pidns: Place under CONFIG_EXPERIMENTAL
PATCH 1/1 pidns: Place under CONFIG_EXPERIMENTAL
I finally found a chance to review the pid namespace implementation in detail and currently it is much to easy to find issues where the kernel does the wrong thing outside of the initial pid namespace. At the same time the pid namespace code we have does appear to be a good base to build on.
Therefore until the dust settles and we are certain we have the ABI and the implementation as correct as humanly possible let's hide the availability of process ID namespaces behind CONFIG_EXPERIMENTAL.
Allowing users to avoid bugs, and removing a guarantee of bug compatibility. Allowing any issues that may be found to be fixed properly.
If CONFIG_PID_NS=N this patch will cause copy_pid_ns to unconditionally return -EINVAL removing the availability of multiple pid namespaces.
--- init/Kconfig | 12 ++++++++++++ kernel/pid.c | 4 ++++ 2 files changed, 16 insertions(+), 0 deletions(-)
init/Kconfig +12 -0
@@ -215,6 +215,18 @@ config USER_NS vservers, to use user namespaces to provide different user info for different servers. If unsure, say N. +config PID_NS + bool "PID Namespaces (EXPERIMENTAL)" + default n + depends on EXPERIMENTAL + help + Suport process id namespaces. This allows having multiple + process with the same pid as long as they are in different + pid namespaces. This is a building block of containers. + + Unless you want to work with an experimental feature + say N here. + config AUDIT bool "Auditing support" depends on NET
kernel/pid.c +4 -0
@@ -599,6 +599,10 @@ struct pid_namespace *copy_pid_ns(unsigned long flags, struct pid_namespace *oldif (flags &CLONE_THREAD ) goto out_put; +#ifndef CONFIG_PID_NS + goto out_put; +#endif + new_ns =create_pid_namespace (old_ns->level +1 );if (!IS_ERR (new_ns)) new_ns->parent =get_pid_ns (old_ns);
Signature · 1 line
1.5.3.rc6.17.g1911
A message that is not in the archive (sent off-list or not yet received): m1640t3has.fsf-T1Yj925okcoyDheHMi7gv2pdwda3JcWeAL8bYrjMMd8@public.gmane.org
kernel/pid.c struct pid_namespace *copy_pid_ns(unsigned long flags, struct pid_namespace *old · line 605
Quoted 5 lines · Eric W. Biederman
kernel/pid.c +4
-0+#ifndef CONFIG_PID_NS + goto out_put; +#endif +
No. We don't do crap like this. That's just horrible.
If this is conditional, then we should have conditional versions of "create/destroy_pid_namespace()" or something.
Linus
PATCH 1/1 pidns: Place under CONFIG_EXPERIMENTAL (take 2)
This is my trivial patch to swat innumerable little bugs with a single blow.
After some intensive review (my apologies for not having gotten to this sooner) what we have looks like a good base to build on with the current pid namespace code but it is not complete, and it is still much to simple to find issues where the kernel does the wrong thing outside of the initial pid namespace.
Until the dust settles and we are certain we have the ABI and the implementation is as correct as humanly possible let's keep process ID namespaces behind CONFIG_EXPERIMENTAL.
Allowing us the option of fixing any ABI or other bugs we find as long as they are minor.
Allowing users of the kernel to avoid those bugs simply by ensuring their kernel does not have support for multiple pid namespaces.
