Skip to contentSkip to conversation

mm: use only pidfd for process_madvise syscall

PATCH In review Minchan Kim 4 participants 8 messages May 16, 2020 – May 19, 2020

Subsystems (18)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mm: use only pidfd for process_madvise syscall

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +13 -29
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-				       int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+				int behavior, unsigned int flags)
 {
 	ssize_t ret;
 	struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	if (flags != 0)
 		return -EINVAL;
 
-	switch (which) {
-	case P_PID:
-		if (upid <= 0)
-			return -EINVAL;
-
-		pid = find_get_pid(upid);
-		if (!pid)
-			return -ESRCH;
-		break;
-	case P_PIDFD:
-		if (upid < 0)
-			return -EINVAL;
-
-		pid = pidfd_get_pid(upid);
-		if (IS_ERR(pid))
-			return PTR_ERR(pid);
-		break;
-	default:
+	if (pidfd < 0)
 		return -EINVAL;
-	}
+
+	pid = pidfd_get_pid(pidfd);
+	if (IS_ERR(pid))
+		return PTR_ERR(pid);
 
 	task = get_pid_task(pid, PIDTYPE_PID);
 	if (!task) {
@@ -1292,9 +1278,8 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	return ret;
 }
 
-SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
-		const struct iovec __user *, vec, unsigned long, vlen,
-		int, behavior, unsigned long, flags)
+SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
+		unsigned long, vlen, int, behavior, unsigned int, flags)
 {
 	ssize_t ret;
 	struct iovec iovstack[UIO_FASTIOV];
@@ -1303,19 +1288,18 @@ SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
 
 	ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(which, upid, &iter, behavior, flags);
+		ret = do_process_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
 }
 
 #ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
-			compat_pid_t, upid,
+COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
 			const struct compat_iovec __user *, vec,
 			compat_ulong_t, vlen,
 			compat_int_t, behavior,
-			compat_ulong_t, flags)
+			compat_int_t, flags)
 
 {
 	ssize_t ret;
@@ -1326,7 +1310,7 @@ COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
 	ret = compat_import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack),
 				&iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(which, upid, &iter, behavior, flags);
+		ret = do_process_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
Signature · 1 line
2.26.2.761.g0e0b3e54be-goog
Reviewed-by on v1 1/1
#1 lore raw
Quoted 27 lines · Minchan Kim

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

Quoted 7 lines

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and

There is no idtype parameter anymore, so maybe just "if the target process is external"?

Quoted 120 lines

id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +13 -30
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
        return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-                                      int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+                               int behavior, unsigned int flags)
 {
        ssize_t ret;
        struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
        if (flags != 0)
                return -EINVAL;
 
-       switch (which) {
-       case P_PID:
-               if (upid <= 0)
-                       return -EINVAL;
-
-               pid = find_get_pid(upid);
-               if (!pid)
-                       return -ESRCH;
-               break;
-       case P_PIDFD:
-               if (upid < 0)
-                       return -EINVAL;
-
-               pid = pidfd_get_pid(upid);
-               if (IS_ERR(pid))
-                       return PTR_ERR(pid);
-               break;
-       default:
+       if (pidfd < 0)
                return -EINVAL;
-       }
+
+       pid = pidfd_get_pid(pidfd);
+       if (IS_ERR(pid))
+               return PTR_ERR(pid);
 
        task = get_pid_task(pid, PIDTYPE_PID);
        if (!task) {
@@ -1292,9 +1278,8 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
        return ret;
 }
 
-SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
-               const struct iovec __user *, vec, unsigned long, vlen,
-               int, behavior, unsigned long, flags)
+SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
+               unsigned long, vlen, int, behavior, unsigned int, flags)
 {
        ssize_t ret;
        struct iovec iovstack[UIO_FASTIOV];
@@ -1303,19 +1288,18 @@ SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
 
        ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
        if (ret >= 0) {
-               ret = do_process_madvise(which, upid, &iter, behavior, flags);
+               ret = do_process_madvise(pidfd, &iter, behavior, flags);
                kfree(iov);
        }
        return ret;
 }
 
 #ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
-                       compat_pid_t, upid,
+COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
                        const struct compat_iovec __user *, vec,
                        compat_ulong_t, vlen,
                        compat_int_t, behavior,
-                       compat_ulong_t, flags)
+                       compat_int_t, flags)
 
 {
        ssize_t ret;
@@ -1326,7 +1310,7 @@ COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
        ret = compat_import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack),
                                &iov, &iter);
        if (ret >= 0) {
-               ret = do_process_madvise(which, upid, &iter, behavior, flags);
+               ret = do_process_madvise(pidfd, &iter, behavior, flags);
                kfree(iov);
        }
        return ret;
--
Quoted 2 lines

2.26.2.761.g0e0b3e54be-goog

author Reviewed-by Suren Baghdasaryan
#2 lore raw
Quoted 31 lines · Suren Baghdasaryan
Quoted 30 lines · Minchan Kim

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

Quoted 7 lines
Quoted 4 lines

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and

There is no idtype parameter anymore, so maybe just "if the target process is external"?

Fixed.

Quoted 109 lines
Quoted 109 lines

id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +12 -28
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
        return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-                                      int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+                               int behavior, unsigned int flags)
 {
        ssize_t ret;
        struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
        if (flags != 0)
                return -EINVAL;
 
-       switch (which) {
-       case P_PID:
-               if (upid <= 0)
-                       return -EINVAL;
-
-               pid = find_get_pid(upid);
-               if (!pid)
-                       return -ESRCH;
-               break;
-       case P_PIDFD:
-               if (upid < 0)
-                       return -EINVAL;
-
-               pid = pidfd_get_pid(upid);
-               if (IS_ERR(pid))
-                       return PTR_ERR(pid);
-               break;
-       default:
+       if (pidfd < 0)
                return -EINVAL;
-       }
+
+       pid = pidfd_get_pid(pidfd);
+       if (IS_ERR(pid))
+               return PTR_ERR(pid);
 
        task = get_pid_task(pid, PIDTYPE_PID);
        if (!task) {
@@ -1292,9 +1278,8 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
        return ret;
 }
 
-SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
-               const struct iovec __user *, vec, unsigned long, vlen,
-               int, behavior, unsigned long, flags)
+SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
+               unsigned long, vlen, int, behavior, unsigned int, flags)
 {
        ssize_t ret;
        struct iovec iovstack[UIO_FASTIOV];
@@ -1303,19 +1288,18 @@ SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
 
        ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
        if (ret >= 0) {
-               ret = do_process_madvise(which, upid, &iter, behavior, flags);
+               ret = do_process_madvise(pidfd, &iter, behavior, flags);
                kfree(iov);
        }
        return ret;
 }
 
 #ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
-                       compat_pid_t, upid,
+COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
                        const struct compat_iovec __user *, vec,
                        compat_ulong_t, vlen,
                        compat_int_t, behavior,
-                       compat_ulong_t, flags)
+                       compat_int_t, flags)
 
Quoted 16 lines
Quoted 12 lines
mm/madvise.c +1 -2
 {
        ssize_t ret;
@@ -1326,7 +1310,7 @@ COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
        ret = compat_import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack),
                                &iov, &iter);
        if (ret >= 0) {
-               ret = do_process_madvise(which, upid, &iter, behavior, flags);
+               ret = do_process_madvise(pidfd, &iter, behavior, flags);
                kfree(iov);
        }
        return ret;
--
Quoted 2 lines

2.26.2.761.g0e0b3e54be-goog

Thanks for the review, Suren.

