plugin/amdgpu: Support for open dmabuf handles

criu/cr-dump.c

@@ -2225,6 +2225,9 @@ int cr_dump_tasks(pid_t pid)
 			goto err;
 	}
 
+	if(run_plugins(DUMP_DEVICE_LATE, pid))
+		goto err;
+
 	if (parent_ie) {
 		inventory_entry__free_unpacked(parent_ie, NULL);
 		parent_ie = NULL;

criu/cr-restore.c

@@ -1651,6 +1651,9 @@ static int __restore_task_with_children(void *_arg)
 	if (open_transport_socket())
 		goto err;
 
+	if (run_plugins(RESUME_DEVICES_EARLY, current->pid->real))
+		goto err;
+
 	timing_start(TIME_FORK);
 
 	if (create_children_and_session())

criu/files-ext.c

@@ -45,10 +45,11 @@ static int open_fd(struct file_desc *d, int *new_fd)
 {
 	struct ext_file_info *xfi;
 	int fd;
+	bool retry_needed;
 
 	xfi = container_of(d, struct ext_file_info, d);
 
-	fd = run_plugins(RESTORE_EXT_FILE, xfi->xfe->id);
+	fd = run_plugins(RESTORE_EXT_FILE, xfi->xfe->id, &retry_needed);
 	if (fd < 0) {
 		pr_err("Unable to restore %#x\n", xfi->xfe->id);
 		return -1;
@@ -57,8 +58,11 @@ static int open_fd(struct file_desc *d, int *new_fd)
 	if (restore_fown(fd, xfi->xfe->fown))
 		return -1;
 
-	*new_fd = fd;
-	return 0;
+	if (!retry_needed)
+		*new_fd = fd;
+	else
+		*new_fd = -1;
+	return retry_needed;
 }
 
 static struct file_desc_ops ext_desc_ops = {
@@ -83,14 +87,14 @@ struct collect_image_info ext_file_cinfo = {
 	.collect = collect_one_ext,
 };
 
-int dump_unsupp_fd(struct fd_parms *p, int lfd, char *more, char *info, FdinfoEntry *e)
+int dump_unsupp_fd(struct fd_parms *p, int lfd, char *more, char *info, FdinfoEntry *e, bool force)
 {
 	int ret;
 
-	ret = do_dump_gen_file(p, lfd, &ext_dump_ops, e);
+	ret = do_dump_gen_file(p, lfd, &ext_dump_ops, e, force);
 	if (ret == 0)
 		return 0;
 	if (ret == -ENOTSUP)
 		pr_err("Can't dump file %d of that type [%o] (%s %s)\n", p->fd, p->stat.st_mode, more, info);
-	return -1;
+	return ret;
 }

criu/files.c

@@ -329,7 +329,9 @@ uint32_t make_gen_id(uint32_t st_dev, uint32_t st_ino, uint64_t pos)
 	return st_dev ^ st_ino ^ pos_hi ^ pos_low;
 }
 
-int do_dump_gen_file(struct fd_parms *p, int lfd, const struct fdtype_ops *ops, FdinfoEntry *e)
+/* Use "force" to override cache */
+int do_dump_gen_file(struct fd_parms *p, int lfd,
+					const struct fdtype_ops *ops, FdinfoEntry *e, bool force)
 {
 	int ret = -1;
 
@@ -339,7 +341,7 @@ int do_dump_gen_file(struct fd_parms *p, int lfd, const struct fdtype_ops *ops,
 	e->flags = p->fd_flags;
 
 	ret = fd_id_generate(p->pid, e, p);
-	if (ret == 1) /* new ID generated */
+	if (ret == 1 || force) /* new ID generated */
 		ret = ops->dump(lfd, e->id, p);
 	else
 		/* Remove locks generated by the fd before going to the next */
@@ -484,19 +486,19 @@ static int dump_chrdev(struct fd_parms *p, int lfd, FdinfoEntry *e)
 		}
 
 		sprintf(more, "%d:%d", maj, minor(p->stat.st_rdev));
-		err = dump_unsupp_fd(p, lfd, "chr", more, e);
+		err = dump_unsupp_fd(p, lfd, "chr", more, e, false);
 		p->link = link_old;
 		return err;
 	}
 	}
 
