*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
- * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ * Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
+ * MA 02110-1301, USA.
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
+#include <sys/types.h>
+#include <sys/stat.h>
#include <sys/mman.h>
+#include <linux/mman.h>
+#include <linux/unistd.h>
#include "qemu.h"
+#include "qemu-common.h"
//#define DEBUG_MMAP
+#if defined(USE_NPTL)
+pthread_mutex_t mmap_mutex;
+static int __thread mmap_lock_count;
+
+void mmap_lock(void)
+{
+ if (mmap_lock_count++ == 0) {
+ pthread_mutex_lock(&mmap_mutex);
+ }
+}
+
+void mmap_unlock(void)
+{
+ if (--mmap_lock_count == 0) {
+ pthread_mutex_unlock(&mmap_mutex);
+ }
+}
+
+/* Grab lock to make sure things are in a consistent state after fork(). */
+void mmap_fork_start(void)
+{
+ if (mmap_lock_count)
+ abort();
+ pthread_mutex_lock(&mmap_mutex);
+}
+
+void mmap_fork_end(int child)
+{
+ if (child)
+ pthread_mutex_init(&mmap_mutex, NULL);
+ else
+ pthread_mutex_unlock(&mmap_mutex);
+}
+#else
+/* We aren't threadsafe to start with, so no need to worry about locking. */
+void mmap_lock(void)
+{
+}
+
+void mmap_unlock(void)
+{
+}
+#endif
+
+void *qemu_vmalloc(size_t size)
+{
+ void *p;
+ unsigned long addr;
+ mmap_lock();
+ /* Use map and mark the pages as used. */
+ p = mmap(NULL, size, PROT_READ | PROT_WRITE,
+ MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+
+ addr = (unsigned long)p;
+ if (addr == (target_ulong) addr) {
+ /* Allocated region overlaps guest address space.
+ This may recurse. */
+ page_set_flags(addr & TARGET_PAGE_MASK, TARGET_PAGE_ALIGN(addr + size),
+ PAGE_RESERVED);
+ }
+
+ mmap_unlock();
+ return p;
+}
+
+void *qemu_malloc(size_t size)
+{
+ char * p;
+ size += 16;
+ p = qemu_vmalloc(size);
+ *(size_t *)p = size;
+ return p + 16;
+}
+
+/* We use map, which is always zero initialized. */
+void * qemu_mallocz(size_t size)
+{
+ return qemu_malloc(size);
+}
+
+void qemu_free(void *ptr)
+{
+ /* FIXME: We should unmark the reserved pages here. However this gets
+ complicated when one target page spans multiple host pages, so we
+ don't bother. */
+ size_t *p;
+ p = (size_t *)((char *)ptr - 16);
+ munmap(p, *p);
+}
+
+void *qemu_realloc(void *ptr, size_t size)
+{
+ size_t old_size, copy;
+ void *new_ptr;
+
+ if (!ptr)
+ return qemu_malloc(size);
+ old_size = *(size_t *)((char *)ptr - 16);
+ copy = old_size < size ? old_size : size;
+ new_ptr = qemu_malloc(size);
+ memcpy(new_ptr, ptr, copy);
+ qemu_free(ptr);
+ return new_ptr;
+}
+
/* NOTE: all the constants are the HOST ones, but addresses are target. */
int target_mprotect(abi_ulong start, abi_ulong len, int prot)
{
#ifdef DEBUG_MMAP
printf("mprotect: start=0x" TARGET_FMT_lx
- "len=0x" TARGET_FMT_lx " prot=%c%c%c\n", start, len,
+ "len=0x" TARGET_FMT_lx " prot=%c%c%c\n", start, len,
prot & PROT_READ ? 'r' : '-',
prot & PROT_WRITE ? 'w' : '-',
prot & PROT_EXEC ? 'x' : '-');
end = start + len;
if (end < start)
return -EINVAL;
- if (prot & ~(PROT_READ | PROT_WRITE | PROT_EXEC))
- return -EINVAL;
+ prot &= PROT_READ | PROT_WRITE | PROT_EXEC;
if (len == 0)
return 0;
+ mmap_lock();
host_start = start & qemu_host_page_mask;
host_end = HOST_PAGE_ALIGN(end);
if (start > host_start) {
}
ret = mprotect(g2h(host_start), qemu_host_page_size, prot1 & PAGE_BITS);
if (ret != 0)
- return ret;
+ goto error;
host_start += qemu_host_page_size;
}
if (end < host_end) {
ret = mprotect(g2h(host_end - qemu_host_page_size), qemu_host_page_size,
prot1 & PAGE_BITS);
if (ret != 0)
- return ret;
+ goto error;
