Skip to contentSkip to conversation

seccomp: hoist out filter resolving logic

PATCH In review Tycho Andersen 3 participants 6 messages Nov 15, 2017 – Feb 21, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v4 1/2 seccomp: hoist out filter resolving logic

Hoist out the nth filter resolving logic that ptrace uses into a new function. We'll use this in the next patch to implement the new PTRACE_SECCOMP_GET_FILTER_FLAGS command.

v3: * significantly revamp get_nth_filter logic (Oleg)
* rebase on 4.14-rc4, using the new __{get,put}_seccomp_filter primitives
v4: * resend

CC Kees Cook <[email protected]>
CC Andy Lutomirski <[email protected]>
CC Oleg Nesterov <[email protected]>
---
 kernel/seccomp.c | 77 +++++++++++++++++++++++++++++++++-----------------------
 1 file changed, 45 insertions(+), 32 deletions(-)
kernel/seccomp.c +45 -32
index 418a1c045933..bad457862ee0 100644
@@ -978,49 +978,68 @@ long prctl_set_seccomp(unsigned long seccomp_mode, char __user *filter)
 }
 
 #if defined(CONFIG_SECCOMP_FILTER) && defined(CONFIG_CHECKPOINT_RESTORE)
-long seccomp_get_filter(struct task_struct *task, unsigned long filter_off,
-			void __user *data)
+static struct seccomp_filter *get_nth_filter(struct task_struct *task,
+					     unsigned long filter_off)
 {
-	struct seccomp_filter *filter;
-	struct sock_fprog_kern *fprog;
-	long ret;
-	unsigned long count = 0;
-
-	if (!capable(CAP_SYS_ADMIN) ||
-	    current->seccomp.mode != SECCOMP_MODE_DISABLED) {
-		return -EACCES;
-	}
+	struct seccomp_filter *orig, *filter;
+	unsigned long count;
 
+	/*
+	 * Note: this is only correct because the caller should be the (ptrace)
+	 * tracer of the task, otherwise lock_task_sighand is needed.
+	 */
 	spin_lock_irq(&task->sighand->siglock);
+
 	if (task->seccomp.mode != SECCOMP_MODE_FILTER) {
-		ret = -EINVAL;
-		goto out;
+		spin_unlock_irq(&task->sighand->siglock);
+		return ERR_PTR(-EINVAL);
 	}
 
-	filter = task->seccomp.filter;
-	while (filter) {
-		filter = filter->prev;
+	orig = task->seccomp.filter;
+	__get_seccomp_filter(orig);
+	spin_unlock_irq(&task->sighand->siglock);
+
+	count = 0;
+	for (filter = orig; filter; filter = filter->prev)
 		count++;
-	}
 
 	if (filter_off >= count) {
-		ret = -ENOENT;
+		filter = ERR_PTR(-ENOENT);
 		goto out;
 	}
-	count -= filter_off;
 
-	filter = task->seccomp.filter;
-	while (filter && count > 1) {
-		filter = filter->prev;
+	count -= filter_off;
+	for (filter = orig; filter && count > 1; filter = filter->prev)
 		count--;
-	}
 
 	if (WARN_ON(count != 1 || !filter)) {
-		/* The filter tree shouldn't shrink while we're using it. */
-		ret = -ENOENT;
+		filter = ERR_PTR(-ENOENT);
 		goto out;
 	}
 
+	__get_seccomp_filter(filter);
+
+out:
+	__put_seccomp_filter(orig);
+	return filter;
+}
+
+long seccomp_get_filter(struct task_struct *task, unsigned long filter_off,
+			void __user *data)
+{
+	struct seccomp_filter *filter;
+	struct sock_fprog_kern *fprog;
+	long ret;
+
+	if (!capable(CAP_SYS_ADMIN) ||
+	    current->seccomp.mode != SECCOMP_MODE_DISABLED) {
+		return -EACCES;
+	}
+
+	filter = get_nth_filter(task, filter_off);
+	if (IS_ERR(filter))
+		return PTR_ERR(filter);
+
 	fprog = filter->prog->orig_prog;
 	if (!fprog) {
 		/* This must be a new non-cBPF filter, since we save
@@ -1035,17 +1054,11 @@ long seccomp_get_filter(struct task_struct *task, unsigned long filter_off,
 	if (!data)
 		goto out;
 
-	__get_seccomp_filter(filter);
-	spin_unlock_irq(&task->sighand->siglock);
-
 	if (copy_to_user(data, fprog->filter, bpf_classic_proglen(fprog)))
 		ret = -EFAULT;
 
-	__put_seccomp_filter(filter);
-	return ret;
-
 out:
-	spin_unlock_irq(&task->sighand->siglock);
+	__put_seccomp_filter(filter);
 	return ret;
 }
 #endif
Signature · 1 line
2.14.1
#1 lore raw

PATCH v4 2/2 ptrace, seccomp: add support for retrieving seccomp metadata

With the new SECCOMP_FILTER_FLAG_LOG, we need to be able to extract these flags for checkpoint restore, since they describe the state of a filter.

