criu: Support C/R of pidfds

Process file descriptors (pidfds) were introduced to provide a stable
handle on a process. They solve the problem of pid recycling.

For a detailed explanation, see https://lwn.net/Articles/801319/ and
http://www.corsix.org/content/what-is-a-pidfd

Before Linux 6.9, anonymous inodes were used for the implementation of
pidfds. So, we detect them in a fashion similiar to other fd types that
use anonymous inodes by calling `readlink()`.
After 6.9, pidfs (a file system for pidfds) was introduced.
In 6.9 `S_ISREG()` returned true for pidfds, but this again changed with
6.10.
(https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/pidfs.c?h=v6.11-rc2#n285)
After this change, pidfs inodes have no file type in st_mode in
userspace.
We use `PID_FS_MAGIC` to detect pidfds for kernel >= 6.9
Hence, check for pidfds occurs before the check for regular files.

For pidfds that refer to dead processes, we lose the pid of the process
as the Pid and NSpid fields in /proc/<pid>/fdinfo/<pidfd> change to -1.
So, we create a temporary process for each unique inode and open pidfds
that refer to this process. After all pidfds have been opened we kill
this temporary process.

This commit does not include support for pidfds that point to a specific
thread, i.e pidfds opened with `PIDFD_THREAD` flag.

Fixes: checkpoint-restore#2258

Signed-off-by: Bhavik Sachdev <[email protected]>

criu/Makefile.crtools

@@ -101,6 +101,7 @@ obj-$(CONFIG_COMPAT)	+= vdso-compat.o
 CFLAGS_REMOVE_vdso-compat.o	+= $(CFLAGS-ASAN) $(CFLAGS-GCOV)
 obj-y			+= pidfd-store.o
 obj-y			+= hugetlb.o
+obj-y			+= pidfd.o
 
 PROTOBUF_GEN := scripts/protobuf-gen.sh
 

criu/cr-restore.c

@@ -79,6 +79,7 @@
 #include "timens.h"
 #include "bpfmap.h"
 #include "apparmor.h"
+#include "pidfd.h"
 
 #include "parasite-syscall.h"
 #include "files-reg.h"
@@ -280,7 +281,7 @@ static struct collect_image_info *cinfos_files[] = {
 	&unix_sk_cinfo,	      &fifo_cinfo,     &pipe_cinfo,    &nsfile_cinfo,	    &packet_sk_cinfo,
 	&netlink_sk_cinfo,    &eventfd_cinfo,  &epoll_cinfo,   &epoll_tfd_cinfo,    &signalfd_cinfo,
 	&tunfile_cinfo,	      &timerfd_cinfo,  &inotify_cinfo, &inotify_mark_cinfo, &fanotify_cinfo,
-	&fanotify_mark_cinfo, &ext_file_cinfo, &memfd_cinfo,
+	&fanotify_mark_cinfo, &ext_file_cinfo, &memfd_cinfo, &pidfd_cinfo
 };
 
 /* These images are required to restore namespaces */

criu/files.c

@@ -49,6 +49,7 @@
 #include "kerndat.h"
 #include "fdstore.h"
 #include "bpfmap.h"
+#include "pidfd.h"
 
 #include "protobuf.h"
 #include "util.h"
@@ -544,6 +545,8 @@ static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts,
 			ops = &signalfd_dump_ops;
 		else if (is_timerfd_link(link))
 			ops = &timerfd_dump_ops;
+		else if (is_pidfd_link(link))
+			ops = &pidfd_dump_ops;
 #ifdef CONFIG_HAS_LIBBPF
 		else if (is_bpfmap_link(link))
 			ops = &bpfmap_dump_ops;
@@ -554,6 +557,11 @@ static int dump_one_file(struct pid *pid, int fd, int lfd, struct fd_opts *opts,
 		return do_dump_gen_file(&p, lfd, ops, e);
 	}
 