host_end -= qemu_host_page_size;
}
if (host_start < host_end) {
ret = mprotect(g2h(host_start), host_end - host_start, prot);
if (ret != 0)
- return ret;
+ goto error;
}
page_set_flags(start, start + len, prot | PAGE_VALID);
+ mmap_unlock();
return 0;
+error:
+ mmap_unlock();
+ return ret;
}
/* map an incomplete host page */
abi_ulong start, abi_ulong end,
int prot, int flags, int fd, abi_ulong offset)
{
- abi_ulong real_end, ret, addr;
+ abi_ulong real_end, addr;
void *host_start;
int prot1, prot_new;
if (prot1 == 0) {
/* no page was there, so we allocate one */
- ret = (long)mmap(host_start, qemu_host_page_size, prot,
- flags | MAP_ANONYMOUS, -1, 0);
- if (ret == -1)
- return ret;
+ void *p = mmap(host_start, qemu_host_page_size, prot,
+ flags | MAP_ANONYMOUS, -1, 0);
+ if (p == MAP_FAILED)
+ return -1;
prot1 = prot;
}
prot1 &= PAGE_BITS;
return 0;
}
+#if defined(__CYGWIN__)
+/* Cygwin doesn't have a whole lot of address space. */
+static abi_ulong mmap_next_start = 0x18000000;
+#else
+static abi_ulong mmap_next_start = 0x40000000;
+#endif
+
+unsigned long last_brk;
+
+/* find a free memory area of size 'size'. The search starts at
+ 'start'. If 'start' == 0, then a default start address is used.
+ Return -1 if error.
+*/
+/* page_init() marks pages used by the host as reserved to be sure not
+ to use them. */
+static abi_ulong mmap_find_vma(abi_ulong start, abi_ulong size)
+{
+ abi_ulong addr, addr1, addr_start;
+ int prot;
+ unsigned long new_brk;
+
+ new_brk = (unsigned long)sbrk(0);
+ if (last_brk && last_brk < new_brk && last_brk == (target_ulong)last_brk) {
+ /* This is a hack to catch the host allocating memory with brk().
+ If it uses mmap then we loose.
+ FIXME: We really want to avoid the host allocating memory in
+ the first place, and maybe leave some slack to avoid switching
+ to mmap. */
+ page_set_flags(last_brk & TARGET_PAGE_MASK,
+ TARGET_PAGE_ALIGN(new_brk),
+ PAGE_RESERVED);
+ }
+ last_brk = new_brk;
+
+ size = HOST_PAGE_ALIGN(size);
+ start = start & qemu_host_page_mask;
+ addr = start;
+ if (addr == 0)
+ addr = mmap_next_start;
+ addr_start = addr;
+ for(;;) {
+ prot = 0;
+ for(addr1 = addr; addr1 < (addr + size); addr1 += TARGET_PAGE_SIZE) {
+ prot |= page_get_flags(addr1);
+ }
+ if (prot == 0)
+ break;
+ addr += qemu_host_page_size;
+ /* we found nothing */
+ if (addr == addr_start)
+ return (abi_ulong)-1;
+ }
+ if (start == 0)
+ mmap_next_start = addr + size;
+ return addr;
+}
+
/* NOTE: all the constants are the HOST ones */
abi_long target_mmap(abi_ulong start, abi_ulong len, int prot,
int flags, int fd, abi_ulong offset)
{
abi_ulong ret, end, real_start, real_end, retaddr, host_offset, host_len;
unsigned long host_start;
-#if defined(__alpha__) || defined(__sparc__) || defined(__x86_64__) || \
- defined(__ia64) || defined(__mips__)
- static abi_ulong last_start = 0x40000000;
-#elif defined(__CYGWIN__)
- /* Cygwin doesn't have a whole lot of address space. */
- static abi_ulong last_start = 0x18000000;
-#endif
+ mmap_lock();
#ifdef DEBUG_MMAP
{
printf("mmap: start=0x" TARGET_FMT_lx
- " len=0x" TARGET_FMT_lx " prot=%c%c%c flags=",
+ " len=0x" TARGET_FMT_lx " prot=%c%c%c flags=",
start, len,
prot & PROT_READ ? 'r' : '-',
prot & PROT_WRITE ? 'w' : '-',
if (offset & ~TARGET_PAGE_MASK) {
errno = EINVAL;
- return -1;
+ goto fail;
}
len = TARGET_PAGE_ALIGN(len);
if (len == 0)
- return start;
+ goto the_end;
real_start = start & qemu_host_page_mask;
+ /* When mapping files into a memory area larger than the file, accesses
+ to pages beyond the file size will cause a SIGBUS.
