Skip to contentSkip to conversation

make 'user_access_begin()' do 'access_ok()'

PATCH New Schmid, Carsten 2 participants 6 messages Apr 1, 2020 1 message not in the archive

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 make 'user_access_begin()' do 'access_ok()'

From eb5a13ddc30824c20f1e2b662d2c821ad3808526 Mon Sep 17 00:00:00 2001
From: Linus Torvalds <[email protected]>
Date: Fri, 4 Jan 2019 12:56:09 -0800
Subject: [PATCH] make 'user_access_begin()' do 'access_ok()'

[ Upstream commit 594cc251fdd0d231d342d88b2fdff4bc42fb0690 ]

Fixes CVE-2018-20669
Backported from v5.0-rc1
Patch 1/1

Originally, the rule used to be that you'd have to do access_ok() separately, and then user_access_begin() before actually doing the direct (optimized) user access.

But experience has shown that people then decide not to do access_ok() at all, and instead rely on it being implied by other operations or similar. Which makes it very hard to verify that the access has actually been range-checked.

If you use the unsafe direct user accesses, hardware features (either SMAP - Supervisor Mode Access Protection - on x86, or PAN - Privileged Access Never - on ARM) do force you to use user_access_begin(). But nothing really forces the range check.

By putting the range check into user_access_begin(), we actually force people to do the right thing (tm), and the range check vill be visible near the actual accesses. We have way too long a history of people trying to avoid them.

---
Rationale:
When working on stability and security for a project with 4.14 kernel, i backported patches from upstream.
Want to give this work back to the community, as 4.14 is a SLTS.

---
 arch/x86/include/asm/uaccess.h             |  9 ++++++++-
 drivers/gpu/drm/i915/i915_gem_execbuffer.c | 15 +++++++++++++--
 include/linux/uaccess.h                    |  2 +-
 kernel/compat.c                            |  6 ++----
 kernel/exit.c                              |  6 ++----
 lib/strncpy_from_user.c                    |  9 +++++----
 lib/strnlen_user.c                         |  9 +++++----
 7 files changed, 36 insertions(+), 20 deletions(-)
arch/x86/include/asm/uaccess.h +8 -1
index 971830341061..fd00c5fba059 100644
@@ -711,7 +711,14 @@ extern struct movsl_mask {
  * checking before using them, but you have to surround them with the
  * user_access_begin/end() pair.
  */
-#define user_access_begin()    __uaccess_begin()
+static __must_check inline bool user_access_begin(const void __user *ptr, size_t len)
+{
+       if (unlikely(!access_ok(VERIFY_READ,ptr,len)))
+               return 0;
+       __uaccess_begin();
+       return 1;
+}
+#define user_access_begin(a,b) user_access_begin(a,b)
 #define user_access_end()       __uaccess_end()
 
 #define unsafe_put_user(x, ptr, err_label)                                      \
drivers/gpu/drm/i915/i915_gem_execbuffer.c +13 -2
index d99d05a91032..3a65a45184ad 100644
@@ -1566,7 +1566,9 @@ static int eb_copy_relocations(const struct i915_execbuffer *eb)
                  * happened we would make the mistake of assuming that the
                  * relocations were valid.
                  */
-               user_access_begin();
+               if (!user_access_begin(urelocs, size))
+                       goto end_user;
+
                 for (copied = 0; copied < nreloc; copied++)
                         unsafe_put_user(-1,
                                         &urelocs[copied].presumed_offset,
@@ -2649,7 +2651,16 @@ i915_gem_execbuffer2(struct drm_device *dev, void *data,
                 unsigned int i;
 
                 /* Copy the new buffer offsets back to the user's exec list. */
-               user_access_begin();
+               /*
+                * Note: args->buffer_count * sizeof(*user_exec_list) does not overflow,
+                * because we checked this on entry.
+                *
+                * And this range already got effectively checked earlier
+                * when we did the "copy_from_user()" above.
+                */
+               if (!user_access_begin(user_exec_list, args->buffer_count * sizeof(*user_exec_list)))
+                       goto end_user;
+
                 for (i = 0; i < args->buffer_count; i++) {
                         if (!(exec2_list[i].offset & UPDATE))
                                 continue;
include/linux/uaccess.h +1 -1
index 251e655d407f..76407748701b 100644
@@ -267,7 +267,7 @@ extern long strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count);
         probe_kernel_read(&retval, addr, sizeof(retval))
 
