[dpdk-dev] [PATCH 01/41] eal: move get_virtual_area out of linuxapp eal_memory.c

[Anatoly Burakov <anatoly.burakov@intel.com>]

Move get_virtual_area out of linuxapp EAL memory and make it
common to EAL, so that other code could reserve virtual areas
as well.

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
---
 lib/librte_eal/common/eal_common_memory.c | 101 ++++++++++++++++++++++
 lib/librte_eal/common/eal_private.h       |  33 +++++++
 lib/librte_eal/linuxapp/eal/eal_memory.c  | 137 ++++++------------------------
 3 files changed, 161 insertions(+), 110 deletions(-)

diff --git a/lib/librte_eal/common/eal_common_memory.c b/lib/librte_eal/common/eal_common_memory.c
index 852f3bb..042881b 100644
--- a/lib/librte_eal/common/eal_common_memory.c
+++ b/lib/librte_eal/common/eal_common_memory.c
@@ -2,10 +2,12 @@
  * Copyright(c) 2010-2014 Intel Corporation
  */
 
+#include <errno.h>
 #include <stdio.h>
 #include <stdint.h>
 #include <stdlib.h>
 #include <stdarg.h>
+#include <string.h>
 #include <unistd.h>
 #include <inttypes.h>
 #include <sys/mman.h>
@@ -14,12 +16,111 @@
 #include <rte_memory.h>
 #include <rte_eal.h>
 #include <rte_eal_memconfig.h>
+#include <rte_errno.h>
 #include <rte_log.h>
 
 #include "eal_private.h"
 #include "eal_internal_cfg.h"
 
 /*
+ * Try to mmap *size bytes in /dev/zero. If it is successful, return the
+ * pointer to the mmap'd area and keep *size unmodified. Else, retry
+ * with a smaller zone: decrease *size by hugepage_sz until it reaches
+ * 0. In this case, return NULL. Note: this function returns an address
+ * which is a multiple of hugepage size.
+ */
+
+static uint64_t baseaddr_offset;
+static uint64_t system_page_sz;
+
+void *
+eal_get_virtual_area(void *requested_addr, uint64_t *size,
+		uint64_t page_sz, int flags, int mmap_flags)
+{
+	bool addr_is_hint, allow_shrink, unmap, no_align;
+	uint64_t map_sz;
+	void *mapped_addr, *aligned_addr;
+
+	if (system_page_sz == 0)
+		system_page_sz = sysconf(_SC_PAGESIZE);
+
+	mmap_flags |= MAP_PRIVATE | MAP_ANONYMOUS;
+
+	RTE_LOG(DEBUG, EAL, "Ask a virtual area of 0x%zx bytes\n", *size);
+
+	addr_is_hint = (flags & EAL_VIRTUAL_AREA_ADDR_IS_HINT) > 0;
+	allow_shrink = (flags & EAL_VIRTUAL_AREA_ALLOW_SHRINK) > 0;
+	unmap = (flags & EAL_VIRTUAL_AREA_UNMAP) > 0;
+
+	if (requested_addr == NULL && internal_config.base_virtaddr != 0) {
+		requested_addr = (void *) (internal_config.base_virtaddr +
+				baseaddr_offset);
+		requested_addr = RTE_PTR_ALIGN(requested_addr, page_sz);
+		addr_is_hint = true;
+	}
+
+	/* if requested address is not aligned by page size, or if requested
+	 * address is NULL, add page size to requested length as we may get an
+	 * address that's aligned by system page size, which can be smaller than
+	 * our requested page size. additionally, we shouldn't try to align if
+	 * system page size is the same as requested page size.
+	 */
+	no_align = (requested_addr != NULL &&
+		((uintptr_t)requested_addr & (page_sz - 1)) == 0) ||
+		page_sz == system_page_sz;
+
+	do {
+		map_sz = no_align ? *size : *size + page_sz;
+
+		mapped_addr = mmap(requested_addr, map_sz, PROT_READ,
+				mmap_flags, -1, 0);
+		if (mapped_addr == MAP_FAILED && allow_shrink)
+			*size -= page_sz;
+	} while (allow_shrink && mapped_addr == MAP_FAILED && *size > 0);
+
+	/* align resulting address - if map failed, we will ignore the value
+	 * anyway, so no need to add additional checks.
+	 */
+	aligned_addr = no_align ? mapped_addr :
+			RTE_PTR_ALIGN(mapped_addr, page_sz);
+
+	if (*size == 0) {
+		RTE_LOG(ERR, EAL, "Cannot get a virtual area of any size: %s\n",
+			strerror(errno));
+		rte_errno = errno;
+		return NULL;
+	} else if (mapped_addr == MAP_FAILED) {
+		RTE_LOG(ERR, EAL, "Cannot get a virtual area: %s\n",
+			strerror(errno));
+		/* pass errno up the call chain */
+		rte_errno = errno;
+		return NULL;
+	} else if (requested_addr != NULL && !addr_is_hint &&
+			aligned_addr != requested_addr) {
+		RTE_LOG(ERR, EAL, "Cannot get a virtual area at requested address: %p (got %p)\n",
+			requested_addr, aligned_addr);
+		munmap(mapped_addr, map_sz);
+		rte_errno = EADDRNOTAVAIL;
+		return NULL;
+	} else if (requested_addr != NULL && addr_is_hint &&
+			aligned_addr != requested_addr) {
+		RTE_LOG(WARNING, EAL, "WARNING! Base virtual address hint (%p != %p) not respected!\n",
+			requested_addr, aligned_addr);
+		RTE_LOG(WARNING, EAL, "   This may cause issues with mapping memory into secondary processes\n");
+	}
+
+	if (unmap)
+		munmap(mapped_addr, map_sz);
+
+	RTE_LOG(DEBUG, EAL, "Virtual area found at %p (size = 0x%zx)\n",
+		aligned_addr, *size);
+
+	baseaddr_offset += *size;
+
+	return aligned_addr;
+}
+
+/*
  * Return a pointer to a read-only table of struct rte_physmem_desc
  * elements, containing the layout of all addressable physical
  * memory. The last element of the table contains a NULL address.
diff --git a/lib/librte_eal/common/eal_private.h b/lib/librte_eal/common/eal_private.h
index 0b28770..96cebb7 100644
--- a/lib/librte_eal/common/eal_private.h
+++ b/lib/librte_eal/common/eal_private.h
@@ -127,6 +127,39 @@ int rte_eal_alarm_init(void);
 int rte_eal_check_module(const char *module_name);
 
