path: root/system/easy-kernel/0250-expose-per-process-ksm.patch

From: Oleksandr Natalenko <oleksandr@natalenko.name>
Date: Mon, 8 May 2023 22:21:53 +0200
Subject: [PATCH] mm: expose per-process KSM control via syscalls

d7597f59d1d3 added a new API to enable per-process KSM control. It
however uses prctl, which doesn't allow controlling KSM from outside of
the current process.

Hence, expose this API via 3 syscalls: process_ksm_enable,
process_ksm_disable and process_ksm_status. Given sufficient privileges,
auto-KSM can be enable by another process.

Since these syscalls are not in the upstream kernel, also expose their
numbers under /sys/kernel/process_ksm so that userspace tooling like
uksmd knows how to use them.

Signed-off-by: Oleksandr Natalenko <oleksandr@natalenko.name>
---
 arch/alpha/kernel/syscalls/syscall.tbl      |   3 +
 arch/arm/tools/syscall.tbl                  |   3 +
 arch/arm64/include/asm/unistd.h             |   2 +-
 arch/arm64/include/asm/unistd32.h           |   6 +
 arch/ia64/kernel/syscalls/syscall.tbl       |   3 +
 arch/m68k/kernel/syscalls/syscall.tbl       |   3 +
 arch/microblaze/kernel/syscalls/syscall.tbl |   3 +
 arch/mips/kernel/syscalls/syscall_n32.tbl   |   3 +
 arch/mips/kernel/syscalls/syscall_n64.tbl   |   3 +
 arch/mips/kernel/syscalls/syscall_o32.tbl   |   3 +
 arch/parisc/kernel/syscalls/syscall.tbl     |   3 +
 arch/powerpc/kernel/syscalls/syscall.tbl    |   3 +
 arch/s390/kernel/syscalls/syscall.tbl       |   3 +
 arch/sh/kernel/syscalls/syscall.tbl         |   3 +
 arch/sparc/kernel/syscalls/syscall.tbl      |   3 +
 arch/x86/entry/syscalls/syscall_32.tbl      |   3 +
 arch/x86/entry/syscalls/syscall_64.tbl      |   3 +
 arch/xtensa/kernel/syscalls/syscall.tbl     |   3 +
 include/linux/syscalls.h                    |   3 +
 include/uapi/asm-generic/unistd.h           |  11 +-
 kernel/sys.c                                | 147 ++++++++++++++++++++
 kernel/sys_ni.c                             |   3 +
 22 files changed, 218 insertions(+), 2 deletions(-)

diff --git a/arch/alpha/kernel/syscalls/syscall.tbl b/arch/alpha/kernel/syscalls/syscall.tbl
index ad37569d0..9f4311e21 100644
--- a/arch/alpha/kernel/syscalls/syscall.tbl
+++ b/arch/alpha/kernel/syscalls/syscall.tbl
@@ -492,3 +492,6 @@
 560	common	set_mempolicy_home_node		sys_ni_syscall
 561	common	cachestat			sys_cachestat
 562	common	fchmodat2			sys_fchmodat2
+563	common	process_ksm_enable		sys_process_ksm_enable
+564	common	process_ksm_disable		sys_process_ksm_disable
+565	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/arm/tools/syscall.tbl b/arch/arm/tools/syscall.tbl
index c572d6c3d..8d40b3a45 100644
--- a/arch/arm/tools/syscall.tbl
+++ b/arch/arm/tools/syscall.tbl
@@ -466,3 +466,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/arm64/include/asm/unistd.h b/arch/arm64/include/asm/unistd.h
index bd77253b6..f33190f17 100644
--- a/arch/arm64/include/asm/unistd.h
+++ b/arch/arm64/include/asm/unistd.h
@@ -39,7 +39,7 @@
 #define __ARM_NR_compat_set_tls		(__ARM_NR_COMPAT_BASE + 5)
 #define __ARM_NR_COMPAT_END		(__ARM_NR_COMPAT_BASE + 0x800)
 
