fork: extend clone3() to support setting a PID
Subsystems (6)
PATCH v9 1/2 fork: extend clone3() to support setting a PID
The main motivation to add set_tid to clone3() is CRIU.
To restore a process with the same PID/TID CRIU currently uses /proc/sys/kernel/ns_last_pid. It writes the desired (PID - 1) to ns_last_pid and then (quickly) does a clone(). This works most of the time, but it is racy. It is also slow as it requires multiple syscalls.
Extending clone3() to support *set_tid makes it possible restore a process using CRIU without accessing /proc/sys/kernel/ns_last_pid and race free (as long as the desired PID/TID is available).
This clone3() extension places the same restrictions (CAP_SYS_ADMIN) on clone3() with *set_tid as they are currently in place for ns_last_pid.
The original version of this change was using a single value for set_tid. At the 2019 LPC, after presenting set_tid, it was, however, decided to change set_tid to an array to enable setting the PID of a process in multiple PID namespaces at the same time. If a process is created in a PID namespace it is possible to influence the PID inside and outside of the PID namespace. Details also in the corresponding selftest.
To create a process with the following PIDs:
PID NS level Requested PID
0 (host) 31496
1 42
2 1
For that example the two newly introduced parameters to struct clone_args (set_tid and set_tid_size) would need to be:
set_tid[0] = 1; set_tid[1] = 42; set_tid[2] = 31496; set_tid_size = 3;
If only the PIDs of the two innermost nested PID namespaces should be defined it would look like this:
set_tid[0] = 1; set_tid[1] = 42; set_tid_size = 2;
The PID of the newly created process would then be the next available free PID in the PID namespace level 0 (host) and 42 in the PID namespace at level 1 and the PID of the process in the innermost PID namespace would be 1.
The set_tid array is used to specify the PID of a process starting from the innermost nested PID namespaces up to set_tid_size PID namespaces.
set_tid_size cannot be larger then the current PID namespace level.
---
v2:
- Removed (size < sizeof(struct clone_args)) as discussed with Christian and Dmitry
- Added comment to ((set_tid != 1) && idr_get_cursor() <= 1) (Oleg)
- Use idr_alloc() instead of idr_alloc_cyclic() (Oleg)
v3:
- Return EEXIST if PID is already in use (Christian)
- Drop CLONE_SET_TID (Christian and Oleg)
- Use idr_is_empty() instead of idr_get_cursor() (Oleg)
- Handle different struct clone_args sizes (Dmitry)
v4:
- Rework struct size check with defines (Christian)
- Reduce number of set_tid checks (Oleg)
- Less parentheses and more robust code (Oleg)
- Do ns_capable() on correct user_ns (Oleg, Christian)
v5:
- make set_tid checks earlier in alloc_pid() (Christian)
- remove unnecessary comment and struct size check (Christian)
v6:
- remove CLONE_SET_TID from description (Christian)
- add clone3() tests for different clone_args sizes (Christian)
- move more set_tid checks to alloc_pid() (Oleg)
- make all set_tid checks lockless (Oleg)
v7:
- changed set_tid to be an array to set the PID of a process in multiple nested PID namespaces at the same time as discussed at LPC 2019 (container MC)
v8:
- skip unnecessary memset() (Rasmus)
- replace set_tid copy loop with a single copy (Christian)
- more parameter error checking (Christian)
- cache set_tid in alloc_pid() (Oleg)
- move code in "else" branch (Oleg)
v9:
- added kernel-doc to include/uapi/linux/sched.h (Christian)
- moved a variable to limit its scope; keep all set_tid_size related changes in one place (Oleg)