+
+ For example, if mmaping a file of 100 bytes on a host with 4K pages
+ emulating a target with 8K pages, the target expects to be able to
+ access the first 8K. But the host will trap us on any access beyond
+ 4K.
+
+ When emulating a target with a larger page-size than the hosts, we
+ may need to truncate file maps at EOF and add extra anonymous pages
+ up to the targets page boundary. */
+
+ if ((qemu_real_host_page_size < TARGET_PAGE_SIZE)
+ && !(flags & MAP_ANONYMOUS)) {
+ struct stat sb;
+
+ if (fstat (fd, &sb) == -1)
+ goto fail;
+
+ /* Are we trying to create a map beyond EOF?. */
+ if (offset + len > sb.st_size) {
+ /* If so, truncate the file map at eof aligned with
+ the hosts real pagesize. Additional anonymous maps
+ will be created beyond EOF. */
+ len = (sb.st_size - offset);
+ len += qemu_real_host_page_size - 1;
+ len &= ~(qemu_real_host_page_size - 1);
+ }
+ }
+
if (!(flags & MAP_FIXED)) {
-#if defined(__alpha__) || defined(__sparc__) || defined(__x86_64__) || \
- defined(__ia64) || defined(__mips__) || defined(__CYGWIN__)
- /* tell the kernel to search at the same place as i386 */
- if (real_start == 0) {
- real_start = last_start;
- last_start += HOST_PAGE_ALIGN(len);
+ abi_ulong mmap_start;
+ void *p;
+ host_offset = offset & qemu_host_page_mask;
+ host_len = len + offset - host_offset;
+ host_len = HOST_PAGE_ALIGN(host_len);
+ mmap_start = mmap_find_vma(real_start, host_len);
+ if (mmap_start == (abi_ulong)-1) {
+ errno = ENOMEM;
+ goto fail;
}
-#endif
- host_offset = offset & qemu_host_page_mask;
- host_len = len + offset - host_offset;
-
- if (qemu_host_page_size > qemu_real_host_page_size) {
- /*
- * The guest expects to see mmapped areas aligned to it's pagesize.
- * If the host's real page size is smaller than the guest's, we need
- * to fixup the maps. It is done by allocating a larger area,
- * displacing the map (if needed) and finally chopping off the spare
- * room at the edges.
- */
-
- /*
- * We assume qemu_host_page_size is always the same as
- * TARGET_PAGE_SIZE, see exec.c. qemu_real_host_page_size is the
- * hosts real page size.
- */
- abi_ulong host_end;
- unsigned long host_aligned_start;
-
- host_len = HOST_PAGE_ALIGN(host_len + qemu_host_page_size
- - qemu_real_host_page_size);
- host_start = (unsigned long) mmap(real_start ?
- g2h(real_start) : NULL,
- host_len, prot, flags,
- fd, host_offset);
- if (host_start == -1)
- return -1;
-
- host_end = host_start + host_len;
-
- /* Find start and end, aligned to the targets pagesize with-in the
- large mmaped area. */
- host_aligned_start = TARGET_PAGE_ALIGN(host_start);
- if (!(flags & MAP_ANONYMOUS))
- host_aligned_start += offset - host_offset;
-
- start = h2g(host_aligned_start);
- end = start + TARGET_PAGE_ALIGN(len);
-
- /* Chop off the leftovers, if any. */
- if (host_aligned_start > host_start)
- munmap((void *)host_start, host_aligned_start - host_start);
- if (end < host_end)
- munmap((void *)g2h(end), host_end - end);
-
- goto the_end1;
- } else {
- /* if not fixed, no need to do anything */
- host_start = (long)mmap(real_start ? g2h(real_start) : NULL,
- host_len, prot, flags, fd, host_offset);
- if (host_start == -1)
- return -1;
- /* update start so that it points to the file position at 'offset' */
- if (!(flags & MAP_ANONYMOUS))
- host_start += offset - host_offset;
- start = h2g(host_start);
- goto the_end1;
+ /* Note: we prefer to control the mapping address. It is
+ especially important if qemu_host_page_size >
+ qemu_real_host_page_size */
+ p = mmap(g2h(mmap_start),
+ host_len, prot, flags | MAP_FIXED | MAP_ANONYMOUS, -1, 0);
