Skip to contentSkip to conversation

compat: remove compat_alloc_user_space

PATCH Applied Arnd Bergmann 5 participants 20 messages Jul 27, 2021 – Jul 30, 2021

Subsystems (14)
Memory management x86 PowerPC MIPS +10 more subsystems
lore
Share & tools
lore

Arnd Bergmann

author
#0 lore raw

v5 cover compat: remove compat_alloc_user_space

From: Arnd Bergmann <[email protected]>

Going through compat_alloc_user_space() to convert indirect system call arguments tends to add complexity compared to handling the native and compat logic in the same code.

Out of the other remaining callers, the linux-media series went into v5.14, and the network ioctl handling is now fixed in net-next, so these are the last remaining users, and I now include the final patch to remove the definitions as well.

Since these patches are now all that remains, it would be nice to merge it all through Andrew's Linux-mm tree, which is already based on top of linux-next.

Arnd
---

Changes in v4:

- Rebase on top of net-next
- Split up and improve the kexec patch based on Christoph's suggestions
- Include final patch to remove compat_alloc_user_space
- Cc compat architecture maintainers

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

Changes in v3:

- fix whitespace as pointed out by Christoph Hellwig
- minor build fixes
- rebase to v5.13-rc1

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

Changes in v2:

- address review comments from Christoph Hellwig
- split syscall removal into a separate patch
- replace __X32_COND_SYSCALL() with individual macros for x32

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

Arnd Bergmann (6):
kexec: move locking into do_kexec_load
kexec: avoid compat_alloc_user_space
mm: simplify compat_sys_move_pages
mm: simplify compat numa syscalls
compat: remove some compat entry points
arch: remove compat_alloc_user_space

 arch/arm64/include/asm/compat.h           |   5 -
 arch/arm64/include/asm/uaccess.h          |  11 --
 arch/arm64/include/asm/unistd32.h         |  10 +-
 arch/arm64/lib/Makefile                   |   2 +-
 arch/arm64/lib/copy_in_user.S             |  77 ---------
 arch/mips/cavium-octeon/octeon-memcpy.S   |   2 -
 arch/mips/include/asm/compat.h            |   8 -
 arch/mips/include/asm/uaccess.h           |  26 ---
 arch/mips/kernel/syscalls/syscall_n32.tbl |  10 +-
 arch/mips/kernel/syscalls/syscall_o32.tbl |  10 +-
 arch/mips/lib/memcpy.S                    |  11 --
 arch/parisc/include/asm/compat.h          |   6 -
 arch/parisc/include/asm/uaccess.h         |   2 -
 arch/parisc/kernel/syscalls/syscall.tbl   |   8 +-
 arch/parisc/lib/memcpy.c                  |   9 -
 arch/powerpc/include/asm/compat.h         |  16 --
 arch/powerpc/kernel/syscalls/syscall.tbl  |  10 +-
 arch/s390/include/asm/compat.h            |  10 --
 arch/s390/include/asm/uaccess.h           |   3 -
 arch/s390/kernel/syscalls/syscall.tbl     |  10 +-
 arch/s390/lib/uaccess.c                   |  63 -------
 arch/sparc/include/asm/compat.h           |  19 ---
 arch/sparc/kernel/process_64.c            |   2 +-
 arch/sparc/kernel/signal32.c              |  12 +-
 arch/sparc/kernel/signal_64.c             |   8 +-
 arch/sparc/kernel/syscalls/syscall.tbl    |  10 +-
 arch/x86/entry/syscalls/syscall_32.tbl    |   4 +-
 arch/x86/entry/syscalls/syscall_64.tbl    |   2 +-
 arch/x86/include/asm/compat.h             |  13 --
 arch/x86/include/asm/uaccess_64.h         |   7 -
 include/linux/compat.h                    |  39 +----
 include/linux/uaccess.h                   |  10 --
 include/uapi/asm-generic/unistd.h         |  10 +-
 kernel/compat.c                           |  21 ---
 kernel/kexec.c                            | 103 +++++-------
 kernel/sys_ni.c                           |   5 -
 mm/mempolicy.c                            | 196 +++++-----------------
 mm/migrate.c                              |  50 +++---
 38 files changed, 175 insertions(+), 645 deletions(-)
 delete mode 100644 arch/arm64/lib/copy_in_user.S
Signature · 34 lines
2.29.2

Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Thomas Bogendoerfer <[email protected]>
Cc: "James E.J. Bottomley" <[email protected]>
Cc: Helge Deller <[email protected]>
Cc: Michael Ellerman <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Heiko Carstens <[email protected]>
Cc: Vasily Gorbik <[email protected]>
Cc: Christian Borntraeger <[email protected]>
Cc: "David S. Miller" <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Borislav Petkov <[email protected]>
Cc: [email protected]
Cc: "H. Peter Anvin" <[email protected]>
Cc: Al Viro <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: "Eric W. Biederman" <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: Feng Tang <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
author
#1 lore raw

PATCH v5 1/6 kexec: move locking into do_kexec_load

From: Arnd Bergmann <[email protected]>

The locking is the same between the native and compat version of sys_kexec_load(), so it can be done in the common implementation to reduce duplication.

Co-developed-by Eric Biederman <[email protected]>
Co-developed-by Christoph Hellwig <[email protected]>
Signed-off-by Arnd Bergmann <[email protected]>
---
 kernel/kexec.c | 44 ++++++++++++++++----------------------------
 1 file changed, 16 insertions(+), 28 deletions(-)
kernel/kexec.c +16 -28
index c82c6c06f051..9c7aef8f4bb6 100644
@@ -110,6 +110,17 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 	unsigned long i;
 	int ret;
 
+	/*
+	 * Because we write directly to the reserved memory region when loading
+	 * crash kernels we need a mutex here to prevent multiple crash kernels
+	 * from attempting to load simultaneously, and to prevent a crash kernel
+	 * from loading over the top of a in use crash kernel.
+	 *
+	 * KISS: always take the mutex.
+	 */
+	if (!mutex_trylock(&kexec_mutex))
+		return -EBUSY;
+
 	if (flags & KEXEC_ON_CRASH) {
 		dest_image = &kexec_crash_image;
 		if (kexec_crash_image)

@@ -121,7 +132,8 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 	if (nr_segments == 0) {
 		/* Uninstall image */
 		kimage_free(xchg(dest_image, NULL));
-		return 0;
+		ret = 0;
+		goto out_unlock;
 	}
 	if (flags & KEXEC_ON_CRASH) {
 		/*

@@ -134,7 +146,7 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 
 	ret = kimage_alloc_init(&image, entry, nr_segments, segments, flags);
 	if (ret)
-		return ret;
+		goto out_unlock;
 
 	if (flags & KEXEC_PRESERVE_CONTEXT)
 		image->preserve_context = 1;

@@ -171,6 +183,8 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 		arch_kexec_protect_crashkres();
 
 	kimage_free(image);
+out_unlock:
+	mutex_unlock(&kexec_mutex);
 	return ret;
 }
 

@@ -247,21 +261,8 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 		((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH_DEFAULT))
 		return -EINVAL;
 
-	/* Because we write directly to the reserved memory
-	 * region when loading crash kernels we need a mutex here to
-	 * prevent multiple crash  kernels from attempting to load
-	 * simultaneously, and to prevent a crash kernel from loading
-	 * over the top of a in use crash kernel.
-	 *
-	 * KISS: always take the mutex.
-	 */
-	if (!mutex_trylock(&kexec_mutex))
-		return -EBUSY;
-
 	result = do_kexec_load(entry, nr_segments, segments, flags);
 
-	mutex_unlock(&kexec_mutex);
-
 	return result;
 }
 

@@ -301,21 +302,8 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 			return -EFAULT;
 	}
 
-	/* Because we write directly to the reserved memory
-	 * region when loading crash kernels we need a mutex here to
-	 * prevent multiple crash  kernels from attempting to load
-	 * simultaneously, and to prevent a crash kernel from loading
-	 * over the top of a in use crash kernel.
-	 *
-	 * KISS: always take the mutex.
-	 */
-	if (!mutex_trylock(&kexec_mutex))
-		return -EBUSY;
-
 	result = do_kexec_load(entry, nr_segments, ksegments, flags);
 
-	mutex_unlock(&kexec_mutex);
-
 	return result;
 }
 #endif
Signature · 1 line
2.29.2
Acked-by on v5 1/6
#2 lore raw

Re: [PATCH v5 1/6] kexec: move locking into do_kexec_load

Quoted 5 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

The locking is the same between the native and compat version of sys_kexec_load(), so it can be done in the common implementation to reduce duplication.

Acked-by "Eric W. Biederman" <[email protected]>
Quoted 102 lines
Co-developed-by Eric Biederman <[email protected]>
Co-developed-by Christoph Hellwig <[email protected]>
Signed-off-by Arnd Bergmann <[email protected]>
---
 kernel/kexec.c | 44 ++++++++++++++++----------------------------
 1 file changed, 16 insertions(+), 28 deletions(-)
kernel/kexec.c +16 -28
index c82c6c06f051..9c7aef8f4bb6 100644
@@ -110,6 +110,17 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 	unsigned long i;
 	int ret;
 
+	/*
+	 * Because we write directly to the reserved memory region when loading
+	 * crash kernels we need a mutex here to prevent multiple crash kernels
+	 * from attempting to load simultaneously, and to prevent a crash kernel
+	 * from loading over the top of a in use crash kernel.
+	 *
+	 * KISS: always take the mutex.
+	 */
+	if (!mutex_trylock(&kexec_mutex))
+		return -EBUSY;
+
 	if (flags & KEXEC_ON_CRASH) {
 		dest_image = &kexec_crash_image;
 		if (kexec_crash_image)

@@ -121,7 +132,8 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 	if (nr_segments == 0) {
 		/* Uninstall image */
 		kimage_free(xchg(dest_image, NULL));
-		return 0;
+		ret = 0;
+		goto out_unlock;
 	}
 	if (flags & KEXEC_ON_CRASH) {
 		/*

@@ -134,7 +146,7 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 
 	ret = kimage_alloc_init(&image, entry, nr_segments, segments, flags);
 	if (ret)
-		return ret;
+		goto out_unlock;
 
 	if (flags & KEXEC_PRESERVE_CONTEXT)
 		image->preserve_context = 1;

@@ -171,6 +183,8 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 		arch_kexec_protect_crashkres();
 
 	kimage_free(image);
+out_unlock:
+	mutex_unlock(&kexec_mutex);
 	return ret;
 }
 

@@ -247,21 +261,8 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 		((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH_DEFAULT))
 		return -EINVAL;
 
-	/* Because we write directly to the reserved memory
-	 * region when loading crash kernels we need a mutex here to
-	 * prevent multiple crash  kernels from attempting to load
-	 * simultaneously, and to prevent a crash kernel from loading
-	 * over the top of a in use crash kernel.
-	 *
-	 * KISS: always take the mutex.
-	 */
-	if (!mutex_trylock(&kexec_mutex))
-		return -EBUSY;
-
 	result = do_kexec_load(entry, nr_segments, segments, flags);
 
-	mutex_unlock(&kexec_mutex);
-
 	return result;
 }
 

@@ -301,21 +302,8 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 			return -EFAULT;
 	}
 
-	/* Because we write directly to the reserved memory
-	 * region when loading crash kernels we need a mutex here to
-	 * prevent multiple crash  kernels from attempting to load
-	 * simultaneously, and to prevent a crash kernel from loading
-	 * over the top of a in use crash kernel.
-	 *
-	 * KISS: always take the mutex.
-	 */
-	if (!mutex_trylock(&kexec_mutex))
-		return -EBUSY;
-
 	result = do_kexec_load(entry, nr_segments, ksegments, flags);
 
-	mutex_unlock(&kexec_mutex);
-
 	return result;
 }
 #endif
author
#3 lore raw

PATCH v5 2/6 kexec: avoid compat_alloc_user_space

In reply to Arnd Bergmann

From: Arnd Bergmann <[email protected]>

kimage_alloc_init() expects a __user pointer, so compat_sys_kexec_load() uses compat_alloc_user_space() to convert the layout and put it back onto the user space caller stack.

