/*
* QEMU low level functions
- *
+ *
* Copyright (c) 2003 Fabrice Bellard
- *
+ *
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
#include <string.h>
#include <errno.h>
#include <unistd.h>
+#include <fcntl.h>
-#include "cpu.h"
-#if defined(USE_KQEMU)
-#include "vl.h"
-#endif
-
-#if defined(__i386__) && !defined(CONFIG_SOFTMMU) && !defined(CONFIG_USER_ONLY)
+/* Needed early for CONFIG_BSD etc. */
+#include "config-host.h"
+#if defined(CONFIG_MADVISE) || defined(CONFIG_POSIX_MADVISE)
#include <sys/mman.h>
-#include <sys/ipc.h>
-
-/* When not using soft mmu, libc independant functions are needed for
- the CPU core because it needs to use alternates stacks and
- libc/thread incompatibles settings */
-
-#include <linux/unistd.h>
-
-#define QEMU_SYSCALL0(name) \
-{ \
-long __res; \
-__asm__ volatile ("int $0x80" \
- : "=a" (__res) \
- : "0" (__NR_##name)); \
-return __res; \
-}
-
-#define QEMU_SYSCALL1(name,arg1) \
-{ \
-long __res; \
-__asm__ volatile ("int $0x80" \
- : "=a" (__res) \
- : "0" (__NR_##name),"b" ((long)(arg1))); \
-return __res; \
-}
-
-#define QEMU_SYSCALL2(name,arg1,arg2) \
-{ \
-long __res; \
-__asm__ volatile ("int $0x80" \
- : "=a" (__res) \
- : "0" (__NR_##name),"b" ((long)(arg1)),"c" ((long)(arg2))); \
-return __res; \
-}
-
-#define QEMU_SYSCALL3(name,arg1,arg2,arg3) \
-{ \
-long __res; \
-__asm__ volatile ("int $0x80" \
- : "=a" (__res) \
- : "0" (__NR_##name),"b" ((long)(arg1)),"c" ((long)(arg2)), \
- "d" ((long)(arg3))); \
-return __res; \
-}
-
-#define QEMU_SYSCALL4(name,arg1,arg2,arg3,arg4) \
-{ \
-long __res; \
-__asm__ volatile ("int $0x80" \
- : "=a" (__res) \
- : "0" (__NR_##name),"b" ((long)(arg1)),"c" ((long)(arg2)), \
- "d" ((long)(arg3)),"S" ((long)(arg4))); \
-return __res; \
-}
-
-#define QEMU_SYSCALL5(name,arg1,arg2,arg3,arg4,arg5) \
-{ \
-long __res; \
-__asm__ volatile ("int $0x80" \
- : "=a" (__res) \
- : "0" (__NR_##name),"b" ((long)(arg1)),"c" ((long)(arg2)), \
- "d" ((long)(arg3)),"S" ((long)(arg4)),"D" ((long)(arg5))); \
-return __res; \
-}
+#endif
-#define QEMU_SYSCALL6(name,arg1,arg2,arg3,arg4,arg5,arg6) \
-{ \
-long __res; \
-__asm__ volatile ("push %%ebp ; movl %%eax,%%ebp ; movl %1,%%eax ; int $0x80 ; pop %%ebp" \
- : "=a" (__res) \
- : "i" (__NR_##name),"b" ((long)(arg1)),"c" ((long)(arg2)), \
- "d" ((long)(arg3)),"S" ((long)(arg4)),"D" ((long)(arg5)), \
- "0" ((long)(arg6))); \
-return __res; \
-}
+#ifdef CONFIG_SOLARIS
+#include <sys/types.h>
+#include <sys/statvfs.h>
+/* See MySQL bug #7156 (http://bugs.mysql.com/bug.php?id=7156) for
+ discussion about Solaris header problems */
+extern int madvise(caddr_t, size_t, int);
+#endif
-/****************************************************************/
-/* shmat replacement */
+#include "qemu-common.h"
+#include "trace.h"
+#include "qemu_socket.h"
-int qemu_ipc(int call, unsigned long first,
- unsigned long second, unsigned long third,
- void *ptr, unsigned long fifth)
+int socket_set_cork(int fd, int v)
{
- QEMU_SYSCALL6(ipc, call, first, second, third, ptr, fifth);
+#if defined(SOL_TCP) && defined(TCP_CORK)
+ return setsockopt(fd, SOL_TCP, TCP_CORK, &v, sizeof(v));
+#else
+ return 0;
+#endif
}
-#define SHMAT 21
-
-/* we must define shmat so that a specific address will be used when
- mapping the X11 ximage */
-void *shmat(int shmid, const void *shmaddr, int shmflg)
+int qemu_madvise(void *addr, size_t len, int advice)
{
- void *ptr;
- int ret;
- /* we give an address in the right memory area */
- if (!shmaddr)
- shmaddr = get_mmap_addr(8192 * 1024);
- ret = qemu_ipc(SHMAT, shmid, shmflg, (unsigned long)&ptr, (void *)shmaddr, 0);
- if (ret < 0)
- return NULL;
- return ptr;
+ if (advice == QEMU_MADV_INVALID) {
+ errno = EINVAL;
+ return -1;
+ }
+#if defined(CONFIG_MADVISE)
+ return madvise(addr, len, advice);
+#elif defined(CONFIG_POSIX_MADVISE)
+ return posix_madvise(addr, len, advice);
+#else
+ errno = EINVAL;
+ return -1;
+#endif
}
-/****************************************************************/
-/* sigaction bypassing the threads */
-
-static int kernel_sigaction(int signum, const struct qemu_sigaction *act,
- struct qemu_sigaction *oldact,
- int sigsetsize)
-{
- QEMU_SYSCALL4(rt_sigaction, signum, act, oldact, sigsetsize);
-}
-int qemu_sigaction(int signum, const struct qemu_sigaction *act,
- struct qemu_sigaction *oldact)
+/*
+ * Opens a file with FD_CLOEXEC set
+ */
+int qemu_open(const char *name, int flags, ...)