+ if (p == MAP_FAILED)
+ goto fail;
+ /* update start so that it points to the file position at 'offset' */
+ host_start = (unsigned long)p;
+ if (!(flags & MAP_ANONYMOUS)) {
+ p = mmap(g2h(mmap_start), len, prot,
+ flags | MAP_FIXED, fd, host_offset);
+ host_start += offset - host_offset;
}
- }
-
- if (start & ~TARGET_PAGE_MASK) {
- errno = EINVAL;
- return -1;
- }
- end = start + len;
- real_end = HOST_PAGE_ALIGN(end);
+ start = h2g(host_start);
+ } else {
+ int flg;
+ target_ulong addr;
- /* worst case: we cannot map the file because the offset is not
- aligned, so we read it */
- if (!(flags & MAP_ANONYMOUS) &&
- (offset & ~qemu_host_page_mask) != (start & ~qemu_host_page_mask)) {
- /* msync() won't work here, so we return an error if write is
- possible while it is a shared mapping */
- if ((flags & MAP_TYPE) == MAP_SHARED &&
- (prot & PROT_WRITE)) {
+ if (start & ~TARGET_PAGE_MASK) {
errno = EINVAL;
- return -1;
+ goto fail;
}
- retaddr = target_mmap(start, len, prot | PROT_WRITE,
- MAP_FIXED | MAP_PRIVATE | MAP_ANONYMOUS,
- -1, 0);
- if (retaddr == -1)
- return -1;
- pread(fd, g2h(start), len, offset);
- if (!(prot & PROT_WRITE)) {
- ret = target_mprotect(start, len, prot);
- if (ret != 0)
- return ret;
+ end = start + len;
+ real_end = HOST_PAGE_ALIGN(end);
+
+ /*
+ * Test if requested memory area fits target address space
+ * It can fail only on 64-bit host with 32-bit target.
+ * On any other target/host host mmap() handles this error correctly.
+ */
+ if ((unsigned long)start + len - 1 > (abi_ulong) -1) {
+ errno = EINVAL;
+ goto fail;
}
- goto the_end;
- }
- /* handle the start of the mapping */
- if (start > real_start) {
- if (real_end == real_start + qemu_host_page_size) {
- /* one single host page */
- ret = mmap_frag(real_start, start, end,
+ for(addr = real_start; addr < real_end; addr += TARGET_PAGE_SIZE) {
+ flg = page_get_flags(addr);
+ if (flg & PAGE_RESERVED) {
+ errno = ENXIO;
+ goto fail;
+ }
+ }
+
+ /* worst case: we cannot map the file because the offset is not
+ aligned, so we read it */
+ if (!(flags & MAP_ANONYMOUS) &&
+ (offset & ~qemu_host_page_mask) != (start & ~qemu_host_page_mask)) {
+ /* msync() won't work here, so we return an error if write is
+ possible while it is a shared mapping */
+ if ((flags & MAP_TYPE) == MAP_SHARED &&
+ (prot & PROT_WRITE)) {
+ errno = EINVAL;
+ goto fail;
+ }
+ retaddr = target_mmap(start, len, prot | PROT_WRITE,
+ MAP_FIXED | MAP_PRIVATE | MAP_ANONYMOUS,
+ -1, 0);
+ if (retaddr == -1)
+ goto fail;
+ pread(fd, g2h(start), len, offset);
+ if (!(prot & PROT_WRITE)) {
+ ret = target_mprotect(start, len, prot);
+ if (ret != 0) {
+ start = ret;
+ goto the_end;
+ }
+ }
+ goto the_end;
+ }
+
+ /* handle the start of the mapping */
+ if (start > real_start) {
+ if (real_end == real_start + qemu_host_page_size) {
+ /* one single host page */
+ ret = mmap_frag(real_start, start, end,
+ prot, flags, fd, offset);
+ if (ret == -1)
+ goto fail;
+ goto the_end1;
+ }
+ ret = mmap_frag(real_start, start, real_start + qemu_host_page_size,
prot, flags, fd, offset);
if (ret == -1)
- return ret;
- goto the_end1;
+ goto fail;
+ real_start += qemu_host_page_size;
+ }
+ /* handle the end of the mapping */
+ if (end < real_end) {
+ ret = mmap_frag(real_end - qemu_host_page_size,
+ real_end - qemu_host_page_size, real_end,
+ prot, flags, fd,
+ offset + real_end - qemu_host_page_size - start);
+ if (ret == -1)
+ goto fail;
+ real_end -= qemu_host_page_size;
}
- ret = mmap_frag(real_start, start, real_start + qemu_host_page_size,
- prot, flags, fd, offset);
- if (ret == -1)