--- include/linux/pid.h | 3 +- include/linux/pid_namespace.h | 2 + include/linux/sched/task.h | 3 ++ include/uapi/linux/sched.h | 52 +++++++++++++++++--------- kernel/fork.c | 24 +++++++++++- kernel/pid.c | 69 +++++++++++++++++++++++++++-------- kernel/pid_namespace.c | 2 - 7 files changed, 118 insertions(+), 37 deletions(-)
include/linux/pid.h +2 -1
@@ -124,7 +124,8 @@ extern struct pid *find_vpid(int nr);extern struct pid *find_get_pid (int nr);extern struct pid *find_ge_pid (int nr,struct pid_namespace *); -extern struct pid *alloc_pid (struct pid_namespace *ns); +extern struct pid *alloc_pid (struct pid_namespace *ns,pid_t *set_tid, +size_t set_tid_size);extern void free_pid (struct pid *pid);extern void disable_pid_allocation (struct pid_namespace *ns);
include/linux/pid_namespace.h +2 -0
@@ -12,6 +12,8 @@#include <linux/ns_common.h> #include <linux/idr.h> +/* MAX_PID_NS_LEVEL is needed for limiting size of 'struct pid' */ +#define MAX_PID_NS_LEVEL 32struct fs_pin ;
include/linux/sched/task.h +3 -0
@@ -26,6 +26,9 @@ struct kernel_clone_args {unsigned long stack;unsigned long stack_size;unsigned long tls; +pid_t *set_tid; +/* Number of elements in *set_tid */ +size_t set_tid_size; };/*
include/uapi/linux/sched.h +34 -18
64 lines not on this page · raw message
kernel/fork.c +23 -1
@@ -2087,7 +2087,8 @@ static __latent_entropy struct task_struct *copy_process(stackleak_task_init (p);if (pid != &init_struct_pid) { - pid =alloc_pid (p->nsproxy->pid_ns_for_children); + pid =alloc_pid (p->nsproxy->pid_ns_for_children, args->set_tid, + args->set_tid_size);if (IS_ERR (pid)) { retval =PTR_ERR (pid); goto bad_fork_cleanup_thread;
@@ -2590,6 +2591,7 @@ noinline static int copy_clone_args_from_user(struct kernel_clone_args *kargs, {int err;struct clone_args args; +pid_t *kset_tid = kargs->set_tid;if (unlikely (usize >PAGE_SIZE ))return -E2BIG ;
@@ -2600,6 +2602,15 @@ noinline static int copy_clone_args_from_user(struct kernel_clone_args *kargs,if (err)return err; +if (unlikely (args.set_tid_size >MAX_PID_NS_LEVEL )) +return -EINVAL ; + +if (unlikely (!args.set_tid && args.set_tid_size >0 )) +return -EINVAL ; + +if (unlikely (args.set_tid && args.set_tid_size ==0 )) +return -EINVAL ; +/* * Verify that higher 32bits of exit_signal are unset and that * it is a valid signal
@@ -2617,8 +2628,16 @@ noinline static int copy_clone_args_from_user(struct kernel_clone_args *kargs, .stack = args.stack, .stack_size = args.stack_size, .tls = args.tls, + .set_tid_size = args.set_tid_size, }; +if (args.set_tid && +copy_from_user (kset_tid,u64_to_user_ptr (args.set_tid), + (kargs->set_tid_size *sizeof (pid_t)))) +return -EFAULT ; + + kargs->set_tid = kset_tid; +return 0 ; }
@@ -2662,6 +2681,9 @@ SYSCALL_DEFINE2(clone3, struct clone_args __user *, uargs, size_t, size)int err;struct kernel_clone_args kargs; +pid_t set_tid[MAX_PID_NS_LEVEL ]; + + kargs.set_tid = set_tid; err =copy_clone_args_from_user (&kargs, uargs, size);if (err)
kernel/pid.c +54 -15
91 lines not on this page · raw message
kernel/pid_namespace.c +0 -2
@@ -26,8 +26,6 @@static DEFINE_MUTEX (pid_caches_mutex);static struct kmem_cache *pid_ns_cachep; -/* MAX_PID_NS_LEVEL is needed for limiting size of 'struct pid' */ -#define MAX_PID_NS_LEVEL 32/* Write once array, filled from the beginning. */ static struct kmem_cache *pid_cache[MAX_PID_NS_LEVEL ];
base-commit: 7acdfe534e726450fe8051abc2f36380fb2c2c0e
Signature · 1 line
2.23.0
PATCH v9 2/2 selftests: add tests for clone3() with *set_tid
This tests clone3() with *set_tid to see if all desired PIDs are working as expected. The tests are trying multiple invalid input parameters as well as creating processes while specifying a certain PID in multiple PID namespaces at the same time.