-	err = do_dump_gen_file(p, lfd, ops, e);
+	err = do_dump_gen_file(p, lfd, ops, e, false);
 	p->link = link_old;
 	return err;
 }
 
 static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts, struct parasite_ctl *ctl,
-			 FdinfoEntry *e, struct parasite_drain_fd *dfds)
+			 FdinfoEntry *e, struct parasite_drain_fd *dfds, bool force)
 {
 	struct fd_parms p = FD_PARMS_INIT;
 	const struct fdtype_ops *ops;
@@ -552,14 +554,14 @@ static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts,
 			ops = &bpfmap_dump_ops;
 #endif
 		else
-			return dump_unsupp_fd(&p, lfd, "anon", link, e);
+			return dump_unsupp_fd(&p, lfd, "anon", link, e, force);
 
-		return do_dump_gen_file(&p, lfd, ops, e);
+		return do_dump_gen_file(&p, lfd, ops, e, force);
 	}
 
 	if (p.fs_type == PID_FS_MAGIC) {
 		ops = &pidfd_dump_ops;
-		return do_dump_gen_file(&p, lfd, ops, e);
+		return do_dump_gen_file(&p, lfd, ops, e, force);
 	}
 
 	if (S_ISREG(p.stat.st_mode) || S_ISDIR(p.stat.st_mode) || S_ISLNK(p.stat.st_mode)) {
@@ -576,9 +578,9 @@ static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts,
 		else if (check_ns_proc(&link))
 			ops = &nsfile_dump_ops;
 		else
-			return dump_unsupp_fd(&p, lfd, "reg", link.name + 1, e);
+			return dump_unsupp_fd(&p, lfd, "reg", link.name + 1, e, force);
 
-		return do_dump_gen_file(&p, lfd, ops, e);
+		return do_dump_gen_file(&p, lfd, ops, e, force);
 	}
 
 	if (S_ISFIFO(p.stat.st_mode)) {
@@ -587,7 +589,7 @@ static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts,
 		else
 			ops = &fifo_dump_ops;
 
-		return do_dump_gen_file(&p, lfd, ops, e);
+		return do_dump_gen_file(&p, lfd, ops, e, force);
 	}
 
 	/*
@@ -598,7 +600,7 @@ static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts,
 	if (fill_fdlink(lfd, &p, &link))
 		memzero(&link, sizeof(link));
 
-	return dump_unsupp_fd(&p, lfd, "unknown", link.name + 1, e);
+	return dump_unsupp_fd(&p, lfd, "unknown", link.name + 1, e, force);
 }
 
 int dump_my_file(int lfd, u32 *id, int *type)
@@ -610,7 +612,7 @@ int dump_my_file(int lfd, u32 *id, int *type)
 	me.real = getpid();
 	me.ns[0].virt = -1; /* FIXME */
 
-	if (dump_one_file(&me, lfd, lfd, &fdo, NULL, &e, NULL))
+	if (dump_one_file(&me, lfd, lfd, &fdo, NULL, &e, NULL, false))
 		return -1;
 
 	*id = e.id;
@@ -625,6 +627,8 @@ int dump_task_files_seized(struct parasite_ctl *ctl, struct pstree_item *item, s
 	struct fd_opts *opts = NULL;
 	int i, ret = -1;
 	int off, nr_fds = min((int)PARASITE_MAX_FDS, dfds->nr_fds);
+	int *retry_indices = NULL;
+	int retry_count = 0;
 
 	pr_info("\n");
 	pr_info("Dumping opened files (pid: %d)\n", item->pid->real);
@@ -642,6 +646,10 @@ int dump_task_files_seized(struct parasite_ctl *ctl, struct pstree_item *item, s
 	if (!img)
 		goto err;
 
+	retry_indices = xmalloc(dfds->nr_fds * sizeof(int)); /* Allocate memory for retry indices*/
+	if (!retry_indices)
+		goto err;
+
 	ret = 0; /* Don't fail if nr_fds == 0 */
 	for (off = 0; ret == 0 && off < dfds->nr_fds; off += nr_fds) {
 		if (nr_fds + off > dfds->nr_fds)
@@ -654,9 +662,30 @@ int dump_task_files_seized(struct parasite_ctl *ctl, struct pstree_item *item, s
 		for (i = 0; i < nr_fds; i++) {
 			FdinfoEntry e = FDINFO_ENTRY__INIT;
 
