Revert "Cleanup: ring buffer: remove lib_ring_buffer_switch_new_end()"
authorMathieu Desnoyers <mathieu.desnoyers@efficios.com>
Sun, 30 Jun 2013 22:36:47 +0000 (18:36 -0400)
committerMathieu Desnoyers <mathieu.desnoyers@efficios.com>
Sun, 30 Jun 2013 22:48:41 +0000 (18:48 -0400)
This reverts commit 019fbcb65ec7eec2c86019ad1afce8fa47dc1269.

Breaks regression tests of lttng-tools.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
libringbuffer/frontend_api.h
libringbuffer/ring_buffer_frontend.c

index b5406b0ef9d3678d44f706b23d66117b5659fd9e..a2a9af39aea777998c5d96cdde5f632f500c8cf7 100644 (file)
@@ -126,6 +126,14 @@ int lib_ring_buffer_try_reserve(const struct lttng_ust_lib_ring_buffer_config *c
         * boundary. It's safe to write.
         */
        *o_end = *o_begin + ctx->slot_size;
+
+       if (caa_unlikely((subbuf_offset(*o_end, chan)) == 0))
+               /*
+                * The offset_end will fall at the very beginning of the next
+                * subbuffer.
+                */
+               return 1;
+
        return 0;
 }
 
index 84c6726e97a2d66ec612bc083aaa4ca36ab6225a..c177f337da01aa4f457fbea66add969a837d691f 100644 (file)
 struct switch_offsets {
        unsigned long begin, end, old;
        size_t pre_header_padding, size;
-       unsigned int switch_new_start:1, switch_old_start:1, switch_old_end:1;
+       unsigned int switch_new_start:1, switch_new_end:1, switch_old_start:1,
+                    switch_old_end:1;
 };
 
 DEFINE_URCU_TLS(unsigned int, lib_ring_buffer_nesting);
@@ -1414,6 +1415,42 @@ void lib_ring_buffer_switch_new_start(struct lttng_ust_lib_ring_buffer *buf,
                                             handle);
 }
 
+/*
+ * lib_ring_buffer_switch_new_end: finish switching current subbuffer
+ *
+ * The only remaining threads could be the ones with pending commits. They will
+ * have to do the deliver themselves.
+ */
+static
+void lib_ring_buffer_switch_new_end(struct lttng_ust_lib_ring_buffer *buf,
+                                   struct channel *chan,
+                                   struct switch_offsets *offsets,
+                                   uint64_t tsc,
+                                   struct lttng_ust_shm_handle *handle)
+{
+       const struct lttng_ust_lib_ring_buffer_config *config = &chan->backend.config;
+       unsigned long endidx = subbuf_index(offsets->end - 1, chan);
+       unsigned long commit_count, padding_size, data_size;
+
+       data_size = subbuf_offset(offsets->end - 1, chan) + 1;
+       padding_size = chan->backend.subbuf_size - data_size;
+       subbuffer_set_data_size(config, &buf->backend, endidx, data_size,
+                               handle);
+
+       /*
+        * Order all writes to buffer before the commit count update that will
+        * determine that the subbuffer is full.
+        */
+       cmm_smp_wmb();
+       v_add(config, padding_size, &shmp_index(handle, buf->commit_hot, endidx)->cc);
+       commit_count = v_read(config, &shmp_index(handle, buf->commit_hot, endidx)->cc);
+       lib_ring_buffer_check_deliver(config, buf, chan, offsets->end - 1,
+                                 commit_count, endidx, handle);
+       lib_ring_buffer_write_commit_counter(config, buf, chan, endidx,
+                                            offsets->end, commit_count,
+                                            padding_size, handle);
+}
+
 /*
  * Returns :
  * 0 if ok
@@ -1566,6 +1603,7 @@ int lib_ring_buffer_try_reserve_slow(struct lttng_ust_lib_ring_buffer *buf,
        offsets->begin = v_read(config, &buf->offset);
        offsets->old = offsets->begin;
        offsets->switch_new_start = 0;
+       offsets->switch_new_end = 0;
        offsets->switch_old_end = 0;
        offsets->pre_header_padding = 0;
 
@@ -1697,6 +1735,14 @@ int lib_ring_buffer_try_reserve_slow(struct lttng_ust_lib_ring_buffer *buf,
                 */
        }
        offsets->end = offsets->begin + offsets->size;
+
+       if (caa_unlikely(subbuf_offset(offsets->end, chan) == 0)) {
+               /*
+                * The offset_end will fall at the very beginning of the next
+                * subbuffer.
+                */
+               offsets->switch_new_end = 1;    /* For offsets->begin */
+       }
        return 0;
 }
 
@@ -1770,6 +1816,9 @@ int lib_ring_buffer_reserve_slow(struct lttng_ust_lib_ring_buffer_ctx *ctx)
        if (caa_unlikely(offsets.switch_new_start))
                lib_ring_buffer_switch_new_start(buf, chan, &offsets, ctx->tsc, handle);
 
+       if (caa_unlikely(offsets.switch_new_end))
+               lib_ring_buffer_switch_new_end(buf, chan, &offsets, ctx->tsc, handle);
+
        ctx->slot_size = offsets.size;
        ctx->pre_offset = offsets.begin;
        ctx->buf_offset = offsets.begin + offsets.pre_header_padding;
This page took 0.02702 seconds and 4 git commands to generate.