Additionally this moves common clone3() test code into clone3_selftests.h.
---
v9:
- applied all changes from Christian's review (except using the
NSpid: parsing code from selftests/pidfd/pidfd_fdinfo_test.c)
--- tools/testing/selftests/clone3/.gitignore | 1 + tools/testing/selftests/clone3/Makefile | 2 +- tools/testing/selftests/clone3/clone3.c | 8 +- .../selftests/clone3/clone3_clear_sighand.c | 20 +- .../selftests/clone3/clone3_selftests.h | 29 ++ .../testing/selftests/clone3/clone3_set_tid.c | 355 ++++++++++++++++++ 6 files changed, 389 insertions(+), 26 deletions(-) create mode 100644 tools/testing/selftests/clone3/clone3_selftests.h create mode 100644 tools/testing/selftests/clone3/clone3_set_tid.c
tools/testing/selftests/clone3/.gitignore +1 -0
@@ -1,2 +1,3 @@ clone3 clone3_clear_sighand +clone3_set_tid
tools/testing/selftests/clone3/Makefile +1 -1
@@ -1,6 +1,6 @@ # SPDX-License-Identifier: GPL-2.0 CFLAGS += -g -I../../../../usr/include/ -TEST_GEN_PROGS := clone3 clone3_clear_sighand +TEST_GEN_PROGS := clone3 clone3_clear_sighand clone3_set_tid include ../lib.mk
tools/testing/selftests/clone3/clone3.c +2 -6
@@ -18,6 +18,7 @@#include <sched.h> #include "../kselftest.h" +#include "clone3_selftests.h" /* * Different sizes of struct clone_args
@@ -35,11 +36,6 @@ enum test_mode { CLONE3_ARGS_INVAL_EXIT_SIGNAL_NSIG, }; -static pid_t raw_clone (struct clone_args *args,size_t size) -{ -return syscall (__NR_clone3, args, size); -} -static int call_clone3 (uint64_t flags,size_t size,enum test_mode test_mode) {struct clone_args args = {
@@ -83,7 +79,7 @@ static int call_clone3(uint64_t flags, size_t size, enum test_mode test_mode)memcpy (&args_ext.args, &args,sizeof (struct clone_args )); - pid =raw_clone ((struct clone_args *)&args_ext, size); + pid =sys_clone3 ((struct clone_args *)&args_ext, size);if (pid <0 ) {ksft_print_msg ("%s - Failed to create new process\n" ,strerror (errno));
tools/testing/selftests/clone3/clone3_clear_sighand.c +1 -19
@@ -14,30 +14,12 @@#include <sys/wait.h> #include "../kselftest.h" +#include "clone3_selftests.h" #ifndef CLONE_CLEAR_SIGHAND #define CLONE_CLEAR_SIGHAND 0x100000000ULL#endif -#ifndef __NR_clone3 -#define __NR_clone3 -1 -struct clone_args { -__aligned_u64 flags; -__aligned_u64 pidfd; -__aligned_u64 child_tid; -__aligned_u64 parent_tid; -__aligned_u64 exit_signal; -__aligned_u64 stack; -__aligned_u64 stack_size; -__aligned_u64 tls; -}; -#endif - -static pid_t sys_clone3 (struct clone_args *args,size_t size) -{ -return syscall (__NR_clone3, args, size); -} -static void test_clone3_supported (void ) {pid_t pid;
tools/testing/selftests/clone3/clone3_selftests.h new file +29 -0