Moving the user space access into the syscall handler directly actually makes the code simpler, as the conversion for compat mode can now be done on kernel memory.

Co-developed-by Eric Biederman <[email protected]>
Co-developed-by Christoph Hellwig <[email protected]>
Link https://lore.kernel.org/lkml/YPbtsU4GX6PL7%[email protected]/ open here
Link https://lore.kernel.org/lkml/[email protected]/ open here
Signed-off-by Arnd Bergmann <[email protected]>
---
 kernel/kexec.c | 61 +++++++++++++++++++++-----------------------------
 1 file changed, 25 insertions(+), 36 deletions(-)
kernel/kexec.c +25 -36
index 9c7aef8f4bb6..b5e40f069768 100644
@@ -19,26 +19,9 @@
 
 #include "kexec_internal.h"
 
-static int copy_user_segment_list(struct kimage *image,
-				  unsigned long nr_segments,
-				  struct kexec_segment __user *segments)
-{
-	int ret;
-	size_t segment_bytes;
-
-	/* Read in the segments */
-	image->nr_segments = nr_segments;
-	segment_bytes = nr_segments * sizeof(*segments);
-	ret = copy_from_user(image->segment, segments, segment_bytes);
-	if (ret)
-		ret = -EFAULT;
-
-	return ret;
-}
-
 static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 			     unsigned long nr_segments,
-			     struct kexec_segment __user *segments,
+			     struct kexec_segment *segments,
 			     unsigned long flags)
 {
 	int ret;

@@ -58,10 +41,8 @@ static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 		return -ENOMEM;
 
 	image->start = entry;
-
-	ret = copy_user_segment_list(image, nr_segments, segments);
-	if (ret)
-		goto out_free_image;
+	image->nr_segments = nr_segments;
+	memcpy(image->segment, segments, nr_segments * sizeof(*segments));
 
 	if (kexec_on_panic) {
 		/* Enable special crash kernel control page alloc policy. */

@@ -104,7 +85,7 @@ static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 }
 
 static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
-		struct kexec_segment __user *segments, unsigned long flags)
+		struct kexec_segment *segments, unsigned long flags)
 {
 	struct kimage **dest_image, *image;
 	unsigned long i;

@@ -250,7 +231,8 @@ static inline int kexec_load_check(unsigned long nr_segments,
 SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 		struct kexec_segment __user *, segments, unsigned long, flags)
 {
-	int result;
+	struct kexec_segment *ksegments;
+	unsigned long result;
 
 	result = kexec_load_check(nr_segments, flags);
 	if (result)

@@ -261,7 +243,12 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 		((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH_DEFAULT))
 		return -EINVAL;
 
-	result = do_kexec_load(entry, nr_segments, segments, flags);
+	ksegments = memdup_user(segments, nr_segments * sizeof(ksegments[0]));
+	if (IS_ERR(ksegments))
+		return PTR_ERR(ksegments);
+
+	result = do_kexec_load(entry, nr_segments, ksegments, flags);
+	kfree(ksegments);
 
 	return result;
 }

@@ -273,7 +260,7 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 		       compat_ulong_t, flags)
 {
 	struct compat_kexec_segment in;
-	struct kexec_segment out, __user *ksegments;
+	struct kexec_segment *ksegments;
 	unsigned long i, result;
 
 	result = kexec_load_check(nr_segments, flags);

@@ -286,24 +273,26 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 	if ((flags & KEXEC_ARCH_MASK) == KEXEC_ARCH_DEFAULT)
 		return -EINVAL;
 
-	ksegments = compat_alloc_user_space(nr_segments * sizeof(out));
+	ksegments = kmalloc_array(nr_segments, sizeof(ksegments[0]),
+			GFP_KERNEL);
+	if (!ksegments)
+		return -ENOMEM;
+
 	for (i = 0; i < nr_segments; i++) {
 		result = copy_from_user(&in, &segments[i], sizeof(in));
 		if (result)
-			return -EFAULT;
+			goto fail;
 
-		out.buf   = compat_ptr(in.buf);
-		out.bufsz = in.bufsz;
-		out.mem   = in.mem;
-		out.memsz = in.memsz;
-
-		result = copy_to_user(&ksegments[i], &out, sizeof(out));
-		if (result)
-			return -EFAULT;
+		ksegments[i].buf   = compat_ptr(in.buf);
+		ksegments[i].bufsz = in.bufsz;
+		ksegments[i].mem   = in.mem;
+		ksegments[i].memsz = in.memsz;
 	}
 
 	result = do_kexec_load(entry, nr_segments, ksegments, flags);
 
+fail:
+	kfree(ksegments);
 	return result;
 }
 #endif
Signature · 1 line
2.29.2
Acked-by on v5 2/6
#4 lore raw

Re: [PATCH v5 2/6] kexec: avoid compat_alloc_user_space

Quoted 9 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

kimage_alloc_init() expects a __user pointer, so compat_sys_kexec_load() uses compat_alloc_user_space() to convert the layout and put it back onto the user space caller stack.

Moving the user space access into the syscall handler directly actually makes the code simpler, as the conversion for compat mode can now be done on kernel memory.

Acked-by "Eric W. Biederman" <[email protected]>
Quoted 134 lines
Co-developed-by Eric Biederman <[email protected]>
Co-developed-by Christoph Hellwig <[email protected]>
Link https://lore.kernel.org/lkml/YPbtsU4GX6PL7%[email protected]/ open here
Link https://lore.kernel.org/lkml/[email protected]/ open here
Signed-off-by Arnd Bergmann <[email protected]>
---
 kernel/kexec.c | 61 +++++++++++++++++++++-----------------------------
 1 file changed, 25 insertions(+), 36 deletions(-)
kernel/kexec.c +25 -36
index 9c7aef8f4bb6..b5e40f069768 100644
@@ -19,26 +19,9 @@
 
 #include "kexec_internal.h"
 
-static int copy_user_segment_list(struct kimage *image,
-				  unsigned long nr_segments,
-				  struct kexec_segment __user *segments)
-{
-	int ret;
-	size_t segment_bytes;
-
-	/* Read in the segments */
-	image->nr_segments = nr_segments;
-	segment_bytes = nr_segments * sizeof(*segments);
-	ret = copy_from_user(image->segment, segments, segment_bytes);
-	if (ret)
-		ret = -EFAULT;
-
-	return ret;
-}
-
 static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 			     unsigned long nr_segments,
-			     struct kexec_segment __user *segments,
+			     struct kexec_segment *segments,
 			     unsigned long flags)
 {
 	int ret;

@@ -58,10 +41,8 @@ static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 		return -ENOMEM;
 
 	image->start = entry;
-
-	ret = copy_user_segment_list(image, nr_segments, segments);
-	if (ret)
-		goto out_free_image;
+	image->nr_segments = nr_segments;
+	memcpy(image->segment, segments, nr_segments * sizeof(*segments));
 
 	if (kexec_on_panic) {
 		/* Enable special crash kernel control page alloc policy. */

@@ -104,7 +85,7 @@ static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 }
 
 static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
-		struct kexec_segment __user *segments, unsigned long flags)
+		struct kexec_segment *segments, unsigned long flags)
 {
 	struct kimage **dest_image, *image;
 	unsigned long i;

@@ -250,7 +231,8 @@ static inline int kexec_load_check(unsigned long nr_segments,
 SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 		struct kexec_segment __user *, segments, unsigned long, flags)
 {
-	int result;
+	struct kexec_segment *ksegments;
+	unsigned long result;
 
 	result = kexec_load_check(nr_segments, flags);
 	if (result)

@@ -261,7 +243,12 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 		((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH_DEFAULT))
 		return -EINVAL;
 
-	result = do_kexec_load(entry, nr_segments, segments, flags);
+	ksegments = memdup_user(segments, nr_segments * sizeof(ksegments[0]));
+	if (IS_ERR(ksegments))
+		return PTR_ERR(ksegments);
+
+	result = do_kexec_load(entry, nr_segments, ksegments, flags);
+	kfree(ksegments);
 
 	return result;
 }

@@ -273,7 +260,7 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 		       compat_ulong_t, flags)
 {
 	struct compat_kexec_segment in;
-	struct kexec_segment out, __user *ksegments;
+	struct kexec_segment *ksegments;
 	unsigned long i, result;
 
 	result = kexec_load_check(nr_segments, flags);

@@ -286,24 +273,26 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 	if ((flags & KEXEC_ARCH_MASK) == KEXEC_ARCH_DEFAULT)
 		return -EINVAL;
 
-	ksegments = compat_alloc_user_space(nr_segments * sizeof(out));
+	ksegments = kmalloc_array(nr_segments, sizeof(ksegments[0]),
+			GFP_KERNEL);
+	if (!ksegments)
+		return -ENOMEM;
+
 	for (i = 0; i < nr_segments; i++) {
 		result = copy_from_user(&in, &segments[i], sizeof(in));
 		if (result)
-			return -EFAULT;
+			goto fail;
 
-		out.buf   = compat_ptr(in.buf);
-		out.bufsz = in.bufsz;
-		out.mem   = in.mem;
-		out.memsz = in.memsz;
-
-		result = copy_to_user(&ksegments[i], &out, sizeof(out));
-		if (result)
-			return -EFAULT;
+		ksegments[i].buf   = compat_ptr(in.buf);
+		ksegments[i].bufsz = in.bufsz;
+		ksegments[i].mem   = in.mem;
+		ksegments[i].memsz = in.memsz;
 	}
 
 	result = do_kexec_load(entry, nr_segments, ksegments, flags);
 
+fail:
+	kfree(ksegments);
 	return result;
 }
 #endif
author
#5 lore raw

PATCH v5 3/6 mm: simplify compat_sys_move_pages

In reply to Arnd Bergmann

From: Arnd Bergmann <[email protected]>

The compat move_pages() implementation uses compat_alloc_user_space() for converting the pointer array. Moving the compat handling into the function itself is a bit simpler and lets us avoid the compat_alloc_user_space() call.