-#define __NR_compat_syscalls		453
+#define __NR_compat_syscalls		456
 #endif
 
 #define __ARCH_WANT_SYS_CLONE
diff --git a/arch/arm64/include/asm/unistd32.h b/arch/arm64/include/asm/unistd32.h
index 78b68311e..8c8b1c749 100644
--- a/arch/arm64/include/asm/unistd32.h
+++ b/arch/arm64/include/asm/unistd32.h
@@ -911,6 +911,12 @@ __SYSCALL(__NR_set_mempolicy_home_node, sys_set_mempolicy_home_node)
 __SYSCALL(__NR_cachestat, sys_cachestat)
 #define __NR_fchmodat2 452
 __SYSCALL(__NR_fchmodat2, sys_fchmodat2)
+#define __NR_process_ksm_enable 453
+__SYSCALL(__NR_process_ksm_enable, sys_process_ksm_enable)
+#define __NR_process_ksm_disable 454
+__SYSCALL(__NR_process_ksm_disable, sys_process_ksm_disable)
+#define __NR_process_ksm_status 455
+__SYSCALL(__NR_process_ksm_status, sys_process_ksm_status)
 
 /*
  * Please add new compat syscalls above this comment and update
diff --git a/arch/ia64/kernel/syscalls/syscall.tbl b/arch/ia64/kernel/syscalls/syscall.tbl
index 83d8609ae..2c370b569 100644
--- a/arch/ia64/kernel/syscalls/syscall.tbl
+++ b/arch/ia64/kernel/syscalls/syscall.tbl
@@ -373,3 +373,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/m68k/kernel/syscalls/syscall.tbl b/arch/m68k/kernel/syscalls/syscall.tbl
index 259ceb125..346033761 100644
--- a/arch/m68k/kernel/syscalls/syscall.tbl
+++ b/arch/m68k/kernel/syscalls/syscall.tbl
@@ -452,3 +452,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/microblaze/kernel/syscalls/syscall.tbl b/arch/microblaze/kernel/syscalls/syscall.tbl
index a3798c263..3d550ff34 100644
--- a/arch/microblaze/kernel/syscalls/syscall.tbl
+++ b/arch/microblaze/kernel/syscalls/syscall.tbl
@@ -458,3 +458,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/mips/kernel/syscalls/syscall_n32.tbl b/arch/mips/kernel/syscalls/syscall_n32.tbl
index 152034b8e..49bedc66c 100644
--- a/arch/mips/kernel/syscalls/syscall_n32.tbl
+++ b/arch/mips/kernel/syscalls/syscall_n32.tbl
@@ -391,3 +391,6 @@
 450	n32	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	n32	cachestat			sys_cachestat
 452	n32	fchmodat2			sys_fchmodat2
+453	n32	process_ksm_enable		sys_process_ksm_enable
+454	n32	process_ksm_disable		sys_process_ksm_disable
+455	n32	process_ksm_status		sys_process_ksm_status
diff --git a/arch/mips/kernel/syscalls/syscall_n64.tbl b/arch/mips/kernel/syscalls/syscall_n64.tbl
index cb5e757f6..5c17a0a34 100644
--- a/arch/mips/kernel/syscalls/syscall_n64.tbl
+++ b/arch/mips/kernel/syscalls/syscall_n64.tbl
@@ -367,3 +367,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	n64	cachestat			sys_cachestat
 452	n64	fchmodat2			sys_fchmodat2
+453	n64	process_ksm_enable		sys_process_ksm_enable
+454	n64	process_ksm_disable		sys_process_ksm_disable
+455	n64	process_ksm_status		sys_process_ksm_status
diff --git a/arch/mips/kernel/syscalls/syscall_o32.tbl b/arch/mips/kernel/syscalls/syscall_o32.tbl
index 1a646813a..95dbd2c60 100644
--- a/arch/mips/kernel/syscalls/syscall_o32.tbl
+++ b/arch/mips/kernel/syscalls/syscall_o32.tbl
@@ -440,3 +440,6 @@
 450	o32	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	o32	cachestat			sys_cachestat