@@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: GPL-2.0 */ + +#ifndef _CLONE3_SELFTESTS_H +#define _CLONE3_SELFTESTS_H + +#include <stdint.h> + +#define ptr_to_u64 (ptr) ((__u64)((uintptr_t)(ptr))) + +#ifndef __NR_clone3 +#define __NR_clone3 -1 +struct clone_args { +__aligned_u64 flags; +__aligned_u64 pidfd; +__aligned_u64 child_tid; +__aligned_u64 parent_tid; +__aligned_u64 exit_signal; +__aligned_u64 stack; +__aligned_u64 stack_size; +__aligned_u64 tls; +}; +#endif + +static pid_t sys_clone3 (struct clone_args *args,size_t size) +{ +return syscall (__NR_clone3, args, size); +} + +#endif /* _CLONE3_SELFTESTS_H */
tools/testing/selftests/clone3/clone3_set_tid.c new file +355 -0
355 lines not on this page · raw message
Signature · 1 line
2.23.0
Re: [PATCH v9 2/2] selftests: add tests for clone3() with *set_tid
Quoted 5 lines · Adrian Reber
+++
b/tools/testing/selftests/clone3/clone3_selftests.htools/testing/selftests/clone3/clone3_selftests.h new file +3
-0@@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: GPL-2.0 */ + +#ifndef _CLONE3_SELFTESTS_H
tools/testing/selftests/clone3/clone3_selftests.h line 13
Quoted 10 lines
tools/testing/selftests/clone3/clone3_selftests.h new file +10
-0+#define _CLONE3_SELFTESTS_H + +#include <stdint.h> + +#define ptr_to_u64 (ptr) ((__u64)((uintptr_t)(ptr))) + +#ifndef __NR_clone3 +#define __NR_clone3 -1 +struct clone_args { + __aligned_u64 flags;
Hugh, doesn't that complain about missing definition of __aligned_u64? I thought that was declared in linux/types.h
tools/testing/selftests/clone3/clone3_selftests.h line 26
Quoted 13 lines
tools/testing/selftests/clone3/clone3_selftests.h new file +13
-0+__aligned_u64 pidfd; +__aligned_u64 child_tid; +__aligned_u64 parent_tid; +__aligned_u64 exit_signal; +__aligned_u64 stack; +__aligned_u64 stack_size; + __aligned_u64 tls; +}; +#endif + +static pid_t sys_clone3 (struct clone_args *args,size_t size) +{ +return syscall (__NR_clone3, args, size);
And doesn't that one complain about missing syscall() defined in sys/syscall.h?
I would rather make the header sm like:
#define _GNU_SOURCE
#include <sched.h>
#include <stdint.h>
#include <syscall.h>
#include <linux/types.h>
so that the order of inclusion doesn't matter. You're getting away with this because you include syscall.h, etc. before clone3_selftests.h in clone3_set_tid.c below. I'd rather not rely on such ordering.
/* stuff */
tools/testing/selftests/clone3/clone3_set_tid.c line 87
Quoted 12 lines
tools/testing/selftests/clone3/clone3_selftests.h new file +3
-0+} + +#endif /* _CLONE3_SELFTESTS_H */ tools/testing/selftests/clone3/clone3_set_tid.c +8
-0+if (expected_pid ==0 || expected_pid == pid) { +ksft_print_msg ("I am the parent (%d). My child's pid is %d\n" , +getpid (), pid); + }else { +ksft_print_msg ( +"Expected child pid %d does not match actual pid %d\n" , + expected_pid, pid); + ret =-1 ;
What's that for?
I think that whole ret variable can go, no?