Andrew, I sent this patch without folding into previous syscall introducing patches because it could be arguable. If you want to fold it into each patchset(i.e., introdcuing process_madvise syscall and introducing compat_syscall), let me know it. I will send partial diff to each patchset.

From 1f5d3c22426b79b76690145102c6ec905a405790 Mon Sep 17 00:00:00 2001
From: Minchan Kim <[email protected]>
Date: Fri, 15 May 2020 17:48:08 -0700
Subject: [PATCH v2] mm: use only pidfd for process_madvise syscall

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +13 -29
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-				       int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+				int behavior, unsigned int flags)
 {
 	ssize_t ret;
 	struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	if (flags != 0)
 		return -EINVAL;
 
-	switch (which) {
-	case P_PID:
-		if (upid <= 0)
-			return -EINVAL;
-
-		pid = find_get_pid(upid);
-		if (!pid)
-			return -ESRCH;
-		break;
-	case P_PIDFD:
-		if (upid < 0)
-			return -EINVAL;
-
-		pid = pidfd_get_pid(upid);
-		if (IS_ERR(pid))
-			return PTR_ERR(pid);
-		break;
-	default:
+	if (pidfd < 0)
 		return -EINVAL;
-	}
+
+	pid = pidfd_get_pid(pidfd);
+	if (IS_ERR(pid))
+		return PTR_ERR(pid);
 
 	task = get_pid_task(pid, PIDTYPE_PID);
 	if (!task) {
@@ -1292,9 +1278,8 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	return ret;
 }
 
-SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
-		const struct iovec __user *, vec, unsigned long, vlen,
-		int, behavior, unsigned long, flags)
+SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
+		unsigned long, vlen, int, behavior, unsigned int, flags)
 {
 	ssize_t ret;
 	struct iovec iovstack[UIO_FASTIOV];
@@ -1303,19 +1288,18 @@ SYSCALL_DEFINE6(process_madvise, int, which, pid_t, upid,
 
 	ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(which, upid, &iter, behavior, flags);
+		ret = do_process_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
 }
 
 #ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
-			compat_pid_t, upid,
+COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
 			const struct compat_iovec __user *, vec,
 			compat_ulong_t, vlen,
 			compat_int_t, behavior,
-			compat_ulong_t, flags)
+			compat_int_t, flags)
 
 {
 	ssize_t ret;
@@ -1326,7 +1310,7 @@ COMPAT_SYSCALL_DEFINE6(process_madvise, compat_int_t, which,
 	ret = compat_import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack),
 				&iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(which, upid, &iter, behavior, flags);
+		ret = do_process_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
Signature · 1 line
2.26.2.761.g0e0b3e54be-goog
maintainer
#3 lore raw
Quoted 5 lines · On Mon, 18 May 2020 14:13:50 -0700 Minchan Kim <[email protected]> wrote:

Andrew, I sent this patch without folding into previous syscall introducing patches because it could be arguable. If you want to fold it into each patchset(i.e., introdcuing process_madvise syscall and introducing compat_syscall), let me know it. I will send partial diff to each patchset.

It doesn't seem necessary - I believe we'll get a clean result if I squish all of these:

mm-support-vector-address-ranges-for-process_madvise-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix-fix-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix-fix-fix-fix.patch mm-use-only-pidfd-for-process_madvise-syscall.patch

into mm-support-vector-address-ranges-for-process_madvise.patch and make the appropriate changelog adjustments?

author Reviewed-by Suren Baghdasaryan
#4 lore raw

Hi Andrew,

Quoted 18 lines · Andrew Morton
Quoted 5 lines · On Mon, 18 May 2020 14:13:50 -0700 Minchan Kim <[email protected]> wrote:

Andrew, I sent this patch without folding into previous syscall introducing patches because it could be arguable. If you want to fold it into each patchset(i.e., introdcuing process_madvise syscall and introducing compat_syscall), let me know it. I will send partial diff to each patchset.

It doesn't seem necessary - I believe we'll get a clean result if I squish all of these:

mm-support-vector-address-ranges-for-process_madvise-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix-fix-fix.patch mm-support-vector-address-ranges-for-process_madvise-fix-fix-fix-fix-fix.patch mm-use-only-pidfd-for-process_madvise-syscall.patch

Quoted 3 lines

into mm-support-vector-address-ranges-for-process_madvise.patch and make the appropriate changelog adjustments?

If you want to fold them all, please use the description below for mm-support-vector-address-ranges-for-process_madvise.patch.

Thanks.

============== &< ===================

Subject: [PATCH] mm: support vector address ranges for process_madvise

This patch changes process_madvise interface:
a) support vector address ranges in a system call
b) support the vector address ranges to local process as well as external process
c) remove pid but keep only pidfd in argument - [1][2]
d) change type of flags with unsgined int

Android app has thousands of vmas due to zygote so it's totally waste of CPU and power if we should call the syscall one by one for each vma. (With testing 2000-vma syscall vs 1-vector syscall, it showed 15% performance improvement. I think it would be bigger in real practice because the testing ran very cache friendly environment).

Another potential use case for the vector range is to amortize the cost of TLB shootdowns for multiple ranges when using MADV_DONTNEED; this could benefit users like TCP receive zerocopy and malloc implementations. In future, we could find more usecases for other advises so let's make it happens as API since we introduce a new syscall at this moment. With that, existing madvise(2) user could replace it with process_madvise(2) with their own pid if they want to have batch address ranges support feature.

So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by pidfd is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

Reviewed-by Suren Baghdasaryan on v1 1/1
#5 lore raw
In reply to Minchan Kim
Quoted 54 lines · Minchan Kim

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

Quoted 3 lines

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

Thanks for the ping Minchan, and sorry for not replying earlier to this.