 /**
+ * Get virtual area of specified size from the OS.
+ *
+ * This function is private to the EAL.
+ *
+ * @param requested_addr
+ *   Address where to request address space.
+ * @param size
+ *   Size of requested area.
+ * @param page_sz
+ *   Page size on which to align requested virtual area.
+ * @param flags
+ *   EAL_VIRTUAL_AREA_* flags.
+ * @param mmap_flags
+ *   Extra flags passed directly to mmap().
+ *
+ * @return
+ *   Virtual area address if successful.
+ *   NULL if unsuccessful.
+ */
+
+#define EAL_VIRTUAL_AREA_ADDR_IS_HINT (1 << 0)
+/**< don't fail if cannot get exact requested address. */
+#define EAL_VIRTUAL_AREA_ALLOW_SHRINK (1 << 1)
+/**< try getting smaller sized (decrement by page size) virtual areas if cannot
+ * get area of requested size.
+ */
+#define EAL_VIRTUAL_AREA_UNMAP (1 << 2)
+/**< immediately unmap reserved virtual area. */
+void *
+eal_get_virtual_area(void *requested_addr, uint64_t *size,
+		uint64_t page_sz, int flags, int mmap_flags);
+
+/**
  * Get cpu core_id.
  *
  * This function is private to the EAL.
diff --git a/lib/librte_eal/linuxapp/eal/eal_memory.c b/lib/librte_eal/linuxapp/eal/eal_memory.c
index 38853b7..5c11d77 100644
--- a/lib/librte_eal/linuxapp/eal/eal_memory.c
+++ b/lib/librte_eal/linuxapp/eal/eal_memory.c
@@ -28,6 +28,7 @@
 #include <numaif.h>
 #endif
 
+#include <rte_errno.h>
 #include <rte_log.h>
 #include <rte_memory.h>
 #include <rte_launch.h>
@@ -57,8 +58,6 @@
  * zone as well as a physical contiguous zone.
  */
 
-static uint64_t baseaddr_offset;
-
 static bool phys_addrs_available = true;
 