 #ifndef user_access_begin
-#define user_access_begin() do { } while (0)
+#define user_access_begin(ptr,len) access_ok(ptr, len)
 #define user_access_end() do { } while (0)
 #define unsafe_get_user(x, ptr, err) do { if (unlikely(__get_user(x, ptr))) goto err; } while (0)
 #define unsafe_put_user(x, ptr, err) do { if (unlikely(__put_user(x, ptr))) goto err; } while (0)
kernel/compat.c +2 -4
index 7e83733d4c95..a9f5de63dc90 100644
@@ -437,10 +437,9 @@ long compat_get_bitmap(unsigned long *mask, const compat_ulong_t __user *umask,
         bitmap_size = ALIGN(bitmap_size, BITS_PER_COMPAT_LONG);
         nr_compat_longs = BITS_TO_COMPAT_LONGS(bitmap_size);
 
-       if (!access_ok(VERIFY_READ, umask, bitmap_size / 8))
+       if (!user_access_begin(umask, bitmap_size / 8))
                 return -EFAULT;
 
-       user_access_begin();
         while (nr_compat_longs > 1) {
                 compat_ulong_t l1, l2;
                 unsafe_get_user(l1, umask++, Efault);
@@ -467,10 +466,9 @@ long compat_put_bitmap(compat_ulong_t __user *umask, unsigned long *mask,
         bitmap_size = ALIGN(bitmap_size, BITS_PER_COMPAT_LONG);
         nr_compat_longs = BITS_TO_COMPAT_LONGS(bitmap_size);
 
-       if (!access_ok(VERIFY_WRITE, umask, bitmap_size / 8))
+       if (!user_access_begin(umask, bitmap_size / 8))
                 return -EFAULT;
 
-       user_access_begin();
         while (nr_compat_longs > 1) {
                 unsigned long m = *mask++;
                 unsafe_put_user((compat_ulong_t)m, umask++, Efault);
kernel/exit.c +2 -4
index d1baf9c96c3e..c3ad546ba7c2 100644
@@ -1597,10 +1597,9 @@ SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
         if (!infop)
                 return err;
 
-       if (!access_ok(VERIFY_WRITE, infop, sizeof(*infop)))
+       if (!user_access_begin(infop, sizeof(*infop)))
                 return -EFAULT;
 
-       user_access_begin();
         unsafe_put_user(signo, &infop->si_signo, Efault);
         unsafe_put_user(0, &infop->si_errno, Efault);
         unsafe_put_user(info.cause, &infop->si_code, Efault);
@@ -1725,10 +1724,9 @@ COMPAT_SYSCALL_DEFINE5(waitid,
         if (!infop)
                 return err;
 
-       if (!access_ok(VERIFY_WRITE, infop, sizeof(*infop)))
+       if (!user_access_begin(infop, sizeof(*infop)))
                 return -EFAULT;
 
-       user_access_begin();
         unsafe_put_user(signo, &infop->si_signo, Efault);
         unsafe_put_user(0, &infop->si_errno, Efault);
         unsafe_put_user(info.cause, &infop->si_code, Efault);
lib/strncpy_from_user.c +5 -4
index e304b54c9c7d..023ba9f3b99f 100644
@@ -115,10 +115,11 @@ long strncpy_from_user(char *dst, const char __user *src, long count)
 
                 kasan_check_write(dst, count);
                 check_object_size(dst, count, false);
-               user_access_begin();
-               retval = do_strncpy_from_user(dst, src, count, max);
-               user_access_end();
-               return retval;
+               if (user_access_begin(src, max)) {
+                       retval = do_strncpy_from_user(dst, src, count, max);
+                       user_access_end();
+                       return retval;
+               }
         }
         return -EFAULT;
 }
lib/strnlen_user.c +5 -4
index 184f80f7bacf..7f2db3fe311f 100644
@@ -114,10 +114,11 @@ long strnlen_user(const char __user *str, long count)
                 unsigned long max = max_addr - src_addr;
                 long retval;
 
-               user_access_begin();
-               retval = do_strnlen_user(str, count, max);
-               user_access_end();
-               return retval;
+               if (user_access_begin(str, max)) {
+                       retval = do_strnlen_user(str, count, max);
+                       user_access_end();
+                       return retval;
+               }
         }
         return 0;
 }

--
2.17.1
-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Alexander Walter

#1 lore raw

PATCH 1/1 make 'user_access_begin()' do 'access_ok()'

From eb5a13ddc30824c20f1e2b662d2c821ad3808526 Mon Sep 17 00:00:00 2001
From: Linus Torvalds <[email protected]>
Date: Fri, 4 Jan 2019 12:56:09 -0800
Subject: [PATCH] make 'user_access_begin()' do 'access_ok()'

[ Upstream commit 594cc251fdd0d231d342d88b2fdff4bc42fb0690 ]

Fixes CVE-2018-20669
Backported from v5.0-rc1
Patch 1/1

Originally, the rule used to be that you'd have to do access_ok() separately, and then user_access_begin() before actually doing the direct (optimized) user access.