Also, sorry that i delayed the patch but this here really seems a way cleaner api to me and feels less hackish. In general this patch seems fine to me.
But two comments below:

Quoted 33 lines
---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +2 -12
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-				       int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+				int behavior, unsigned int flags)
 {
 	ssize_t ret;
 	struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	if (flags != 0)
 		return -EINVAL;
 
-	switch (which) {
-	case P_PID:
-		if (upid <= 0)
-			return -EINVAL;
-
-		pid = find_get_pid(upid);
-		if (!pid)
-			return -ESRCH;
-		break;
-	case P_PIDFD:

mm/madvise.c static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter, · line 1245

Quoted 10 lines
mm/madvise.c +1 -8
-		if (upid < 0)
-			return -EINVAL;
-
-		pid = pidfd_get_pid(upid);
-		if (IS_ERR(pid))
-			return PTR_ERR(pid);
-		break;
-	default:
+	if (pidfd < 0)
 		return -EINVAL;

When garbage file descriptors are passed EBADF needs to be returned, not EINVAL. That's the case with most apis and also with pidfds, compare:

int main(void)
{

	if (syscall(__NR_openat, -1, "tmp", O_RDONLY) < 0)
		printf("%m - openat(-1)\n");

	if (syscall(__NR_fcntl, -1, F_GETFL) < 0)
		printf("%m - fcntl(-1)\n");

	if (syscall(__NR_dup, -1) < 0)
		printf("%m - dup(-1)\n");

	if (syscall(__NR_close, -1) < 0)
		printf("%m - close(-1)\n");

	if (syscall(__NR_pidfd_getfd, -1, 0) < 0)
		printf("%m - pidfd_getfd(-1)\n");

	if (syscall(__NR_pidfd_send_signal, -1, 0, NULL, 0) < 0)
		printf("%m - pidfd_getfd(-1)\n");

	exit(EXIT_SUCCESS);
}

which all give:

Bad file descriptor - openat(-1)
Bad file descriptor - fcntl(-1)
Bad file descriptor - dup(-1)
Bad file descriptor - close(-1)
Bad file descriptor - pidfd_getfd(-1)
Bad file descriptor - pidfd_send_signal(-1)

In addition, I have one more request/proposal. I really think that consistent api naming is something we should consider so I'd propose:

From eef84697f1f19be4de1416d7d86975c319620569 Mon Sep 17 00:00:00 2001
From: Christian Brauner <[email protected]>
Date: Tue, 19 May 2020 08:26:07 +0200
Subject: [PATCH] mm: retain consistent pidfd syscall naming

I think there's value in consistent api naming. So far we have pidfd_open(), pidfd_send_signal(), pidfd_getfd(), likely there's going to be a pidfd_addfd() in the near future. process_madvise() would currently be the only syscall not fitting in that set. It probably doesn't matter much whether we call it process_madvise() or pidfd_madvise() and as long as noone has serious problems with renaming this I'd vote to keep naming consistent. This is also way nicer for userspace.

I tested this patch on the akpm branch:
ssh://[email protected]/pub/scm/linux/kernel/git/next/linux-next.git akpm

It can be applied directly on top of Minchan's patch:

https://lore.kernel.org/lkml/[email protected]

Here's also the adapted commit message:

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t pidfd_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The pidfd_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The pidfd_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use pidfd_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, pidfd_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

---
 arch/alpha/kernel/syscalls/syscall.tbl      |  2 +-
 arch/arm/tools/syscall.tbl                  |  2 +-
 arch/arm64/include/asm/unistd32.h           |  4 ++--
 arch/ia64/kernel/syscalls/syscall.tbl       |  2 +-
 arch/m68k/kernel/syscalls/syscall.tbl       |  2 +-
 arch/microblaze/kernel/syscalls/syscall.tbl |  2 +-
 arch/mips/kernel/syscalls/syscall_n32.tbl   |  2 +-
 arch/mips/kernel/syscalls/syscall_n64.tbl   |  2 +-
 arch/mips/kernel/syscalls/syscall_o32.tbl   |  2 +-
 arch/parisc/kernel/syscalls/syscall.tbl     |  2 +-
 arch/powerpc/kernel/syscalls/syscall.tbl    |  2 +-
 arch/s390/kernel/syscalls/syscall.tbl       |  2 +-
 arch/sh/kernel/syscalls/syscall.tbl         |  2 +-
 arch/sparc/kernel/syscalls/syscall.tbl      |  2 +-
 arch/x86/entry/syscalls/syscall_32.tbl      |  2 +-
 arch/x86/entry/syscalls/syscall_64.tbl      |  4 ++--
 arch/xtensa/kernel/syscalls/syscall.tbl     |  2 +-
 include/linux/compat.h                      |  2 +-
 include/linux/syscalls.h                    |  2 +-
 include/uapi/asm-generic/unistd.h           |  6 +++---
 kernel/sys_ni.c                             |  4 ++--
 mm/madvise.c                                | 18 +++++++++---------
 22 files changed, 35 insertions(+), 35 deletions(-)
arch/alpha/kernel/syscalls/syscall.tbl +1 -1
index b249824c8267..2e125f20e182 100644
@@ -481,4 +481,4 @@
 550	common	watch_mount			sys_watch_mount
 551	common	watch_sb			sys_watch_sb
 552	common	fsinfo				sys_fsinfo
-553	common	process_madvise			sys_process_madvise
+553	common	pidfd_madvise			sys_pidfd_madvise
arch/arm/tools/syscall.tbl +1 -1
index 7b3832d920ee..e65949d0019a 100644
@@ -455,4 +455,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/arm64/include/asm/unistd32.h +2 -2
index f8dafe900cb9..0f18d84f5ec0 100644
@@ -891,8 +891,8 @@ __SYSCALL(__NR_watch_mount, sys_watch_mount)
 __SYSCALL(__NR_watch_sb, sys_watch_sb)
 #define __NR_fsinfo 442
 __SYSCALL(__NR_fsinfo, sys_fsinfo)
-#define __NR_process_madvise 443
-__SYSCALL(__NR_process_madvise, compat_sys_process_madvise)
+#define __NR_pidfd_madvise 443
+__SYSCALL(__NR_pidfd_madvise, compat_sys_pidfd_madvise)
 
