linux/debian/patches/patchset-pf/pksm/0001-mm-expose-per-process-KSM-control-via-syscalls.patch

From 4eb6615c1498cb8bff76c31e9596b585410f507d Mon Sep 17 00:00:00 2001
From: Oleksandr Natalenko <oleksandr@natalenko.name>
Date: Mon, 30 Sep 2024 08:58:38 +0200
Subject: 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/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             |   9 +-
 kernel/sys.c                                  | 138 ++++++++++++++++++
 kernel/sys_ni.c                               |   3 +
 scripts/syscall.tbl                           |   3 +
 .../arch/powerpc/entry/syscalls/syscall.tbl   |   3 +
 .../perf/arch/s390/entry/syscalls/syscall.tbl |   3 +
 22 files changed, 206 insertions(+), 1 deletion(-)

--- a/arch/alpha/kernel/syscalls/syscall.tbl
+++ b/arch/alpha/kernel/syscalls/syscall.tbl
@@ -502,3 +502,6 @@
 570	common	lsm_set_self_attr		sys_lsm_set_self_attr
 571	common	lsm_list_modules		sys_lsm_list_modules
 572	common  mseal				sys_mseal
+573	common	process_ksm_enable		sys_process_ksm_enable
+574	common	process_ksm_disable		sys_process_ksm_disable
+575	common	process_ksm_status		sys_process_ksm_status
--- a/arch/arm/tools/syscall.tbl
+++ b/arch/arm/tools/syscall.tbl
@@ -477,3 +477,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/m68k/kernel/syscalls/syscall.tbl
+++ b/arch/m68k/kernel/syscalls/syscall.tbl
@@ -462,3 +462,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/microblaze/kernel/syscalls/syscall.tbl
+++ b/arch/microblaze/kernel/syscalls/syscall.tbl
@@ -468,3 +468,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/mips/kernel/syscalls/syscall_n32.tbl
+++ b/arch/mips/kernel/syscalls/syscall_n32.tbl
@@ -401,3 +401,6 @@
 460	n32	lsm_set_self_attr		sys_lsm_set_self_attr
 461	n32	lsm_list_modules		sys_lsm_list_modules
 462	n32	mseal				sys_mseal
+463	n32	process_ksm_enable		sys_process_ksm_enable
+464	n32	process_ksm_disable		sys_process_ksm_disable
+465	n32	process_ksm_status		sys_process_ksm_status
--- a/arch/mips/kernel/syscalls/syscall_n64.tbl
+++ b/arch/mips/kernel/syscalls/syscall_n64.tbl
@@ -377,3 +377,6 @@
 460	n64	lsm_set_self_attr		sys_lsm_set_self_attr
 461	n64	lsm_list_modules		sys_lsm_list_modules
 462	n64	mseal				sys_mseal
+463	n64	process_ksm_enable		sys_process_ksm_enable
+464	n64	process_ksm_disable		sys_process_ksm_disable
+465	n64	process_ksm_status		sys_process_ksm_status
--- a/arch/mips/kernel/syscalls/syscall_o32.tbl
+++ b/arch/mips/kernel/syscalls/syscall_o32.tbl
@@ -450,3 +450,6 @@
 460	o32	lsm_set_self_attr		sys_lsm_set_self_attr
 461	o32	lsm_list_modules		sys_lsm_list_modules
 462	o32	mseal				sys_mseal
+463	o32	process_ksm_enable		sys_process_ksm_enable
+464	o32	process_ksm_disable		sys_process_ksm_disable
+465	o32	process_ksm_status		sys_process_ksm_status
--- a/arch/parisc/kernel/syscalls/syscall.tbl
+++ b/arch/parisc/kernel/syscalls/syscall.tbl
@@ -461,3 +461,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/powerpc/kernel/syscalls/syscall.tbl
+++ b/arch/powerpc/kernel/syscalls/syscall.tbl
@@ -553,3 +553,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/s390/kernel/syscalls/syscall.tbl
+++ b/arch/s390/kernel/syscalls/syscall.tbl
@@ -465,3 +465,6 @@
 460  common	lsm_set_self_attr	sys_lsm_set_self_attr		sys_lsm_set_self_attr
 461  common	lsm_list_modules	sys_lsm_list_modules		sys_lsm_list_modules
 462  common	mseal			sys_mseal			sys_mseal
+463  common	process_ksm_enable	sys_process_ksm_enable		sys_process_ksm_enable
+464  common	process_ksm_disable	sys_process_ksm_disable		sys_process_ksm_disable
+465  common	process_ksm_status	sys_process_ksm_status		sys_process_ksm_status
--- a/arch/sh/kernel/syscalls/syscall.tbl
+++ b/arch/sh/kernel/syscalls/syscall.tbl
@@ -466,3 +466,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/sparc/kernel/syscalls/syscall.tbl
+++ b/arch/sparc/kernel/syscalls/syscall.tbl
@@ -508,3 +508,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal 				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/arch/x86/entry/syscalls/syscall_32.tbl
+++ b/arch/x86/entry/syscalls/syscall_32.tbl
@@ -468,3 +468,6 @@
 460	i386	lsm_set_self_attr	sys_lsm_set_self_attr
 461	i386	lsm_list_modules	sys_lsm_list_modules
 462	i386	mseal 			sys_mseal
+463	i386	process_ksm_enable		sys_process_ksm_enable
+464	i386	process_ksm_disable		sys_process_ksm_disable
+465	i386	process_ksm_status		sys_process_ksm_status
--- a/arch/x86/entry/syscalls/syscall_64.tbl
+++ b/arch/x86/entry/syscalls/syscall_64.tbl
@@ -386,6 +386,9 @@
 460	common	lsm_set_self_attr	sys_lsm_set_self_attr
 461	common	lsm_list_modules	sys_lsm_list_modules
 462 	common  mseal			sys_mseal
+463	common	process_ksm_enable	sys_process_ksm_enable
+464	common	process_ksm_disable	sys_process_ksm_disable
+465	common	process_ksm_status	sys_process_ksm_status
 