Reviewed-by Christoph Hellwig <[email protected]>
Signed-off-by Arnd Bergmann <[email protected]>
---
 mm/migrate.c | 45 ++++++++++++++++++++++++++++++---------------
 1 file changed, 30 insertions(+), 15 deletions(-)
mm/migrate.c +30 -15
index 34a9ad3e0a4f..60634d2653b9 100644
@@ -1820,6 +1820,23 @@ static void do_pages_stat_array(struct mm_struct *mm, unsigned long nr_pages,
 	mmap_read_unlock(mm);
 }
 
+static int get_compat_pages_array(const void __user *chunk_pages[],
+				  const void __user * __user *pages,
+				  unsigned long chunk_nr)
+{
+	compat_uptr_t __user *pages32 = (compat_uptr_t __user *)pages;
+	compat_uptr_t p;
+	int i;
+
+	for (i = 0; i < chunk_nr; i++) {
+		if (get_user(p, pages32 + i))
+			return -EFAULT;
+		chunk_pages[i] = compat_ptr(p);
+	}
+
+	return 0;
+}
+
 /*
  * Determine the nodes of a user array of pages and store it in
  * a user array of status.

@@ -1839,8 +1856,15 @@ static int do_pages_stat(struct mm_struct *mm, unsigned long nr_pages,
 		if (chunk_nr > DO_PAGES_STAT_CHUNK_NR)
 			chunk_nr = DO_PAGES_STAT_CHUNK_NR;
 
-		if (copy_from_user(chunk_pages, pages, chunk_nr * sizeof(*chunk_pages)))
-			break;
+		if (in_compat_syscall()) {
+			if (get_compat_pages_array(chunk_pages, pages,
+						   chunk_nr))
+				break;
+		} else {
+			if (copy_from_user(chunk_pages, pages,
+				      chunk_nr * sizeof(*chunk_pages)))
+				break;
+		}
 
 		do_pages_stat_array(mm, chunk_nr, chunk_pages, chunk_status);
 

@@ -1945,23 +1969,14 @@ SYSCALL_DEFINE6(move_pages, pid_t, pid, unsigned long, nr_pages,
 
 #ifdef CONFIG_COMPAT
 COMPAT_SYSCALL_DEFINE6(move_pages, pid_t, pid, compat_ulong_t, nr_pages,
-		       compat_uptr_t __user *, pages32,
+		       compat_uptr_t __user *, pages,
 		       const int __user *, nodes,
 		       int __user *, status,
 		       int, flags)
 {
-	const void __user * __user *pages;
-	int i;
-
-	pages = compat_alloc_user_space(nr_pages * sizeof(void *));
-	for (i = 0; i < nr_pages; i++) {
-		compat_uptr_t p;
-
-		if (get_user(p, pages32 + i) ||
-			put_user(compat_ptr(p), pages + i))
-			return -EFAULT;
-	}
-	return kernel_move_pages(pid, nr_pages, pages, nodes, status, flags);
+	return kernel_move_pages(pid, nr_pages,
+				 (const void __user *__user *)pages,
+				 nodes, status, flags);
 }
 #endif /* CONFIG_COMPAT */
Signature · 1 line
2.29.2
author
#6 lore raw

PATCH v5 4/6 mm: simplify compat numa syscalls

In reply to Arnd Bergmann

From: Arnd Bergmann <[email protected]>

The compat implementations for mbind, get_mempolicy, set_mempolicy and migrate_pages are just there to handle the subtly different layout of bitmaps on 32-bit hosts.

The compat implementation however lacks some of the checks that are present in the native one, in particular for checking that the extra bits are all zero when user space has a larger mask size than the kernel. Worse, those extra bits do not get cleared when copying in or out of the kernel, which can lead to incorrect data as well.

Unify the implementation to handle the compat bitmap layout directly in the get_nodes() and copy_nodes_to_user() helpers. Splitting out the get_bitmap() helper from get_nodes() also helps readability of the native case.

On x86, two additional problems are addressed by this: compat tasks can pass a bitmap at the end of a mapping, causing a fault when reading across the page boundary for a 64-bit word. x32 tasks might also run into problems with get_mempolicy corrupting data when an odd number of 32-bit words gets passed.

On parisc the migrate_pages() system call apparently had the wrong calling convention, as big-endian architectures expect the words inside of a bitmap to be swapped. This is not a problem though since parisc has no NUMA support.

Reviewed-by Christoph Hellwig <[email protected]>
Signed-off-by Arnd Bergmann <[email protected]>
---
 include/linux/compat.h |  17 ++--
 mm/mempolicy.c         | 175 +++++++++++++----------------------------
 2 files changed, 63 insertions(+), 129 deletions(-)
include/linux/compat.h +9 -8
index 8e0598c7d1d1..3a2ac5afee30 100644
@@ -395,14 +395,6 @@ struct compat_kexec_segment;
 struct compat_mq_attr;
 struct compat_msgbuf;
 
-#define BITS_PER_COMPAT_LONG    (8*sizeof(compat_long_t))
-
-#define BITS_TO_COMPAT_LONGS(bits) DIV_ROUND_UP(bits, BITS_PER_COMPAT_LONG)
-
-long compat_get_bitmap(unsigned long *mask, const compat_ulong_t __user *umask,
-		       unsigned long bitmap_size);
-long compat_put_bitmap(compat_ulong_t __user *umask, unsigned long *mask,
-		       unsigned long bitmap_size);
 void copy_siginfo_to_external32(struct compat_siginfo *to,
 		const struct kernel_siginfo *from);
 int copy_siginfo_from_user32(kernel_siginfo_t *to,

@@ -976,6 +968,15 @@ static inline bool in_compat_syscall(void) { return false; }
 
 #endif /* CONFIG_COMPAT */
 
+#define BITS_PER_COMPAT_LONG    (8*sizeof(compat_long_t))
+
+#define BITS_TO_COMPAT_LONGS(bits) DIV_ROUND_UP(bits, BITS_PER_COMPAT_LONG)
+
+long compat_get_bitmap(unsigned long *mask, const compat_ulong_t __user *umask,
+		       unsigned long bitmap_size);
+long compat_put_bitmap(compat_ulong_t __user *umask, unsigned long *mask,
+		       unsigned long bitmap_size);
+
 /*
  * Some legacy ABIs like the i386 one use less than natural alignment for 64-bit
  * types, and will need special compat treatment for that.  Most architectures
mm/mempolicy.c +54 -121

235 lines not on this page · raw message

Signature · 1 line
2.29.2
maintainer
#7 lore raw

Re: [PATCH v5 4/6] mm: simplify compat numa syscalls

Quoted 4 lines · Arnd Bergmann
---
 include/linux/compat.h |  17 ++--
 mm/mempolicy.c         | 175 +++++++++++++----------------------------
 2 files changed, 63 insertions(+), 129 deletions(-)

...

mm/mempolicy.c static long do_mbind(unsigned long start, unsigned long len, · line 1374

Quoted 10 lines
mm/mempolicy.c +10 -0
+static int get_bitmap(unsigned long *mask, const unsigned long __user *nmask,
+		      unsigned long maxnode)
+{
+	unsigned long nlongs = BITS_TO_LONGS(maxnode);
+	int ret;
+
+	if (in_compat_syscall())
+		ret = compat_get_bitmap(mask,
+					(const compat_ulong_t __user *)nmask,
+					maxnode);

compat_ptr() conversion for e.g. nmask is missing with the next patch which removes the compat system calls.
Is that intended or am I missing something?

author
#8 lore raw

Re: [PATCH v5 4/6] mm: simplify compat numa syscalls

Quoted 7 lines · Heiko Carstens
Quoted 4 lines · Arnd Bergmann
---
 include/linux/compat.h |  17 ++--
 mm/mempolicy.c         | 175 +++++++++++++----------------------------
 2 files changed, 63 insertions(+), 129 deletions(-)

...

Quoted 14 lines

mm/mempolicy.c static long do_mbind(unsigned long start, unsigned long len, · line 1374

Quoted 10 lines
mm/mempolicy.c +10 -0
+static int get_bitmap(unsigned long *mask, const unsigned long __user *nmask,
+                   unsigned long maxnode)
+{
+     unsigned long nlongs = BITS_TO_LONGS(maxnode);
+     int ret;
+
+     if (in_compat_syscall())
+             ret = compat_get_bitmap(mask,
+                                     (const compat_ulong_t __user *)nmask,
+                                     maxnode);

compat_ptr() conversion for e.g. nmask is missing with the next patch which removes the compat system calls.
Is that intended or am I missing something?

I don't think it's needed here, since the pointer comes from the system call argument, which has the compat_ptr() conversion applied in arch/s390/include/asm/syscall_wrapper.h, not from a compat_uptr_t that gets passed indirectly. The compat_get_bitmap() conversion is only needed for byte order adjustment, not for converting pointers.

It's also possible that I'm the one who's missing something.

Arnd

maintainer
#9 lore raw

Re: [PATCH v5 4/6] mm: simplify compat numa syscalls

Quoted 16 lines · Arnd Bergmann
Quoted 6 lines · Heiko Carstens
Quoted 6 lines
mm/mempolicy.c +6 -0
+static int get_bitmap(unsigned long *mask, const unsigned long __user *nmask,
+                   unsigned long maxnode)
+{
+     unsigned long nlongs = BITS_TO_LONGS(maxnode);
+     int ret;
+
Quoted 8 lines

mm/mempolicy.c static long do_mbind(unsigned long start, unsigned long len, · line 1374

Quoted 4 lines
mm/mempolicy.c +4 -0
+     if (in_compat_syscall())
+             ret = compat_get_bitmap(mask,
+                                     (const compat_ulong_t __user *)nmask,
+                                     maxnode);

compat_ptr() conversion for e.g. nmask is missing with the next patch which removes the compat system calls.
Is that intended or am I missing something?

