tee: simplify shm pool handling

Replaces the shared memory pool based on two pools with a single pool.
The alloc() function pointer in struct tee_shm_pool_ops gets another
parameter, align. This makes it possible to make less than page aligned
allocations from the optional reserved shared memory pool while still
making user space allocations page aligned. With in practice unchanged
behaviour using only a single pool for bookkeeping.

The allocation algorithm in the static OP-TEE shared memory pool is
changed from best-fit to first-fit since only the latter supports an
alignment parameter. The best-fit algorithm was previously the default
choice and not a conscious one.

The optee and amdtee drivers are updated as needed to work with this
changed pool handling.

This also removes OPTEE_SHM_NUM_PRIV_PAGES which becomes obsolete with
this change as the private pages can be mixed with the payload pages.

The OP-TEE driver changes minimum alignment for argument struct from 8
bytes to 512 bytes. A typical OP-TEE private shm allocation is 224 bytes
(argument struct with 6 parameters, needed for open session). So with an
alignment of 512 well waste a bit more than 50%. Before this we had a
single page reserved for this so worst case usage compared to that would
be 3 pages instead of 1 page. However, this worst case only occurs if
there is a high pressure from multiple threads on secure world. All in
all this should scale up and down better than fixed boundaries.

Reviewed-by: Sumit Garg <sumit.garg@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

5 files changed, 53 insertions, 133 deletions

diff --git a/drivers/tee/optee/Kconfig b/drivers/tee/optee/Kconfig
index 3ca71e3812ed..f121c224e682 100644
--- a/drivers/tee/optee/Kconfig
+++ b/drivers/tee/optee/Kconfig
@@ -7,11 +7,3 @@ config OPTEE
 	help
 	  This implements the OP-TEE Trusted Execution Environment (TEE)
 	  driver.
-
-config OPTEE_SHM_NUM_PRIV_PAGES
-	int "Private Shared Memory Pages"
-	default 1
-	depends on OPTEE
-	help
-	  This sets the number of private shared memory pages to be
-	  used by OP-TEE TEE driver.
diff --git a/drivers/tee/optee/core.c b/drivers/tee/optee/core.c
index 17a6f51d3089..f4bccb5f0e93 100644
--- a/drivers/tee/optee/core.c
+++ b/drivers/tee/optee/core.c
@@ -18,8 +18,8 @@
 #include <linux/workqueue.h>
 #include "optee_private.h"
 
-int optee_pool_op_alloc_helper(struct tee_shm_pool_mgr *poolm,
-			       struct tee_shm *shm, size_t size,
+int optee_pool_op_alloc_helper(struct tee_shm_pool *pool, struct tee_shm *shm,
+			       size_t size, size_t align,
 			       int (*shm_register)(struct tee_context *ctx,
 						   struct tee_shm *shm,
 						   struct page **pages,
@@ -30,6 +30,10 @@ int optee_pool_op_alloc_helper(struct tee_shm_pool_mgr *poolm,
 	struct page *page;
 	int rc = 0;
 
+	/*
+	 * Ignore alignment since this is already going to be page aligned
+	 * and there's no need for any larger alignment.
+	 */
 	page = alloc_pages(GFP_KERNEL | __GFP_ZERO, order);
 	if (!page)
 		return -ENOMEM;
@@ -51,7 +55,6 @@ int optee_pool_op_alloc_helper(struct tee_shm_pool_mgr *poolm,
 		for (i = 0; i < nr_pages; i++)
 			pages[i] = page + i;
 
-		shm->flags |= TEE_SHM_REGISTER;
 		rc = shm_register(shm->ctx, shm, pages, nr_pages,
 				  (unsigned long)shm->kaddr);
 		kfree(pages);
@@ -62,7 +65,7 @@ int optee_pool_op_alloc_helper(struct tee_shm_pool_mgr *poolm,
 	return 0;
 
 err:
-	__free_pages(page, order);
+	free_pages((unsigned long)shm->kaddr, order);
 	return rc;
 }
 
diff --git a/drivers/tee/optee/ffa_abi.c b/drivers/tee/optee/ffa_abi.c
index 545f61af1248..91dd80945bf1 100644
--- a/drivers/tee/optee/ffa_abi.c
+++ b/drivers/tee/optee/ffa_abi.c
@@ -369,14 +369,14 @@ static int optee_ffa_shm_unregister_supp(struct tee_context *ctx,
  * The main function is optee_ffa_shm_pool_alloc_pages().
  */
 