{
- return kernel_sigaction(signum, act, oldact, 8);
-}
-
-/****************************************************************/
-/* memory allocation */
-
-//#define DEBUG_MALLOC
-
-#define MALLOC_BASE 0xab000000
-#define PHYS_RAM_BASE 0xac000000
+ int ret;
+ int mode = 0;
-#define MALLOC_ALIGN 16
-#define BLOCK_HEADER_SIZE 16
+ if (flags & O_CREAT) {
+ va_list ap;
-typedef struct MemoryBlock {
- struct MemoryBlock *next;
- unsigned long size; /* size of block, including header */
-} MemoryBlock;
+ va_start(ap, flags);
+ mode = va_arg(ap, int);
+ va_end(ap);
+ }
-static MemoryBlock *first_free_block;
-static unsigned long malloc_addr = MALLOC_BASE;
+#ifdef O_CLOEXEC
+ ret = open(name, flags | O_CLOEXEC, mode);
+#else
+ ret = open(name, flags, mode);
+ if (ret >= 0) {
+ qemu_set_cloexec(ret);
+ }
+#endif
-static void *malloc_get_space(size_t size)
-{
- void *ptr;
- size = TARGET_PAGE_ALIGN(size);
- ptr = mmap((void *)malloc_addr, size,
- PROT_WRITE | PROT_READ,
- MAP_PRIVATE | MAP_FIXED | MAP_ANON, -1, 0);
- if (ptr == MAP_FAILED)
- return NULL;
- malloc_addr += size;
- return ptr;
+ return ret;
}
-void *qemu_malloc(size_t size)
+/*
+ * A variant of write(2) which handles partial write.
+ *
+ * Return the number of bytes transferred.
+ * Set errno if fewer than `count' bytes are written.
+ *
+ * This function don't work with non-blocking fd's.
+ * Any of the possibilities with non-bloking fd's is bad:
+ * - return a short write (then name is wrong)
+ * - busy wait adding (errno == EAGAIN) to the loop
+ */
+ssize_t qemu_write_full(int fd, const void *buf, size_t count)
{
- MemoryBlock *mb, *mb1, **pmb;
- void *ptr;
- size_t size1, area_size;
-
- if (size == 0)
- return NULL;
+ ssize_t ret = 0;
+ ssize_t total = 0;
- size = (size + BLOCK_HEADER_SIZE + MALLOC_ALIGN - 1) & ~(MALLOC_ALIGN - 1);
- pmb = &first_free_block;
- for(;;) {
- mb = *pmb;
- if (mb == NULL)
+ while (count) {
+ ret = write(fd, buf, count);
+ if (ret < 0) {
+ if (errno == EINTR)
+ continue;
break;
- if (size <= mb->size)
- goto found;
- pmb = &mb->next;
- }
- /* no big enough blocks found: get new space */
- area_size = TARGET_PAGE_ALIGN(size);
- mb = malloc_get_space(area_size);
- if (!mb)
- return NULL;
- size1 = area_size - size;
- if (size1 > 0) {
- /* create a new free block */
- mb1 = (MemoryBlock *)((uint8_t *)mb + size);
- mb1->next = NULL;
- mb1->size = size1;
- *pmb = mb1;
- }
- goto the_end;
- found:
- /* a free block was found: use it */
- size1 = mb->size - size;
- if (size1 > 0) {
- /* create a new free block */
- mb1 = (MemoryBlock *)((uint8_t *)mb + size);
- mb1->next = mb->next;
