snapshot_trigger_lock.
> > > + */
> > > + spin_lock(&tr->snapshot_trigger_lock);
> > > +
> > > + if (tr->snapshot || tr->mapped == UINT_MAX)
> > > + err = -EBUSY;
> > > + else
> > > + tr->mapped++;
> > > +
> > > + spin_unlock(&tr->snapshot_trigger_lock);
> > > +
> > > + /* Wait for update_max_tr() to observe iter->tr->mapped */
> > > + if (tr->mapped == 1)
> > > + synchronize_rcu();
> > > +
> > > + return err;
> > > +
> > > +}
> > > +static void put_snapshot_map(struct trace_array *tr)
> > > +{
> > > + spin_lock(&tr->snapshot_trigger_lock);
> > > + if (!WARN_ON(!tr->mapped))
> > > + tr->mapped--;
> > > + spin_unlock(&tr->snapshot_trigger_lock);
> > > +}
> > > +#else
> > > +static inline int get_snapshot_map(struct trace_array *tr) { return 0; }
> > > +static inline void put_snapshot_map(struct trace_array *tr) { }
> > > +#endif
> > > +
> > > +static void tracing_buffers_mmap_close(struct vm_area_struct *vma)
> > > +{
> > > + struct ftrace_buffer_info *info = vma->vm_file->private_data;
> > > + struct trace_iterator *iter = &info->iter;
> > > +
> > > + WARN_ON(ring_buffer_unmap(iter->array_buffer->buffer, iter->cpu_file));
> > > + put_snapshot_map(iter->tr);
> > > +}
> > > +
> > > +static void tracing_buffers_mmap_open(struct vm_area_struct *vma) { }
> >
> > Same for the open.
> >
> >
> > > +
> > > +static const struct vm_operations_struct tracing_buffers_vmops = {
> > > + .open = tracing_buffers_mmap_open,
> > > + .close = tracing_buffers_mmap_close,
> > > + .fault = tracing_buffers_mmap_fault,
> > > +};
> >
> > I replaced this with:
> >
> > static const struct vm_operations_struct tracing_buffers_vmops = {
> > .close = tracing_buffers_mmap_close,
> > };
> >
> > And it appears to work just fine. The mm code handles the NULL cases for
> > .open and .fault.
> >
> > Is there any reason to do something different than the mm defaults?
No other reason here than my own ignorance. I will remove.
>
> Hi Vincent,
>
> Do you plan on sending out a v20 series?
Of course, let me spin that this week! Got also few typos to fix in the doc and
I believe an include missing for riscv.
>
> -- Steve
On Fri, 29 Mar 2024 14:40:55 -0400
Steven Rostedt <[email protected]> wrote:
> > +static vm_fault_t tracing_buffers_mmap_fault(struct vm_fault *vmf)
> > +{
> > + return VM_FAULT_SIGBUS;
> > +}
>
> If this is all it does, I don't believe it's needed.
>
> > +
> > +#ifdef CONFIG_TRACER_MAX_TRACE
> > +static int get_snapshot_map(struct trace_array *tr)
> > +{
> > + int err = 0;
> > +
> > + /*
> > + * Called with mmap_lock held. lockdep would be unhappy if we would now
> > + * take trace_types_lock. Instead use the specific
> > + * snapshot_trigger_lock.
> > + */
> > + spin_lock(&tr->snapshot_trigger_lock);
> > +
> > + if (tr->snapshot || tr->mapped == UINT_MAX)
> > + err = -EBUSY;
> > + else
> > + tr->mapped++;
> > +
> > + spin_unlock(&tr->snapshot_trigger_lock);
> > +
> > + /* Wait for update_max_tr() to observe iter->tr->mapped */
> > + if (tr->mapped == 1)
> > + synchronize_rcu();
> > +
> > + return err;
> > +
> > +}
> > +static void put_snapshot_map(struct trace_array *tr)
> > +{
> > + spin_lock(&tr->snapshot_trigger_lock);
> > + if (!WARN_ON(!tr->mapped))
> > + tr->mapped--;
> > + spin_unlock(&tr->snapshot_trigger_lock);
> > +}
> > +#else
> > +static inline int get_snapshot_map(struct trace_array *tr) { return 0; }
> > +static inline void put_snapshot_map(struct trace_array *tr) { }
> > +#endif
> > +
> > +static void tracing_buffers_mmap_close(struct vm_area_struct *vma)
> > +{
> > + struct ftrace_buffer_info *info = vma->vm_file->private_data;
> > + struct trace_iterator *iter = &info->iter;
> > +
> > + WARN_ON(ring_buffer_unmap(iter->array_buffer->buffer, iter->cpu_file));
> > + put_snapshot_map(iter->tr);
> > +}
> > +
> > +static void tracing_buffers_mmap_open(struct vm_area_struct *vma) { }
>
> Same for the open.
>
>
> > +
> > +static const struct vm_operations_struct tracing_buffers_vmops = {
> > + .open = tracing_buffers_mmap_open,
> > + .close = tracing_buffers_mmap_close,
> > + .fault = tracing_buffers_mmap_fault,
> > +};
>
> I replaced this with:
>
> static const struct vm_operations_struct tracing_buffers_vmops = {
> .close = tracing_buffers_mmap_close,
> };
>
> And it appears to work just fine. The mm code handles the NULL cases for
> .open and .fault.
>
> Is there any reason to do something different than the mm defaults?
Hi Vincent,
Do you plan on sending out a v20 series?
-- Steve
On Wed, 3 Apr 2024 15:39:44 +0100
Vincent Donnefort <[email protected]> wrote:
> > Do you plan on sending out a v20 series?
>
> Of course, let me spin that this week! Got also few typos to fix in the doc and
> I believe an include missing for riscv.
No rush, I'll be on PTO until next Tuesday, and will not get to it before then.
-- Steve