Quoted 7 lines

I don't think it's needed here, since the pointer comes from the system call argument, which has the compat_ptr() conversion applied in arch/s390/include/asm/syscall_wrapper.h, not from a compat_uptr_t that gets passed indirectly. The compat_get_bitmap() conversion is only needed for byte order adjustment, not for converting pointers.

It's also possible that I'm the one who's missing something.

What I was trying to say: this patch on its own is ok. However with the next patch you remove the compat system calls and map the regular system calls instead.

That is:

-COMPAT_SYSCALL_DEFINE6(mbind, compat_ulong_t, start, compat_ulong_t, len,
- compat_ulong_t, mode, compat_ulong_t __user *, nmask,
- compat_ulong_t, maxnode, compat_ulong_t, flags)
-{
- return kernel_mbind(start, len, mode, (unsigned long __user *)nmask,
- maxnode, flags);
-}

and this:

-268  common	mbind			sys_mbind			compat_sys_mbind
-269  common	get_mempolicy		sys_get_mempolicy		compat_sys_get_mempolicy
-270  common	set_mempolicy		sys_set_mempolicy		compat_sys_set_mempolicy
+268  common	mbind			sys_mbind			sys_mbind
+269  common	get_mempolicy		sys_get_mempolicy		sys_get_mempolicy
+270  common	set_mempolicy		sys_set_mempolicy		sys_set_mempolicy

would remove compat_ptr() conversion from nmask above if I'm not mistaken.

author
#10 lore raw

Re: [PATCH v5 4/6] mm: simplify compat numa syscalls

Quoted 9 lines · Heiko Carstens
-268  common    mbind                   sys_mbind                       compat_sys_mbind
-269  common    get_mempolicy           sys_get_mempolicy               compat_sys_get_mempolicy
-270  common    set_mempolicy           sys_set_mempolicy               compat_sys_set_mempolicy
+268  common    mbind                   sys_mbind                       sys_mbind
+269  common    get_mempolicy           sys_get_mempolicy               sys_get_mempolicy
+270  common    set_mempolicy           sys_set_mempolicy               sys_set_mempolicy

would remove compat_ptr() conversion from nmask above if I'm not mistaken.

Maybe I'm misremembering how compat syscalls work on s390. Doesn't SYSCALL_DEFINEx(sys_mbind) still create two entry points __s390x_sys_mbind() and __s390_sys_mbind() with different argument conversion (__SC_CAST vs __SC_COMPAT_CAST)? I thought that was the whole point of the macros.

Arnd

maintainer
#11 lore raw

Re: [PATCH v5 4/6] mm: simplify compat numa syscalls

Quoted 9 lines · Arnd Bergmann
Quoted 8 lines · Heiko Carstens
-268  common    mbind                   sys_mbind                       compat_sys_mbind
-269  common    get_mempolicy           sys_get_mempolicy               compat_sys_get_mempolicy
-270  common    set_mempolicy           sys_set_mempolicy               compat_sys_set_mempolicy
+268  common    mbind                   sys_mbind                       sys_mbind
+269  common    get_mempolicy           sys_get_mempolicy               sys_get_mempolicy
+270  common    set_mempolicy           sys_set_mempolicy               sys_set_mempolicy
Quoted 6 lines
Quoted 1 line

would remove compat_ptr() conversion from nmask above if I'm not mistaken.

Maybe I'm misremembering how compat syscalls work on s390. Doesn't SYSCALL_DEFINEx(sys_mbind) still create two entry points __s390x_sys_mbind() and __s390_sys_mbind() with different argument conversion (__SC_CAST vs __SC_COMPAT_CAST)? I thought that was the whole point of the macros.

You are remembering correctly, probably because you implemented it ;) I totally forgot - sorry for the noise!

author
#12 lore raw

PATCH v5 5/6 compat: remove some compat entry points

In reply to Arnd Bergmann

From: Arnd Bergmann <[email protected]>

These are all handled correctly when calling the native system call entry point, so remove the special cases.

Reviewed-by Christoph Hellwig <[email protected]>
Signed-off-by Arnd Bergmann <[email protected]>
---
 arch/arm64/include/asm/unistd32.h         | 10 +++---
 arch/mips/kernel/syscalls/syscall_n32.tbl | 10 +++---
 arch/mips/kernel/syscalls/syscall_o32.tbl | 10 +++---
 arch/parisc/kernel/syscalls/syscall.tbl   |  8 ++---
 arch/powerpc/kernel/syscalls/syscall.tbl  | 10 +++---
 arch/s390/kernel/syscalls/syscall.tbl     | 10 +++---
 arch/sparc/kernel/syscalls/syscall.tbl    | 10 +++---
 arch/x86/entry/syscalls/syscall_32.tbl    |  4 +--
 arch/x86/entry/syscalls/syscall_64.tbl    |  2 +-
 include/linux/compat.h                    | 20 ------------
 include/uapi/asm-generic/unistd.h         | 10 +++---
 kernel/sys_ni.c                           |  5 ---
 mm/mempolicy.c                            | 37 -----------------------
 mm/migrate.c                              | 13 --------
 14 files changed, 42 insertions(+), 117 deletions(-)
arch/arm64/include/asm/unistd32.h +5 -5
index 99ffcafc736c..8a5cf6daf7c0 100644
@@ -649,11 +649,11 @@ __SYSCALL(__NR_inotify_add_watch, sys_inotify_add_watch)
 #define __NR_inotify_rm_watch 318
 __SYSCALL(__NR_inotify_rm_watch, sys_inotify_rm_watch)
 #define __NR_mbind 319
-__SYSCALL(__NR_mbind, compat_sys_mbind)
+__SYSCALL(__NR_mbind, sys_mbind)
 #define __NR_get_mempolicy 320
-__SYSCALL(__NR_get_mempolicy, compat_sys_get_mempolicy)
+__SYSCALL(__NR_get_mempolicy, sys_get_mempolicy)
 #define __NR_set_mempolicy 321
-__SYSCALL(__NR_set_mempolicy, compat_sys_set_mempolicy)
+__SYSCALL(__NR_set_mempolicy, sys_set_mempolicy)
 #define __NR_openat 322
 __SYSCALL(__NR_openat, compat_sys_openat)
 #define __NR_mkdirat 323

@@ -699,7 +699,7 @@ __SYSCALL(__NR_tee, sys_tee)
 #define __NR_vmsplice 343
 __SYSCALL(__NR_vmsplice, sys_vmsplice)
 #define __NR_move_pages 344
-__SYSCALL(__NR_move_pages, compat_sys_move_pages)
+__SYSCALL(__NR_move_pages, sys_move_pages)
 #define __NR_getcpu 345
 __SYSCALL(__NR_getcpu, sys_getcpu)
 #define __NR_epoll_pwait 346

@@ -811,7 +811,7 @@ __SYSCALL(__NR_rseq, sys_rseq)
 #define __NR_io_pgetevents 399
 __SYSCALL(__NR_io_pgetevents, compat_sys_io_pgetevents)
 #define __NR_migrate_pages 400
-__SYSCALL(__NR_migrate_pages, compat_sys_migrate_pages)
+__SYSCALL(__NR_migrate_pages, sys_migrate_pages)
 #define __NR_kexec_file_load 401
 __SYSCALL(__NR_kexec_file_load, sys_kexec_file_load)
 /* 402 is unused */
arch/mips/kernel/syscalls/syscall_n32.tbl +5 -5
index c2d2e19abea8..c42378dbb1c2 100644
@@ -239,9 +239,9 @@
 228	n32	clock_nanosleep			sys_clock_nanosleep_time32
 229	n32	tgkill				sys_tgkill
 230	n32	utimes				sys_utimes_time32
-231	n32	mbind				compat_sys_mbind
-232	n32	get_mempolicy			compat_sys_get_mempolicy
-233	n32	set_mempolicy			compat_sys_set_mempolicy
+231	n32	mbind				sys_mbind
+232	n32	get_mempolicy			sys_get_mempolicy
+233	n32	set_mempolicy			sys_set_mempolicy
 234	n32	mq_open				compat_sys_mq_open
 235	n32	mq_unlink			sys_mq_unlink
 236	n32	mq_timedsend			sys_mq_timedsend_time32

@@ -258,7 +258,7 @@
 247	n32	inotify_init			sys_inotify_init
 248	n32	inotify_add_watch		sys_inotify_add_watch
 249	n32	inotify_rm_watch		sys_inotify_rm_watch
-250	n32	migrate_pages			compat_sys_migrate_pages
+250	n32	migrate_pages			sys_migrate_pages
 251	n32	openat				sys_openat
 252	n32	mkdirat				sys_mkdirat
 253	n32	mknodat				sys_mknodat

@@ -279,7 +279,7 @@
 268	n32	sync_file_range			sys_sync_file_range
 269	n32	tee				sys_tee
 270	n32	vmsplice			sys_vmsplice
-271	n32	move_pages			compat_sys_move_pages
+271	n32	move_pages			sys_move_pages
 272	n32	set_robust_list			compat_sys_set_robust_list
 273	n32	get_robust_list			compat_sys_get_robust_list
 274	n32	kexec_load			compat_sys_kexec_load
arch/mips/kernel/syscalls/syscall_o32.tbl +5 -5
index 253f2cd70b6b..5ff8ef9d8a04 100644
@@ -279,9 +279,9 @@
 265	o32	clock_nanosleep			sys_clock_nanosleep_time32
 266	o32	tgkill				sys_tgkill
 267	o32	utimes				sys_utimes_time32
-268	o32	mbind				sys_mbind			compat_sys_mbind
-269	o32	get_mempolicy			sys_get_mempolicy		compat_sys_get_mempolicy
-270	o32	set_mempolicy			sys_set_mempolicy		compat_sys_set_mempolicy
+268	o32	mbind				sys_mbind
+269	o32	get_mempolicy			sys_get_mempolicy
+270	o32	set_mempolicy			sys_set_mempolicy
 271	o32	mq_open				sys_mq_open			compat_sys_mq_open
 272	o32	mq_unlink			sys_mq_unlink
 273	o32	mq_timedsend			sys_mq_timedsend_time32

@@ -298,7 +298,7 @@
 284	o32	inotify_init			sys_inotify_init
 285	o32	inotify_add_watch		sys_inotify_add_watch
 286	o32	inotify_rm_watch		sys_inotify_rm_watch
-287	o32	migrate_pages			sys_migrate_pages		compat_sys_migrate_pages
+287	o32	migrate_pages			sys_migrate_pages
 288	o32	openat				sys_openat			compat_sys_openat
 289	o32	mkdirat				sys_mkdirat
 290	o32	mknodat				sys_mknodat