Quoted 42 lines
tools/testing/selftests/clone3/clone3_set_tid.c +42
-0+ } + +if (wait (&status) <0 ) { +ksft_print_msg ("Child returned %s\n" ,strerror (errno)); +return -errno; + } + +if (!WIFEXITED (status)) +return -1 ; + +return WEXITSTATUS (status); +} + +static void test_clone3_set_tid (pid_t *set_tid, +size_t set_tid_size, +int flags, +int expected, +int expected_pid, +bool wait_for_it) +{ +int ret; + +ksft_print_msg ( +"[%d] Trying clone3() with CLONE_SET_TID to %d and 0x%x\n" , +getpid (), set_tid[0 ], flags); + ret =call_clone3_set_tid (set_tid, set_tid_size, flags, expected_pid, + wait_for_it); +ksft_print_msg ( +"[%d] clone3() with CLONE_SET_TID %d says :%d - expected %d\n" , +getpid (), set_tid[0 ], ret, expected); +if (ret != expected) +ksft_test_result_fail ( +"[%d] Result (%d) is different than expected (%d)\n" , +getpid (), ret, expected); +else +ksft_test_result_pass ( +"[%d] Result (%d) matches expectation (%d)\n" , +getpid (), ret, expected); +} +int main (int argc,char *argv[]) +{ +FILE *f;
tools/testing/selftests/clone3/clone3_set_tid.c line 139
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+char buf; +char *line; +int status; +int ret =-1 ; +size_t len =0 ; +int pid_max =0 ; +uid_t uid =getuid (); +char proc_path[100 ] = {0 }; +pid_t pid, ns1, ns2, ns3, ns_pid; +pid_t set_tid[MAX_PID_NS_LEVEL *2 ];
reverse-reverse Christmas tree ordering I see :)
Quoted 14 lines
tools/testing/selftests/clone3/clone3_set_tid.c +14
-0+ +if (pipe (pipe_1) <0 ||pipe (pipe_2) <0 ) +ksft_exit_fail_msg ("pipe() failed\n" ); + +ksft_print_header (); +ksft_set_plan (26 ); + + f =fopen ("/proc/sys/kernel/pid_max" ,"r" ); +if (f ==NULL ) +ksft_exit_fail_msg ( +"%s - Could not open /proc/sys/kernel/pid_max\n" , +strerror (errno)); +fscanf (f,"%d" , &pid_max); +fclose (f);
tools/testing/selftests/clone3/clone3_set_tid.c line 163
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+ksft_print_msg ("/proc/sys/kernel/pid_max %d\n" , pid_max); + +/* Try invalid settings */ +memset (&set_tid,0 ,sizeof (set_tid)); +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL +1 ,0 , -EINVAL ,0 ,0 ); + +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL *2 ,0 , -EINVAL ,0 ,0 ); + +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL *2 +1 ,0 , + -EINVAL ,0 ,0 );
\n
Quoted 5 lines
tools/testing/selftests/clone3/clone3_set_tid.c +5
-0+test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL *42 ,0 , -EINVAL ,0 ,0 ); + +/* +* This can actually work if this test running in a MAX_PID_NS_LEVEL - 1 +* nested PID namespace.
tools/testing/selftests/clone3/clone3_set_tid.c line 178
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+*/ +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL -1 ,0 , -EINVAL ,0 ,0 ); + +memset (&set_tid,0xff ,sizeof (set_tid)); +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL +1 ,0 , -EINVAL ,0 ,0 ); + +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL *2 ,0 , -EINVAL ,0 ,0 ); + +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL *2 +1 ,0 , + -EINVAL ,0 ,0 );
\n
tools/testing/selftests/clone3/clone3_set_tid.c
Quoted 1 line
tools/testing/selftests/clone3/clone3_set_tid.c +1
-0+test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL *42 ,0 , -EINVAL ,0 ,0 );
\n
tools/testing/selftests/clone3/clone3_set_tid.c line 189