--- include/linux/pid_namespace.h | 22 ++++++++++++++++++++++ init/Kconfig | 12 ++++++++++++ kernel/pid.c | 2 ++ 3 files changed, 36 insertions(+), 0 deletions(-)
include/linux/pid_namespace.h +22 -0
@@ -29,6 +29,7 @@ struct pid_namespace {extern struct pid_namespace init_pid_ns; +#ifdef CONFIG_PID_NS static inline struct pid_namespace *get_pid_ns (struct pid_namespace *ns) {if (ns != &init_pid_ns)
@@ -45,6 +46,27 @@ static inline void put_pid_ns(struct pid_namespace *ns) kref_put(&ns->kref, free_pid_ns); } +#else /* !CONFIG_PID_NS */ +#include <linux/err.h> + +static inline struct pid_namespace *get_pid_ns (struct pid_namespace *ns) +{ +return ns; +} + +static inline struct pid_namespace *copy_pid_ns (unsigned long flags,struct pid_namespace *ns) +{ +if (flags &CLONE_NEWPID ) + ns =ERR_PTR (-EINVAL ); +return ns; +} + +static inline void put_pid_ns (struct pid_namespace *ns) +{ +} + +#endif /* CONFIG_PID_NS */ +static inline struct pid_namespace *task_active_pid_ns (struct task_struct *tsk) {return tsk->nsproxy->pid_ns;
init/Kconfig +12 -0
@@ -215,6 +215,18 @@ config USER_NS vservers, to use user namespaces to provide different user info for different servers. If unsure, say N. +config PID_NS + bool "PID Namespaces (EXPERIMENTAL)" + default n + depends on EXPERIMENTAL + help + Suport process id namespaces. This allows having multiple + process with the same pid as long as they are in different + pid namespaces. This is a building block of containers. + + Unless you want to work with an experimental feature + say N here. + config AUDIT bool "Auditing support" depends on NET
kernel/pid.c +2 -0
@@ -537,6 +537,7 @@ err_alloc:return NULL ; } +#ifdef CONFIG_PID_NS static struct pid_namespace *create_pid_namespace (int level) {struct pid_namespace *ns;
@@ -621,6 +622,7 @@ void free_pid_ns(struct kref *kref)if (parent !=NULL ) put_pid_ns(parent); } +#endif /* CONFIG_PID_NS */ void zap_pid_ns_processes (struct pid_namespace *pid_ns) {
Signature · 1 line
1.5.3.rc6.17.g1911
Re: [PATCH] pidns: Place under CONFIG_EXPERIMENTAL (take 2)
Can you please send namespace related patches to containers@ ML first before sending them to Linus/Andrew?
Thanks,
Kirill
Quoted 111 lines · Eric W. Biederman
This is my trivial patch to swat innumerable little bugs with a single blow.
After some intensive review (my apologies for not having gotten to this sooner) what we have looks like a good base to build on with the current pid namespace code but it is not complete, and it is still much to simple to find issues where the kernel does the wrong thing outside of the initial pid namespace.
Until the dust settles and we are certain we have the ABI and the implementation is as correct as humanly possible let's keep process ID namespaces behind
CONFIG_EXPERIMENTAL.Allowing us the option of fixing any ABI or other bugs we find as long as they are minor.
Allowing users of the kernel to avoid those bugs simply by ensuring their kernel does not have support for multiple pid namespaces.
Signed-off-by Eric W. Biederman <[email protected]>--- include/linux/pid_namespace.h | 22 ++++++++++++++++++++++ init/Kconfig | 12 ++++++++++++ kernel/pid.c | 2 ++ 3 files changed, 36 insertions(+), 0 deletions(-)include/linux/pid_namespace.h +22
-0@@ -29,6 +29,7 @@ struct pid_namespace {extern struct pid_namespace init_pid_ns; +#ifdef CONFIG_PID_NS static inline struct pid_namespace *get_pid_ns (struct pid_namespace *ns) {if (ns != &init_pid_ns)@@ -45,6 +46,27 @@ static inline void put_pid_ns(struct pid_namespace *ns) kref_put(&ns->kref, free_pid_ns); } +#else /* !CONFIG_PID_NS */ +#include <linux/err.h> + +static inline struct pid_namespace *get_pid_ns (struct pid_namespace *ns) +{ +return ns; +} + +static inline struct pid_namespace *copy_pid_ns (unsigned long flags,struct pid_namespace *ns) +{ +if (flags &CLONE_NEWPID ) + ns =ERR_PTR (-EINVAL ); +return ns; +} + +static inline void put_pid_ns (struct pid_namespace *ns) +{ +} + +#endif /* CONFIG_PID_NS */ +static inline struct pid_namespace *task_active_pid_ns (struct task_struct *tsk) {return tsk->nsproxy->pid_ns;init/Kconfig +12