- mb1->size = size1;
- *pmb = mb1;
- } else {
- /* suppress the first block */
- *pmb = mb->next;
- }
- the_end:
- mb->size = size;
- mb->next = NULL;
- ptr = ((uint8_t *)mb + BLOCK_HEADER_SIZE);
-#ifdef DEBUG_MALLOC
- qemu_printf("malloc: size=0x%x ptr=0x%lx\n", size, (unsigned long)ptr);
-#endif
- return ptr;
-}
+ }
-void qemu_free(void *ptr)
-{
- MemoryBlock *mb;
+ count -= ret;
+ buf += ret;
+ total += ret;
+ }
- if (!ptr)
- return;
- mb = (MemoryBlock *)((uint8_t *)ptr - BLOCK_HEADER_SIZE);
- mb->next = first_free_block;
- first_free_block = mb;
+ return total;
}
-/****************************************************************/
-/* virtual memory allocation */
-
-unsigned long mmap_addr = PHYS_RAM_BASE;
-
-void *get_mmap_addr(unsigned long size)
+/*
+ * Opens a socket with FD_CLOEXEC set
+ */
+int qemu_socket(int domain, int type, int protocol)
{
- unsigned long addr;
- addr = mmap_addr;
- mmap_addr += ((size + 4095) & ~4095) + 4096;
- return (void *)addr;
-}
-
-#else
+ int ret;
-#ifdef _WIN32
-#include <windows.h>
-#elif defined(_BSD)
-#include <stdlib.h>
-#else
-#include <malloc.h>
+#ifdef SOCK_CLOEXEC
+ ret = socket(domain, type | SOCK_CLOEXEC, protocol);
+ if (ret != -1 || errno != EINVAL) {
+ return ret;
+ }
#endif
+ ret = socket(domain, type, protocol);
+ if (ret >= 0) {
+ qemu_set_cloexec(ret);
+ }
-void *get_mmap_addr(unsigned long size)
-{
- return NULL;
-}
-
-void qemu_free(void *ptr)
-{
- free(ptr);
+ return ret;
}
-void *qemu_malloc(size_t size)
+/*
+ * Accept a connection and set FD_CLOEXEC
+ */
+int qemu_accept(int s, struct sockaddr *addr, socklen_t *addrlen)
{
- return malloc(size);
-}
-
-#if defined(_WIN32)
+ int ret;
-void *qemu_vmalloc(size_t size)
-{
- /* FIXME: this is not exactly optimal solution since VirtualAlloc
- has 64Kb granularity, but at least it guarantees us that the
- memory is page aligned. */
- return VirtualAlloc(NULL, size, MEM_COMMIT, PAGE_READWRITE);
-}
+#ifdef CONFIG_ACCEPT4
+ ret = accept4(s, addr, addrlen, SOCK_CLOEXEC);
+ if (ret != -1 || errno != ENOSYS) {
+ return ret;
+ }
+#endif
+ ret = accept(s, addr, addrlen);
+ if (ret >= 0) {
+ qemu_set_cloexec(ret);
+ }
-void qemu_vfree(void *ptr)
-{
- VirtualFree(ptr, 0, MEM_RELEASE);
+ return ret;
}
-#else
-
-#if defined(USE_KQEMU)
-
-#include <sys/vfs.h>
-#include <sys/mman.h>
-#include <fcntl.h>
-
-void *kqemu_vmalloc(size_t size)
+/*
+ * A variant of send(2) which handles partial write.
+ *
+ * Return the number of bytes transferred, which is only
+ * smaller than `count' if there is an error.
+ *
+ * This function won't work with non-blocking fd's.