+	if (p.fs_type == PID_FS_MAGIC) {
+		ops = &pidfd_dump_ops;
+		return do_dump_gen_file(&p, lfd, ops, e);
+	}
+
 	if (S_ISREG(p.stat.st_mode) || S_ISDIR(p.stat.st_mode) || S_ISLNK(p.stat.st_mode)) {
 		if (fill_fdlink(lfd, &p, &link))
 			return -1;
@@ -1778,6 +1786,9 @@ static int collect_one_file(void *o, ProtobufCMessage *base, struct cr_img *i)
 	case FD_TYPES__MEMFD:
 		ret = collect_one_file_entry(fe, fe->memfd->id, &fe->memfd->base, &memfd_cinfo);
 		break;
+	case FD_TYPES__PIDFD:
+		ret = collect_one_file_entry(fe, fe->pidfd->id, &fe->pidfd->base, &pidfd_cinfo);
+		break;
 #ifdef CONFIG_HAS_LIBBPF
 	case FD_TYPES__BPFMAP:
 		ret = collect_one_file_entry(fe, fe->bpf->id, &fe->bpf->base, &bpfmap_cinfo);
@@ -1800,5 +1811,11 @@ int prepare_files(void)
 {
 	init_fdesc_hash();
 	init_sk_info_hash();
+
+	if (init_dead_pidfd_hash()) {
+		pr_err("Could not initialise hash map for dead pidfds\n");
+		return -1;
+	}
+
 	return collect_image(&files_cinfo);
 }

criu/image-desc.c

@@ -107,6 +107,7 @@ struct cr_fd_desc_tmpl imgset_template[CR_FD_MAX] = {
 	FD_ENTRY_F(BPFMAP_FILE,	"bpfmap-file", O_NOBUF),
 	FD_ENTRY_F(BPFMAP_DATA,	"bpfmap-data", O_NOBUF),
 	FD_ENTRY(APPARMOR,	"apparmor"),
+	FD_ENTRY(PIDFD,		"pidfd"),
 
 	[CR_FD_STATS] = {
 		.fmt	= "stats-%s",

criu/include/fs-magic.h

@@ -57,4 +57,8 @@
 #define OVERLAYFS_SUPER_MAGIC 0x794c7630
 #endif
 
+#ifndef PID_FS_MAGIC
+#define PID_FS_MAGIC 0x50494446
+#endif
+
 #endif /* __CR_FS_MAGIC_H__ */

criu/include/image-desc.h

@@ -113,6 +113,7 @@ enum {
 	CR_FD_PIPES,
 	CR_FD_TTY_FILES,
 	CR_FD_MEMFD_FILE,
+	CR_FD_PIDFD,
 
 	CR_FD_AUTOFS,
 

criu/include/magic.h

@@ -100,6 +100,7 @@
 #define BPFMAP_FILE_MAGIC    0x57506142 /* Alapayevsk */
 #define BPFMAP_DATA_MAGIC    0x64324033 /* Arkhangelsk */
 #define APPARMOR_MAGIC	     0x59423047 /* Nikolskoye */
+#define PIDFD_MAGIC	     0x54435556 /* Ufa */
 
 #define IFADDR_MAGIC	RAW_IMAGE_MAGIC
 #define ROUTE_MAGIC	RAW_IMAGE_MAGIC

criu/include/pidfd.h

@@ -0,0 +1,16 @@
+#ifndef __CR_PIDFD_H__
+#define __CR_PIDFD_H__
+
+#include "files.h"
+#include "pidfd.pb-c.h"
+
+extern const struct fdtype_ops pidfd_dump_ops;
+extern struct collect_image_info pidfd_cinfo;
+extern int is_pidfd_link(char *link);
+extern int init_dead_pidfd_hash(void);
+struct pidfd_dump_info {
+	PidfdEntry pidfe;
+	pid_t pid;
+};
+
+#endif /* __CR_PIDFD_H__ */

criu/include/protobuf-desc.h

@@ -70,6 +70,7 @@ enum {
 	PB_BPFMAP_FILE,
 	PB_BPFMAP_DATA,
 	PB_APPARMOR,
+	PB_PIDFD,
 
 	/* PB_AUTOGEN_STOP */