Merge tag 'kern-priv-shm-for-v6.8' of https://git.linaro.org/people/jens.wiklander/linux-tee into soc/drivers

OP-TEE kernel private shared memory optimizations

Optimize OP-TEE driver private shared memory allocated as dynamic shared
memory. Both to handle larger than one page allocations and for more
efficient memory usage.

* tag 'kern-priv-shm-for-v6.8' of https://git.linaro.org/people/jens.wiklander/linux-tee:
  optee: allocate shared memory with alloc_pages_exact()
  optee: add page list to kernel private shared memory

Link: https://lore.kernel.org/r/20231211115815.GA616539@rayden
Signed-off-by: Arnd Bergmann <arnd@arndb.de>

2 files changed, 44 insertions, 45 deletions

diff --git a/drivers/tee/optee/core.c b/drivers/tee/optee/core.c
index d01ca47f7bde..0c7336077a7b 100644
--- a/drivers/tee/optee/core.c
+++ b/drivers/tee/optee/core.c
@@ -26,46 +26,46 @@ int optee_pool_op_alloc_helper(struct tee_shm_pool *pool, struct tee_shm *shm,
 						   size_t num_pages,
 						   unsigned long start))
 {
-	unsigned int order = get_order(size);
-	struct page *page;
+	size_t nr_pages = roundup(size, PAGE_SIZE) / PAGE_SIZE;
+	struct page **pages;
+	unsigned int i;
 	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)
+	shm->kaddr = alloc_pages_exact(nr_pages * PAGE_SIZE,
+				       GFP_KERNEL | __GFP_ZERO);
+	if (!shm->kaddr)
 		return -ENOMEM;
 
-	shm->kaddr = page_address(page);
-	shm->paddr = page_to_phys(page);
-	shm->size = PAGE_SIZE << order;
+	shm->paddr = virt_to_phys(shm->kaddr);
+	shm->size = nr_pages * PAGE_SIZE;
 
-	if (shm_register) {
-		unsigned int nr_pages = 1 << order, i;
-		struct page **pages;
+	pages = kcalloc(nr_pages, sizeof(*pages), GFP_KERNEL);
+	if (!pages) {
+		rc = -ENOMEM;
+		goto err;
+	}
 
-		pages = kcalloc(nr_pages, sizeof(*pages), GFP_KERNEL);
-		if (!pages) {
-			rc = -ENOMEM;
-			goto err;
-		}
+	for (i = 0; i < nr_pages; i++)
+		pages[i] = virt_to_page((u8 *)shm->kaddr + i * PAGE_SIZE);
 
-		for (i = 0; i < nr_pages; i++)
-			pages[i] = page + i;
+	shm->pages = pages;
+	shm->num_pages = nr_pages;
 
+	if (shm_register) {
 		rc = shm_register(shm->ctx, shm, pages, nr_pages,
 				  (unsigned long)shm->kaddr);
-		kfree(pages);
 		if (rc)
 			goto err;
 	}
 
 	return 0;
-
 err:
-	free_pages((unsigned long)shm->kaddr, order);
+	free_pages_exact(shm->kaddr, shm->size);
+	shm->kaddr = NULL;
 	return rc;
 }
 
@@ -75,8 +75,10 @@ void optee_pool_op_free_helper(struct tee_shm_pool *pool, struct tee_shm *shm,
 {
 	if (shm_unregister)
 		shm_unregister(shm->ctx, shm);
-	free_pages((unsigned long)shm->kaddr, get_order(shm->size));
+	free_pages_exact(shm->kaddr, shm->size);
 	shm->kaddr = NULL;
+	kfree(shm->pages);
+	shm->pages = NULL;
 }
 
 static void optee_bus_scan(struct work_struct *work)
diff --git a/drivers/tee/optee/smc_abi.c b/drivers/tee/optee/smc_abi.c
index e158f3136c26..f934467bea44 100644
--- a/drivers/tee/optee/smc_abi.c
+++ b/drivers/tee/optee/smc_abi.c
@@ -678,10 +678,11 @@ static void handle_rpc_func_cmd_shm_alloc(struct tee_context *ctx,
 					  struct optee_msg_arg *arg,
 					  struct optee_call_ctx *call_ctx)
 {
-	phys_addr_t pa;
 	struct tee_shm *shm;
 	size_t sz;
 	size_t n;
+	struct page **pages;
+	size_t page_count;
 
 	arg->ret_origin = TEEC_ORIGIN_COMMS;
 
@@ -716,32 +717,23 @@ static void handle_rpc_func_cmd_shm_alloc(struct tee_context *ctx,
 		return;
 	}
 
-	if (tee_shm_get_pa(shm, 0, &pa)) {
-		arg->ret = TEEC_ERROR_BAD_PARAMETERS;
-		goto bad;
-	}
-
-	sz = tee_shm_get_size(shm);
-
-	if (tee_shm_is_dynamic(shm)) {
-		struct page **pages;
+	/*
+	 * If there are pages it's dynamically allocated shared memory (not
+	 * from the reserved shared memory pool) and needs to be
+	 * registered.
+	 */
+	pages = tee_shm_get_pages(shm, &page_count);
+	if (pages) {
 		u64 *pages_list;
-		size_t page_num;
-
-		pages = tee_shm_get_pages(shm, &page_num);
-		if (!pages || !page_num) {
-			arg->ret = TEEC_ERROR_OUT_OF_MEMORY;
-			goto bad;
-		}
 
-		pages_list = optee_allocate_pages_list(page_num);
+		pages_list = optee_allocate_pages_list(page_count);
 		if (!pages_list) {
 			arg->ret = TEEC_ERROR_OUT_OF_MEMORY;
 			goto bad;
 		}
 
 		call_ctx->pages_list = pages_list;
-		call_ctx->num_entries = page_num;
+		call_ctx->num_entries = page_count;
 
 		arg->params[0].attr = OPTEE_MSG_ATTR_TYPE_TMEM_OUTPUT |
 				      OPTEE_MSG_ATTR_NONCONTIG;
@@ -752,17 +744,22 @@ static void handle_rpc_func_cmd_shm_alloc(struct tee_context *ctx,
 		arg->params[0].u.tmem.buf_ptr = virt_to_phys(pages_list) |
 			(tee_shm_get_page_offset(shm) &
 			 (OPTEE_MSG_NONCONTIG_PAGE_SIZE - 1));
-		arg->params[0].u.tmem.size = tee_shm_get_size(shm);
-		arg->params[0].u.tmem.shm_ref = (unsigned long)shm;
 
-		optee_fill_pages_list(pages_list, pages, page_num,
+		optee_fill_pages_list(pages_list, pages, page_count,
 				      tee_shm_get_page_offset(shm));
 	} else {
+		phys_addr_t pa;
+
+		if (tee_shm_get_pa(shm, 0, &pa)) {
+			arg->ret = TEEC_ERROR_BAD_PARAMETERS;
+			goto bad;
+		}
+
 		arg->params[0].attr = OPTEE_MSG_ATTR_TYPE_TMEM_OUTPUT;
 		arg->params[0].u.tmem.buf_ptr = pa;
-		arg->params[0].u.tmem.size = sz;
-		arg->params[0].u.tmem.shm_ref = (unsigned long)shm;
 	}
+	arg->params[0].u.tmem.size = tee_shm_get_size(shm);
+	arg->params[0].u.tmem.shm_ref = (unsigned long)shm;
 
 	arg->ret = TEEC_SUCCESS;
 	return;