+ * Any of the possibilities with non-bloking fd's is bad:
+ * - return a short write (then name is wrong)
+ * - busy wait adding (errno == EAGAIN) to the loop
+ */
+ssize_t qemu_send_full(int fd, const void *buf, size_t count, int flags)
{
- static int phys_ram_fd = -1;
- static int phys_ram_size = 0;
- const char *tmpdir;
- char phys_ram_file[1024];
- void *ptr;
- struct statfs stfs;
-
- if (phys_ram_fd < 0) {
- tmpdir = getenv("QEMU_TMPDIR");
- if (!tmpdir)
- tmpdir = "/dev/shm";
- if (statfs(tmpdir, &stfs) == 0) {
- int64_t free_space;
- int ram_mb;
+ ssize_t ret = 0;
+ ssize_t total = 0;
- extern int ram_size;
- free_space = (int64_t)stfs.f_bavail * stfs.f_bsize;
- if ((ram_size + 8192 * 1024) >= free_space) {
- ram_mb = (ram_size / (1024 * 1024));
- fprintf(stderr,
- "You do not have enough space in '%s' for the %d MB of QEMU virtual RAM.\n",
- tmpdir, ram_mb);
- if (strcmp(tmpdir, "/dev/shm") == 0) {
- fprintf(stderr, "To have more space available provided you have enough RAM and swap, do as root:\n"
- "umount /dev/shm\n"
- "mount -t tmpfs -o size=%dm none /dev/shm\n",
- ram_mb + 16);
- } else {
- fprintf(stderr,
- "Use the '-m' option of QEMU to diminish the amount of virtual RAM or use the\n"
- "QEMU_TMPDIR environment variable to set another directory where the QEMU\n"
- "temporary RAM file will be opened.\n");
- }
- fprintf(stderr, "Or disable the accelerator module with -no-kqemu\n");
- exit(1);
+ while (count) {
+ ret = send(fd, buf, count, flags);
+ if (ret < 0) {
+ if (errno == EINTR) {
+ continue;
}
+ break;
}
- snprintf(phys_ram_file, sizeof(phys_ram_file), "%s/qemuXXXXXX",
- tmpdir);
- if (mkstemp(phys_ram_file) < 0) {
- fprintf(stderr,
- "warning: could not create temporary file in '%s'.\n"
- "Use QEMU_TMPDIR to select a directory in a tmpfs filesystem.\n"
- "Using '/tmp' as fallback.\n",
- tmpdir);
- snprintf(phys_ram_file, sizeof(phys_ram_file), "%s/qemuXXXXXX",
- "/tmp");
- if (mkstemp(phys_ram_file) < 0) {
- fprintf(stderr, "Could not create temporary memory file '%s'\n",
- phys_ram_file);
- exit(1);
- }
- }
- phys_ram_fd = open(phys_ram_file, O_CREAT | O_TRUNC | O_RDWR, 0600);
- if (phys_ram_fd < 0) {
- fprintf(stderr, "Could not open temporary memory file '%s'\n",
- phys_ram_file);
- exit(1);
- }
- unlink(phys_ram_file);
- }
- size = (size + 4095) & ~4095;
- ftruncate(phys_ram_fd, phys_ram_size + size);
- ptr = mmap(NULL,
- size,
- PROT_WRITE | PROT_READ, MAP_SHARED,
- phys_ram_fd, phys_ram_size);
- if (ptr == MAP_FAILED) {
- fprintf(stderr, "Could not map physical memory\n");
- exit(1);
- }
- phys_ram_size += size;
- return ptr;
-}
-
-void kqemu_vfree(void *ptr)
-{
- /* may be useful some day, but currently we do not need to free */
-}
-#endif
+ count -= ret;
+ buf += ret;
+ total += ret;
+ }
-/* alloc shared memory pages */
-void *qemu_vmalloc(size_t size)
-{
-#if defined(USE_KQEMU)
- if (kqemu_allowed)
- return kqemu_vmalloc(size);
-#endif
-#ifdef _BSD
- return valloc(size);
-#else
- return memalign(4096, size);
-#endif
+ return total;
}
-void qemu_vfree(void *ptr)
+/*
+ * A variant of recv(2) which handles partial write.
+ *
+ * Return the number of bytes transferred, which is only
+ * smaller than `count' if there is an error.
+ *
+ * This function won't work with non-blocking fd's.
+ * Any of the possibilities with non-bloking fd's is bad:
+ * - return a short write (then name is wrong)
+ * - busy wait adding (errno == EAGAIN) to the loop
+ */
+ssize_t qemu_recv_full(int fd, void *buf, size_t count, int flags)
{
-#if defined(USE_KQEMU)
- if (kqemu_allowed)
- kqemu_vfree(ptr);
-#endif
- free(ptr);
-}
+ ssize_t ret = 0;
+ ssize_t total = 0;
-#endif
+ while (count) {
+ ret = qemu_recv(fd, buf, count, flags);
+ if (ret <= 0) {
+ if (ret < 0 && errno == EINTR) {
+ continue;
+ }
+ break;
+ }
-#endif
+ count -= ret;
+ buf += ret;
+ total += ret;
+ }
-void *qemu_mallocz(size_t size)
-{
- void *ptr;
- ptr = qemu_malloc(size);
- if (!ptr)
- return NULL;
- memset(ptr, 0, size);
- return ptr;
+ return total;
}
-char *qemu_strdup(const char *str)
-{
- char *ptr;
- ptr = qemu_malloc(strlen(str) + 1);
- if (!ptr)
- return NULL;
- strcpy(ptr, str);
- return ptr;
-}