 /*
  * Please add new compat syscalls above this comment and update
arch/ia64/kernel/syscalls/syscall.tbl +1 -1
index 6636a1aef36b..494b8af8207d 100644
@@ -362,4 +362,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/m68k/kernel/syscalls/syscall.tbl +1 -1
index 8cd84a7fbf57..e7397ab813d3 100644
@@ -441,4 +441,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/microblaze/kernel/syscalls/syscall.tbl +1 -1
index f581a02257ec..f70a0abcbe99 100644
@@ -447,4 +447,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/mips/kernel/syscalls/syscall_n32.tbl +1 -1
index c85bdc3abe55..b6a992f2a255 100644
@@ -380,4 +380,4 @@
 440	n32	watch_mount			sys_watch_mount
 441	n32	watch_sb			sys_watch_sb
 442	n32	fsinfo				sys_fsinfo
-443	n32	process_madvise			compat_sys_process_madvise
+443	n32	pidfd_madvise			compat_sys_pidfd_madvise
arch/mips/kernel/syscalls/syscall_n64.tbl +1 -1
index 9e08c40d3cb9..8638beaf724d 100644
@@ -356,4 +356,4 @@
 440	n64	watch_mount			sys_watch_mount
 441	n64	watch_sb			sys_watch_sb
 442	n64	fsinfo				sys_fsinfo
-443	n64	process_madvise			sys_process_madvise
+443	n64	pidfd_madvise			sys_pidfd_madvise
arch/mips/kernel/syscalls/syscall_o32.tbl +1 -1
index a2b591d7c874..7228e9c91263 100644
@@ -429,4 +429,4 @@
 440	o32	watch_mount			sys_watch_mount
 441	o32	watch_sb			sys_watch_sb
 442	o32	fsinfo				sys_fsinfo
-443	o32	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	o32	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/parisc/kernel/syscalls/syscall.tbl +1 -1
index 98e7442e4d7a..14991103f5e2 100644
@@ -439,4 +439,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/powerpc/kernel/syscalls/syscall.tbl +1 -1
index 72fb9dd88e16..a4a286c222aa 100644
@@ -531,4 +531,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/s390/kernel/syscalls/syscall.tbl +1 -1
index b731fcb03f1b..ec11df661285 100644
@@ -444,4 +444,4 @@
 440	common	watch_mount		sys_watch_mount			sys_watch_mount
 441	common	watch_sb		sys_watch_sb			sys_watch_sb
 442  common	fsinfo			sys_fsinfo			sys_fsinfo
-443  common	process_madvise		sys_process_madvise		compat_sys_process_madvise
+443  common	pidfd_madvise		sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/sh/kernel/syscalls/syscall.tbl +1 -1
index e7a4804e23cf..62c4926e3150 100644
@@ -444,4 +444,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/sparc/kernel/syscalls/syscall.tbl +1 -1
index d6126ee2f51d..50ccfc41d3a0 100644
@@ -487,4 +487,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/x86/entry/syscalls/syscall_32.tbl +1 -1
index 686d59df1e12..6b16b95c58a0 100644
@@ -446,4 +446,4 @@
 440	i386	watch_mount		sys_watch_mount
 441	i386	watch_sb		sys_watch_sb
 442	i386	fsinfo			sys_fsinfo
-443	i386	process_madvise		sys_process_madvise		compat_sys_process_madvise
+443	i386	pidfd_madvise		sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/x86/entry/syscalls/syscall_64.tbl +2 -2
index b345b35b3308..3662a0f4b682 100644
@@ -363,7 +363,7 @@
 440	common	watch_mount		sys_watch_mount
 441	common	watch_sb		sys_watch_sb
 442	common	fsinfo			sys_fsinfo
-443	64	process_madvise		sys_process_madvise
+443	64	pidfd_madvise		sys_pidfd_madvise
 
 #
 # x32-specific system call numbers start at 512 to avoid cache impact
@@ -407,4 +407,4 @@
 545	x32	execveat		compat_sys_execveat
 546	x32	preadv2			compat_sys_preadv64v2
 547	x32	pwritev2		compat_sys_pwritev64v2
-548	x32	process_madvise		compat_sys_process_madvise
+548	x32	pidfd_madvise		compat_sys_pidfd_madvise
arch/xtensa/kernel/syscalls/syscall.tbl +1 -1
index 96cb07095e4c..a2510c81ab35 100644
@@ -412,4 +412,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
include/linux/compat.h +1 -1
index 2e2f0a2700ab..f9eca29244f5 100644
@@ -827,7 +827,7 @@ asmlinkage long compat_sys_pwritev64v2(unsigned long fd,
 		unsigned long vlen, loff_t pos, rwf_t flags);
 #endif
 
-asmlinkage ssize_t compat_sys_process_madvise(compat_int_t which,
+asmlinkage ssize_t compat_sys_pidfd_madvise(compat_int_t which,
 		compat_pid_t upid, const struct compat_iovec __user *vec,
 		compat_ulong_t vlen, compat_int_t behavior,
 		compat_ulong_t flags);
include/linux/syscalls.h +1 -1
index 2cbd3660e8e6..5c4949606502 100644
@@ -879,7 +879,7 @@ asmlinkage long sys_munlockall(void);
 asmlinkage long sys_mincore(unsigned long start, size_t len,
 				unsigned char __user * vec);
 asmlinkage long sys_madvise(unsigned long start, size_t len, int behavior);
-asmlinkage long sys_process_madvise(int which, pid_t upid,
+asmlinkage long sys_pidfd_madvise(int which, pid_t upid,
 		const struct iovec __user *vec, unsigned long vlen,
 		int behavior, unsigned long flags);
 asmlinkage long sys_remap_file_pages(unsigned long start, unsigned long size,
include/uapi/asm-generic/unistd.h +3 -3
index 6a5856c9d9b4..40dd1b66858e 100644
@@ -863,9 +863,9 @@ __SYSCALL(__NR_watch_mount, sys_watch_mount)
 __SYSCALL(__NR_watch_sb, sys_watch_sb)
 #define __NR_fsinfo 442
 __SYSCALL(__NR_fsinfo, sys_fsinfo)
-#define __NR_process_madvise 443
-__SC_COMP(__NR_process_madvise, sys_process_madvise, \
-		compat_sys_process_madvise)
+#define __NR_pidfd_madvise 443
+__SC_COMP(__NR_pidfd_madvise, sys_pidfd_madvise, \
+		compat_sys_pidfd_madvise)
 
 #undef __NR_syscalls
 #define __NR_syscalls 444
kernel/sys_ni.c +2 -2
index fad48ace1123..6b7cc69617f8 100644
@@ -287,8 +287,8 @@ COND_SYSCALL(mlockall);
 COND_SYSCALL(munlockall);
 COND_SYSCALL(mincore);
 COND_SYSCALL(madvise);
-COND_SYSCALL(process_madvise);
-COND_SYSCALL_COMPAT(process_madvise);
+COND_SYSCALL(pidfd_madvise);
+COND_SYSCALL_COMPAT(pidfd_madvise);
 COND_SYSCALL(remap_file_pages);
 COND_SYSCALL(mbind);
 COND_SYSCALL_COMPAT(mbind);
mm/madvise.c +9 -9
index 35c9b220146a..d8b9e9c54275 100644
@@ -1008,7 +1008,7 @@ madvise_behavior_valid(int behavior)
 }
 
 static bool
-process_madvise_behavior_valid(int behavior)
+pidfd_madvise_behavior_valid(int behavior)
 {
 	switch (behavior) {
 	case MADV_COLD:
@@ -1211,7 +1211,7 @@ SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
 	return do_madvise(current, current->mm, start, len_in, behavior);
 }
 