-static int pool_ffa_op_alloc(struct tee_shm_pool_mgr *poolm,
-			     struct tee_shm *shm, size_t size)
+static int pool_ffa_op_alloc(struct tee_shm_pool *pool,
+			     struct tee_shm *shm, size_t size, size_t align)
 {
-	return optee_pool_op_alloc_helper(poolm, shm, size,
+	return optee_pool_op_alloc_helper(pool, shm, size, align,
 					  optee_ffa_shm_register);
 }
 
-static void pool_ffa_op_free(struct tee_shm_pool_mgr *poolm,
+static void pool_ffa_op_free(struct tee_shm_pool *pool,
 			     struct tee_shm *shm)
 {
 	optee_ffa_shm_unregister(shm->ctx, shm);
@@ -384,15 +384,15 @@ static void pool_ffa_op_free(struct tee_shm_pool_mgr *poolm,
 	shm->kaddr = NULL;
 }
 
-static void pool_ffa_op_destroy_poolmgr(struct tee_shm_pool_mgr *poolm)
+static void pool_ffa_op_destroy_pool(struct tee_shm_pool *pool)
 {
-	kfree(poolm);
+	kfree(pool);
 }
 
-static const struct tee_shm_pool_mgr_ops pool_ffa_ops = {
+static const struct tee_shm_pool_ops pool_ffa_ops = {
 	.alloc = pool_ffa_op_alloc,
 	.free = pool_ffa_op_free,
-	.destroy_poolmgr = pool_ffa_op_destroy_poolmgr,
+	.destroy_pool = pool_ffa_op_destroy_pool,
 };
 
 /**
@@ -401,16 +401,16 @@ static const struct tee_shm_pool_mgr_ops pool_ffa_ops = {
  * This pool is used with OP-TEE over FF-A. In this case command buffers
  * and such are allocated from kernel's own memory.
  */
-static struct tee_shm_pool_mgr *optee_ffa_shm_pool_alloc_pages(void)
+static struct tee_shm_pool *optee_ffa_shm_pool_alloc_pages(void)
 {
-	struct tee_shm_pool_mgr *mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
+	struct tee_shm_pool *pool = kzalloc(sizeof(*pool), GFP_KERNEL);
 
-	if (!mgr)
+	if (!pool)
 		return ERR_PTR(-ENOMEM);
 
-	mgr->ops = &pool_ffa_ops;
+	pool->ops = &pool_ffa_ops;
 
-	return mgr;
+	return pool;
 }
 
 /*
@@ -691,33 +691,6 @@ static bool optee_ffa_exchange_caps(struct ffa_device *ffa_dev,
 	return true;
 }
 
-static struct tee_shm_pool *optee_ffa_config_dyn_shm(void)
-{
-	struct tee_shm_pool_mgr *priv_mgr;
-	struct tee_shm_pool_mgr *dmabuf_mgr;
-	void *rc;
-
-	rc = optee_ffa_shm_pool_alloc_pages();
-	if (IS_ERR(rc))
-		return rc;
-	priv_mgr = rc;
-
-	rc = optee_ffa_shm_pool_alloc_pages();
-	if (IS_ERR(rc)) {
-		tee_shm_pool_mgr_destroy(priv_mgr);
-		return rc;
-	}
-	dmabuf_mgr = rc;
-
-	rc = tee_shm_pool_alloc(priv_mgr, dmabuf_mgr);
-	if (IS_ERR(rc)) {
-		tee_shm_pool_mgr_destroy(priv_mgr);
-		tee_shm_pool_mgr_destroy(dmabuf_mgr);
-	}
-
-	return rc;
-}
-
 static void optee_ffa_get_version(struct tee_device *teedev,
 				  struct tee_ioctl_version_data *vers)
 {
@@ -815,7 +788,7 @@ static int optee_ffa_probe(struct ffa_device *ffa_dev)
 	if (!optee)
 		return -ENOMEM;
 
-	pool = optee_ffa_config_dyn_shm();
+	pool = optee_ffa_shm_pool_alloc_pages();
 	if (IS_ERR(pool)) {
 		rc = PTR_ERR(pool);
 		goto err_free_optee;
diff --git a/drivers/tee/optee/optee_private.h b/drivers/tee/optee/optee_private.h
index 92bc47bef95f..df3a483bbf46 100644
--- a/drivers/tee/optee/optee_private.h
+++ b/drivers/tee/optee/optee_private.h
@@ -229,8 +229,8 @@ int optee_cancel_req(struct tee_context *ctx, u32 cancel_id, u32 session);
 int optee_enumerate_devices(u32 func);
 void optee_unregister_devices(void);
 
