summaryrefslogtreecommitdiff
path: root/sound/soc/fsl/fsl_dsp_pool.c
diff options
context:
space:
mode:
authorDaniel Baluta <daniel.baluta@nxp.com>2018-10-04 16:34:29 +0300
committerDong Aisheng <aisheng.dong@nxp.com>2019-11-25 15:51:56 +0800
commit17aa1ae7609a2dbb1e705fb6b4766451662afc49 (patch)
treebaa4cd1a9a2ea7cd9070c68a4b0aff03e0518058 /sound/soc/fsl/fsl_dsp_pool.c
parent1aeb3e481a292ae78906e5cb0f4f0d8b57f05ad3 (diff)
MLK-18497-4: ASoC: fsl: dsp: Add message buffer pool API
Memory is allocated to clients from memory pools. A memory pool allocation is requested to DSP framework via XF_ALLOC command and freed via XF_FREE. Memory pool allocation API offers two functions: * xf_pool_alloc, allocate a number of buffers of given length * xf_pool_free, free memory area allocated for a pool. Once a buffer pool is allocated users can handle buffers using the following API: * xf_buffer_get(pool), gets a buffer from a pool * xf_buffer_put(buf), puts back a buffer into its pool Reviewed-by: Cosmin-Gabriel Samoila <cosmin.samoila@nxp.com> Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
Diffstat (limited to 'sound/soc/fsl/fsl_dsp_pool.c')
-rw-r--r--sound/soc/fsl/fsl_dsp_pool.c150
1 files changed, 150 insertions, 0 deletions
diff --git a/sound/soc/fsl/fsl_dsp_pool.c b/sound/soc/fsl/fsl_dsp_pool.c
new file mode 100644
index 000000000000..637454d97231
--- /dev/null
+++ b/sound/soc/fsl/fsl_dsp_pool.c
@@ -0,0 +1,150 @@
+// SPDX-License-Identifier: GPL-2.0+
+//
+// Xtensa buffer pool API
+//
+// Copyright 2018 NXP
+// Copyright (c) 2012-2013 by Tensilica Inc.
+
+#include <linux/slab.h>
+
+#include "fsl_dsp_pool.h"
+#include "fsl_dsp.h"
+
+/* ...allocate buffer pool */
+int xf_pool_alloc(struct xf_client *client, struct xf_proxy *proxy,
+ u32 number, u32 length, xf_pool_type_t type,
+ struct xf_pool **pool)
+{
+ struct xf_pool *p;
+ struct xf_buffer *b;
+ void *data;
+ struct xf_message msg;
+ struct xf_message *rmsg;
+
+ /* ...basic sanity checks; number of buffers is positive */
+ if (number <=0)
+ return -EINVAL;
+
+ /* ...get properly aligned buffer length */
+ length = ALIGN(length, XF_PROXY_ALIGNMENT);
+
+ p = kzalloc(offsetof(struct xf_pool, buffer) +
+ number * sizeof(struct xf_buffer), GFP_KERNEL);
+ if(!p)
+ return -ENOMEM;
+
+ /* ...prepare command parameters */
+ msg.id = __XF_MSG_ID(__XF_AP_PROXY(0), __XF_DSP_PROXY(0));
+ msg.id = XF_MSG_AP_FROM_USER(msg.id, client->id);
+ msg.opcode = XF_ALLOC;
+ msg.length = length * number;
+ msg.buffer = NULL;
+ msg.ret = 0;
+
+ /* ...execute command synchronously */
+ rmsg = xf_cmd_send_recv_complete(client, proxy, msg.id, msg.opcode,
+ msg.buffer, msg.length, &client->work,
+ &client->compr_complete);
+ if (IS_ERR(rmsg)) {
+ kfree(p);
+ return PTR_ERR(rmsg);
+ }
+
+ p->p = rmsg->buffer;
+ /* TODO: review cleanup */
+ /* xf_msg_free(proxy, rmsg);
+ * xf_unlock(&proxy->lock); */
+
+ /* ...if operation is failed, do cleanup */
+ /* ...set pool parameters */
+ p->number = number, p->length = length;
+ p->proxy = proxy;
+
+ /* ...create individual buffers and link them into free list */
+ for (p->free = b = &p->buffer[0], data = p->p; number > 0;
+ number--, b++) {
+ /* ...set address of the buffer (no length there) */
+ b->address = data;
+
+ /* ...file buffer into the free list */
+ b->link.next = b + 1;
+
+ /* ...advance data pointer in contiguous buffer */
+ data += length;
+ }
+
+ /* ...terminate list of buffers (not too good - tbd) */
+ b[-1].link.next = NULL;
+
+ /* ...return buffer pointer */
+ *pool = p;
+
+ return 0;
+}
+/* ...buffer pool destruction */
+int xf_pool_free(struct xf_client *client, struct xf_pool *pool)
+{
+ struct xf_proxy *proxy;
+ struct xf_message msg;
+ struct xf_message *rmsg;
+
+ /* ...basic sanity checks; pool is positive */
+ if (pool == NULL)
+ return -EINVAL;
+
+ /* ...get proxy pointer */
+ if ((proxy = pool->proxy) == NULL)
+ return -EINVAL;
+
+ /* ...prepare command parameters */
+ msg.id = __XF_MSG_ID(__XF_AP_PROXY(0), __XF_DSP_PROXY(0));
+ msg.id = XF_MSG_AP_FROM_USER(msg.id, client->id);
+ msg.opcode = XF_FREE;
+ msg.length = pool->length * pool->number;
+ msg.buffer = pool->p;
+ msg.ret = 0;
+
+ /* ...execute command synchronously */
+ rmsg = xf_cmd_send_recv_complete(client, proxy, msg.id, msg.opcode,
+ msg.buffer, msg.length, &client->work,
+ &client->compr_complete);
+ kfree(pool);
+ if (IS_ERR(rmsg))
+ return PTR_ERR(rmsg);
+
+ /* TODO: review cleanup */
+ /* xf_msg_free(proxy, rmsg);
+ * xf_unlock(&proxy->lock); */
+
+ return 0;
+}
+
+/* ...get new buffer from a pool */
+struct xf_buffer *xf_buffer_get(struct xf_pool *pool)
+{
+ struct xf_buffer *b;
+
+ xf_lock(&pool->proxy->lock);
+ /* ...take buffer from a head of the free list */
+ b = pool->free;
+ if (b) {
+ /* ...advance free list head */
+ pool->free = b->link.next, b->link.pool = pool;
+ }
+
+ xf_unlock(&pool->proxy->lock);
+ return b;
+}
+
+/* ...return buffer back to pool */
+void xf_buffer_put(struct xf_buffer *buffer)
+{
+ struct xf_pool *pool = buffer->link.pool;
+
+ xf_lock(&pool->proxy->lock);
+ /* ...use global proxy lock for pool operations protection */
+ /* ...put buffer back to a pool */
+ buffer->link.next = pool->free, pool->free = buffer;
+
+ xf_unlock(&pool->proxy->lock);
+}