-static int process_madvise_vec(struct task_struct *target_task,
+static int pidfd_madvise_vec(struct task_struct *target_task,
 		struct mm_struct *mm, struct iov_iter *iter, int behavior)
 {
 	struct iovec iovec;
@@ -1229,7 +1229,7 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+static ssize_t do_pidfd_madvise(int pidfd, struct iov_iter *iter,
 				int behavior, unsigned int flags)
 {
 	ssize_t ret;
@@ -1255,7 +1255,7 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 	}
 
 	if (task->mm != current->mm &&
-			!process_madvise_behavior_valid(behavior)) {
+			!pidfd_madvise_behavior_valid(behavior)) {
 		ret = -EINVAL;
 		goto release_task;
 	}
@@ -1266,7 +1266,7 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 		goto release_task;
 	}
 
-	ret = process_madvise_vec(task, mm, iter, behavior);
+	ret = pidfd_madvise_vec(task, mm, iter, behavior);
 	if (ret >= 0)
 		ret = total_len - iov_iter_count(iter);
 
@@ -1278,7 +1278,7 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 	return ret;
 }
 
-SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
+SYSCALL_DEFINE5(pidfd_madvise, int, pidfd, const struct iovec __user *, vec,
 		unsigned long, vlen, int, behavior, unsigned int, flags)
 {
 	ssize_t ret;
@@ -1288,14 +1288,14 @@ SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
 
 	ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(pidfd, &iter, behavior, flags);
+		ret = do_pidfd_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
 }
 
 #ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
+COMPAT_SYSCALL_DEFINE5(pidfd_madvise, compat_int_t, pidfd,
 			const struct compat_iovec __user *, vec,
 			compat_ulong_t, vlen,
 			compat_int_t, behavior,
@@ -1310,7 +1310,7 @@ COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
 	ret = compat_import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack),
 				&iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(pidfd, &iter, behavior, flags);
+		ret = do_pidfd_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
Signature · 1 line
2.26.2
#6 lore raw

Hi Christian,

Quoted 59 lines · Christian Brauner
Quoted 54 lines · Minchan Kim

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

Quoted 3 lines

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

Quoted 5 lines

Thanks for the ping Minchan, and sorry for not replying earlier to this.

Also, sorry that i delayed the patch but this here really seems a way cleaner api to me and feels less hackish. In general this patch seems fine to me.

Thanks.

Quoted 40 lines

But two comments below:

Quoted 38 lines
---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +2 -17
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-				       int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+				int behavior, unsigned int flags)
 {
 	ssize_t ret;
 	struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	if (flags != 0)
 		return -EINVAL;
 
-	switch (which) {
-	case P_PID:
-		if (upid <= 0)
-			return -EINVAL;
-
-		pid = find_get_pid(upid);
-		if (!pid)
-			return -ESRCH;
-		break;
-	case P_PIDFD:
-		if (upid < 0)
-			return -EINVAL;
-
-		pid = pidfd_get_pid(upid);
-		if (IS_ERR(pid))
Quoted 8 lines

mm/madvise.c static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter, · line 1245

Quoted 5 lines
mm/madvise.c +1 -3
-			return PTR_ERR(pid);
-		break;
-	default:
+	if (pidfd < 0)
 		return -EINVAL;

When garbage file descriptors are passed EBADF needs to be returned, not EINVAL. That's the case with most apis and also with pidfds, compare:

True. Let me cook a patch for that.

Quoted 42 lines

int main(void)
{

	if (syscall(__NR_openat, -1, "tmp", O_RDONLY) < 0)
		printf("%m - openat(-1)\n");

	if (syscall(__NR_fcntl, -1, F_GETFL) < 0)
		printf("%m - fcntl(-1)\n");

	if (syscall(__NR_dup, -1) < 0)
		printf("%m - dup(-1)\n");

	if (syscall(__NR_close, -1) < 0)
		printf("%m - close(-1)\n");

	if (syscall(__NR_pidfd_getfd, -1, 0) < 0)
		printf("%m - pidfd_getfd(-1)\n");

	if (syscall(__NR_pidfd_send_signal, -1, 0, NULL, 0) < 0)
		printf("%m - pidfd_getfd(-1)\n");

	exit(EXIT_SUCCESS);
}

which all give:

Bad file descriptor - openat(-1)
Bad file descriptor - fcntl(-1)
Bad file descriptor - dup(-1)
Bad file descriptor - close(-1)
Bad file descriptor - pidfd_getfd(-1)
Bad file descriptor - pidfd_send_signal(-1)

In addition, I have one more request/proposal. I really think that consistent api naming is something we should consider so I'd propose:

From eef84697f1f19be4de1416d7d86975c319620569 Mon Sep 17 00:00:00 2001
From: Christian Brauner <[email protected]>
Date: Tue, 19 May 2020 08:26:07 +0200
Subject: [PATCH] mm: retain consistent pidfd syscall naming

Quoted 8 lines

I think there's value in consistent api naming. So far we have pidfd_open(), pidfd_send_signal(), pidfd_getfd(), likely there's going to be a pidfd_addfd() in the near future. process_madvise() would currently be the only syscall not fitting in that set. It probably doesn't matter much whether we call it process_madvise() or pidfd_madvise() and as long as noone has serious problems with renaming this I'd vote to keep naming consistent. This is also way nicer for userspace.

I agree if the syscall aimed for creating the particular fd(e.g., pidfd_open, pidfd_getfd, timefd_create, memfd_create, eventfd, signalfd). However, this process_madvise is not the case. Since fd is general concept to cover various type's resources's handle, API naming doesn't follow the particular type since they are created/opened in general.
I think it's more consistent with other APIs in the Linux since we follow general concept.