So, let's add PTRACE_SECCOMP_GET_METADATA, similar to ..._GET_FILTER, which returns the metadata of the nth filter (right now, just the flags). Hopefully this will be future proof, and new per-filter metadata can be added to this struct.

v3: * use GET_METADATA instead of GET_FLAGS
v4: * resend

CC Kees Cook <[email protected]>
CC Andy Lutomirski <[email protected]>
CC Oleg Nesterov <[email protected]>
---
 include/linux/seccomp.h     |  8 ++++++++
 include/uapi/linux/ptrace.h |  6 ++++++
 kernel/ptrace.c             |  4 ++++
 kernel/seccomp.c            | 34 ++++++++++++++++++++++++++++++++++
 4 files changed, 52 insertions(+)
include/linux/seccomp.h +8 -0
index 10f25f7e4304..c723a5c4e3ff 100644
@@ -95,11 +95,19 @@ static inline void get_seccomp_filter(struct task_struct *tsk)
 #if defined(CONFIG_SECCOMP_FILTER) && defined(CONFIG_CHECKPOINT_RESTORE)
 extern long seccomp_get_filter(struct task_struct *task,
 			       unsigned long filter_off, void __user *data);
+extern long seccomp_get_metadata(struct task_struct *task,
+				 unsigned long filter_off, void __user *data);
 #else
 static inline long seccomp_get_filter(struct task_struct *task,
 				      unsigned long n, void __user *data)
 {
 	return -EINVAL;
 }
+static inline long seccomp_get_metadata(struct task_struct *task,
+					unsigned long filter_off,
+					void __user *data)
+{
+	return -EINVAL;
+}
 #endif /* CONFIG_SECCOMP_FILTER && CONFIG_CHECKPOINT_RESTORE */
 #endif /* _LINUX_SECCOMP_H */
include/uapi/linux/ptrace.h +6 -0
index e3939e00980b..e46d82b91166 100644
@@ -66,6 +66,12 @@ struct ptrace_peeksiginfo_args {
 #define PTRACE_SETSIGMASK	0x420b
 
 #define PTRACE_SECCOMP_GET_FILTER	0x420c
+#define PTRACE_SECCOMP_GET_METADATA	0x420d
+
+struct seccomp_metadata {
+	unsigned long filter_off;	/* Input: which filter */
+	unsigned int flags;		/* Output: filter's flags */
+};
 
 /* Read signals from a shared (process wide) queue */
 #define PTRACE_PEEKSIGINFO_SHARED	(1 << 0)
kernel/ptrace.c +4 -0
index 84b1367935e4..58291e9f3276 100644
@@ -1092,6 +1092,10 @@ int ptrace_request(struct task_struct *child, long request,
 		ret = seccomp_get_filter(child, addr, datavp);
 		break;
 
+	case PTRACE_SECCOMP_GET_METADATA:
+		ret = seccomp_get_metadata(child, addr, datavp);
+		break;
+
 	default:
 		break;
 	}
kernel/seccomp.c +34 -0
index bad457862ee0..7f1f2f3ea549 100644
@@ -1061,6 +1061,40 @@ long seccomp_get_filter(struct task_struct *task, unsigned long filter_off,
 	__put_seccomp_filter(filter);
 	return ret;
 }
+
+long seccomp_get_metadata(struct task_struct *task,
+			  unsigned long size, void __user *data)
+{
+	long ret;
+	struct seccomp_filter *filter;
+	struct seccomp_metadata kmd = {};
+
+	if (!capable(CAP_SYS_ADMIN) ||
+	    current->seccomp.mode != SECCOMP_MODE_DISABLED) {
+		return -EACCES;
+	}
+
+	size = min_t(unsigned long, size, sizeof(kmd));
+
+	if (copy_from_user(&kmd, data, size))
+		return -EFAULT;
+
+	filter = get_nth_filter(task, kmd.filter_off);
+	if (IS_ERR(filter))
+		return PTR_ERR(filter);
+
+	memset(&kmd, 0, sizeof(kmd));
+	if (filter->log)
+		kmd.flags |= SECCOMP_FILTER_FLAG_LOG;
+
+	ret = size;
+	if (copy_to_user(data, &kmd, size))
+		ret = -EFAULT;
+
+	__put_seccomp_filter(filter);
+	return ret;
+
+}
 #endif
 
 #ifdef CONFIG_SYSCTL
Signature · 1 line
2.14.1
#2 lore raw

Re: [PATCH v4 2/2] ptrace, seccomp: add support for retrieving seccomp metadata

Hi,

Quoted 15 lines · Tycho Andersen

With the new SECCOMP_FILTER_FLAG_LOG, we need to be able to extract these flags for checkpoint restore, since they describe the state of a filter.

So, let's add PTRACE_SECCOMP_GET_METADATA, similar to ..._GET_FILTER, which returns the metadata of the nth filter (right now, just the flags). Hopefully this will be future proof, and new per-filter metadata can be added to this struct.

v3: * use GET_METADATA instead of GET_FLAGS
v4: * resend

CC Kees Cook <[email protected]>
CC Andy Lutomirski <[email protected]>
CC Oleg Nesterov <[email protected]>

[...]

Quoted 5 lines
include/uapi/linux/ptrace.h +0 -0
index e3939e00980b..e46d82b91166 100644

include/uapi/linux/ptrace.h struct ptrace_peeksiginfo_args { · line 74

Quoted 9 lines
include/uapi/linux/ptrace.h +6 -0
@@ -66,6 +66,12 @@ struct ptrace_peeksiginfo_args {
 #define PTRACE_SETSIGMASK	0x420b
 
 #define PTRACE_SECCOMP_GET_FILTER	0x420c
+#define PTRACE_SECCOMP_GET_METADATA	0x420d
+
+struct seccomp_metadata {
+	unsigned long filter_off;	/* Input: which filter */
+	unsigned int flags;		/* Output: filter's flags */
+};

This "unsigned long" field is unacceptable unless you are willing to implement a compat layer for this arch-dependent type.

Please use an arch-independent type instead.
See e.g. struct ptrace_peeksiginfo_args or struct struct seccomp_data.

Signature · 1 line
ldv
#3 lore raw

Re: [PATCH v4 2/2] ptrace, seccomp: add support for retrieving seccomp metadata

Quoted 10 lines · Dmitry V. Levin

include/uapi/linux/ptrace.h struct ptrace_peeksiginfo_args { · line 74

Quoted 4 lines
include/uapi/linux/ptrace.h +4 -0
+struct seccomp_metadata {
+	unsigned long filter_off;	/* Input: which filter */
+	unsigned int flags;		/* Output: filter's flags */
+};

This "unsigned long" field is unacceptable unless you are willing to implement a compat layer for this arch-dependent type.

Please use an arch-independent type instead.
See e.g. struct ptrace_peeksiginfo_args or struct struct seccomp_data.

Whoops, sorry about that. I believe I've addressed it in
https://lkml.org/lkml/2018/2/20/755

Cheers,

Tycho

#4 lore raw

Re: [PATCH v4 2/2] ptrace, seccomp: add support for retrieving seccomp metadata

In reply to Tycho Andersen
Quoted 94 lines · Tycho Andersen

With the new SECCOMP_FILTER_FLAG_LOG, we need to be able to extract these flags for checkpoint restore, since they describe the state of a filter.

So, let's add PTRACE_SECCOMP_GET_METADATA, similar to ..._GET_FILTER, which returns the metadata of the nth filter (right now, just the flags). Hopefully this will be future proof, and new per-filter metadata can be added to this struct.

v3: * use GET_METADATA instead of GET_FLAGS
v4: * resend

CC Kees Cook <[email protected]>
CC Andy Lutomirski <[email protected]>
CC Oleg Nesterov <[email protected]>
---
 include/linux/seccomp.h     |  8 ++++++++
 include/uapi/linux/ptrace.h |  6 ++++++
 kernel/ptrace.c             |  4 ++++
 kernel/seccomp.c            | 34 ++++++++++++++++++++++++++++++++++
 4 files changed, 52 insertions(+)
include/linux/seccomp.h +8 -0
index 10f25f7e4304..c723a5c4e3ff 100644
@@ -95,11 +95,19 @@ static inline void get_seccomp_filter(struct task_struct *tsk)
 #if defined(CONFIG_SECCOMP_FILTER) && defined(CONFIG_CHECKPOINT_RESTORE)
 extern long seccomp_get_filter(struct task_struct *task,
 			       unsigned long filter_off, void __user *data);
+extern long seccomp_get_metadata(struct task_struct *task,
+				 unsigned long filter_off, void __user *data);
 #else
 static inline long seccomp_get_filter(struct task_struct *task,
 				      unsigned long n, void __user *data)
 {
 	return -EINVAL;
 }
+static inline long seccomp_get_metadata(struct task_struct *task,
+					unsigned long filter_off,
+					void __user *data)
+{
+	return -EINVAL;
+}
 #endif /* CONFIG_SECCOMP_FILTER && CONFIG_CHECKPOINT_RESTORE */
 #endif /* _LINUX_SECCOMP_H */
include/uapi/linux/ptrace.h +6 -0
index e3939e00980b..e46d82b91166 100644
@@ -66,6 +66,12 @@ struct ptrace_peeksiginfo_args {
 #define PTRACE_SETSIGMASK	0x420b
 
 #define PTRACE_SECCOMP_GET_FILTER	0x420c
+#define PTRACE_SECCOMP_GET_METADATA	0x420d
+
+struct seccomp_metadata {
+	unsigned long filter_off;	/* Input: which filter */
+	unsigned int flags;		/* Output: filter's flags */
+};
 
 /* Read signals from a shared (process wide) queue */
 #define PTRACE_PEEKSIGINFO_SHARED	(1 << 0)
kernel/ptrace.c +4 -0
index 84b1367935e4..58291e9f3276 100644
@@ -1092,6 +1092,10 @@ int ptrace_request(struct task_struct *child, long request,
 		ret = seccomp_get_filter(child, addr, datavp);
 		break;
 
+	case PTRACE_SECCOMP_GET_METADATA:
+		ret = seccomp_get_metadata(child, addr, datavp);
+		break;
+
 	default:
 		break;
 	}
kernel/seccomp.c +8 -0
index bad457862ee0..7f1f2f3ea549 100644
@@ -1061,6 +1061,40 @@ long seccomp_get_filter(struct task_struct *task, unsigned long filter_off,
 	__put_seccomp_filter(filter);
 	return ret;
 }
+
+long seccomp_get_metadata(struct task_struct *task,
+			  unsigned long size, void __user *data)
+{
+	long ret;
+	struct seccomp_filter *filter;
+	struct seccomp_metadata kmd = {};
+

kernel/seccomp.c long seccomp_get_filter(struct task_struct *task, unsigned long filter_off, · line 1081

Quoted 10 lines
kernel/seccomp.c +9 -0
+	if (!capable(CAP_SYS_ADMIN) ||
+	    current->seccomp.mode != SECCOMP_MODE_DISABLED) {
+		return -EACCES;
+	}
+
 
+	size = min_t(unsigned long, size, sizeof(kmd));
+
+	if (copy_from_user(&kmd, data, size))
+		return -EFAULT;

This will work kinda funny if size is less than sizeof(kmd.filter_off). And I see no reason for copying the whole structure when only filter_off is needed.

kernel/seccomp.c long seccomp_get_filter(struct task_struct *task, unsigned long filter_off, · line 1086

Quoted 6 lines
kernel/seccomp.c +6 -0
+
+	filter = get_nth_filter(task, kmd.filter_off);
+	if (IS_ERR(filter))
+		return PTR_ERR(filter);
+
+	memset(&kmd, 0, sizeof(kmd));

Is rewriting the filter_off field to 0 intentional?

Quoted 17 lines
kernel/seccomp.c +11 -0
+	if (filter->log)
+		kmd.flags |= SECCOMP_FILTER_FLAG_LOG;
+
+	ret = size;
+	if (copy_to_user(data, &kmd, size))
+		ret = -EFAULT;
+
+	__put_seccomp_filter(filter);
+	return ret;
+
+}
 #endif
 

#ifdef CONFIG_SYSCTL

Signature · 1 line
2.14.1
#5 lore raw

Re: [PATCH v4 2/2] ptrace, seccomp: add support for retrieving seccomp metadata

Quoted 101 lines · Eugene Syromiatnikov
Quoted 99 lines · Tycho Andersen

With the new SECCOMP_FILTER_FLAG_LOG, we need to be able to extract these flags for checkpoint restore, since they describe the state of a filter.

So, let's add PTRACE_SECCOMP_GET_METADATA, similar to ..._GET_FILTER, which returns the metadata of the nth filter (right now, just the flags). Hopefully this will be future proof, and new per-filter metadata can be added to this struct.

v3: * use GET_METADATA instead of GET_FLAGS
v4: * resend

CC Kees Cook <[email protected]>
CC Andy Lutomirski <[email protected]>
CC Oleg Nesterov <[email protected]>
---
 include/linux/seccomp.h     |  8 ++++++++
 include/uapi/linux/ptrace.h |  6 ++++++
 kernel/ptrace.c             |  4 ++++
 kernel/seccomp.c            | 34 ++++++++++++++++++++++++++++++++++
 4 files changed, 52 insertions(+)
include/linux/seccomp.h +8 -0
index 10f25f7e4304..c723a5c4e3ff 100644
@@ -95,11 +95,19 @@ static inline void get_seccomp_filter(struct task_struct *tsk)
 #if defined(CONFIG_SECCOMP_FILTER) && defined(CONFIG_CHECKPOINT_RESTORE)
 extern long seccomp_get_filter(struct task_struct *task,
 			       unsigned long filter_off, void __user *data);
+extern long seccomp_get_metadata(struct task_struct *task,
+				 unsigned long filter_off, void __user *data);
 #else
 static inline long seccomp_get_filter(struct task_struct *task,
 				      unsigned long n, void __user *data)
 {
 	return -EINVAL;
 }
+static inline long seccomp_get_metadata(struct task_struct *task,
+					unsigned long filter_off,
+					void __user *data)
+{
+	return -EINVAL;
+}
 #endif /* CONFIG_SECCOMP_FILTER && CONFIG_CHECKPOINT_RESTORE */
 #endif /* _LINUX_SECCOMP_H */
include/uapi/linux/ptrace.h +6 -0
index e3939e00980b..e46d82b91166 100644
@@ -66,6 +66,12 @@ struct ptrace_peeksiginfo_args {
 #define PTRACE_SETSIGMASK	0x420b
 
 #define PTRACE_SECCOMP_GET_FILTER	0x420c
+#define PTRACE_SECCOMP_GET_METADATA	0x420d
+
+struct seccomp_metadata {
+	unsigned long filter_off;	/* Input: which filter */
+	unsigned int flags;		/* Output: filter's flags */
+};
 
 /* Read signals from a shared (process wide) queue */
 #define PTRACE_PEEKSIGINFO_SHARED	(1 << 0)
kernel/ptrace.c +4 -0
index 84b1367935e4..58291e9f3276 100644
@@ -1092,6 +1092,10 @@ int ptrace_request(struct task_struct *child, long request,
 		ret = seccomp_get_filter(child, addr, datavp);
 		break;
 
+	case PTRACE_SECCOMP_GET_METADATA:
+		ret = seccomp_get_metadata(child, addr, datavp);
+		break;
+
 	default:
 		break;
 	}
kernel/seccomp.c +13 -0
index bad457862ee0..7f1f2f3ea549 100644
@@ -1061,6 +1061,40 @@ long seccomp_get_filter(struct task_struct *task, unsigned long filter_off,
 	__put_seccomp_filter(filter);
 	return ret;
 }
+
+long seccomp_get_metadata(struct task_struct *task,
+			  unsigned long size, void __user *data)
+{
+	long ret;
+	struct seccomp_filter *filter;
+	struct seccomp_metadata kmd = {};
+
+	if (!capable(CAP_SYS_ADMIN) ||
+	    current->seccomp.mode != SECCOMP_MODE_DISABLED) {
+		return -EACCES;
+	}
+
Quoted 8 lines

kernel/seccomp.c long seccomp_get_filter(struct task_struct *task, unsigned long filter_off, · line 1080

Quoted 4 lines
kernel/seccomp.c +4 -0
+	size = min_t(unsigned long, size, sizeof(kmd));
+
+	if (copy_from_user(&kmd, data, size))
+		return -EFAULT;

This will work kinda funny if size is less than sizeof(kmd.filter_off). And I see no reason for copying the whole structure when only filter_off is needed.

Yes, agreed. EINVAL seems better.

Quoted 8 lines

kernel/seccomp.c long seccomp_get_filter(struct task_struct *task, unsigned long filter_off, · line 1086

Quoted 6 lines
kernel/seccomp.c +6 -0
+
+	filter = get_nth_filter(task, kmd.filter_off);
+	if (IS_ERR(filter))
+		return PTR_ERR(filter);
+
+	memset(&kmd, 0, sizeof(kmd));

Is rewriting the filter_off field to 0 intentional?

It "shouldn't" matter, but it's probably better not to; I've fixed it in https://lkml.org/lkml/2018/2/20/755

Thanks!

Tycho