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+/* +* This can actually work if this test running in a MAX_PID_NS_LEVEL - 1 +* nested PID namespace. +*/ +test_clone3_set_tid (set_tid,MAX_PID_NS_LEVEL -1 ,0 , -EINVAL ,0 ,0 ); + +memset (&set_tid,0 ,sizeof (set_tid)); +/* Try with an invalid PID */ + set_tid[0 ] =0 ; +test_clone3_set_tid (set_tid,1 ,0 , -EINVAL ,0 ,0 );
\n
tools/testing/selftests/clone3/clone3_set_tid.c
Quoted 2 lines
tools/testing/selftests/clone3/clone3_set_tid.c +2
-0+ set_tid[0 ] =-1 ; +test_clone3_set_tid (set_tid,1 ,0 , -EINVAL ,0 ,0 );
\n
tools/testing/selftests/clone3/clone3_set_tid.c line 193
Quoted 2 lines
tools/testing/selftests/clone3/clone3_set_tid.c +2
-0+/* Claim that the set_tid array actually contains 2 elements. */ +test_clone3_set_tid (set_tid,2 ,0 , -EINVAL ,0 ,0 );
\n
tools/testing/selftests/clone3/clone3_set_tid.c line 201
Quoted 8 lines
tools/testing/selftests/clone3/clone3_set_tid.c +8
-0+/* Try it in a new PID namespace */ +if (uid ==0 ) +test_clone3_set_tid (set_tid,1 ,CLONE_NEWPID , -EINVAL ,0 ,0 ); +else +ksft_test_result_skip ("Clone3() with set_tid requires root\n" ); + +/* +* Try with a valid PID (1) but as non-root. This should fail
Where are you trying this as non-root? I think that comment is obsolete, no? :)
Quoted 8 lines
tools/testing/selftests/clone3/clone3_set_tid.c +8
-0+* with -EPERM if running in the initial user namespace. +* As root it should tell us -EEXIST. +*/ + set_tid[0 ] =1 ; +if (uid ==0 ) +test_clone3_set_tid (set_tid,1 ,0 , -EEXIST ,0 ,0 ); +else +ksft_test_result_skip ("Clone3() with set_tid requires root\n" );
tools/testing/selftests/clone3/clone3_set_tid.c line 219
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+ +/* Try it in a new PID namespace */ +if (uid ==0 ) +test_clone3_set_tid (set_tid,1 ,CLONE_NEWPID ,0 ,0 ,0 ); +else +ksft_test_result_skip ("Clone3() with set_tid requires root\n" ); + +/* pid_max should fail everywhere */ + set_tid[0 ] = pid_max; +test_clone3_set_tid (set_tid,1 ,0 , -EINVAL ,0 ,0 );
\n
Quoted 44 lines
tools/testing/selftests/clone3/clone3_set_tid.c +44
-0+if (uid ==0 ) +test_clone3_set_tid (set_tid,1 ,CLONE_NEWPID , -EINVAL ,0 ,0 ); +else +ksft_test_result_skip ("Clone3() with set_tid requires root\n" ); + +if (uid !=0 ) { +/* +* All remaining tests require root. Tell the framework +* that all those tests are skipped as non-root. +*/ + ksft_cnt.ksft_xskip += ksft_plan -ksft_test_num (); + goto out; + } + +/* Find the current active PID */ + pid =fork (); +if (pid ==0 ) { +ksft_print_msg ("Child has PID %d\n" ,getpid ()); +_exit (EXIT_SUCCESS ); + } + (void )wait (NULL ); + +/* After the child has finished, its PID should be free. */ + set_tid[0 ] = pid; +test_clone3_set_tid (set_tid,1 ,0 ,0 ,0 ,0 ); + +/* This should fail as there is no PID 1 in that namespace */ +test_clone3_set_tid (set_tid,1 ,CLONE_NEWPID , -EINVAL ,0 ,0 ); + +/* +* Creating a process with PID 1 in the newly created most nested +* PID namespace and PID 'pid' in the parent PID namespace. This +* needs to work. +*/ + set_tid[0 ] =1 ; + set_tid[1 ] = pid; +test_clone3_set_tid (set_tid,2 ,CLONE_NEWPID ,0 , pid,0 ); + +ksft_print_msg ("unshare PID namespace\n" ); +if (unshare (CLONE_NEWPID ) ==-1 ) +ksft_exit_fail_msg ("unshare(CLONE_NEWPID) failed: %s\n" , +strerror (errno)); + + set_tid[0 ] = pid;
tools/testing/selftests/clone3/clone3_set_tid.c line 273