Quoted 424 lines

I tested this patch on the akpm branch:
ssh://[email protected]/pub/scm/linux/kernel/git/next/linux-next.git akpm

It can be applied directly on top of Minchan's patch:

https://lore.kernel.org/lkml/[email protected]

Here's also the adapted commit message:

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t pidfd_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The pidfd_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The pidfd_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use pidfd_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, pidfd_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

---
 arch/alpha/kernel/syscalls/syscall.tbl      |  2 +-
 arch/arm/tools/syscall.tbl                  |  2 +-
 arch/arm64/include/asm/unistd32.h           |  4 ++--
 arch/ia64/kernel/syscalls/syscall.tbl       |  2 +-
 arch/m68k/kernel/syscalls/syscall.tbl       |  2 +-
 arch/microblaze/kernel/syscalls/syscall.tbl |  2 +-
 arch/mips/kernel/syscalls/syscall_n32.tbl   |  2 +-
 arch/mips/kernel/syscalls/syscall_n64.tbl   |  2 +-
 arch/mips/kernel/syscalls/syscall_o32.tbl   |  2 +-
 arch/parisc/kernel/syscalls/syscall.tbl     |  2 +-
 arch/powerpc/kernel/syscalls/syscall.tbl    |  2 +-
 arch/s390/kernel/syscalls/syscall.tbl       |  2 +-
 arch/sh/kernel/syscalls/syscall.tbl         |  2 +-
 arch/sparc/kernel/syscalls/syscall.tbl      |  2 +-
 arch/x86/entry/syscalls/syscall_32.tbl      |  2 +-
 arch/x86/entry/syscalls/syscall_64.tbl      |  4 ++--
 arch/xtensa/kernel/syscalls/syscall.tbl     |  2 +-
 include/linux/compat.h                      |  2 +-
 include/linux/syscalls.h                    |  2 +-
 include/uapi/asm-generic/unistd.h           |  6 +++---
 kernel/sys_ni.c                             |  4 ++--
 mm/madvise.c                                | 18 +++++++++---------
 22 files changed, 35 insertions(+), 35 deletions(-)
arch/alpha/kernel/syscalls/syscall.tbl +1 -1
index b249824c8267..2e125f20e182 100644
@@ -481,4 +481,4 @@
 550	common	watch_mount			sys_watch_mount
 551	common	watch_sb			sys_watch_sb
 552	common	fsinfo				sys_fsinfo
-553	common	process_madvise			sys_process_madvise
+553	common	pidfd_madvise			sys_pidfd_madvise
arch/arm/tools/syscall.tbl +1 -1
index 7b3832d920ee..e65949d0019a 100644
@@ -455,4 +455,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/arm64/include/asm/unistd32.h +2 -2
index f8dafe900cb9..0f18d84f5ec0 100644
@@ -891,8 +891,8 @@ __SYSCALL(__NR_watch_mount, sys_watch_mount)
 __SYSCALL(__NR_watch_sb, sys_watch_sb)
 #define __NR_fsinfo 442
 __SYSCALL(__NR_fsinfo, sys_fsinfo)
-#define __NR_process_madvise 443
-__SYSCALL(__NR_process_madvise, compat_sys_process_madvise)
+#define __NR_pidfd_madvise 443
+__SYSCALL(__NR_pidfd_madvise, compat_sys_pidfd_madvise)
 
 /*
  * Please add new compat syscalls above this comment and update
arch/ia64/kernel/syscalls/syscall.tbl +1 -1
index 6636a1aef36b..494b8af8207d 100644
@@ -362,4 +362,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/m68k/kernel/syscalls/syscall.tbl +1 -1
index 8cd84a7fbf57..e7397ab813d3 100644
@@ -441,4 +441,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/microblaze/kernel/syscalls/syscall.tbl +1 -1
index f581a02257ec..f70a0abcbe99 100644
@@ -447,4 +447,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/mips/kernel/syscalls/syscall_n32.tbl +1 -1
index c85bdc3abe55..b6a992f2a255 100644
@@ -380,4 +380,4 @@
 440	n32	watch_mount			sys_watch_mount
 441	n32	watch_sb			sys_watch_sb
 442	n32	fsinfo				sys_fsinfo
-443	n32	process_madvise			compat_sys_process_madvise
+443	n32	pidfd_madvise			compat_sys_pidfd_madvise
arch/mips/kernel/syscalls/syscall_n64.tbl +1 -1
index 9e08c40d3cb9..8638beaf724d 100644
@@ -356,4 +356,4 @@
 440	n64	watch_mount			sys_watch_mount
 441	n64	watch_sb			sys_watch_sb
 442	n64	fsinfo				sys_fsinfo
-443	n64	process_madvise			sys_process_madvise
+443	n64	pidfd_madvise			sys_pidfd_madvise
arch/mips/kernel/syscalls/syscall_o32.tbl +1 -1
index a2b591d7c874..7228e9c91263 100644
@@ -429,4 +429,4 @@
 440	o32	watch_mount			sys_watch_mount
 441	o32	watch_sb			sys_watch_sb
 442	o32	fsinfo				sys_fsinfo
-443	o32	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	o32	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/parisc/kernel/syscalls/syscall.tbl +1 -1
index 98e7442e4d7a..14991103f5e2 100644
@@ -439,4 +439,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/powerpc/kernel/syscalls/syscall.tbl +1 -1
index 72fb9dd88e16..a4a286c222aa 100644
@@ -531,4 +531,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/s390/kernel/syscalls/syscall.tbl +1 -1
index b731fcb03f1b..ec11df661285 100644
@@ -444,4 +444,4 @@
 440	common	watch_mount		sys_watch_mount			sys_watch_mount
 441	common	watch_sb		sys_watch_sb			sys_watch_sb
 442  common	fsinfo			sys_fsinfo			sys_fsinfo
-443  common	process_madvise		sys_process_madvise		compat_sys_process_madvise
+443  common	pidfd_madvise		sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/sh/kernel/syscalls/syscall.tbl +1 -1
index e7a4804e23cf..62c4926e3150 100644
@@ -444,4 +444,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
arch/sparc/kernel/syscalls/syscall.tbl +1 -1
index d6126ee2f51d..50ccfc41d3a0 100644
@@ -487,4 +487,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise		compat_sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/x86/entry/syscalls/syscall_32.tbl +1 -1
index 686d59df1e12..6b16b95c58a0 100644
@@ -446,4 +446,4 @@
 440	i386	watch_mount		sys_watch_mount
 441	i386	watch_sb		sys_watch_sb
 442	i386	fsinfo			sys_fsinfo
-443	i386	process_madvise		sys_process_madvise		compat_sys_process_madvise
+443	i386	pidfd_madvise		sys_pidfd_madvise		compat_sys_pidfd_madvise
arch/x86/entry/syscalls/syscall_64.tbl +2 -2
index b345b35b3308..3662a0f4b682 100644
@@ -363,7 +363,7 @@
 440	common	watch_mount		sys_watch_mount
 441	common	watch_sb		sys_watch_sb
 442	common	fsinfo			sys_fsinfo
-443	64	process_madvise		sys_process_madvise
+443	64	pidfd_madvise		sys_pidfd_madvise
 
