migrate: Update ram_block_discard_range for shared

The choice of call to discard a block is getting more complicated
for other cases.   We use fallocate PUNCH_HOLE in any file cases;
it works for both hugepage and for tmpfs.
We use the DONTNEED for non-hugepage cases either where they're
anonymous or where they're private.

Care should be taken when trying other backing files.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This commit is contained in:
Dr. David Alan Gilbert 2018-03-12 17:20:56 +00:00 committed by Michael S. Tsirkin
parent 9578f8cc3e
commit db144f7000
2 changed files with 50 additions and 17 deletions

64
exec.c
View File

@ -3721,6 +3721,7 @@ int ram_block_discard_range(RAMBlock *rb, uint64_t start, size_t length)
} }
if ((start + length) <= rb->used_length) { if ((start + length) <= rb->used_length) {
bool need_madvise, need_fallocate;
uint8_t *host_endaddr = host_startaddr + length; uint8_t *host_endaddr = host_startaddr + length;
if ((uintptr_t)host_endaddr & (rb->page_size - 1)) { if ((uintptr_t)host_endaddr & (rb->page_size - 1)) {
error_report("ram_block_discard_range: Unaligned end address: %p", error_report("ram_block_discard_range: Unaligned end address: %p",
@ -3730,29 +3731,60 @@ int ram_block_discard_range(RAMBlock *rb, uint64_t start, size_t length)
errno = ENOTSUP; /* If we are missing MADVISE etc */ errno = ENOTSUP; /* If we are missing MADVISE etc */
if (rb->page_size == qemu_host_page_size) { /* The logic here is messy;
#if defined(CONFIG_MADVISE) * madvise DONTNEED fails for hugepages
/* Note: We need the madvise MADV_DONTNEED behaviour of definitely * fallocate works on hugepages and shmem
* freeing the page. */
*/ need_madvise = (rb->page_size == qemu_host_page_size);
ret = madvise(host_startaddr, length, MADV_DONTNEED); need_fallocate = rb->fd != -1;
#endif if (need_fallocate) {
} else { /* For a file, this causes the area of the file to be zero'd
/* Huge page case - unfortunately it can't do DONTNEED, but * if read, and for hugetlbfs also causes it to be unmapped
* it can do the equivalent by FALLOC_FL_PUNCH_HOLE in the * so a userfault will trigger.
* huge page file.
*/ */
#ifdef CONFIG_FALLOCATE_PUNCH_HOLE #ifdef CONFIG_FALLOCATE_PUNCH_HOLE
ret = fallocate(rb->fd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, ret = fallocate(rb->fd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
start, length); start, length);
#endif if (ret) {
} ret = -errno;
if (ret) { error_report("ram_block_discard_range: Failed to fallocate "
ret = -errno; "%s:%" PRIx64 " +%zx (%d)",
error_report("ram_block_discard_range: Failed to discard range " rb->idstr, start, length, ret);
goto err;
}
#else
ret = -ENOSYS;
error_report("ram_block_discard_range: fallocate not available/file"
"%s:%" PRIx64 " +%zx (%d)", "%s:%" PRIx64 " +%zx (%d)",
rb->idstr, start, length, ret); rb->idstr, start, length, ret);
goto err;
#endif
} }
if (need_madvise) {
/* For normal RAM this causes it to be unmapped,
* for shared memory it causes the local mapping to disappear
* and to fall back on the file contents (which we just
* fallocate'd away).
*/
#if defined(CONFIG_MADVISE)
ret = madvise(host_startaddr, length, MADV_DONTNEED);
if (ret) {
ret = -errno;
error_report("ram_block_discard_range: Failed to discard range "
"%s:%" PRIx64 " +%zx (%d)",
rb->idstr, start, length, ret);
goto err;
}
#else
ret = -ENOSYS;
error_report("ram_block_discard_range: MADVISE not available"
"%s:%" PRIx64 " +%zx (%d)",
rb->idstr, start, length, ret);
goto err;
#endif
}
trace_ram_block_discard_range(rb->idstr, host_startaddr, length,
need_madvise, need_fallocate, ret);
} else { } else {
error_report("ram_block_discard_range: Overrun block '%s' (%" PRIu64 error_report("ram_block_discard_range: Overrun block '%s' (%" PRIu64
"/%zx/" RAM_ADDR_FMT")", "/%zx/" RAM_ADDR_FMT")",

View File

@ -55,9 +55,10 @@ dma_complete(void *dbs, int ret, void *cb) "dbs=%p ret=%d cb=%p"
dma_blk_cb(void *dbs, int ret) "dbs=%p ret=%d" dma_blk_cb(void *dbs, int ret) "dbs=%p ret=%d"
dma_map_wait(void *dbs) "dbs=%p" dma_map_wait(void *dbs) "dbs=%p"
# # exec.c # exec.c
find_ram_offset(uint64_t size, uint64_t offset) "size: 0x%" PRIx64 " @ 0x%" PRIx64 find_ram_offset(uint64_t size, uint64_t offset) "size: 0x%" PRIx64 " @ 0x%" PRIx64
find_ram_offset_loop(uint64_t size, uint64_t candidate, uint64_t offset, uint64_t next, uint64_t mingap) "trying size: 0x%" PRIx64 " @ 0x%" PRIx64 ", offset: 0x%" PRIx64" next: 0x%" PRIx64 " mingap: 0x%" PRIx64 find_ram_offset_loop(uint64_t size, uint64_t candidate, uint64_t offset, uint64_t next, uint64_t mingap) "trying size: 0x%" PRIx64 " @ 0x%" PRIx64 ", offset: 0x%" PRIx64" next: 0x%" PRIx64 " mingap: 0x%" PRIx64
ram_block_discard_range(const char *rbname, void *hva, size_t length, bool need_madvise, bool need_fallocate, int ret) "%s@%p + 0x%zx: madvise: %d fallocate: %d ret: %d"
# memory.c # memory.c
memory_region_ops_read(int cpu_index, void *mr, uint64_t addr, uint64_t value, unsigned size) "cpu %d mr %p addr 0x%"PRIx64" value 0x%"PRIx64" size %u" memory_region_ops_read(int cpu_index, void *mr, uint64_t addr, uint64_t value, unsigned size) "cpu %d mr %p addr 0x%"PRIx64" value 0x%"PRIx64" size %u"