@@ -319,7 +319,7 @@
 305	o32	sync_file_range			sys_sync_file_range		sys32_sync_file_range
 306	o32	tee				sys_tee
 307	o32	vmsplice			sys_vmsplice
-308	o32	move_pages			sys_move_pages			compat_sys_move_pages
+308	o32	move_pages			sys_move_pages
 309	o32	set_robust_list			sys_set_robust_list		compat_sys_set_robust_list
 310	o32	get_robust_list			sys_get_robust_list		compat_sys_get_robust_list
 311	o32	kexec_load			sys_kexec_load			compat_sys_kexec_load
arch/parisc/kernel/syscalls/syscall.tbl +4 -4
index e26187b9ab87..21be3d200e7b 100644
@@ -292,9 +292,9 @@
 258	32	clock_nanosleep		sys_clock_nanosleep_time32
 258	64	clock_nanosleep		sys_clock_nanosleep
 259	common	tgkill			sys_tgkill
-260	common	mbind			sys_mbind			compat_sys_mbind
-261	common	get_mempolicy		sys_get_mempolicy		compat_sys_get_mempolicy
-262	common	set_mempolicy		sys_set_mempolicy		compat_sys_set_mempolicy
+260	common	mbind			sys_mbind
+261	common	get_mempolicy		sys_get_mempolicy
+262	common	set_mempolicy		sys_set_mempolicy
 # 263 was vserver
 264	common	add_key			sys_add_key
 265	common	request_key		sys_request_key

@@ -331,7 +331,7 @@
 292	64	sync_file_range		sys_sync_file_range
 293	common	tee			sys_tee
 294	common	vmsplice		sys_vmsplice
-295	common	move_pages		sys_move_pages			compat_sys_move_pages
+295	common	move_pages		sys_move_pages
 296	common	getcpu			sys_getcpu
 297	common	epoll_pwait		sys_epoll_pwait			compat_sys_epoll_pwait
 298	common	statfs64		sys_statfs64			compat_sys_statfs64
arch/powerpc/kernel/syscalls/syscall.tbl +5 -5
index aef2a290e71a..9e237232d108 100644
@@ -330,10 +330,10 @@
 256	64	sys_debug_setcontext		sys_ni_syscall
 256	spu	sys_debug_setcontext		sys_ni_syscall
 # 257 reserved for vserver
-258	nospu	migrate_pages			sys_migrate_pages		compat_sys_migrate_pages
-259	nospu	mbind				sys_mbind			compat_sys_mbind
-260	nospu	get_mempolicy			sys_get_mempolicy		compat_sys_get_mempolicy
-261	nospu	set_mempolicy			sys_set_mempolicy		compat_sys_set_mempolicy
+258	nospu	migrate_pages			sys_migrate_pages
+259	nospu	mbind				sys_mbind
+260	nospu	get_mempolicy			sys_get_mempolicy
+261	nospu	set_mempolicy			sys_set_mempolicy
 262	nospu	mq_open				sys_mq_open			compat_sys_mq_open
 263	nospu	mq_unlink			sys_mq_unlink
 264	32	mq_timedsend			sys_mq_timedsend_time32

@@ -381,7 +381,7 @@
 298	common	faccessat			sys_faccessat
 299	common	get_robust_list			sys_get_robust_list		compat_sys_get_robust_list
 300	common	set_robust_list			sys_set_robust_list		compat_sys_set_robust_list
-301	common	move_pages			sys_move_pages			compat_sys_move_pages
+301	common	move_pages			sys_move_pages
 302	common	getcpu				sys_getcpu
 303	nospu	epoll_pwait			sys_epoll_pwait			compat_sys_epoll_pwait
 304	32	utimensat			sys_utimensat_time32
arch/s390/kernel/syscalls/syscall.tbl +5 -5
index 64d51ab5a8b4..81bff2ffe6c0 100644
@@ -274,9 +274,9 @@
 265  common	statfs64		sys_statfs64			compat_sys_statfs64
 266  common	fstatfs64		sys_fstatfs64			compat_sys_fstatfs64
 267  common	remap_file_pages	sys_remap_file_pages		sys_remap_file_pages
-268  common	mbind			sys_mbind			compat_sys_mbind
-269  common	get_mempolicy		sys_get_mempolicy		compat_sys_get_mempolicy
-270  common	set_mempolicy		sys_set_mempolicy		compat_sys_set_mempolicy
+268  common	mbind			sys_mbind			sys_mbind
+269  common	get_mempolicy		sys_get_mempolicy		sys_get_mempolicy
+270  common	set_mempolicy		sys_set_mempolicy		sys_set_mempolicy
 271  common	mq_open			sys_mq_open			compat_sys_mq_open
 272  common	mq_unlink		sys_mq_unlink			sys_mq_unlink
 273  common	mq_timedsend		sys_mq_timedsend		sys_mq_timedsend_time32

@@ -293,7 +293,7 @@
 284  common	inotify_init		sys_inotify_init		sys_inotify_init
 285  common	inotify_add_watch	sys_inotify_add_watch		sys_inotify_add_watch
 286  common	inotify_rm_watch	sys_inotify_rm_watch		sys_inotify_rm_watch
-287  common	migrate_pages		sys_migrate_pages		compat_sys_migrate_pages
+287  common	migrate_pages		sys_migrate_pages		sys_migrate_pages
 288  common	openat			sys_openat			compat_sys_openat
 289  common	mkdirat			sys_mkdirat			sys_mkdirat
 290  common	mknodat			sys_mknodat			sys_mknodat

@@ -317,7 +317,7 @@
 307  common	sync_file_range		sys_sync_file_range		compat_sys_s390_sync_file_range
 308  common	tee			sys_tee				sys_tee
 309  common	vmsplice		sys_vmsplice			sys_vmsplice
-310  common	move_pages		sys_move_pages			compat_sys_move_pages
+310  common	move_pages		sys_move_pages			sys_move_pages
 311  common	getcpu			sys_getcpu			sys_getcpu
 312  common	epoll_pwait		sys_epoll_pwait			compat_sys_epoll_pwait
 313  common	utimes			sys_utimes			sys_utimes_time32
arch/sparc/kernel/syscalls/syscall.tbl +5 -5
index 603f5a821502..ea1660a933c0 100644
@@ -365,12 +365,12 @@
 299	common	unshare			sys_unshare
 300	common	set_robust_list		sys_set_robust_list		compat_sys_set_robust_list
 301	common	get_robust_list		sys_get_robust_list		compat_sys_get_robust_list
-302	common	migrate_pages		sys_migrate_pages		compat_sys_migrate_pages
-303	common	mbind			sys_mbind			compat_sys_mbind
-304	common	get_mempolicy		sys_get_mempolicy		compat_sys_get_mempolicy
-305	common	set_mempolicy		sys_set_mempolicy		compat_sys_set_mempolicy
+302	common	migrate_pages		sys_migrate_pages
+303	common	mbind			sys_mbind
+304	common	get_mempolicy		sys_get_mempolicy
+305	common	set_mempolicy		sys_set_mempolicy
 306	common	kexec_load		sys_kexec_load			compat_sys_kexec_load
-307	common	move_pages		sys_move_pages			compat_sys_move_pages
+307	common	move_pages		sys_move_pages
 308	common	getcpu			sys_getcpu
 309	common	epoll_pwait		sys_epoll_pwait			compat_sys_epoll_pwait
 310	32	utimensat		sys_utimensat_time32
arch/x86/entry/syscalls/syscall_32.tbl +2 -2
index ce763a12311c..f54c006603d6 100644
@@ -286,7 +286,7 @@
 272	i386	fadvise64_64		sys_ia32_fadvise64_64
 273	i386	vserver
 274	i386	mbind			sys_mbind
-275	i386	get_mempolicy		sys_get_mempolicy		compat_sys_get_mempolicy
+275	i386	get_mempolicy		sys_get_mempolicy
 276	i386	set_mempolicy		sys_set_mempolicy
 277	i386	mq_open			sys_mq_open			compat_sys_mq_open
 278	i386	mq_unlink		sys_mq_unlink

@@ -328,7 +328,7 @@
 314	i386	sync_file_range		sys_ia32_sync_file_range
 315	i386	tee			sys_tee
 316	i386	vmsplice		sys_vmsplice
-317	i386	move_pages		sys_move_pages			compat_sys_move_pages
+317	i386	move_pages		sys_move_pages
 318	i386	getcpu			sys_getcpu
 319	i386	epoll_pwait		sys_epoll_pwait
 320	i386	utimensat		sys_utimensat_time32
arch/x86/entry/syscalls/syscall_64.tbl +1 -1
index f6b57799c1ea..f3dfb5b97e15 100644
@@ -397,7 +397,7 @@
 530	x32	set_robust_list		compat_sys_set_robust_list
 531	x32	get_robust_list		compat_sys_get_robust_list
 532	x32	vmsplice		sys_vmsplice
-533	x32	move_pages		compat_sys_move_pages
+533	x32	move_pages		sys_move_pages
 534	x32	preadv			compat_sys_preadv64
 535	x32	pwritev			compat_sys_pwritev64
 536	x32	rt_tgsigqueueinfo	compat_sys_rt_tgsigqueueinfo
include/linux/compat.h +0 -20
index 3a2ac5afee30..2d42cebd1fb8 100644
@@ -799,26 +799,6 @@ asmlinkage long compat_sys_execve(const char __user *filename, const compat_uptr
 /* mm/fadvise.c: No generic prototype for fadvise64_64 */
 