-			ret = dump_one_file(item->pid, dfds->fds[i + off], lfds[i], opts + i, ctl, &e, dfds);
+			ret = dump_one_file(item->pid, dfds->fds[i + off], lfds[i], opts + i, ctl, &e, dfds, false);
+			if (ret == -EAGAIN) {
+				retry_indices[retry_count++] = i;
+				ret = 0; //* Reset ret to continue the loop */
+				continue;
+			} else if (ret)
+				break;
+
+			ret = pb_write_one(img, &e, PB_FDINFO);
 			if (ret)
 				break;
+		}
+		/* Some handles like dmabuf handles should be dealt with after
+		drm handles. Do that here, if necessary. */
+		for (i = 0; i < retry_count; i++) {
+			int idx = retry_indices[i];
+			FdinfoEntry e = FDINFO_ENTRY__INIT;
+
+			ret = dump_one_file(item->pid, dfds->fds[idx + off], lfds[idx],
+											opts + idx, ctl, &e, dfds, true);
+			if (ret) {
+				pr_err("Retry failed for fd index %d\n", idx);
+				break;
+			}
 
 			ret = pb_write_one(img, &e, PB_FDINFO);
 			if (ret)
@@ -673,6 +702,8 @@ int dump_task_files_seized(struct parasite_ctl *ctl, struct pstree_item *item, s
 		close_image(img);
 	xfree(opts);
 	xfree(lfds);
+	xfree(retry_indices);
+
 	return ret;
 }
 
@@ -836,6 +867,9 @@ struct fdinfo_list_entry *collect_fd_to(int pid, FdinfoEntry *e, struct rst_info
 {
 	struct fdinfo_list_entry *new_le;
 
+	if (fdesc->ops->type == FD_TYPES__EXT)
+		run_plugins(COLLECT_FILE, pid, fdesc->id);
+
 	new_le = alloc_fle(pid, e);
 	if (new_le) {
 		new_le->fake = (!!fake);

criu/include/criu-plugin.h

@@ -60,6 +60,12 @@ enum {
 
 	CR_PLUGIN_HOOK__CHECKPOINT_DEVICES = 11,
 
+	CR_PLUGIN_HOOK__RESUME_DEVICES_EARLY = 12,
+
+	CR_PLUGIN_HOOK__COLLECT_FILE = 13,
+
+	CR_PLUGIN_HOOK__DUMP_DEVICE_LATE = 14,
+
 	CR_PLUGIN_HOOK__MAX
 };
 
@@ -68,7 +74,7 @@ enum {
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__DUMP_UNIX_SK, int fd, int id);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__RESTORE_UNIX_SK, int id);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__DUMP_EXT_FILE, int fd, int id);
-DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__RESTORE_EXT_FILE, int id);
+DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__RESTORE_EXT_FILE, int id, bool *retry_needed);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__DUMP_EXT_MOUNT, char *mountpoint, int id);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__RESTORE_EXT_MOUNT, int id, char *mountpoint, char *old_root, int *is_file);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__DUMP_EXT_LINK, int index, int type, char *kind);
@@ -78,6 +84,9 @@ DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__UPDATE_VMA_MAP, const char *path, const
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__RESUME_DEVICES_LATE, int pid);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__PAUSE_DEVICES, int pid);
 DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__CHECKPOINT_DEVICES, int pid);
+DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__RESUME_DEVICES_EARLY, int pid);
+DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__COLLECT_FILE, int pid, int fd);
+DECLARE_PLUGIN_HOOK_ARGS(CR_PLUGIN_HOOK__DUMP_DEVICE_LATE, int id);
 
 enum {
 	CR_PLUGIN_STAGE__DUMP,

criu/include/files.h

@@ -138,7 +138,8 @@ struct fdtype_ops {
 struct cr_img;
 
 extern int dump_my_file(int lfd, u32 *, int *type);
-extern int do_dump_gen_file(struct fd_parms *p, int lfd, const struct fdtype_ops *ops, FdinfoEntry *e);
+extern int do_dump_gen_file(struct fd_parms *p, int lfd,
+					const struct fdtype_ops *ops, FdinfoEntry *e, bool force);
 struct parasite_drain_fd;
 int dump_task_files_seized(struct parasite_ctl *ctl, struct pstree_item *item, struct parasite_drain_fd *dfds);
 int predump_task_files(int pid);
@@ -177,7 +178,8 @@ extern int close_old_fds(void);
 extern int shared_fdt_prepare(struct pstree_item *item);
 