-0@@ -215,6 +215,18 @@ config USER_NS vservers, to use user namespaces to provide different user info for different servers. If unsure, say N. +config PID_NS + bool "PID Namespaces (EXPERIMENTAL)" + default n + depends on EXPERIMENTAL + help + Suport process id namespaces. This allows having multiple + process with the same pid as long as they are in different + pid namespaces. This is a building block of containers. + + Unless you want to work with an experimental feature + say N here. + config AUDIT bool "Auditing support" depends on NETkernel/pid.c +2
-0@@ -537,6 +537,7 @@ err_alloc:return NULL ; } +#ifdef CONFIG_PID_NS static struct pid_namespace *create_pid_namespace (int level) {struct pid_namespace *ns;@@ -621,6 +622,7 @@ void free_pid_ns(struct kref *kref)if (parent !=NULL ) put_pid_ns(parent); } +#endif /* CONFIG_PID_NS */ void zap_pid_ns_processes (struct pid_namespace *pid_ns) {
Re: [PATCH] pidns: Place under CONFIG_EXPERIMENTAL (take 2)
Quoted 2 lines · Kirill Korotaev
Can you please send namespace related patches to containers@ ML first before sending them to Linus/Andrew?
If you are so anxious to review my patches can you please review them? I'd love to see an acked-by or an actual bug found.
I only did what I always do when I am focusing on fixing a bug in the stable kernel. I fix the bug if I can with a simple obviously correct patch and send that patch off to the maintainer. Copying other people that I know of who have a chance of reviewing the patch.
Eric
Quoted 10 lines · Eric W. Biederman
I finally found a chance to review the pid namespace implementation in detail and currently it is much to easy to find issues where the kernel does the wrong thing outside of the initial pid namespace. At the same time the pid namespace code we have does appear to be a good base to build on.
Therefore until the dust settles and we are certain we have the ABI and the implementation as correct as humanly possible let's hide the availability of process ID namespaces behind
CONFIG_EXPERIMENTAL.
Sorry for the late answer - I was out for a conference and the organization committee didn't provide an internet access.
I currently have a set of patches that move all the namespaces cloning code under the config option. This is done to help embedded people have a small kernel.
I was planning to wait with this set untill 2.6.24-rc-mm1 kernel, but since (as I see) this is required rather badly I will send this set in a couple of days.
Thanks,
Pavel
Quoted 52 lines
Allowing users to avoid bugs, and removing a guarantee of bug compatibility. Allowing any issues that may be found to be fixed properly.
If
CONFIG_PID_NS=Nthis patch will causecopy_pid_nsto unconditionally return-EINVALremoving the availability of multiple pid namespaces.--- init/Kconfig | 12 ++++++++++++ kernel/pid.c | 4 ++++ 2 files changed, 16 insertions(+), 0 deletions(-)init/Kconfig +12
-0@@ -215,6 +215,18 @@ config USER_NS vservers, to use user namespaces to provide different user info for different servers. If unsure, say N. +config PID_NS + bool "PID Namespaces (EXPERIMENTAL)" + default n + depends on EXPERIMENTAL + help + Suport process id namespaces. This allows having multiple + process with the same pid as long as they are in different + pid namespaces. This is a building block of containers. + + Unless you want to work with an experimental feature + say N here. + config AUDIT bool "Auditing support" depends on NETkernel/pid.c +4
-0@@ -599,6 +599,10 @@ struct pid_namespace *copy_pid_ns(unsigned long flags, struct pid_namespace *oldif (flags &CLONE_THREAD ) goto out_put; +#ifndef CONFIG_PID_NS + goto out_put; +#endif + new_ns =create_pid_namespace (old_ns->level +1 );if (!IS_ERR (new_ns)) new_ns->parent =get_pid_ns (old_ns);