But experience has shown that people then decide not to do access_ok() at all, and instead rely on it being implied by other operations or similar. Which makes it very hard to verify that the access has actually been range-checked.

If you use the unsafe direct user accesses, hardware features (either SMAP - Supervisor Mode Access Protection - on x86, or PAN - Privileged Access Never - on ARM) do force you to use user_access_begin(). But nothing really forces the range check.

By putting the range check into user_access_begin(), we actually force people to do the right thing (tm), and the range check vill be visible near the actual accesses. We have way too long a history of people trying to avoid them.

---
Rationale:
When working on stability and security for a project with 4.14 kernel, i backported patches from upstream.
Want to give this work back to the community, as 4.14 is a SLTS.

---
 arch/x86/include/asm/uaccess.h             |  9 ++++++++-
 drivers/gpu/drm/i915/i915_gem_execbuffer.c | 15 +++++++++++++--
 include/linux/uaccess.h                    |  2 +-
 kernel/compat.c                            |  6 ++----
 kernel/exit.c                              |  6 ++----
 lib/strncpy_from_user.c                    |  9 +++++----
 lib/strnlen_user.c                         |  9 +++++----
 7 files changed, 36 insertions(+), 20 deletions(-)
arch/x86/include/asm/uaccess.h +8 -1
index 971830341061..fd00c5fba059 100644
@@ -711,7 +711,14 @@ extern struct movsl_mask {
  * checking before using them, but you have to surround them with the
  * user_access_begin/end() pair.
  */
-#define user_access_begin()    __uaccess_begin()
+static __must_check inline bool user_access_begin(const void __user *ptr, size_t len)
+{
+       if (unlikely(!access_ok(VERIFY_READ,ptr,len)))
+               return 0;
+       __uaccess_begin();
+       return 1;
+}
+#define user_access_begin(a,b) user_access_begin(a,b)
 #define user_access_end()       __uaccess_end()
 
 #define unsafe_put_user(x, ptr, err_label)                                      \
drivers/gpu/drm/i915/i915_gem_execbuffer.c +13 -2
index d99d05a91032..3a65a45184ad 100644
@@ -1566,7 +1566,9 @@ static int eb_copy_relocations(const struct i915_execbuffer *eb)
                  * happened we would make the mistake of assuming that the
                  * relocations were valid.
                  */
-               user_access_begin();
+               if (!user_access_begin(urelocs, size))
+                       goto end_user;
+
                 for (copied = 0; copied < nreloc; copied++)
                         unsafe_put_user(-1,
                                         &urelocs[copied].presumed_offset,
@@ -2649,7 +2651,16 @@ i915_gem_execbuffer2(struct drm_device *dev, void *data,
                 unsigned int i;
 
                 /* Copy the new buffer offsets back to the user's exec list. */
-               user_access_begin();
+               /*
+                * Note: args->buffer_count * sizeof(*user_exec_list) does not overflow,
+                * because we checked this on entry.
+                *
+                * And this range already got effectively checked earlier
+                * when we did the "copy_from_user()" above.
+                */
+               if (!user_access_begin(user_exec_list, args->buffer_count * sizeof(*user_exec_list)))
+                       goto end_user;
+
                 for (i = 0; i < args->buffer_count; i++) {
                         if (!(exec2_list[i].offset & UPDATE))
                                 continue;
include/linux/uaccess.h +1 -1
index 251e655d407f..76407748701b 100644
@@ -267,7 +267,7 @@ extern long strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count);
         probe_kernel_read(&retval, addr, sizeof(retval))
 
 #ifndef user_access_begin
-#define user_access_begin() do { } while (0)
+#define user_access_begin(ptr,len) access_ok(ptr, len)
 #define user_access_end() do { } while (0)
 #define unsafe_get_user(x, ptr, err) do { if (unlikely(__get_user(x, ptr))) goto err; } while (0)
 #define unsafe_put_user(x, ptr, err) do { if (unlikely(__put_user(x, ptr))) goto err; } while (0)
kernel/compat.c +2 -4
index 7e83733d4c95..a9f5de63dc90 100644
@@ -437,10 +437,9 @@ long compat_get_bitmap(unsigned long *mask, const compat_ulong_t __user *umask,
         bitmap_size = ALIGN(bitmap_size, BITS_PER_COMPAT_LONG);
         nr_compat_longs = BITS_TO_COMPAT_LONGS(bitmap_size);
 