 #
 # x32-specific system call numbers start at 512 to avoid cache impact
@@ -407,4 +407,4 @@
 545	x32	execveat		compat_sys_execveat
 546	x32	preadv2			compat_sys_preadv64v2
 547	x32	pwritev2		compat_sys_pwritev64v2
-548	x32	process_madvise		compat_sys_process_madvise
+548	x32	pidfd_madvise		compat_sys_pidfd_madvise
arch/xtensa/kernel/syscalls/syscall.tbl +1 -1
index 96cb07095e4c..a2510c81ab35 100644
@@ -412,4 +412,4 @@
 440	common	watch_mount			sys_watch_mount
 441	common	watch_sb			sys_watch_sb
 442	common	fsinfo				sys_fsinfo
-443	common	process_madvise			sys_process_madvise
+443	common	pidfd_madvise			sys_pidfd_madvise
include/linux/compat.h +1 -1
index 2e2f0a2700ab..f9eca29244f5 100644
@@ -827,7 +827,7 @@ asmlinkage long compat_sys_pwritev64v2(unsigned long fd,
 		unsigned long vlen, loff_t pos, rwf_t flags);
 #endif
 
-asmlinkage ssize_t compat_sys_process_madvise(compat_int_t which,
+asmlinkage ssize_t compat_sys_pidfd_madvise(compat_int_t which,
 		compat_pid_t upid, const struct compat_iovec __user *vec,
 		compat_ulong_t vlen, compat_int_t behavior,
 		compat_ulong_t flags);
include/linux/syscalls.h +1 -1
index 2cbd3660e8e6..5c4949606502 100644
@@ -879,7 +879,7 @@ asmlinkage long sys_munlockall(void);
 asmlinkage long sys_mincore(unsigned long start, size_t len,
 				unsigned char __user * vec);
 asmlinkage long sys_madvise(unsigned long start, size_t len, int behavior);
-asmlinkage long sys_process_madvise(int which, pid_t upid,
+asmlinkage long sys_pidfd_madvise(int which, pid_t upid,
 		const struct iovec __user *vec, unsigned long vlen,
 		int behavior, unsigned long flags);
 asmlinkage long sys_remap_file_pages(unsigned long start, unsigned long size,
include/uapi/asm-generic/unistd.h +3 -3
index 6a5856c9d9b4..40dd1b66858e 100644
@@ -863,9 +863,9 @@ __SYSCALL(__NR_watch_mount, sys_watch_mount)
 __SYSCALL(__NR_watch_sb, sys_watch_sb)
 #define __NR_fsinfo 442
 __SYSCALL(__NR_fsinfo, sys_fsinfo)
-#define __NR_process_madvise 443
-__SC_COMP(__NR_process_madvise, sys_process_madvise, \
-		compat_sys_process_madvise)
+#define __NR_pidfd_madvise 443
+__SC_COMP(__NR_pidfd_madvise, sys_pidfd_madvise, \
+		compat_sys_pidfd_madvise)
 
 #undef __NR_syscalls
 #define __NR_syscalls 444
kernel/sys_ni.c +2 -2
index fad48ace1123..6b7cc69617f8 100644
@@ -287,8 +287,8 @@ COND_SYSCALL(mlockall);
 COND_SYSCALL(munlockall);
 COND_SYSCALL(mincore);
 COND_SYSCALL(madvise);
-COND_SYSCALL(process_madvise);
-COND_SYSCALL_COMPAT(process_madvise);
+COND_SYSCALL(pidfd_madvise);
+COND_SYSCALL_COMPAT(pidfd_madvise);
 COND_SYSCALL(remap_file_pages);
 COND_SYSCALL(mbind);
 COND_SYSCALL_COMPAT(mbind);
mm/madvise.c +9 -9
index 35c9b220146a..d8b9e9c54275 100644
@@ -1008,7 +1008,7 @@ madvise_behavior_valid(int behavior)
 }
 
 static bool
-process_madvise_behavior_valid(int behavior)
+pidfd_madvise_behavior_valid(int behavior)
 {
 	switch (behavior) {
 	case MADV_COLD:
@@ -1211,7 +1211,7 @@ SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
 	return do_madvise(current, current->mm, start, len_in, behavior);
 }
 
-static int process_madvise_vec(struct task_struct *target_task,
+static int pidfd_madvise_vec(struct task_struct *target_task,
 		struct mm_struct *mm, struct iov_iter *iter, int behavior)
 {
 	struct iovec iovec;
@@ -1229,7 +1229,7 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+static ssize_t do_pidfd_madvise(int pidfd, struct iov_iter *iter,
 				int behavior, unsigned int flags)
 {
 	ssize_t ret;
@@ -1255,7 +1255,7 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 	}
 
 	if (task->mm != current->mm &&
-			!process_madvise_behavior_valid(behavior)) {
+			!pidfd_madvise_behavior_valid(behavior)) {
 		ret = -EINVAL;
 		goto release_task;
 	}
@@ -1266,7 +1266,7 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 		goto release_task;
 	}
 
-	ret = process_madvise_vec(task, mm, iter, behavior);
+	ret = pidfd_madvise_vec(task, mm, iter, behavior);
 	if (ret >= 0)
 		ret = total_len - iov_iter_count(iter);
 
@@ -1278,7 +1278,7 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 	return ret;
 }
 
-SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
+SYSCALL_DEFINE5(pidfd_madvise, int, pidfd, const struct iovec __user *, vec,
 		unsigned long, vlen, int, behavior, unsigned int, flags)
 {
 	ssize_t ret;
@@ -1288,14 +1288,14 @@ SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
 
 	ret = import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(pidfd, &iter, behavior, flags);
+		ret = do_pidfd_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
 }
 
 #ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
+COMPAT_SYSCALL_DEFINE5(pidfd_madvise, compat_int_t, pidfd,
 			const struct compat_iovec __user *, vec,
 			compat_ulong_t, vlen,
 			compat_int_t, behavior,
@@ -1310,7 +1310,7 @@ COMPAT_SYSCALL_DEFINE5(process_madvise, compat_int_t, pidfd,
 	ret = compat_import_iovec(READ, vec, vlen, ARRAY_SIZE(iovstack),
 				&iov, &iter);
 	if (ret >= 0) {
-		ret = do_process_madvise(pidfd, &iter, behavior, flags);
+		ret = do_pidfd_madvise(pidfd, &iter, behavior, flags);
 		kfree(iov);
 	}
 	return ret;
Signature · 1 line
2.26.2
#7 lore raw
Quoted 119 lines · Minchan Kim

Hi Christian,

Quoted 59 lines · Christian Brauner
Quoted 54 lines · Minchan Kim

Based on discussion[1], people didn't feel we need to support both pid and pidfd for every new coming API[2] so this patch keeps only pidfd. This patch also changes flags's type with "unsigned int". So finally, the API is as follows,

ssize_t process_madvise(int pidfd, const struct iovec *iovec,
unsigned long vlen, int advice, unsigned int flags);

DESCRIPTION
The process_madvise() system call is used to give advice or directions to the kernel about the address ranges from external process as well as local process. It provides the advice to address ranges of process described by iovec and vlen. The goal of such advice is to improve system or application performance.