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+ +/* This should fail as there is no PID 1 in that namespace */ +test_clone3_set_tid (set_tid,1 ,0 , -EINVAL ,0 ,0 ); + +/* Let's create a PID 1 */ + ns_pid =fork (); +if (ns_pid ==0 ) { +ksft_print_msg ("Child in PID namespace has PID %d\n" ,getpid ()); + set_tid[0 ] =2 ; +test_clone3_set_tid (set_tid,1 ,0 ,0 ,2 ,0 );
\n
Quoted 5 lines
tools/testing/selftests/clone3/clone3_set_tid.c +5
-0+ set_tid[0 ] =1 ; + set_tid[1 ] =42 ; + set_tid[2 ] = pid; +/* +* This should fail as there are not enough active PID
tools/testing/selftests/clone3/clone3_set_tid.c line 288
Quoted 10 lines
tools/testing/selftests/clone3/clone3_set_tid.c +10
-0+* namespaces. Again assuming this is running in the host's +* PID namespace. Not yet nested. +*/ +test_clone3_set_tid (set_tid,4 ,CLONE_NEWPID , -EINVAL ,0 ,0 ); + +/* +* This should work and from the parent we should see +* something like 'NSpid: pid 42 1'. +*/ +test_clone3_set_tid (set_tid,3 ,CLONE_NEWPID ,0 ,42 ,1 );
Nit: test_clone3_set_tid(set_tid, 3, CLONE_NEWPID, 0, 42, true);
^^^ not 1 :)Quoted 49 lines · Adrian Reber
The main motivation to add
set_tidtoclone3()is CRIU.To restore a process with the same PID/TID CRIU currently uses
/proc/sys/kernel/ns_last_pid. It writes the desired (PID - 1) tons_last_pidand then (quickly) does aclone(). This works most of the time, but it is racy. It is also slow as it requires multiple syscalls.Extending
clone3()to support*set_tidmakes it possible restore a process using CRIU without accessing/proc/sys/kernel/ns_last_pidand race free (as long as the desired PID/TID is available).This
clone3()extension places the same restrictions (CAP_SYS_ADMIN) onclone3()with*set_tidas they are currently in place forns_last_pid.The original version of this change was using a single value for
set_tid. At the 2019 LPC, after presentingset_tid, it was, however, decided to changeset_tidto an array to enable setting the PID of a process in multiple PID namespaces at the same time. If a process is created in a PID namespace it is possible to influence the PID inside and outside of the PID namespace. Details also in the corresponding selftest.To create a process with the following PIDs:
PID NS level Requested PID
0 (host) 31496
1 42
2 1For that example the two newly introduced parameters to struct
clone_args(set_tidandset_tid_size) would need to be:set_tid[0] = 1; set_tid[1] = 42; set_tid[2] = 31496; set_tid_size = 3;If only the PIDs of the two innermost nested PID namespaces should be defined it would look like this:
set_tid[0] = 1; set_tid[1] = 42; set_tid_size = 2;The PID of the newly created process would then be the next available free PID in the PID namespace level 0 (host) and 42 in the PID namespace at level 1 and the PID of the process in the innermost PID namespace would be 1.
Quoted 6 lines
The
set_tidarray is used to specify the PID of a process starting from the innermost nested PID namespaces up toset_tid_sizePID namespaces.
set_tid_sizecannot be larger then the current PID namespace level.Signed-off-by Adrian Reber <[email protected]>
I have no quarrels with the core patch anymore.
Note, once Oleg has said he's fine with this patch too I will likely reword the kernel-doc and the comment in alloc_pid() and the commit message a little before applying; but really just minor things that are not worth resending for.
Thanks!