-       if (!access_ok(VERIFY_READ, umask, bitmap_size / 8))
+       if (!user_access_begin(umask, bitmap_size / 8))
                 return -EFAULT;
 
-       user_access_begin();
         while (nr_compat_longs > 1) {
                 compat_ulong_t l1, l2;
                 unsafe_get_user(l1, umask++, Efault);
@@ -467,10 +466,9 @@ long compat_put_bitmap(compat_ulong_t __user *umask, unsigned long *mask,
         bitmap_size = ALIGN(bitmap_size, BITS_PER_COMPAT_LONG);
         nr_compat_longs = BITS_TO_COMPAT_LONGS(bitmap_size);
 
-       if (!access_ok(VERIFY_WRITE, umask, bitmap_size / 8))
+       if (!user_access_begin(umask, bitmap_size / 8))
                 return -EFAULT;
 
-       user_access_begin();
         while (nr_compat_longs > 1) {
                 unsigned long m = *mask++;
                 unsafe_put_user((compat_ulong_t)m, umask++, Efault);
kernel/exit.c +2 -4
index d1baf9c96c3e..c3ad546ba7c2 100644
@@ -1597,10 +1597,9 @@ SYSCALL_DEFINE5(waitid, int, which, pid_t, upid, struct siginfo __user *,
         if (!infop)
                 return err;
 
-       if (!access_ok(VERIFY_WRITE, infop, sizeof(*infop)))
+       if (!user_access_begin(infop, sizeof(*infop)))
                 return -EFAULT;
 
-       user_access_begin();
         unsafe_put_user(signo, &infop->si_signo, Efault);
         unsafe_put_user(0, &infop->si_errno, Efault);
         unsafe_put_user(info.cause, &infop->si_code, Efault);
@@ -1725,10 +1724,9 @@ COMPAT_SYSCALL_DEFINE5(waitid,
         if (!infop)
                 return err;
 
-       if (!access_ok(VERIFY_WRITE, infop, sizeof(*infop)))
+       if (!user_access_begin(infop, sizeof(*infop)))
                 return -EFAULT;
 
-       user_access_begin();
         unsafe_put_user(signo, &infop->si_signo, Efault);
         unsafe_put_user(0, &infop->si_errno, Efault);
         unsafe_put_user(info.cause, &infop->si_code, Efault);
lib/strncpy_from_user.c +5 -4
index e304b54c9c7d..023ba9f3b99f 100644
@@ -115,10 +115,11 @@ long strncpy_from_user(char *dst, const char __user *src, long count)
 
                 kasan_check_write(dst, count);
                 check_object_size(dst, count, false);
-               user_access_begin();
-               retval = do_strncpy_from_user(dst, src, count, max);
-               user_access_end();
-               return retval;
+               if (user_access_begin(src, max)) {
+                       retval = do_strncpy_from_user(dst, src, count, max);
+                       user_access_end();
+                       return retval;
+               }
         }
         return -EFAULT;
 }
lib/strnlen_user.c +5 -4
index 184f80f7bacf..7f2db3fe311f 100644
@@ -114,10 +114,11 @@ long strnlen_user(const char __user *str, long count)
                 unsigned long max = max_addr - src_addr;
                 long retval;
 
-               user_access_begin();
-               retval = do_strnlen_user(str, count, max);
-               user_access_end();
-               return retval;
+               if (user_access_begin(str, max)) {
+                       retval = do_strnlen_user(str, count, max);
+                       user_access_end();
+                       return retval;
+               }
         }
         return 0;
 }

--
2.17.1
-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Alexander Walter

maintainer
#2 lore raw
Quoted 10 lines · Carsten
Quoted 1 line

From eb5a13ddc30824c20f1e2b662d2c821ad3808526 Mon Sep 17 00:00:00 2001

From: Linus Torvalds <[email protected]>
Date: Fri, 4 Jan 2019 12:56:09 -0800
Subject: [PATCH] make 'user_access_begin()' do 'access_ok()'

[ Upstream commit 594cc251fdd0d231d342d88b2fdff4bc42fb0690 ]

Fixes CVE-2018-20669
Backported from v5.0-rc1
Patch 1/1

What about 4.19?

thanks,

greg k-h

maintainer
#3 lore raw
In reply to Schmid, Carsten
Quoted 10 lines · Carsten
Quoted 1 line

From eb5a13ddc30824c20f1e2b662d2c821ad3808526 Mon Sep 17 00:00:00 2001

From: Linus Torvalds <[email protected]>
Date: Fri, 4 Jan 2019 12:56:09 -0800
Subject: [PATCH] make 'user_access_begin()' do 'access_ok()'

[ Upstream commit 594cc251fdd0d231d342d88b2fdff4bc42fb0690 ]

Fixes CVE-2018-20669
Backported from v5.0-rc1
Patch 1/1

Also, that cve was "supposed" to already be fixed in the 4.19.13 kernel release for some reason, and it's a drm issue, not a core access_ok() issue.

So why is this needed for 4.14?

Quoted 15 lines

Originally, the rule used to be that you'd have to do access_ok() separately, and then user_access_begin() before actually doing the direct (optimized) user access.

But experience has shown that people then decide not to do access_ok() at all, and instead rely on it being implied by other operations or similar. Which makes it very hard to verify that the access has actually been range-checked.

If you use the unsafe direct user accesses, hardware features (either SMAP - Supervisor Mode Access Protection - on x86, or PAN - Privileged Access Never - on ARM) do force you to use user_access_begin(). But nothing really forces the range check.

Quoted 6 lines

By putting the range check into user_access_begin(), we actually force people to do the right thing (tm), and the range check vill be visible near the actual accesses. We have way too long a history of people trying to avoid them.

No s-o-by from you?

Quoted 5 lines

---
Rationale:
When working on stability and security for a project with 4.14 kernel, i backported patches from upstream.
Want to give this work back to the community, as 4.14 is a SLTS.

What is "SLTS"?

thanks,

greg k-h

A message that is not in the archive (sent off-list or not yet received): [email protected]

#5 lore raw
Quoted 11 lines
Quoted 4 lines

Fixes CVE-2018-20669
Backported from v5.0-rc1
Patch 1/1

Also, that cve was "supposed" to already be fixed in the 4.19.13 kernel release for some reason, and it's a drm issue, not a core access_ok() issue.

So why is this needed for 4.14?

See https://access.redhat.com/security/cve/cve-2018-20669
Looks like Linus' fix was attacking this at the root cause, not only for DRM.

Also, i use https://www.linuxkernelcves.com/ as a research source, and they claim that CVE not fixed in 4.19.
(and i'll check for the other LTS kernels as well)

Quoted 4 lines
Quoted 2 lines

No s-o-by from you?

Ops. Will add this in a resend.

Quoted 3 lines
Quoted 1 line

Want to give this work back to the community, as 4.14 is a SLTS.

What is "SLTS"?

Super Long Term Supported kernel - thanks to guys like you :-)
4.14 really is that (Jan. 2024, as of https://www.kernel.org/category/releases.html)

Quoted 4 lines

thanks,

greg k-h

Thanks, and i have some other patches backported to 4.14 as CVE fixes, which i'll propose in the next hours.

BR
Carsten
-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Alexander Walter

maintainer
#6 lore raw
Quoted 5 lines · Carsten
Quoted 5 lines
Quoted 4 lines

Fixes CVE-2018-20669
Backported from v5.0-rc1
Patch 1/1

Quoted 8 lines
Quoted 6 lines

Also, that cve was "supposed" to already be fixed in the 4.19.13 kernel release for some reason, and it's a drm issue, not a core access_ok() issue.

So why is this needed for 4.14?

See https://access.redhat.com/security/cve/cve-2018-20669
Looks like Linus' fix was attacking this at the root cause, not only for DRM.

And are you _sure_ this really is an issue in 4.14? And in 4.19? There was some reason I didn't backport this to 4.19 at the time...

Quoted 2 lines

Also, i use https://www.linuxkernelcves.com/ as a research source, and they claim that CVE not fixed in 4.19.

That disagrees with the "main" CVE database. Not that I really trust any of them further than I can throw their servers...

Quoted 1 line

(and i'll check for the other LTS kernels as well)

Please do.

Quoted 10 lines
Quoted 3 lines
Quoted 1 line

No s-o-by from you?

Ops. Will add this in a resend.

Quoted 3 lines
Quoted 1 line

Want to give this work back to the community, as 4.14 is a SLTS.

What is "SLTS"?

Super Long Term Supported kernel - thanks to guys like you :-)
4.14 really is that (Jan. 2024, as of https://www.kernel.org/category/releases.html)

I don't use that term, don't make new things up where they aren't :)

Quoted 7 lines
Quoted 4 lines

thanks,

greg k-h

Thanks, and i have some other patches backported to 4.14 as CVE fixes, which i'll propose in the next hours.

Make sure that they are really issues, and that they are fixed in all current trees. I can't take patches for an older stable tree without a newer one also having it otherwise people would upgrade and suffer a regression.

thanks,

greg k-h