The pidfd selects the process referred to by the PID file descriptor specified in pidfd. (See pidofd_open(2) for further information)

The pointer iovec points to an array of iovec structures, defined in <sys/uio.h> as:

struct iovec {
void *iov_base; /* starting address */
size_t iov_len; /* number of bytes to be advised */
};

The iovec describes address ranges beginning at address(iov_base) and with size length of bytes(iov_len).

The vlen represents the number of elements in iovec.

The advice is indicated in the advice argument, which is one of the following at this moment if the target process specified by idtype and id is external.

MADV_COLD
MADV_PAGEOUT
MADV_MERGEABLE
MADV_UNMERGEABLE

Permission to provide a hint to external process is governed by a ptrace access mode PTRACE_MODE_ATTACH_FSCREDS check; see ptrace(2).

The process_madvise supports every advice madvise(2) has if target process is in same thread group with calling process so user could use process_madvise(2) to extend existing madvise(2) to support vector address ranges.

RETURN VALUE
On success, process_madvise() returns the number of bytes advised. This return value may be less than the total number of requested bytes, if an error occurred. The caller should check return value to determine whether a partial advice occurred.

Quoted 3 lines

[1] https://lore.kernel.org/linux-mm/20200509124817.xmrvsrq3mla6b76k@wittgenstein/
[2] https://lore.kernel.org/linux-mm/[email protected]/

Quoted 5 lines

Thanks for the ping Minchan, and sorry for not replying earlier to this.

Also, sorry that i delayed the patch but this here really seems a way cleaner api to me and feels less hackish. In general this patch seems fine to me.

Thanks.

Quoted 40 lines

But two comments below:

Quoted 38 lines
---
 mm/madvise.c | 42 +++++++++++++-----------------------------
 1 file changed, 13 insertions(+), 29 deletions(-)
mm/madvise.c +2 -17
index d3fbbe52d230..35c9b220146a 100644
@@ -1229,8 +1229,8 @@ static int process_madvise_vec(struct task_struct *target_task,
 	return ret;
 }
 
-static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
-				       int behavior, unsigned long flags)
+static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
+				int behavior, unsigned int flags)
 {
 	ssize_t ret;
 	struct pid *pid;
@@ -1241,26 +1241,12 @@ static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter,
 	if (flags != 0)
 		return -EINVAL;
 
-	switch (which) {
-	case P_PID:
-		if (upid <= 0)
-			return -EINVAL;
-
-		pid = find_get_pid(upid);
-		if (!pid)
-			return -ESRCH;
-		break;
-	case P_PIDFD:
-		if (upid < 0)
-			return -EINVAL;
-
-		pid = pidfd_get_pid(upid);
-		if (IS_ERR(pid))
Quoted 8 lines

mm/madvise.c static ssize_t do_process_madvise(int which, pid_t upid, struct iov_iter *iter, · line 1245

Quoted 5 lines
mm/madvise.c +1 -3
-			return PTR_ERR(pid);
-		break;
-	default:
+	if (pidfd < 0)
 		return -EINVAL;

When garbage file descriptors are passed EBADF needs to be returned, not EINVAL. That's the case with most apis and also with pidfds, compare:

Quoted 1 line

True. Let me cook a patch for that.

Hi Andrew,

Please fold this patch against on last patch you have for process_madvise. Thanks!

From 939e4c5b7ca12efc5d5eeb8ff55fc02752f70544 Mon Sep 17 00:00:00 2001
From: Minchan Kim <[email protected]>
Date: Tue, 19 May 2020 11:06:57 -0700
Subject: [PATCH] mm: return EBADF if pidfd is invalid

This patch makes returning of EBADF when the fd passed as argument is invalid. The implementaion relies on pidfd_get_pid's error return.

This patch also fixes syscall declare part since we removed pid support.

---
 include/linux/compat.h   | 6 +++---
 include/linux/syscalls.h | 5 ++---
 mm/madvise.c             | 3 ---
 3 files changed, 5 insertions(+), 9 deletions(-)
include/linux/compat.h +3 -3
index 2e2f0a2700ab..86b61e873947 100644
@@ -827,10 +827,10 @@ asmlinkage long compat_sys_pwritev64v2(unsigned long fd,
 		unsigned long vlen, loff_t pos, rwf_t flags);
 #endif
 
-asmlinkage ssize_t compat_sys_process_madvise(compat_int_t which,
-		compat_pid_t upid, const struct compat_iovec __user *vec,
+asmlinkage ssize_t compat_sys_process_madvise(compat_int_t pidfd,
+		const struct compat_iovec __user *vec,
 		compat_ulong_t vlen, compat_int_t behavior,
-		compat_ulong_t flags);
+		compat_int_t flags);
 
 /*
  * Deprecated system calls which are still defined in
include/linux/syscalls.h +2 -3
index 2cbd3660e8e6..63ffa6dc9da3 100644
@@ -879,9 +879,8 @@ asmlinkage long sys_munlockall(void);
 asmlinkage long sys_mincore(unsigned long start, size_t len,
 				unsigned char __user * vec);
 asmlinkage long sys_madvise(unsigned long start, size_t len, int behavior);
-asmlinkage long sys_process_madvise(int which, pid_t upid,
-		const struct iovec __user *vec, unsigned long vlen,
-		int behavior, unsigned long flags);
+asmlinkage long sys_process_madvise(int pidfd, const struct iovec __user *vec,
+		unsigned long vlen, int behavior, unsigned int flags);
 asmlinkage long sys_remap_file_pages(unsigned long start, unsigned long size,
 			unsigned long prot, unsigned long pgoff,
 			unsigned long flags);
mm/madvise.c +0 -3
index 35c9b220146a..3ac1eda1203f 100644
@@ -1241,9 +1241,6 @@ static ssize_t do_process_madvise(int pidfd, struct iov_iter *iter,
 	if (flags != 0)
 		return -EINVAL;
 
-	if (pidfd < 0)
-		return -EINVAL;
-
 	pid = pidfd_get_pid(pidfd);
 	if (IS_ERR(pid))
 		return PTR_ERR(pid);
Signature · 1 line
2.26.2.761.g0e0b3e54be-goog