Quoted 57 lines · Christian Brauner
Quoted 49 lines · Adrian Reber
The main motivation to add
set_tidtoclone3()is CRIU.To restore a process with the same PID/TID CRIU currently uses
/proc/sys/kernel/ns_last_pid. It writes the desired (PID - 1) tons_last_pidand then (quickly) does aclone(). This works most of the time, but it is racy. It is also slow as it requires multiple syscalls.Extending
clone3()to support*set_tidmakes it possible restore a process using CRIU without accessing/proc/sys/kernel/ns_last_pidand race free (as long as the desired PID/TID is available).This
clone3()extension places the same restrictions (CAP_SYS_ADMIN) onclone3()with*set_tidas they are currently in place forns_last_pid.The original version of this change was using a single value for
set_tid. At the 2019 LPC, after presentingset_tid, it was, however, decided to changeset_tidto an array to enable setting the PID of a process in multiple PID namespaces at the same time. If a process is created in a PID namespace it is possible to influence the PID inside and outside of the PID namespace. Details also in the corresponding selftest.To create a process with the following PIDs:
PID NS level Requested PID
0 (host) 31496
1 42
2 1For that example the two newly introduced parameters to struct
clone_args(set_tidandset_tid_size) would need to be:set_tid[0] = 1; set_tid[1] = 42; set_tid[2] = 31496; set_tid_size = 3;If only the PIDs of the two innermost nested PID namespaces should be defined it would look like this:
set_tid[0] = 1; set_tid[1] = 42; set_tid_size = 2;The PID of the newly created process would then be the next available free PID in the PID namespace level 0 (host) and 42 in the PID namespace at level 1 and the PID of the process in the innermost PID namespace would be 1.
Quoted 6 lines
The
set_tidarray is used to specify the PID of a process starting from the innermost nested PID namespaces up toset_tid_sizePID namespaces.
set_tid_sizecannot be larger then the current PID namespace level.Signed-off-by Adrian Reber <[email protected]>
Quoted 8 lines
I have no quarrels with the core patch anymore.
Note, once Oleg has said he's fine with this patch too I will likely reword the kernel-doc and the comment inalloc_pid()and the commit message a little before applying; but really just minor things that are not worth resending for.Thanks!
Reviewed-by Christian Brauner <[email protected]>
And thanks for the --base=<commit> information you've added to this version. :)
Christian
include/uapi/linux/sched.h struct clone_args { · line 89
Quoted 8 lines · Adrian Reber
* The structure is versioned by size and thus extensible.
* New struct members must go at the end of the struct andinclude/uapi/linux/sched.h +2
-0@@ -71,6 +85,8 @@ struct clone_args {__aligned_u64 stack;__aligned_u64 stack_size;__aligned_u64 tls; +__aligned_u64 set_tid; +__aligned_u64 set_tid_size;
Oh, one thing that is missing is the addition of
+#define CLONE_ARGS_SIZE_VER1 80 /* sizeof second published struct */
in sched.h. Please add that, when you resend.
Thanks!
Christian
Quoted 12 lines · Christian Brauner
include/uapi/linux/sched.h struct clone_args { · line 89
Quoted 8 lines · Adrian Reber
* The structure is versioned by size and thus extensible.
* New struct members must go at the end of the struct andinclude/uapi/linux/sched.h +2
-0@@ -71,6 +85,8 @@ struct clone_args {__aligned_u64 stack;__aligned_u64 stack_size;__aligned_u64 tls; +__aligned_u64 set_tid; +__aligned_u64 set_tid_size;Oh, one thing that is missing is the addition of
Quoted 3 lines
+#define
CLONE_ARGS_SIZE_VER180 /* sizeof second published struct */in
sched.h. Please add that, when you resend.
Even if it is unused?
Adrian
Quoted 13 lines · Adrian Reber
Quoted 12 lines · Christian Brauner
include/uapi/linux/sched.h struct clone_args { · line 89
Quoted 8 lines · Adrian Reber
* The structure is versioned by size and thus extensible.
* New struct members must go at the end of the struct andinclude/uapi/linux/sched.h +2
-0@@ -71,6 +85,8 @@ struct clone_args {__aligned_u64 stack;__aligned_u64 stack_size;__aligned_u64 tls; +__aligned_u64 set_tid; +__aligned_u64 set_tid_size;Oh, one thing that is missing is the addition of
Quoted 5 lines
Quoted 3 lines
+#define
CLONE_ARGS_SIZE_VER180 /* sizeof second published struct */in
sched.h. Please add that, when you resend.Even if it is unused?
Yes, I think so. It's nice for userspace to have these macros for versioning purposes. We already started exposing this versioning with _VER0 to userspace.
Christian