 /* mm/, CONFIG_MMU only */
-asmlinkage long compat_sys_mbind(compat_ulong_t start, compat_ulong_t len,
-				 compat_ulong_t mode,
-				 compat_ulong_t __user *nmask,
-				 compat_ulong_t maxnode, compat_ulong_t flags);
-asmlinkage long compat_sys_get_mempolicy(int __user *policy,
-					 compat_ulong_t __user *nmask,
-					 compat_ulong_t maxnode,
-					 compat_ulong_t addr,
-					 compat_ulong_t flags);
-asmlinkage long compat_sys_set_mempolicy(int mode, compat_ulong_t __user *nmask,
-					 compat_ulong_t maxnode);
-asmlinkage long compat_sys_migrate_pages(compat_pid_t pid,
-		compat_ulong_t maxnode, const compat_ulong_t __user *old_nodes,
-		const compat_ulong_t __user *new_nodes);
-asmlinkage long compat_sys_move_pages(pid_t pid, compat_ulong_t nr_pages,
-				      __u32 __user *pages,
-				      const int __user *nodes,
-				      int __user *status,
-				      int flags);
-
 asmlinkage long compat_sys_rt_tgsigqueueinfo(compat_pid_t tgid,
 					compat_pid_t pid, int sig,
 					struct compat_siginfo __user *uinfo);
include/uapi/asm-generic/unistd.h +5 -5
index a9d6fcd95f42..cde0439913b4 100644
@@ -673,15 +673,15 @@ __SYSCALL(__NR_madvise, sys_madvise)
 #define __NR_remap_file_pages 234
 __SYSCALL(__NR_remap_file_pages, sys_remap_file_pages)
 #define __NR_mbind 235
-__SC_COMP(__NR_mbind, sys_mbind, compat_sys_mbind)
+__SYSCALL(__NR_mbind, sys_mbind)
 #define __NR_get_mempolicy 236
-__SC_COMP(__NR_get_mempolicy, sys_get_mempolicy, compat_sys_get_mempolicy)
+__SYSCALL(__NR_get_mempolicy, sys_get_mempolicy)
 #define __NR_set_mempolicy 237
-__SC_COMP(__NR_set_mempolicy, sys_set_mempolicy, compat_sys_set_mempolicy)
+__SYSCALL(__NR_set_mempolicy, sys_set_mempolicy)
 #define __NR_migrate_pages 238
-__SC_COMP(__NR_migrate_pages, sys_migrate_pages, compat_sys_migrate_pages)
+__SYSCALL(__NR_migrate_pages, sys_migrate_pages)
 #define __NR_move_pages 239
-__SC_COMP(__NR_move_pages, sys_move_pages, compat_sys_move_pages)
+__SYSCALL(__NR_move_pages, sys_move_pages)
 #endif
 
 #define __NR_rt_tgsigqueueinfo 240
kernel/sys_ni.c +0 -5
index 30971b1dd4a9..52c27ecf5ff7 100644
@@ -291,15 +291,10 @@ COND_SYSCALL(madvise);
 COND_SYSCALL(process_madvise);
 COND_SYSCALL(remap_file_pages);
 COND_SYSCALL(mbind);
-COND_SYSCALL_COMPAT(mbind);
 COND_SYSCALL(get_mempolicy);
-COND_SYSCALL_COMPAT(get_mempolicy);
 COND_SYSCALL(set_mempolicy);
-COND_SYSCALL_COMPAT(set_mempolicy);
 COND_SYSCALL(migrate_pages);
-COND_SYSCALL_COMPAT(migrate_pages);
 COND_SYSCALL(move_pages);
-COND_SYSCALL_COMPAT(move_pages);
 
 COND_SYSCALL(perf_event_open);
 COND_SYSCALL(accept4);
mm/mempolicy.c +0 -37
index 9bc5dabdb7a7..6fdfc8ab6770 100644
@@ -1647,43 +1647,6 @@ SYSCALL_DEFINE5(get_mempolicy, int __user *, policy,
 	return kernel_get_mempolicy(policy, nmask, maxnode, addr, flags);
 }
 
-#ifdef CONFIG_COMPAT
-
-COMPAT_SYSCALL_DEFINE5(get_mempolicy, int __user *, policy,
-		       compat_ulong_t __user *, nmask,
-		       compat_ulong_t, maxnode,
-		       compat_ulong_t, addr, compat_ulong_t, flags)
-{
-	return kernel_get_mempolicy(policy, (unsigned long __user *)nmask,
-				    maxnode, addr, flags);
-}
-
-COMPAT_SYSCALL_DEFINE3(set_mempolicy, int, mode, compat_ulong_t __user *, nmask,
-		       compat_ulong_t, maxnode)
-{
-	return kernel_set_mempolicy(mode, (unsigned long __user *)nmask, maxnode);
-}
-
-COMPAT_SYSCALL_DEFINE6(mbind, compat_ulong_t, start, compat_ulong_t, len,
-		       compat_ulong_t, mode, compat_ulong_t __user *, nmask,
-		       compat_ulong_t, maxnode, compat_ulong_t, flags)
-{
-	return kernel_mbind(start, len, mode, (unsigned long __user *)nmask,
-			    maxnode, flags);
-}
-
-COMPAT_SYSCALL_DEFINE4(migrate_pages, compat_pid_t, pid,
-		       compat_ulong_t, maxnode,
-		       const compat_ulong_t __user *, old_nodes,
-		       const compat_ulong_t __user *, new_nodes)
-{
-	return kernel_migrate_pages(pid, maxnode,
-				    (const unsigned long __user *)old_nodes,
-				    (const unsigned long __user *)new_nodes);
-}
-
-#endif /* CONFIG_COMPAT */
-
 bool vma_migratable(struct vm_area_struct *vma)
 {
 	if (vma->vm_flags & (VM_IO | VM_PFNMAP))
mm/migrate.c +0 -13
index 60634d2653b9..266ae1837b6a 100644
@@ -1967,19 +1967,6 @@ SYSCALL_DEFINE6(move_pages, pid_t, pid, unsigned long, nr_pages,
 	return kernel_move_pages(pid, nr_pages, pages, nodes, status, flags);
 }
 
-#ifdef CONFIG_COMPAT
-COMPAT_SYSCALL_DEFINE6(move_pages, pid_t, pid, compat_ulong_t, nr_pages,
-		       compat_uptr_t __user *, pages,
-		       const int __user *, nodes,
-		       int __user *, status,
-		       int, flags)
-{
-	return kernel_move_pages(pid, nr_pages,
-				 (const void __user *__user *)pages,
-				 nodes, status, flags);
-}
-#endif /* CONFIG_COMPAT */
-
 #ifdef CONFIG_NUMA_BALANCING
 /*
  * Returns true if this is a safe migration target node for misplaced NUMA
Signature · 1 line
2.29.2
author
#13 lore raw

PATCH v5 6/6 arch: remove compat_alloc_user_space

In reply to Arnd Bergmann

From: Arnd Bergmann <[email protected]>

All users of compat_alloc_user_space() and copy_in_user() have been removed from the kernel, only a few functions in sparc remain that can be changed to calling arch_copy_in_user() instead.

Signed-off-by Arnd Bergmann <[email protected]>
---
 arch/arm64/include/asm/compat.h         |  5 --
 arch/arm64/include/asm/uaccess.h        | 11 ----
 arch/arm64/lib/Makefile                 |  2 +-
 arch/arm64/lib/copy_in_user.S           | 77 -------------------------
 arch/mips/cavium-octeon/octeon-memcpy.S |  2 -
 arch/mips/include/asm/compat.h          |  8 ---
 arch/mips/include/asm/uaccess.h         | 26 ---------
 arch/mips/lib/memcpy.S                  | 11 ----
 arch/parisc/include/asm/compat.h        |  6 --
 arch/parisc/include/asm/uaccess.h       |  2 -
 arch/parisc/lib/memcpy.c                |  9 ---
 arch/powerpc/include/asm/compat.h       | 16 -----
 arch/s390/include/asm/compat.h          | 10 ----
 arch/s390/include/asm/uaccess.h         |  3 -
 arch/s390/lib/uaccess.c                 | 63 --------------------
 arch/sparc/include/asm/compat.h         | 19 ------
 arch/sparc/kernel/process_64.c          |  2 +-
 arch/sparc/kernel/signal32.c            | 12 ++--
 arch/sparc/kernel/signal_64.c           |  8 +--
 arch/x86/include/asm/compat.h           | 13 -----
 arch/x86/include/asm/uaccess_64.h       |  7 ---
 include/linux/compat.h                  |  2 -
 include/linux/uaccess.h                 | 10 ----
 kernel/compat.c                         | 21 -------
 24 files changed, 12 insertions(+), 333 deletions(-)
 delete mode 100644 arch/arm64/lib/copy_in_user.S
arch/arm64/include/asm/compat.h +0 -5
index 79c1a750e357..eaa6ca062d89 100644
@@ -107,11 +107,6 @@ struct compat_statfs {
 #define compat_user_stack_pointer() (user_stack_pointer(task_pt_regs(current)))
 #define COMPAT_MINSIGSTKSZ	2048
 
-static inline void __user *arch_compat_alloc_user_space(long len)
-{
-	return (void __user *)compat_user_stack_pointer() - len;
-}
-
 struct compat_ipc64_perm {
 	compat_key_t key;
 	__compat_uid32_t uid;
arch/arm64/include/asm/uaccess.h +0 -11
index b5f08621fa29..190b494e22ab 100644
@@ -430,17 +430,6 @@ extern unsigned long __must_check __arch_copy_to_user(void __user *to, const voi
 	__actu_ret;							\
 })
 
-extern unsigned long __must_check __arch_copy_in_user(void __user *to, const void __user *from, unsigned long n);
-#define raw_copy_in_user(to, from, n)					\
-({									\
-	unsigned long __aciu_ret;					\
-	uaccess_ttbr0_enable();						\
-	__aciu_ret = __arch_copy_in_user(__uaccess_mask_ptr(to),	\
-				    __uaccess_mask_ptr(from), (n));	\
-	uaccess_ttbr0_disable();					\
-	__aciu_ret;							\
-})
-
 #define INLINE_COPY_TO_USER
 #define INLINE_COPY_FROM_USER
arch/arm64/lib/Makefile +1 -1
index 6dd56a49790a..0941180a86d3 100644
@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0
 lib-y		:= clear_user.o delay.o copy_from_user.o		\
-		   copy_to_user.o copy_in_user.o copy_page.o		\
+		   copy_to_user.o copy_page.o				\
 		   clear_page.o csum.o insn.o memchr.o memcpy.o		\
 		   memset.o memcmp.o strcmp.o strncmp.o strlen.o	\
 		   strnlen.o strchr.o strrchr.o tishift.o
arch/arm64/lib/copy_in_user.S deleted +0 -77

77 lines not on this page · raw message

arch/mips/cavium-octeon/octeon-memcpy.S +0 -2
index 600d018cf354..0a515cde1c18 100644
@@ -154,8 +154,6 @@ FEXPORT(__raw_copy_from_user)
 EXPORT_SYMBOL(__raw_copy_from_user)
 FEXPORT(__raw_copy_to_user)
 EXPORT_SYMBOL(__raw_copy_to_user)
-FEXPORT(__raw_copy_in_user)
-EXPORT_SYMBOL(__raw_copy_in_user)
 	/*
 	 * Note: dst & src may be unaligned, len may be 0
 	 * Temps
arch/mips/include/asm/compat.h +0 -8
index 53f015a1b0a7..bbb3bc5a42fd 100644
@@ -96,14 +96,6 @@ struct compat_statfs {
 
