configure: update compiler probing
[fio.git] / memory.c
index 9124117..04dc3be 100644 (file)
--- a/memory.c
+++ b/memory.c
@@ -33,13 +33,13 @@ int fio_pin_memory(struct thread_data *td)
        dprint(FD_MEM, "pinning %llu bytes\n", td->o.lockmem);
 
        /*
-        * Don't allow mlock of more than real_mem-128MB
+        * Don't allow mlock of more than real_mem-128MiB
         */
        phys_mem = os_phys_mem();
        if (phys_mem) {
                if ((td->o.lockmem + 128 * 1024 * 1024) > phys_mem) {
                        td->o.lockmem = phys_mem - 128 * 1024 * 1024;
-                       log_info("fio: limiting mlocked memory to %lluMB\n",
+                       log_info("fio: limiting mlocked memory to %lluMiB\n",
                                                        td->o.lockmem >> 20);
                }
        }
@@ -138,6 +138,9 @@ static int alloc_mem_mmap(struct thread_data *td, size_t total_mem)
        }
 
        if (td->o.mmapfile) {
+               if (access(td->o.mmapfile, F_OK) == 0)
+                       td->flags |= TD_F_MMAP_KEEP;
+
                td->mmapfd = open(td->o.mmapfile, O_RDWR|O_CREAT, 0644);
 
                if (td->mmapfd < 0) {
@@ -169,7 +172,7 @@ static int alloc_mem_mmap(struct thread_data *td, size_t total_mem)
                td->orig_buffer = NULL;
                if (td->mmapfd != 1 && td->mmapfd != -1) {
                        close(td->mmapfd);
-                       if (td->o.mmapfile)
+                       if (td->o.mmapfile && !(td->flags & TD_F_MMAP_KEEP))
                                unlink(td->o.mmapfile);
                }
 
@@ -187,7 +190,8 @@ static void free_mem_mmap(struct thread_data *td, size_t total_mem)
        if (td->o.mmapfile) {
                if (td->mmapfd != -1)
                        close(td->mmapfd);
-               unlink(td->o.mmapfile);
+               if (!(td->flags & TD_F_MMAP_KEEP))
+                       unlink(td->o.mmapfile);
                free(td->o.mmapfile);
        }
 }
@@ -207,6 +211,78 @@ static void free_mem_malloc(struct thread_data *td)
        free(td->orig_buffer);
 }
 
+static int alloc_mem_cudamalloc(struct thread_data *td, size_t total_mem)
+{
+#ifdef CONFIG_CUDA
+       CUresult ret;
+       char name[128];
+
+       ret = cuInit(0);
+       if (ret != CUDA_SUCCESS) {
+               log_err("fio: failed initialize cuda driver api\n");
+               return 1;
+       }
+
+       ret = cuDeviceGetCount(&td->gpu_dev_cnt);
+       if (ret != CUDA_SUCCESS) {
+               log_err("fio: failed get device count\n");
+               return 1;
+       }
+       dprint(FD_MEM, "found %d GPU devices\n", td->gpu_dev_cnt);
+
+       if (td->gpu_dev_cnt == 0) {
+               log_err("fio: no GPU device found. "
+                       "Can not perform GPUDirect RDMA.\n");
+               return 1;
+       }
+
+       td->gpu_dev_id = td->o.gpu_dev_id;
+       ret = cuDeviceGet(&td->cu_dev, td->gpu_dev_id);
+       if (ret != CUDA_SUCCESS) {
+               log_err("fio: failed get GPU device\n");
+               return 1;
+       }
+
+       ret = cuDeviceGetName(name, sizeof(name), td->gpu_dev_id);
+       if (ret != CUDA_SUCCESS) {
+               log_err("fio: failed get device name\n");
+               return 1;
+       }
+       dprint(FD_MEM, "dev_id = [%d], device name = [%s]\n", \
+              td->gpu_dev_id, name);
+
+       ret = cuCtxCreate(&td->cu_ctx, CU_CTX_MAP_HOST, td->cu_dev);
+       if (ret != CUDA_SUCCESS) {
+               log_err("fio: failed to create cuda context: %d\n", ret);
+               return 1;
+       }
+
+       ret = cuMemAlloc(&td->dev_mem_ptr, total_mem);
+       if (ret != CUDA_SUCCESS) {
+               log_err("fio: cuMemAlloc %zu bytes failed\n", total_mem);
+               return 1;
+       }
+       td->orig_buffer = (void *) td->dev_mem_ptr;
+
+       dprint(FD_MEM, "cudaMalloc %llu %p\n",                          \
+              (unsigned long long) total_mem, td->orig_buffer);
+       return 0;
+#else
+       return -EINVAL;
+#endif
+}
+
+static void free_mem_cudamalloc(struct thread_data *td)
+{
+#ifdef CONFIG_CUDA
+       if (td->dev_mem_ptr != NULL)
+               cuMemFree(td->dev_mem_ptr);
+
+       if (cuCtxDestroy(td->cu_ctx) != CUDA_SUCCESS)
+               log_err("fio: failed to destroy cuda context\n");
+#endif
+}
+
 /*
  * Set up the buffer area we need for io.
  */
@@ -246,6 +322,8 @@ int allocate_io_mem(struct thread_data *td)
        else if (td->o.mem_type == MEM_MMAP || td->o.mem_type == MEM_MMAPHUGE ||
                 td->o.mem_type == MEM_MMAPSHARED)
                ret = alloc_mem_mmap(td, total_mem);
+       else if (td->o.mem_type == MEM_CUDA_MALLOC)
+               ret = alloc_mem_cudamalloc(td, total_mem);
        else {
                log_err("fio: bad mem type: %d\n", td->o.mem_type);
                ret = 1;
@@ -275,6 +353,8 @@ void free_io_mem(struct thread_data *td)
        else if (td->o.mem_type == MEM_MMAP || td->o.mem_type == MEM_MMAPHUGE ||
                 td->o.mem_type == MEM_MMAPSHARED)
                free_mem_mmap(td, total_mem);
+       else if (td->o.mem_type == MEM_CUDA_MALLOC)
+               free_mem_cudamalloc(td);
        else
                log_err("Bad memory type %u\n", td->o.mem_type);