 452	o32	fchmodat2			sys_fchmodat2
+453	o32	process_ksm_enable		sys_process_ksm_enable
+454	o32	process_ksm_disable		sys_process_ksm_disable
+455	o32	process_ksm_status		sys_process_ksm_status
diff --git a/arch/parisc/kernel/syscalls/syscall.tbl b/arch/parisc/kernel/syscalls/syscall.tbl
index e97c175b5..9b325ef36 100644
--- a/arch/parisc/kernel/syscalls/syscall.tbl
+++ b/arch/parisc/kernel/syscalls/syscall.tbl
@@ -451,3 +451,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/powerpc/kernel/syscalls/syscall.tbl b/arch/powerpc/kernel/syscalls/syscall.tbl
index 20e50586e..b62ba0834 100644
--- a/arch/powerpc/kernel/syscalls/syscall.tbl
+++ b/arch/powerpc/kernel/syscalls/syscall.tbl
@@ -539,3 +539,6 @@
 450 	nospu	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/s390/kernel/syscalls/syscall.tbl b/arch/s390/kernel/syscalls/syscall.tbl
index 0122cc156..70165723f 100644
--- a/arch/s390/kernel/syscalls/syscall.tbl
+++ b/arch/s390/kernel/syscalls/syscall.tbl
@@ -455,3 +455,6 @@
 450  common	set_mempolicy_home_node	sys_set_mempolicy_home_node	sys_set_mempolicy_home_node
 451  common	cachestat		sys_cachestat			sys_cachestat
 452  common	fchmodat2		sys_fchmodat2			sys_fchmodat2
+453  common	process_ksm_enable	sys_process_ksm_enable		sys_process_ksm_enable
+454  common	process_ksm_disable	sys_process_ksm_disable		sys_process_ksm_disable
+455  common	process_ksm_status	sys_process_ksm_status		sys_process_ksm_status
diff --git a/arch/sh/kernel/syscalls/syscall.tbl b/arch/sh/kernel/syscalls/syscall.tbl
index e90d585c4..80769b880 100644
--- a/arch/sh/kernel/syscalls/syscall.tbl
+++ b/arch/sh/kernel/syscalls/syscall.tbl
@@ -455,3 +455,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/sparc/kernel/syscalls/syscall.tbl b/arch/sparc/kernel/syscalls/syscall.tbl
index 4ed06c71c..fb3514cce 100644
--- a/arch/sparc/kernel/syscalls/syscall.tbl
+++ b/arch/sparc/kernel/syscalls/syscall.tbl
@@ -498,3 +498,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/arch/x86/entry/syscalls/syscall_32.tbl b/arch/x86/entry/syscalls/syscall_32.tbl
index 2d0b1bd86..80a57f6a8 100644
--- a/arch/x86/entry/syscalls/syscall_32.tbl
+++ b/arch/x86/entry/syscalls/syscall_32.tbl
@@ -457,3 +457,6 @@
 450	i386	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	i386	cachestat		sys_cachestat
 452	i386	fchmodat2		sys_fchmodat2
+453	i386	process_ksm_enable		sys_process_ksm_enable
+454	i386	process_ksm_disable		sys_process_ksm_disable
+455	i386	process_ksm_status		sys_process_ksm_status
diff --git a/arch/x86/entry/syscalls/syscall_64.tbl b/arch/x86/entry/syscalls/syscall_64.tbl
index 1d6eee30e..38faca76e 100644
--- a/arch/x86/entry/syscalls/syscall_64.tbl
+++ b/arch/x86/entry/syscalls/syscall_64.tbl
@@ -375,6 +375,9 @@
 451	common	cachestat		sys_cachestat
 452	common	fchmodat2		sys_fchmodat2
 453	64	map_shadow_stack	sys_map_shadow_stack
+454	common	process_ksm_enable	sys_process_ksm_enable
+455	common	process_ksm_disable	sys_process_ksm_disable
+456	common	process_ksm_status	sys_process_ksm_status
 