 #define COMPAT_OFF_T_MAX	0x7fffffff
 
-static inline void __user *arch_compat_alloc_user_space(long len)
-{
-	struct pt_regs *regs = (struct pt_regs *)
-		((unsigned long) current_thread_info() + THREAD_SIZE - 32) - 1;
-
-	return (void __user *) (regs->regs[29] - len);
-}
-
 struct compat_ipc64_perm {
 	compat_key_t key;
 	__compat_uid32_t uid;
arch/mips/include/asm/uaccess.h +0 -26
index 783fecce65c8..f8f74f9f5883 100644
@@ -428,7 +428,6 @@ do {									\
 
 extern size_t __raw_copy_from_user(void *__to, const void *__from, size_t __n);
 extern size_t __raw_copy_to_user(void *__to, const void *__from, size_t __n);
-extern size_t __raw_copy_in_user(void *__to, const void *__from, size_t __n);
 
 static inline unsigned long
 raw_copy_from_user(void *to, const void __user *from, unsigned long n)

@@ -480,31 +479,6 @@ raw_copy_to_user(void __user *to, const void *from, unsigned long n)
 #define INLINE_COPY_FROM_USER
 #define INLINE_COPY_TO_USER
 
-static inline unsigned long
-raw_copy_in_user(void __user *to, const void __user *from, unsigned long n)
-{
-	register void __user *__cu_to_r __asm__("$4");
-	register const void __user *__cu_from_r __asm__("$5");
-	register long __cu_len_r __asm__("$6");
-
-	__cu_to_r = to;
-	__cu_from_r = from;
-	__cu_len_r = n;
-
-	__asm__ __volatile__(
-		".set\tnoreorder\n\t"
-		__MODULE_JAL(__raw_copy_in_user)
-		".set\tnoat\n\t"
-		__UA_ADDU "\t$1, %1, %2\n\t"
-		".set\tat\n\t"
-		".set\treorder"
-		: "+r" (__cu_to_r), "+r" (__cu_from_r), "+r" (__cu_len_r)
-		:
-		: "$8", "$9", "$10", "$11", "$12", "$14", "$15", "$24", "$31",
-		  DADDI_SCRATCH, "memory");
-	return __cu_len_r;
-}
-
 extern __kernel_size_t __bzero(void __user *addr, __kernel_size_t size);
 
 /*
arch/mips/lib/memcpy.S +0 -11
index e19fb98b5d38..277c32296636 100644
@@ -666,8 +666,6 @@ FEXPORT(__raw_copy_from_user)
 EXPORT_SYMBOL(__raw_copy_from_user)
 FEXPORT(__raw_copy_to_user)
 EXPORT_SYMBOL(__raw_copy_to_user)
-FEXPORT(__raw_copy_in_user)
-EXPORT_SYMBOL(__raw_copy_in_user)
 #endif
 	/* Legacy Mode, user <-> user */
 	__BUILD_COPY_USER LEGACY_MODE USEROP USEROP

@@ -703,13 +701,4 @@ EXPORT_SYMBOL(__raw_copy_to_user)
 __BUILD_COPY_USER EVA_MODE KERNELOP USEROP
 END(__raw_copy_to_user)
 
-/*
- * __copy_in_user (EVA)
- */
-
-LEAF(__raw_copy_in_user)
-EXPORT_SYMBOL(__raw_copy_in_user)
-__BUILD_COPY_USER EVA_MODE USEROP USEROP
-END(__raw_copy_in_user)
-
 #endif
arch/parisc/include/asm/compat.h +0 -6
index b5d90e82b65d..c04f5a637c39 100644
@@ -163,12 +163,6 @@ struct compat_shmid64_ds {
 #define COMPAT_ELF_NGREG 80
 typedef compat_ulong_t compat_elf_gregset_t[COMPAT_ELF_NGREG];
 
-static __inline__ void __user *arch_compat_alloc_user_space(long len)
-{
-	struct pt_regs *regs = &current->thread.regs;
-	return (void __user *)regs->gr[30];
-}
-
 static inline int __is_compat_task(struct task_struct *t)
 {
 	return test_tsk_thread_flag(t, TIF_32BIT);
arch/parisc/include/asm/uaccess.h +0 -2
index ed2cd4fb479b..7c13314aae4a 100644
@@ -215,8 +215,6 @@ unsigned long __must_check raw_copy_to_user(void __user *dst, const void *src,
 					    unsigned long len);
 unsigned long __must_check raw_copy_from_user(void *dst, const void __user *src,
 					    unsigned long len);
-unsigned long __must_check raw_copy_in_user(void __user *dst, const void __user *src,
-					    unsigned long len);
 #define INLINE_COPY_TO_USER
 #define INLINE_COPY_FROM_USER
arch/parisc/lib/memcpy.c +0 -9
index 4b75388190b4..ea70a0e08321 100644
@@ -38,14 +38,6 @@ unsigned long raw_copy_from_user(void *dst, const void __user *src,
 }
 EXPORT_SYMBOL(raw_copy_from_user);
 
-unsigned long raw_copy_in_user(void __user *dst, const void __user *src, unsigned long len)
-{
-	mtsp(get_user_space(), 1);
-	mtsp(get_user_space(), 2);
-	return pa_memcpy((void __force *)dst, (void __force *)src, len);
-}
-
-
 void * memcpy(void * dst,const void *src, size_t count)
 {
 	mtsp(get_kernel_space(), 1);

@@ -54,7 +46,6 @@ void * memcpy(void * dst,const void *src, size_t count)
 	return dst;
 }
 
-EXPORT_SYMBOL(raw_copy_in_user);
 EXPORT_SYMBOL(memcpy);
 
 bool copy_from_kernel_nofault_allowed(const void *unsafe_src, size_t size)
arch/powerpc/include/asm/compat.h +0 -16
index e33dcf134cdd..7afc96fb6524 100644
@@ -83,22 +83,6 @@ struct compat_statfs {
 
 #define COMPAT_OFF_T_MAX	0x7fffffff
 
-static inline void __user *arch_compat_alloc_user_space(long len)
-{
-	struct pt_regs *regs = current->thread.regs;
-	unsigned long usp = regs->gpr[1];
-
-	/*
-	 * We can't access below the stack pointer in the 32bit ABI and
-	 * can access 288 bytes in the 64bit big-endian ABI,
-	 * or 512 bytes with the new ELFv2 little-endian ABI.
-	 */
-	if (!is_32bit_task())
-		usp -= USER_REDZONE_SIZE;
-
-	return (void __user *) (usp - len);
-}
-
 /*
  * ipc64_perm is actually 32/64bit clean but since the compat layer refers to
  * it we may as well define it.
arch/s390/include/asm/compat.h +0 -10
index 8d49505b4a43..cdc7ae72529d 100644
@@ -176,16 +176,6 @@ static inline int is_compat_task(void)
 	return test_thread_flag(TIF_31BIT);
 }
 
-static inline void __user *arch_compat_alloc_user_space(long len)
-{
-	unsigned long stack;
-
-	stack = KSTK_ESP(current);
-	if (is_compat_task())
-		stack &= 0x7fffffffUL;
-	return (void __user *) (stack - len);
-}
-
 #endif
 
 struct compat_ipc64_perm {
arch/s390/include/asm/uaccess.h +0 -3
index 2316f2440881..98a1fce2ce67 100644
@@ -227,9 +227,6 @@ static inline int __get_user_fn(void *x, const void __user *ptr, unsigned long s
 	__get_user(x, ptr);					\
 })
 
-unsigned long __must_check
-raw_copy_in_user(void __user *to, const void __user *from, unsigned long n);
-
 /*
  * Copy a null terminated string from userspace.
  */
arch/s390/lib/uaccess.c +0 -63

69 lines not on this page · raw message

arch/sparc/include/asm/compat.h +0 -19
index 8b63410e830f..bd949fcf9d63 100644
@@ -116,25 +116,6 @@ struct compat_statfs {
 
 #define COMPAT_OFF_T_MAX	0x7fffffff
 
-#ifdef CONFIG_COMPAT
-static inline void __user *arch_compat_alloc_user_space(long len)
-{
-	struct pt_regs *regs = current_thread_info()->kregs;
-	unsigned long usp = regs->u_regs[UREG_I6];
-
-	if (test_thread_64bit_stack(usp))
-		usp += STACK_BIAS;
-
-	if (test_thread_flag(TIF_32BIT))
-		usp &= 0xffffffffUL;
-
-	usp -= len;
-	usp &= ~0x7UL;
-
-	return (void __user *) usp;
-}
-#endif
-
 struct compat_ipc64_perm {
 	compat_key_t key;
 	__compat_uid32_t uid;
arch/sparc/kernel/process_64.c +1 -1
index d33c58a58d4f..4f5967ebf4e5 100644
@@ -458,7 +458,7 @@ static unsigned long clone_stackframe(unsigned long csp, unsigned long psp)
 
 	distance = fp - psp;
 	rval = (csp - distance);
-	if (copy_in_user((void __user *) rval, (void __user *) psp, distance))
+	if (raw_copy_in_user((void __user *)rval, (void __user *)psp, distance))
 		rval = 0;
 	else if (!stack_64bit) {
 		if (put_user(((u32)csp),
arch/sparc/kernel/signal32.c +6 -6
index e9695a06492f..afd8455ea05f 100644
@@ -435,9 +435,9 @@ static int setup_frame32(struct ksignal *ksig, struct pt_regs *regs,
 			      (_COMPAT_NSIG_WORDS - 1) * sizeof(unsigned int));
 
 	if (!wsaved) {
-		err |= copy_in_user((u32 __user *)sf,
-				    (u32 __user *)(regs->u_regs[UREG_FP]),
-				    sizeof(struct reg_window32));
+		err |= raw_copy_in_user((u32 __user *)sf,
+					(u32 __user *)(regs->u_regs[UREG_FP]),
+					sizeof(struct reg_window32));
 	} else {
 		struct reg_window *rp;
 

@@ -567,9 +567,9 @@ static int setup_rt_frame32(struct ksignal *ksig, struct pt_regs *regs,
 	err |= put_compat_sigset(&sf->mask, oldset, sizeof(compat_sigset_t));
 
 	if (!wsaved) {
-		err |= copy_in_user((u32 __user *)sf,
-				    (u32 __user *)(regs->u_regs[UREG_FP]),
-				    sizeof(struct reg_window32));
+		err |= raw_copy_in_user((u32 __user *)sf,
+					(u32 __user *)(regs->u_regs[UREG_FP]),
+					sizeof(struct reg_window32));
 	} else {
 		struct reg_window *rp;
arch/sparc/kernel/signal_64.c +4 -4
index a0eec62c825d..3bc4a25f7c25 100644
@@ -406,10 +406,10 @@ setup_rt_frame(struct ksignal *ksig, struct pt_regs *regs)
 	err |= copy_to_user(&sf->mask, sigmask_to_save(), sizeof(sigset_t));
 
 	if (!wsaved) {
-		err |= copy_in_user((u64 __user *)sf,
-				    (u64 __user *)(regs->u_regs[UREG_FP] +
-						   STACK_BIAS),
-				    sizeof(struct reg_window));
+		err |= raw_copy_in_user((u64 __user *)sf,
+					(u64 __user *)(regs->u_regs[UREG_FP] +
+					   STACK_BIAS),
+					sizeof(struct reg_window));
 	} else {
 		struct reg_window *rp;
arch/x86/include/asm/compat.h +0 -13
index 4ae01cdb99de..7516e4199b3c 100644
@@ -156,19 +156,6 @@ struct compat_shmid64_ds {
 	(!!(task_pt_regs(current)->orig_ax & __X32_SYSCALL_BIT))
 #endif
 