 extern struct collect_image_info ext_file_cinfo;
-extern int dump_unsupp_fd(struct fd_parms *p, int lfd, char *more, char *info, FdinfoEntry *);
+extern int dump_unsupp_fd(struct fd_parms *p, int lfd, char *more,
+								char *info, FdinfoEntry *, bool force);
 
 extern int inherit_fd_parse(char *optarg);
 extern int inherit_fd_add(int fd, char *key);

criu/include/servicefd.h

@@ -47,5 +47,6 @@ extern int install_service_fd(enum sfd_type type, int fd);
 extern int close_service_fd(enum sfd_type type);
 extern void __close_service_fd(enum sfd_type type);
 extern int clone_service_fd(struct pstree_item *me);
+extern int get_unused_high_fd(void);
 
 #endif /* __CR_SERVICE_FD_H__ */

criu/pie/restorer.c

@@ -1920,6 +1920,10 @@ __visible long __export_restore_task(struct task_restore_args *args)
 
 		for (m = 0; m < sizeof(vma_entry->madv) * 8; m++) {
 			if (vma_entry->madv & (1ul << m)) {
+
+				if (!(vma_entry_is(vma_entry, VMA_AREA_REGULAR)))
+					continue;
+
 				ret = sys_madvise(vma_entry->start, vma_entry_len(vma_entry), m);
 				if (ret) {
 					pr_err("madvise(%" PRIx64 ", %" PRIu64 ", %ld) "

criu/plugin.c

@@ -59,6 +59,9 @@ static cr_plugin_desc_t *cr_gen_plugin_desc(void *h, char *path)
 	__assign_hook(RESUME_DEVICES_LATE, "cr_plugin_resume_devices_late");
 	__assign_hook(PAUSE_DEVICES, "cr_plugin_pause_devices");
 	__assign_hook(CHECKPOINT_DEVICES, "cr_plugin_checkpoint_devices");
+	__assign_hook(RESUME_DEVICES_EARLY, "cr_plugin_resume_devices_early");
+	__assign_hook(COLLECT_FILE, "cr_plugin_collect_file");
+	__assign_hook(DUMP_DEVICE_LATE, "cr_plugin_dump_device_late");
 
 #undef __assign_hook
 

criu/servicefd.c

@@ -25,6 +25,7 @@ int service_fd_rlim_cur;
 
 /* Base of current process service fds set */
 static int service_fd_base;
+static int next_high_fd;
 
 /* Id of current process in shared fdt */
 static int service_fd_id = 0;
@@ -312,5 +313,15 @@ int clone_service_fd(struct pstree_item *me)
 	service_fd_id = id;
 	ret = 0;
 
+	next_high_fd = service_fd_base + 1024;
+
 	return ret;
 }
+
+int get_unused_high_fd(void)
+{
+	if (next_high_fd > service_fd_rlim_cur)
+		return -1;
+	next_high_fd += 1;
+	return next_high_fd - 1;
+}

criu/sockets.c

@@ -754,7 +754,7 @@ int dump_socket(struct fd_parms *p, int lfd, FdinfoEntry *e)
 		return -1;
 	}
 
-	return do_dump_gen_file(p, lfd, ops, e);
+	return do_dump_gen_file(p, lfd, ops, e, false);
 }
 
 static int inet_receive_one(struct nlmsghdr *h, struct ns_id *ns, void *arg)

plugins/amdgpu/Makefile

@@ -27,7 +27,7 @@ endif
 criu-amdgpu.pb-c.c: criu-amdgpu.proto
 		protoc-c --proto_path=. --c_out=. criu-amdgpu.proto
 
-amdgpu_plugin.so: amdgpu_plugin.c amdgpu_plugin_drm.c amdgpu_plugin_topology.c amdgpu_plugin_util.c criu-amdgpu.pb-c.c
+amdgpu_plugin.so: amdgpu_plugin.c amdgpu_plugin_drm.c amdgpu_plugin_dmabuf.c amdgpu_plugin_topology.c amdgpu_plugin_util.c criu-amdgpu.pb-c.c
 	$(CC) $(PLUGIN_CFLAGS) $(shell $(COMPEL) includes) $^ -o $@ $(PLUGIN_INCLUDE) $(PLUGIN_LDFLAGS) $(LIBDRM_INC)
 
 amdgpu_plugin_clean: