Re: [PATCH v3 1/4] kexec: simplify compat_sys_kexec_load

Subsystems: kexec, the rest

2 messages, 2 authors, 2021-05-19 · open the first message on its own page

Re: [PATCH v3 1/4] kexec: simplify compat_sys_kexec_load

From: Eric W. Biederman <hidden>
Date: 2021-05-18 22:45:44

Arnd Bergmann [off-list ref] writes:
On Tue, May 18, 2021 at 4:05 PM Arnd Bergmann [off-list ref] wrote:
quoted
On Tue, May 18, 2021 at 3:41 PM Eric W. Biederman [off-list ref] wrote:
quoted
Arnd Bergmann [off-list ref] writes:
quoted
From: Arnd Bergmann <arnd@arndb.de>KEXEC_ARCH_DEFAULT

The compat version of sys_kexec_load() uses compat_alloc_user_space to
convert the user-provided arguments into the native format.

Move the conversion into the regular implementation with
an in_compat_syscall() check to simplify it and avoid the
compat_alloc_user_space() call.

compat_sys_kexec_load() now behaves the same as sys_kexec_load().
Nacked-by: "Eric W. Biederman" [off-list ref]
KEXEC_ARCH_DEFAULT
The patch is wrong.

The logic between the compat entry point and the ordinary entry point
are by necessity different.   This unifies the logic and breaks the compat
entry point.

The fundamentally necessity is that the code being loaded needs to know
which mode the kernel is running in so it can safely transition to the
new kernel.

Given that the two entry points fundamentally need different logic,
and that difference was not preserved and the goal of this patchset
was to unify that which fundamentally needs to be different.  I don't
think this patch series makes any sense for kexec.
Sorry, I'm not following that explanation. Can you clarify what different
modes of the kernel you are referring to here, and how my patch
changes this?
I think something like the untested diff below is enough to get rid of
compat_alloc_user cleanly.

Certainly it should be enough to give any idea what I am thinking.
diff --git a/kernel/kexec.c b/kernel/kexec.c
index c82c6c06f051..ce69a5d68023 100644
--- a/kernel/kexec.c
+++ b/kernel/kexec.c
@@ -19,26 +19,21 @@
 
 #include "kexec_internal.h"
 
-static int copy_user_segment_list(struct kimage *image,
+static void copy_user_segment_list(struct kimage *image,
 				  unsigned long nr_segments,
-				  struct kexec_segment __user *segments)
+				  struct kexec_segment *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;
+	memcpy(image->segment, segments, segment_bytes);
 }
 
 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;
@@ -59,9 +54,7 @@ static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 
 	image->start = entry;
 
-	ret = copy_user_segment_list(image, nr_segments, segments);
-	if (ret)
-		goto out_free_image;
+	copy_user_segment_list(image, nr_segments, segments);
 
 	if (kexec_on_panic) {
 		/* Enable special crash kernel control page alloc policy. */
@@ -103,8 +96,8 @@ static int kimage_alloc_init(struct kimage **rimage, unsigned long entry,
 	return ret;
 }
 
-static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
-		struct kexec_segment __user *segments, unsigned long flags)
+static int do_kexec_load_locked(unsigned long entry, unsigned long nr_segments,
+			struct kexec_segment *segments, unsigned long flags)
 {
 	struct kimage **dest_image, *image;
 	unsigned long i;
@@ -174,6 +167,27 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
 	return ret;
 }
 
+static int do_kexec_load(unsigned long entry, unsigned long nr_segments,
+			struct kexec_segment *segments, unsigned long flags)
+{
+	int result;
+
+	/* 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_locked(entry, nr_segments, segments, flags);
+	mutex_unlock(&kexec_mutex);
+	return result;
+}
+
 /*
  * Exec Kernel system call: for obvious reasons only root may call it.
  *
@@ -224,6 +238,11 @@ static inline int kexec_load_check(unsigned long nr_segments,
 	if ((flags & KEXEC_FLAGS) != (flags & ~KEXEC_ARCH_MASK))
 		return -EINVAL;
 
+	/* Verify we are on the appropriate architecture */
+	if (((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH) &&
+	    ((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH_DEFAULT))
+		return -EINVAL;
+
 	/* Put an artificial cap on the number
 	 * of segments passed to kexec_load.
 	 */
@@ -236,33 +255,29 @@ 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 bytes, result;
 
 	result = kexec_load_check(nr_segments, flags);
 	if (result)
 		return result;
 
-	/* Verify we are on the appropriate architecture */
-	if (((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH) &&
-		((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;
+	bytes = nr_segments * sizeof(ksegments[0]);
+	ksegments = kmalloc(bytes, GFP_KERNEL);
+	if (!ksegments)
+		return -ENOMEM;
 
+	result = copy_from_user(ksegments, segments, bytes);
+	if (result)
+		goto fail;
+	
 	result = do_kexec_load(entry, nr_segments, segments, flags);
+	kfree(ksegments);
 
-	mutex_unlock(&kexec_mutex);
-
+fail:
+	kfree(ksegments);
 	return result;
+	
 }
 
 #ifdef CONFIG_COMPAT
@@ -272,9 +287,9 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry,
 		       compat_ulong_t, flags)
 {
 	struct compat_kexec_segment in;
-	struct kexec_segment out, __user *ksegments;
-	unsigned long i, result;
-
+	struct kexec_segment *ksegments;
+	unsigned long bytes, i, result;
+ 
 	result = kexec_load_check(nr_segments, flags);
 	if (result)
 		return result;
@@ -285,37 +300,26 @@ 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));
+	bytes = nr_segments * sizeof(ksegments[0]);
+	ksegments = kmalloc(bytes, 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;
-
-		out.buf   = compat_ptr(in.buf);
-		out.bufsz = in.bufsz;
-		out.mem   = in.mem;
-		out.memsz = in.memsz;
+			goto fail;
 
-		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;
 	}
 
-	/* 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);
-
+fail:
+	kfree(ksegments);
 	return result;
 }
 #endif
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH v3 1/4] kexec: simplify compat_sys_kexec_load

From: Arnd Bergmann <arnd@kernel.org>
Date: 2021-05-19 09:56:25

On Wed, May 19, 2021 at 12:45 AM Eric W. Biederman
[off-list ref] wrote:
Arnd Bergmann [off-list ref] writes:
quoted
On Tue, May 18, 2021 at 4:05 PM Arnd Bergmann [off-list ref] wrote:
quoted
On Tue, May 18, 2021 at 3:41 PM Eric W. Biederman [off-list ref] wrote:
I think something like the untested diff below is enough to get rid of
compat_alloc_user cleanly.

Certainly it should be enough to give any idea what I am thinking.
Yes, that looks sufficient to me. I had started a slightly different
approach by trying
to move the kimage_alloc_init() into the top-level entry points to
avoid the extra
kmalloc, but that got rather complicated, and your patch is simpler overall.

The allocation could still be combined with kexec_load_check() into a new
function to reduce the number of duplicate lines, but if you think the current
version is ok, then I'll leave this part as it is.

I've fixed a duplicate kfree() and some whitespace damage, and rebased the
rest of my series on top of this to give it a spin on the build test boxes.
I'll send a v4 series once I have made sure there are no build-time regressions.

Can I add your Signed-off-by for the patch?
Is there a set of tests I should run on it?

      Arnd

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help