 #
 # Due to a historical design error, certain syscalls are numbered differently
diff --git a/arch/xtensa/kernel/syscalls/syscall.tbl b/arch/xtensa/kernel/syscalls/syscall.tbl
index fc1a4f3c8..83f5032b2 100644
--- a/arch/xtensa/kernel/syscalls/syscall.tbl
+++ b/arch/xtensa/kernel/syscalls/syscall.tbl
@@ -423,3 +423,6 @@
 450	common	set_mempolicy_home_node		sys_set_mempolicy_home_node
 451	common	cachestat			sys_cachestat
 452	common	fchmodat2			sys_fchmodat2
+453	common	process_ksm_enable		sys_process_ksm_enable
+454	common	process_ksm_disable		sys_process_ksm_disable
+455	common	process_ksm_status		sys_process_ksm_status
diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
index 22bc6bc14..da013ad43 100644
--- a/include/linux/syscalls.h
+++ b/include/linux/syscalls.h
@@ -799,6 +799,9 @@ asmlinkage long sys_madvise(unsigned long start, size_t len, int behavior);
 asmlinkage long sys_process_madvise(int pidfd, const struct iovec __user *vec,
 			size_t vlen, int behavior, unsigned int flags);
 asmlinkage long sys_process_mrelease(int pidfd, unsigned int flags);
+asmlinkage long sys_process_ksm_enable(int pidfd, unsigned int flags);
+asmlinkage long sys_process_ksm_disable(int pidfd, unsigned int flags);
+asmlinkage long sys_process_ksm_status(int pidfd, unsigned int flags);
 asmlinkage long sys_remap_file_pages(unsigned long start, unsigned long size,
 			unsigned long prot, unsigned long pgoff,
 			unsigned long flags);
diff --git a/include/uapi/asm-generic/unistd.h b/include/uapi/asm-generic/unistd.h
index abe087c53..e393422e2 100644
--- a/include/uapi/asm-generic/unistd.h
+++ b/include/uapi/asm-generic/unistd.h
@@ -823,8 +823,17 @@ __SYSCALL(__NR_cachestat, sys_cachestat)
 #define __NR_fchmodat2 452
 __SYSCALL(__NR_fchmodat2, sys_fchmodat2)
 
+#define __NR_process_ksm_enable 453
+__SYSCALL(__NR_process_ksm_enable, sys_process_ksm_enable)
+
+#define __NR_process_ksm_disable 454
+__SYSCALL(__NR_process_ksm_disable, sys_process_ksm_disable)
+
+#define __NR_process_ksm_status 455
+__SYSCALL(__NR_process_ksm_status, sys_process_ksm_status)
+
 #undef __NR_syscalls
-#define __NR_syscalls 453
+#define __NR_syscalls 456
 
 /*
  * 32 bit systems traditionally used different
diff --git a/kernel/sys.c b/kernel/sys.c
index 2410e3999..b0841a2dd 100644
--- a/kernel/sys.c
+++ b/kernel/sys.c
@@ -2727,6 +2727,153 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
 	return error;
 }
 