-int optee_pool_op_alloc_helper(struct tee_shm_pool_mgr *poolm,
-			       struct tee_shm *shm, size_t size,
+int optee_pool_op_alloc_helper(struct tee_shm_pool *pool, struct tee_shm *shm,
+			       size_t size, size_t align,
 			       int (*shm_register)(struct tee_context *ctx,
 						   struct tee_shm *shm,
 						   struct page **pages,
diff --git a/drivers/tee/optee/smc_abi.c b/drivers/tee/optee/smc_abi.c
index bacd1a1d79ee..3dc8cbea1a24 100644
--- a/drivers/tee/optee/smc_abi.c
+++ b/drivers/tee/optee/smc_abi.c
@@ -42,7 +42,15 @@
  * 6. Driver initialization.
  */
 
-#define OPTEE_SHM_NUM_PRIV_PAGES	CONFIG_OPTEE_SHM_NUM_PRIV_PAGES
+/*
+ * A typical OP-TEE private shm allocation is 224 bytes (argument struct
+ * with 6 parameters, needed for open session). So with an alignment of 512
+ * we'll waste a bit more than 50%. However, it's only expected that we'll
+ * have a handful of these structs allocated at a time. Most memory will
+ * be allocated aligned to the page size, So all in all this should scale
+ * up and down quite well.
+ */
+#define OPTEE_MIN_STATIC_POOL_ALIGN    9 /* 512 bytes aligned */
 
 /*
  * 1. Convert between struct tee_param and struct optee_msg_param
@@ -532,20 +540,21 @@ static int optee_shm_unregister_supp(struct tee_context *ctx,
  * The main function is optee_shm_pool_alloc_pages().
  */
 
-static int pool_op_alloc(struct tee_shm_pool_mgr *poolm,
-			 struct tee_shm *shm, size_t size)
+static int pool_op_alloc(struct tee_shm_pool *pool,
+			 struct tee_shm *shm, size_t size, size_t align)
 {
 	/*
 	 * Shared memory private to the OP-TEE driver doesn't need
 	 * to be registered with OP-TEE.
 	 */
 	if (shm->flags & TEE_SHM_PRIV)
-		return optee_pool_op_alloc_helper(poolm, shm, size, NULL);
+		return optee_pool_op_alloc_helper(pool, shm, size, align, NULL);
 
-	return optee_pool_op_alloc_helper(poolm, shm, size, optee_shm_register);
+	return optee_pool_op_alloc_helper(pool, shm, size, align,
+					  optee_shm_register);
 }
 
-static void pool_op_free(struct tee_shm_pool_mgr *poolm,
+static void pool_op_free(struct tee_shm_pool *pool,
 			 struct tee_shm *shm)
 {
 	if (!(shm->flags & TEE_SHM_PRIV))
@@ -555,15 +564,15 @@ static void pool_op_free(struct tee_shm_pool_mgr *poolm,
 	shm->kaddr = NULL;
 }
 
-static void pool_op_destroy_poolmgr(struct tee_shm_pool_mgr *poolm)
+static void pool_op_destroy_pool(struct tee_shm_pool *pool)
 {
-	kfree(poolm);
+	kfree(pool);
 }
 
-static const struct tee_shm_pool_mgr_ops pool_ops = {
+static const struct tee_shm_pool_ops pool_ops = {
 	.alloc = pool_op_alloc,
 	.free = pool_op_free,
-	.destroy_poolmgr = pool_op_destroy_poolmgr,
+	.destroy_pool = pool_op_destroy_pool,
 };
 
 /**
@@ -572,16 +581,16 @@ static const struct tee_shm_pool_mgr_ops pool_ops = {
  * This pool is used when OP-TEE supports dymanic SHM. In this case
  * command buffers and such are allocated from kernel's own memory.
  */
-static struct tee_shm_pool_mgr *optee_shm_pool_alloc_pages(void)
+static struct tee_shm_pool *optee_shm_pool_alloc_pages(void)
 {
-	struct tee_shm_pool_mgr *mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
+	struct tee_shm_pool *pool = kzalloc(sizeof(*pool), GFP_KERNEL);
 
-	if (!mgr)
+	if (!pool)
 		return ERR_PTR(-ENOMEM);
 
-	mgr->ops = &pool_ops;
+	pool->ops = &pool_ops;
 