 #define RANDOMIZE_VA_SPACE_FILE "/proc/sys/kernel/randomize_va_space"
@@ -221,82 +220,6 @@ aslr_enabled(void)
 	}
 }
 
-/*
- * Try to mmap *size bytes in /dev/zero. If it is successful, return the
- * pointer to the mmap'd area and keep *size unmodified. Else, retry
- * with a smaller zone: decrease *size by hugepage_sz until it reaches
- * 0. In this case, return NULL. Note: this function returns an address
- * which is a multiple of hugepage size.
- */
-static void *
-get_virtual_area(size_t *size, size_t hugepage_sz)
-{
-	void *addr;
-	void *addr_hint;
-	int fd;
-	long aligned_addr;
-
-	if (internal_config.base_virtaddr != 0) {
-		int page_size = sysconf(_SC_PAGE_SIZE);
-		addr_hint = (void *) (uintptr_t)
-			(internal_config.base_virtaddr + baseaddr_offset);
-		addr_hint = RTE_PTR_ALIGN_FLOOR(addr_hint, page_size);
-	} else {
-		addr_hint = NULL;
-	}
-
-	RTE_LOG(DEBUG, EAL, "Ask a virtual area of 0x%zx bytes\n", *size);
-
-
-	fd = open("/dev/zero", O_RDONLY);
-	if (fd < 0){
-		RTE_LOG(ERR, EAL, "Cannot open /dev/zero\n");
-		return NULL;
-	}
-	do {
-		addr = mmap(addr_hint, (*size) + hugepage_sz, PROT_READ,
-#ifdef RTE_ARCH_PPC_64
-				MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB,
-#else
-				MAP_PRIVATE,
-#endif
-				fd, 0);
-		if (addr == MAP_FAILED) {
-			*size -= hugepage_sz;
-		} else if (addr_hint != NULL && addr != addr_hint) {
-			RTE_LOG(WARNING, EAL, "WARNING! Base virtual address "
-				"hint (%p != %p) not respected!\n",
-				addr_hint, addr);
-			RTE_LOG(WARNING, EAL, "   This may cause issues with "
-				"mapping memory into secondary processes\n");
-		}
-	} while (addr == MAP_FAILED && *size > 0);
-
-	if (addr == MAP_FAILED) {
-		close(fd);
-		RTE_LOG(ERR, EAL, "Cannot get a virtual area: %s\n",
-			strerror(errno));
-		return NULL;
-	}
-
-	munmap(addr, (*size) + hugepage_sz);
-	close(fd);
-
-	/* align addr to a huge page size boundary */
-	aligned_addr = (long)addr;
-	aligned_addr += (hugepage_sz - 1);
-	aligned_addr &= (~(hugepage_sz - 1));
-	addr = (void *)(aligned_addr);
-
-	RTE_LOG(DEBUG, EAL, "Virtual area found at %p (size = 0x%zx)\n",
-		addr, *size);
-
-	/* increment offset */
-	baseaddr_offset += *size;
-
-	return addr;
-}
-
 static sigjmp_buf huge_jmpenv;
 
 static void huge_sigbus_handler(int signo __rte_unused)
@@ -445,7 +368,16 @@ map_all_hugepages(struct hugepage_file *hugepg_tbl, struct hugepage_info *hpi,
 			/* get the biggest virtual memory area up to
 			 * vma_len. If it fails, vma_addr is NULL, so
 			 * let the kernel provide the address. */
-			vma_addr = get_virtual_area(&vma_len, hpi->hugepage_sz);
+			vma_addr = eal_get_virtual_area(NULL, &vma_len,
+					hpi->hugepage_sz,
+					EAL_VIRTUAL_AREA_ALLOW_SHRINK |
+					EAL_VIRTUAL_AREA_UNMAP,
+#ifdef RTE_ARCH_PPC_64
+					MAP_HUGETLB
+#else
+					0
+#endif
+					);
 			if (vma_addr == NULL)
 				vma_len = hugepage_sz;
 		}
@@ -1339,7 +1271,7 @@ rte_eal_hugepage_attach(void)
 	unsigned i, s = 0; /* s used to track the segment number */
 	unsigned max_seg = RTE_MAX_MEMSEG;
 	off_t size = 0;
-	int fd, fd_zero = -1, fd_hugepage = -1;
+	int fd, fd_hugepage = -1;
 
 	if (aslr_enabled() > 0) {
 		RTE_LOG(WARNING, EAL, "WARNING: Address Space Layout Randomization "
@@ -1350,11 +1282,6 @@ rte_eal_hugepage_attach(void)
 
 	test_phys_addrs_available();
 