 #
 # Due to a historical design error, certain syscalls are numbered differently
--- a/arch/xtensa/kernel/syscalls/syscall.tbl
+++ b/arch/xtensa/kernel/syscalls/syscall.tbl
@@ -433,3 +433,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal 				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/include/linux/syscalls.h
+++ b/include/linux/syscalls.h
@@ -818,6 +818,9 @@ asmlinkage long sys_madvise(unsigned lon
 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);
--- a/include/uapi/asm-generic/unistd.h
+++ b/include/uapi/asm-generic/unistd.h
@@ -841,8 +841,15 @@ __SYSCALL(__NR_lsm_list_modules, sys_lsm
 #define __NR_mseal 462
 __SYSCALL(__NR_mseal, sys_mseal)
 
+#define __NR_process_ksm_enable 463
+__SYSCALL(__NR_process_ksm_enable, sys_process_ksm_enable)
+#define __NR_process_ksm_disable 464
+__SYSCALL(__NR_process_ksm_disable, sys_process_ksm_disable)
+#define __NR_process_ksm_status 465
+__SYSCALL(__NR_process_ksm_status, sys_process_ksm_status)
+
 #undef __NR_syscalls
-#define __NR_syscalls 463
+#define __NR_syscalls 466
 
 /*
  * 32 bit systems traditionally used different
--- a/kernel/sys.c
+++ b/kernel/sys.c
@@ -2791,6 +2791,144 @@ SYSCALL_DEFINE5(prctl, int, option, unsi
 	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 task_struct *task;
+	struct mm_struct *mm;
+	unsigned int f_flags;
+
+	task = pidfd_get_task(pidfd, &f_flags);
+	if (IS_ERR(task)) {
+		ret = PTR_ERR(task);
+		goto out;
+	}
+
+	/* 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);
+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)
 {
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -186,6 +186,9 @@ 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);
--- a/scripts/syscall.tbl
+++ b/scripts/syscall.tbl
@@ -403,3 +403,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable			sys_process_ksm_enable
+464	common	process_ksm_disable			sys_process_ksm_disable
+465	common	process_ksm_status			sys_process_ksm_status
--- a/tools/perf/arch/powerpc/entry/syscalls/syscall.tbl
+++ b/tools/perf/arch/powerpc/entry/syscalls/syscall.tbl
@@ -553,3 +553,6 @@
 460	common	lsm_set_self_attr		sys_lsm_set_self_attr
 461	common	lsm_list_modules		sys_lsm_list_modules
 462	common	mseal				sys_mseal
+463	common	process_ksm_enable		sys_process_ksm_enable
+464	common	process_ksm_disable		sys_process_ksm_disable
+465	common	process_ksm_status		sys_process_ksm_status
--- a/tools/perf/arch/s390/entry/syscalls/syscall.tbl
+++ b/tools/perf/arch/s390/entry/syscalls/syscall.tbl
@@ -465,3 +465,6 @@
 460  common	lsm_set_self_attr	sys_lsm_set_self_attr		sys_lsm_set_self_attr
 461  common	lsm_list_modules	sys_lsm_list_modules		sys_lsm_list_modules
 462  common	mseal			sys_mseal			sys_mseal
+463  common	process_ksm_enable	sys_process_ksm_enable		sys_process_ksm_enable
+464  common	process_ksm_disable	sys_process_ksm_disable		sys_process_ksm_disable
+465  common	process_ksm_status	sys_process_ksm_status		sys_process_ksm_status