-	return mgr;
+	return pool;
 }
 
 /*
@@ -1153,33 +1162,6 @@ static bool optee_msg_exchange_capabilities(optee_invoke_fn *invoke_fn,
 	return true;
 }
 
-static struct tee_shm_pool *optee_config_dyn_shm(void)
-{
-	struct tee_shm_pool_mgr *priv_mgr;
-	struct tee_shm_pool_mgr *dmabuf_mgr;
-	void *rc;
-
-	rc = optee_shm_pool_alloc_pages();
-	if (IS_ERR(rc))
-		return rc;
-	priv_mgr = rc;
-
-	rc = optee_shm_pool_alloc_pages();
-	if (IS_ERR(rc)) {
-		tee_shm_pool_mgr_destroy(priv_mgr);
-		return rc;
-	}
-	dmabuf_mgr = rc;
-
-	rc = tee_shm_pool_alloc(priv_mgr, dmabuf_mgr);
-	if (IS_ERR(rc)) {
-		tee_shm_pool_mgr_destroy(priv_mgr);
-		tee_shm_pool_mgr_destroy(dmabuf_mgr);
-	}
-
-	return rc;
-}
-
 static struct tee_shm_pool *
 optee_config_shm_memremap(optee_invoke_fn *invoke_fn, void **memremaped_shm)
 {
@@ -1193,10 +1175,7 @@ optee_config_shm_memremap(optee_invoke_fn *invoke_fn, void **memremaped_shm)
 	phys_addr_t begin;
 	phys_addr_t end;
 	void *va;
-	struct tee_shm_pool_mgr *priv_mgr;
-	struct tee_shm_pool_mgr *dmabuf_mgr;
 	void *rc;
-	const int sz = OPTEE_SHM_NUM_PRIV_PAGES * PAGE_SIZE;
 
 	invoke_fn(OPTEE_SMC_GET_SHM_CONFIG, 0, 0, 0, 0, 0, 0, 0, &res.smccc);
 	if (res.result.status != OPTEE_SMC_RETURN_OK) {
@@ -1214,11 +1193,6 @@ optee_config_shm_memremap(optee_invoke_fn *invoke_fn, void **memremaped_shm)
 	paddr = begin;
 	size = end - begin;
 
-	if (size < 2 * OPTEE_SHM_NUM_PRIV_PAGES * PAGE_SIZE) {
-		pr_err("too small shared memory area\n");
-		return ERR_PTR(-EINVAL);
-	}
-
 	va = memremap(paddr, size, MEMREMAP_WB);
 	if (!va) {
 		pr_err("shared memory ioremap failed\n");
@@ -1226,35 +1200,13 @@ optee_config_shm_memremap(optee_invoke_fn *invoke_fn, void **memremaped_shm)
 	}
 	vaddr = (unsigned long)va;
 
-	rc = tee_shm_pool_mgr_alloc_res_mem(vaddr, paddr, sz,
-					    3 /* 8 bytes aligned */);
-	if (IS_ERR(rc))
-		goto err_memunmap;
-	priv_mgr = rc;
-
-	vaddr += sz;
-	paddr += sz;
-	size -= sz;
-
-	rc = tee_shm_pool_mgr_alloc_res_mem(vaddr, paddr, size, PAGE_SHIFT);
-	if (IS_ERR(rc))
-		goto err_free_priv_mgr;
-	dmabuf_mgr = rc;
-
-	rc = tee_shm_pool_alloc(priv_mgr, dmabuf_mgr);
+	rc = tee_shm_pool_alloc_res_mem(vaddr, paddr, size,
+					OPTEE_MIN_STATIC_POOL_ALIGN);
 	if (IS_ERR(rc))
-		goto err_free_dmabuf_mgr;
-
-	*memremaped_shm = va;
-
-	return rc;
+		memunmap(va);
+	else
+		*memremaped_shm = va;
 
-err_free_dmabuf_mgr:
-	tee_shm_pool_mgr_destroy(dmabuf_mgr);
-err_free_priv_mgr:
-	tee_shm_pool_mgr_destroy(priv_mgr);
-err_memunmap:
-	memunmap(va);
 	return rc;
 }
 
@@ -1376,7 +1328,7 @@ static int optee_probe(struct platform_device *pdev)
 	 * Try to use dynamic shared memory if possible
 	 */
 	if (sec_caps & OPTEE_SMC_SEC_CAP_DYNAMIC_SHM)
-		pool = optee_config_dyn_shm();
+		pool = optee_shm_pool_alloc_pages();
 
 	/*
 	 * If dynamic shared memory is not available or failed - try static one