-static inline void __user *arch_compat_alloc_user_space(long len)
-{
-	compat_uptr_t sp = task_pt_regs(current)->sp;
-
-	/*
-	 * -128 for the x32 ABI redzone.  For IA32, it is not strictly
-	 * necessary, but not harmful.
-	 */
-	sp -= 128;
-
-	return (void __user *)round_down(sp - len, 16);
-}
-
 static inline bool in_x32_syscall(void)
 {
 #ifdef CONFIG_X86_X32_ABI
arch/x86/include/asm/uaccess_64.h +0 -7
index e7265a552f4f..45697e04d771 100644
@@ -58,13 +58,6 @@ raw_copy_to_user(void __user *dst, const void *src, unsigned long size)
 	return copy_user_generic((__force void *)dst, src, size);
 }
 
-static __always_inline __must_check
-unsigned long raw_copy_in_user(void __user *dst, const void __user *src, unsigned long size)
-{
-	return copy_user_generic((__force void *)dst,
-				 (__force void *)src, size);
-}
-
 extern long __copy_user_nocache(void *dst, const void __user *src,
 				unsigned size, int zerorest);
include/linux/compat.h +0 -2
index 2d42cebd1fb8..1c758b0e0359 100644
@@ -511,8 +511,6 @@ extern long compat_arch_ptrace(struct task_struct *child, compat_long_t request,
 
 struct epoll_event;	/* fortunately, this one is fixed-layout */
 
-extern void __user *compat_alloc_user_space(unsigned long len);
-
 int compat_restore_altstack(const compat_stack_t __user *uss);
 int __compat_save_altstack(compat_stack_t __user *, unsigned long);
 #define unsafe_compat_save_altstack(uss, sp, label) do { \
include/linux/uaccess.h +0 -10
index c05e903cef02..ac0394087f7d 100644
@@ -200,16 +200,6 @@ copy_to_user(void __user *to, const void *from, unsigned long n)
 		n = _copy_to_user(to, from, n);
 	return n;
 }
-#ifdef CONFIG_COMPAT
-static __always_inline unsigned long __must_check
-copy_in_user(void __user *to, const void __user *from, unsigned long n)
-{
-	might_fault();
-	if (access_ok(to, n) && access_ok(from, n))
-		n = raw_copy_in_user(to, from, n);
-	return n;
-}
-#endif
 
 #ifndef copy_mc_to_kernel
 /*
kernel/compat.c +0 -21
index 05adfd6fa8bf..55551989d9da 100644
@@ -269,24 +269,3 @@ get_compat_sigset(sigset_t *set, const compat_sigset_t __user *compat)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(get_compat_sigset);
-
-/*
- * Allocate user-space memory for the duration of a single system call,
- * in order to marshall parameters inside a compat thunk.
- */
-void __user *compat_alloc_user_space(unsigned long len)
-{
-	void __user *ptr;
-
-	/* If len would occupy more than half of the entire compat space... */
-	if (unlikely(len > (((compat_uptr_t)~0) >> 1)))
-		return NULL;
-
-	ptr = arch_compat_alloc_user_space(len);
-
-	if (unlikely(!access_ok(ptr, len)))
-		return NULL;
-
-	return ptr;
-}
-EXPORT_SYMBOL_GPL(compat_alloc_user_space);
Signature · 1 line
2.29.2
maintainer
#16 lore raw
Quoted 6 lines · On Tue, 27 Jul 2021 15:59:55 +0100 Christoph Hellwig <[email protected]> wrote:
Quoted 3 lines · Arnd Bergmann

Since these patches are now all that remains, it would be nice to merge it all through Andrew's Linux-mm tree, which is already based on top of linux-next.

Is it?

the -mm tree is structured as

<90% of stuff>
linux-next.patch
<the other 10% of stuff>

So things like Arnd's series which have a dependency on linux-next material get added to the "other 10%" and are merged behind the linux-next material and all is good.

If possible I'll queue things ahead of linux-next.patch. Those few things which have dependencies on linux-next material get sent to Linus after the required linux-next material is merged into mainline.

author
#17 lore raw
Quoted 16 lines · Andrew Morton
Quoted 6 lines · On Tue, 27 Jul 2021 15:59:55 +0100 Christoph Hellwig <[email protected]> wrote:
Quoted 3 lines · Arnd Bergmann

Since these patches are now all that remains, it would be nice to merge it all through Andrew's Linux-mm tree, which is already based on top of linux-next.

Is it?

the -mm tree is structured as

<90% of stuff>
linux-next.patch
<the other 10% of stuff>

Quoted 7 lines

So things like Arnd's series which have a dependency on linux-next material get added to the "other 10%" and are merged behind the linux-next material and all is good.

If possible I'll queue things ahead of linux-next.patch. Those few things which have dependencies on linux-next material get sent to Linus after the required linux-next material is merged into mainline.

The first five patches in my series should apply cleanly on mainline kernels and make sense by themselves, the last patch is the one that depends on this series as well as another series in the netdev tree, so that has to go behind linux-next.

I suppose I could also merge the first five through my asm-generic tree and send you the last one if you prefer, but then again two of the patches are actually memory management stuff.

Arnd

maintainer
#18 lore raw
In reply to Arnd Bergmann
Quoted 14 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

Going through compat_alloc_user_space() to convert indirect system call arguments tends to add complexity compared to handling the native and compat logic in the same code.

Out of the other remaining callers, the linux-media series went into v5.14, and the network ioctl handling is now fixed in net-next, so these are the last remaining users, and I now include the final patch to remove the definitions as well.

Since these patches are now all that remains, it would be nice to merge it all through Andrew's Linux-mm tree, which is already based on top of linux-next.

...

Quoted 8 lines

Arnd Bergmann (6):
kexec: move locking into do_kexec_load
kexec: avoid compat_alloc_user_space
mm: simplify compat_sys_move_pages
mm: simplify compat numa syscalls
compat: remove some compat entry points
arch: remove compat_alloc_user_space

Our CI reports this with linux-next and running strace selftest in compat mode:

Unable to handle kernel pointer dereference in virtual kernel address space
Failing address: 0000038003e7c000 TEID: 0000038003e7c803
Fault in home space mode while using kernel ASCE.
AS:00000001fb388007 R3:000000008021c007 S:0000000082142000 P:0000000000000400 
Oops: 0011 ilc:3 [#1] SMP 
CPU: 0 PID: 1017495 Comm: get_mempolicy Tainted: G           OE     5.14.0-20210730.rc3.git0.4ccc9e2db7ac.300.fc34.s390x+next #1
Hardware name: IBM 2827 H66 708 (LPAR)
Krnl PSW : 0704e00180000000 00000001f9f11000 (compat_put_bitmap+0x48/0xd0)
           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:2 PM:0 RI:0 EA:3
Krnl GPRS: 0000000000810000 0000000000000000 000000007d9df1c0 0000038003e7c008
           0000000000000004 000000007d9df1c4 0000038003e7be40 0000000000010000
           0000000000008000 0000000000000000 0000000000000390 00000000000001c8
           000000020d6ea000 000002aa00401a48 00000001fa0a85fa 0000038003e7bd50
Krnl Code: 00000001f9f10ff4: a7bb0001            aghi    %r11,1
           00000001f9f10ff8: 41303008            la      %r3,8(%r3)
          #00000001f9f10ffc: 41502004            la      %r5,4(%r2)
          >00000001f9f11000: e3103ff8ff04        lg      %r1,-8(%r3)
           00000001f9f11006: 5010f0a4            st      %r1,164(%r15)
           00000001f9f1100a: a50e0081            llilh   %r0,129
           00000001f9f1100e: c8402000f0a4        mvcos   0(%r2),164(%r15),%r4
           00000001f9f11014: 1799                xr      %r9,%r9
Call Trace:
 [<00000001f9f11000>] compat_put_bitmap+0x48/0xd0 
 [<00000001fa0a85fa>] kernel_get_mempolicy+0x102/0x178 
 [<00000001fa0a86b0>] __s390_sys_get_mempolicy+0x40/0x50 
 [<00000001fa92be30>] __do_syscall+0x1c0/0x1e8 
 [<00000001fa939148>] system_call+0x78/0xa0 
Last Breaking-Event-Address:
 [<0000038003e7bc00>] 0x38003e7bc00
Kernel panic - not syncing: Fatal exception: panic_on_oops

Note: I did not try to bisect this, since it looks to me like this patch series causes the problem. Also, please don't get confused with the kernel version name. The date encoded is the build date, not the linux-next version.
linux-next commit 4ccc9e2db7ac ("Add linux-next specific files for 20210729") was used to build the kernel (s390 defconfig).

author
#19 lore raw
Quoted 4 lines · Heiko Carstens

On Tue, Jul 27, 2021 at 04:48:53PM +0200, Arnd Bergmann wrote:

Our CI reports this with linux-next and running strace selftest in compat mode:

Thanks a lot for the report! I managed track it down based on your output, it turns out that I end up copying data from the stack according to how much the user asked for, and in this case that was much more than the 8 byte nodemask_t, copying all of the kernel stack all the way into the guard page with CONFIG_VMAP_STACK, where it crashed. Without CONFIG_VMAP_STACK, or with user space that asks for less data, it would just be an information leak, so others probably haven't noticed the problem.

The change below should fix that, I'll double-check the other callers as well before sending a proper fixup patch to Andrew.

Arnd

mm/mempolicy.c +0 -0
index 4fabf2dddbc0..0d1f3be32723 100644
@@ -1438,6 +1438,7 @@ static int copy_nodes_to_user(unsigned long

__user *mask, unsigned long maxnode,

                if (clear_user((char __user *)mask + nbytes, copy - nbytes))
                        return -EFAULT;
                copy = nbytes;
+               maxnode = nr_node_ids;
        }

        if (compat)