-	fd_zero = open("/dev/zero", O_RDONLY);
-	if (fd_zero < 0) {
-		RTE_LOG(ERR, EAL, "Could not open /dev/zero\n");
-		goto error;
-	}
 	fd_hugepage = open(eal_hugepage_info_path(), O_RDONLY);
 	if (fd_hugepage < 0) {
 		RTE_LOG(ERR, EAL, "Could not open %s\n", eal_hugepage_info_path());
@@ -1364,6 +1291,8 @@ rte_eal_hugepage_attach(void)
 	/* map all segments into memory to make sure we get the addrs */
 	for (s = 0; s < RTE_MAX_MEMSEG; ++s) {
 		void *base_addr;
+		uint64_t mmap_sz;
+		int mmap_flags = 0;
 
 		/*
 		 * the first memory segment with len==0 is the one that
@@ -1372,35 +1301,26 @@ rte_eal_hugepage_attach(void)
 		if (mcfg->memseg[s].len == 0)
 			break;
 
-		/*
-		 * fdzero is mmapped to get a contiguous block of virtual
-		 * addresses of the appropriate memseg size.
-		 * use mmap to get identical addresses as the primary process.
+		/* get identical addresses as the primary process.
 		 */
-		base_addr = mmap(mcfg->memseg[s].addr, mcfg->memseg[s].len,
-				 PROT_READ,
 #ifdef RTE_ARCH_PPC_64
-				 MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB,
-#else
-				 MAP_PRIVATE,
+		mmap_flags |= MAP_HUGETLB;
 #endif
-				 fd_zero, 0);
-		if (base_addr == MAP_FAILED ||
-		    base_addr != mcfg->memseg[s].addr) {
+		mmap_sz = mcfg->memseg[s].len;
+		base_addr = eal_get_virtual_area(mcfg->memseg[s].addr,
+				&mmap_sz, mcfg->memseg[s].hugepage_sz, 0,
+				mmap_flags);
+		if (base_addr == NULL) {
 			max_seg = s;
-			if (base_addr != MAP_FAILED) {
-				/* errno is stale, don't use */
-				RTE_LOG(ERR, EAL, "Could not mmap %llu bytes "
-					"in /dev/zero at [%p], got [%p] - "
-					"please use '--base-virtaddr' option\n",
+			if (rte_errno == EADDRNOTAVAIL) {
+				RTE_LOG(ERR, EAL, "Could not mmap %llu bytes at [%p] - please use '--base-virtaddr' option\n",
 					(unsigned long long)mcfg->memseg[s].len,
-					mcfg->memseg[s].addr, base_addr);
-				munmap(base_addr, mcfg->memseg[s].len);
+					mcfg->memseg[s].addr);
 			} else {
-				RTE_LOG(ERR, EAL, "Could not mmap %llu bytes "
-					"in /dev/zero at [%p]: '%s'\n",
+				RTE_LOG(ERR, EAL, "Could not mmap %llu bytes at [%p]: '%s'\n",
 					(unsigned long long)mcfg->memseg[s].len,
-					mcfg->memseg[s].addr, strerror(errno));
+					mcfg->memseg[s].addr,
+					rte_strerror(rte_errno));
 			}
 			if (aslr_enabled() > 0) {
 				RTE_LOG(ERR, EAL, "It is recommended to "
@@ -1465,7 +1385,6 @@ rte_eal_hugepage_attach(void)
 	}
 	/* unmap the hugepage config file, since we are done using it */
 	munmap(hp, size);
-	close(fd_zero);
 	close(fd_hugepage);
 	return 0;
 
@@ -1474,8 +1393,6 @@ rte_eal_hugepage_attach(void)
 		munmap(mcfg->memseg[i].addr, mcfg->memseg[i].len);
 	if (hp != NULL && hp != MAP_FAILED)
 		munmap(hp, size);
-	if (fd_zero >= 0)
-		close(fd_zero);
 	if (fd_hugepage >= 0)
 		close(fd_hugepage);
 	return -1;
-- 
2.7.4