+#ifdef CONFIG_KSM
+enum pkc_action {
+	PKSM_ENABLE = 0,
+	PKSM_DISABLE,
+	PKSM_STATUS,
+};
+
+static long do_process_ksm_control(int pidfd, enum pkc_action action)
+{
+	long ret;
+	struct pid *pid;
+	struct task_struct *task;
+	struct mm_struct *mm;
+	unsigned int f_flags;
+
+	pid = pidfd_get_pid(pidfd, &f_flags);
+	if (IS_ERR(pid)) {
+		ret = PTR_ERR(pid);
+		goto out;
+	}
+
+	task = get_pid_task(pid, PIDTYPE_PID);
+	if (!task) {
+		ret = -ESRCH;
+		goto put_pid;
+	}
+
+	/* Require PTRACE_MODE_READ to avoid leaking ASLR metadata. */
+	mm = mm_access(task, PTRACE_MODE_READ_FSCREDS);
+	if (IS_ERR_OR_NULL(mm)) {
+		ret = IS_ERR(mm) ? PTR_ERR(mm) : -ESRCH;
+		goto release_task;
+	}
+
+	/* Require CAP_SYS_NICE for influencing process performance. */
+	if (!capable(CAP_SYS_NICE)) {
+		ret = -EPERM;
+		goto release_mm;
+	}
+
+	if (mmap_write_lock_killable(mm)) {
+		ret = -EINTR;
+		goto release_mm;
+	}
+
+	switch (action) {
+		case PKSM_ENABLE:
+			ret = ksm_enable_merge_any(mm);
+			break;
+		case PKSM_DISABLE:
+			ret = ksm_disable_merge_any(mm);
+			break;
+		case PKSM_STATUS:
+			ret = !!test_bit(MMF_VM_MERGE_ANY, &mm->flags);
+			break;
+	}
+
+	mmap_write_unlock(mm);
+
+release_mm:
+	mmput(mm);
+release_task:
+	put_task_struct(task);
+put_pid:
+	put_pid(pid);
+out:
+	return ret;
+}
+#endif /* CONFIG_KSM */
+
+SYSCALL_DEFINE2(process_ksm_enable, int, pidfd, unsigned int, flags)
+{
+#ifdef CONFIG_KSM
+	if (flags != 0)
+		return -EINVAL;
+
+	return do_process_ksm_control(pidfd, PKSM_ENABLE);
+#else /* CONFIG_KSM */
+	return -ENOSYS;
+#endif /* CONFIG_KSM */
+}
+
+SYSCALL_DEFINE2(process_ksm_disable, int, pidfd, unsigned int, flags)
+{
+#ifdef CONFIG_KSM
+	if (flags != 0)
+		return -EINVAL;
+
+	return do_process_ksm_control(pidfd, PKSM_DISABLE);
+#else /* CONFIG_KSM */
+	return -ENOSYS;
+#endif /* CONFIG_KSM */
+}
+
+SYSCALL_DEFINE2(process_ksm_status, int, pidfd, unsigned int, flags)
+{
+#ifdef CONFIG_KSM
+	if (flags != 0)
+		return -EINVAL;
+
+	return do_process_ksm_control(pidfd, PKSM_STATUS);
+#else /* CONFIG_KSM */
+	return -ENOSYS;
+#endif /* CONFIG_KSM */
+}
+
+#ifdef CONFIG_KSM
+static ssize_t process_ksm_enable_show(struct kobject *kobj,
+		struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%u\n", __NR_process_ksm_enable);
+}
+static struct kobj_attribute process_ksm_enable_attr = __ATTR_RO(process_ksm_enable);
+
+static ssize_t process_ksm_disable_show(struct kobject *kobj,
+		struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%u\n", __NR_process_ksm_disable);
+}
+static struct kobj_attribute process_ksm_disable_attr = __ATTR_RO(process_ksm_disable);
+
+static ssize_t process_ksm_status_show(struct kobject *kobj,
+		struct kobj_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%u\n", __NR_process_ksm_status);
+}
+static struct kobj_attribute process_ksm_status_attr = __ATTR_RO(process_ksm_status);
+
+static struct attribute *process_ksm_sysfs_attrs[] = {
+	&process_ksm_enable_attr.attr,
+	&process_ksm_disable_attr.attr,
+	&process_ksm_status_attr.attr,
+	NULL,
+};
+
+static const struct attribute_group process_ksm_sysfs_attr_group = {
+	.attrs = process_ksm_sysfs_attrs,
+	.name = "process_ksm",
+};
+
+static int __init process_ksm_sysfs_init(void)
+{
+	return sysfs_create_group(kernel_kobj, &process_ksm_sysfs_attr_group);
+}
+subsys_initcall(process_ksm_sysfs_init);
+#endif /* CONFIG_KSM */
+
 SYSCALL_DEFINE3(getcpu, unsigned __user *, cpup, unsigned __user *, nodep,
 		struct getcpu_cache __user *, unused)
 {
diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
index e137c1385..2d9772d11 100644
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -184,6 +184,9 @@ COND_SYSCALL(mincore);
 COND_SYSCALL(madvise);
 COND_SYSCALL(process_madvise);
 COND_SYSCALL(process_mrelease);
+COND_SYSCALL(process_ksm_enable);
+COND_SYSCALL(process_ksm_disable);
+COND_SYSCALL(process_ksm_status);
 COND_SYSCALL(remap_file_pages);
 COND_SYSCALL(mbind);
 COND_SYSCALL(get_mempolicy);