- return ret;
- real_start += qemu_host_page_size;
- }
- /* handle the end of the mapping */
- if (end < real_end) {
- ret = mmap_frag(real_end - qemu_host_page_size,
- real_end - qemu_host_page_size, real_end,
- prot, flags, fd,
- offset + real_end - qemu_host_page_size - start);
- if (ret == -1)
- return -1;
- real_end -= qemu_host_page_size;
- }
- /* map the middle (easier) */
- if (real_start < real_end) {
- unsigned long offset1;
- if (flags & MAP_ANONYMOUS)
- offset1 = 0;
- else
- offset1 = offset + real_start - start;
- ret = (long)mmap(g2h(real_start), real_end - real_start,
- prot, flags, fd, offset1);
- if (ret == -1)
- return -1;
+ /* map the middle (easier) */
+ if (real_start < real_end) {
+ void *p;
+ unsigned long offset1;
+ if (flags & MAP_ANONYMOUS)
+ offset1 = 0;
+ else
+ offset1 = offset + real_start - start;
+ p = mmap(g2h(real_start), real_end - real_start,
+ prot, flags, fd, offset1);
+ if (p == MAP_FAILED)
+ goto fail;
+ }
}
the_end1:
page_set_flags(start, start + len, prot | PAGE_VALID);
the_end:
#ifdef DEBUG_MMAP
- printf("ret=0x%llx\n", start);
+ printf("ret=0x" TARGET_FMT_lx "\n", start);
page_dump(stdout);
printf("\n");
#endif
+ mmap_unlock();
return start;
+fail:
+ mmap_unlock();
+ return -1;
}
int target_munmap(abi_ulong start, abi_ulong len)
len = TARGET_PAGE_ALIGN(len);
if (len == 0)
return -EINVAL;
+ mmap_lock();
end = start + len;
real_start = start & qemu_host_page_mask;
real_end = HOST_PAGE_ALIGN(end);
real_end -= qemu_host_page_size;
}
+ ret = 0;
/* unmap what we can */
if (real_start < real_end) {
ret = munmap(g2h(real_start), real_end - real_start);
- if (ret != 0)
- return ret;
}
- page_set_flags(start, start + len, 0);
- return 0;
+ if (ret == 0)
+ page_set_flags(start, start + len, 0);
+ mmap_unlock();
+ return ret;
}
-/* XXX: currently, we only handle MAP_ANONYMOUS and not MAP_FIXED
- blocks which have been allocated starting on a host page */
abi_long target_mremap(abi_ulong old_addr, abi_ulong old_size,
abi_ulong new_size, unsigned long flags,
abi_ulong new_addr)
{
int prot;
- unsigned long host_addr;
-
- /* XXX: use 5 args syscall */
- host_addr = (long)mremap(g2h(old_addr), old_size, new_size, flags);
- if (host_addr == -1)
- return -1;
- new_addr = h2g(host_addr);
- prot = page_get_flags(old_addr);
- page_set_flags(old_addr, old_addr + old_size, 0);
- page_set_flags(new_addr, new_addr + new_size, prot | PAGE_VALID);
+ void *host_addr;
+
+ mmap_lock();
+
+ if (flags & MREMAP_FIXED)
+ host_addr = (void *) syscall(__NR_mremap, g2h(old_addr),
+ old_size, new_size,
+ flags,
+ new_addr);
+ else if (flags & MREMAP_MAYMOVE) {
+ abi_ulong mmap_start;
+
+ mmap_start = mmap_find_vma(0, new_size);
+
+ if (mmap_start == -1) {
+ errno = ENOMEM;
+ host_addr = MAP_FAILED;
+ } else
+ host_addr = (void *) syscall(__NR_mremap, g2h(old_addr),
+ old_size, new_size,
+ flags | MREMAP_FIXED,
+ g2h(mmap_start));
+ } else {
+ host_addr = mremap(g2h(old_addr), old_size, new_size, flags);
+ /* Check if address fits target address space */
+ if ((unsigned long)host_addr + new_size > (abi_ulong)-1) {
+ /* Revert mremap() changes */
+ host_addr = mremap(g2h(old_addr), new_size, old_size, flags);
+ errno = ENOMEM;
+ host_addr = MAP_FAILED;
+ }
+ }
+
+ if (host_addr == MAP_FAILED) {
+ new_addr = -1;
+ } else {
+ new_addr = h2g(host_addr);
+ prot = page_get_flags(old_addr);
+ page_set_flags(old_addr, old_addr + old_size, 0);
+ page_set_flags(new_addr, new_addr + new_size, prot | PAGE_VALID);
+ }
+ mmap_unlock();
return new_addr;
}
start &= qemu_host_page_mask;
return msync(g2h(start), end - start, flags);
}
-