arm64: untag user pointers passed to the kernel

mm/mprotect.c +1 -0

@@ -578,6 +578,7 @@ static int do_mprotect_pkey(unsigned long start, size_t len,
 SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
 		unsigned long, prot)
 {
+	start = untagged_addr(start);
 	return do_mprotect_pkey(start, len, prot, -1);
 }
 

Quoted 10 lines · Andrey Konovalov

mm/mprotect.c +1 -0

@@ -578,6 +578,7 @@ static int do_mprotect_pkey(unsigned long start, size_t len,
 SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
              unsigned long, prot)
 {
+     start = untagged_addr(start);
      return do_mprotect_pkey(start, len, prot, -1);
 }
 

mm/mprotect.c SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len, · line 592

Quoted 7 lines

mm/mprotect.c +1 -0

@@ -586,6 +587,7 @@ SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
 SYSCALL_DEFINE4(pkey_mprotect, unsigned long, start, size_t, len,
              unsigned long, prot, int, pkey)
 {
+     start = untagged_addr(start);
      return do_mprotect_pkey(start, len, prot, pkey);
 }

This seems to have taken the approach of going as close as possible to the syscall boundary and untagging the pointer there. I guess that's OK, but it does lead to more churn than necessary. For instance, why not just do the untagging in do_mprotect_pkey()?

I think that's an overall design question. I kinda asked the same thing about patching call sites vs. VMA lookup functions.

fs/namespace.c void *copy_mount_options(const void __user * data) · line 2735

Quoted 10 lines · Andrey Konovalov

fs/namespace.c +1 -1

@@ -2730,7 +2730,7 @@ void *copy_mount_options(const void __user * data)
       * the remainder of the page.
       */
      /* copy_from_user cannot cross TASK_SIZE ! */
-     size = TASK_SIZE - (unsigned long)data;
+     size = TASK_SIZE - (unsigned long)untagged_addr(data);
      if (size > PAGE_SIZE)
              size = PAGE_SIZE;

I would have thought that copy_from_user() *is* entirely capable of detecting and returning an error in the case that its arguments cross TASK_SIZE. It will fail and return an error, but that's what it's supposed to do.

I'd question why this code needs to be doing its own checking in the first place. Is there something subtle going on?

userfaultfd_register() and userfaultfd_unregister() use provided user pointers for vma lookups, which can only by done with untagged pointers.

Quoted 2 lines · Andrey Konovalov

userfaultfd_register() and userfaultfd_unregister() use provided user pointers for vma lookups, which can only by done with untagged pointers.

So, we have to patch all these sites before the tagged values get to the point of hitting the vma lookup functions. Dumb question: Why don't we just patch the vma lookup functions themselves instead of all of these callers?

Quoted 5 lines · Dave Hansen

Quoted 2 lines · Andrey Konovalov

userfaultfd_register() and userfaultfd_unregister() use provided user pointers for vma lookups, which can only by done with untagged pointers.

Quoted 4 lines

So, we have to patch all these sites before the tagged values get to the point of hitting the vma lookup functions. Dumb question: Why don't we just patch the vma lookup functions themselves instead of all of these callers?

That might be a working approach as well. We'll still need to fix up places where the vma fields are accessed directly. Catalin, what do you think?

Quoted 7 lines

Quoted 4 lines

So, we have to patch all these sites before the tagged values get to the point of hitting the vma lookup functions. Dumb question: Why don't we just patch the vma lookup functions themselves instead of all of these callers?

That might be a working approach as well. We'll still need to fix up places where the vma fields are accessed directly. Catalin, what do you think?

Most callers of find_vma*() always follow it by a check of vma->vma_start against some tagged address ('end' in the userfaultfd_(un)register()) case. So it's not sufficient to untag it in find_vma().

Quoted 7 lines · Catalin Marinas

Quoted 7 lines

Quoted 4 lines

So, we have to patch all these sites before the tagged values get to the point of hitting the vma lookup functions. Dumb question: Why don't we just patch the vma lookup functions themselves instead of all of these callers?

That might be a working approach as well. We'll still need to fix up places where the vma fields are accessed directly. Catalin, what do you think?

Quoted 4 lines

Most callers of find_vma*() always follow it by a check of vma->vma_start against some tagged address ('end' in the userfaultfd_(un)register()) case. So it's not sufficient to untag it in find_vma().

If that's truly the common case, sounds like we should have a find_vma() that does the vma_end checking as well. Then at least the common case would not have to worry about tagging.

This patchset is meant to be merged together with "arm64 relaxed ABI" [1].

arm64 has a feature called Top Byte Ignore, which allows to embed pointer tags into the top byte of each pointer. Userspace programs (such as HWASan, a memory debugging tool [2]) might use this feature and pass tagged user pointers to the kernel through syscalls or other interfaces.

Right now the kernel is already able to handle user faults with tagged pointers, due to these patches:

1. 81cddd65 ("arm64: traps: fix userspace cache maintenance emulation on a tagged pointer")
2. 7dcd9dd8 ("arm64: hw_breakpoint: fix watchpoint matching for tagged pointers")
3. 276e9327 ("arm64: entry: improve data abort handling of tagged pointers")

This patchset extends tagged pointer support to syscall arguments.

For non-memory syscalls this is done by untaging user pointers when the kernel performs pointer checking to find out whether the pointer comes from userspace (most notably in access_ok). The untagging is done only when the pointer is being checked, the tag is preserved as the pointer makes its way through the kernel.

Since memory syscalls (mmap, mprotect, etc.) don't do memory accesses but rather deal with memory ranges, untagged pointers are better suited to describe memory ranges internally. Thus for memory syscalls we untag pointers completely when they enter the kernel.

One of the alternative approaches to untagging that was considered is to completely strip the pointer tag as the pointer enters the kernel with some kind of a syscall wrapper, but that won't work with the countless number of different ioctl calls. With this approach we would need a custom wrapper for each ioctl variation, which doesn't seem practical.

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

2. Static testing with grep to find parts of the kernel that call find_vma() (and other similar functions) or directly compare against vm_start/vm_end fields of vma.

3. Static testing with grep to find parts of the kernel that compare user pointers with TASK_SIZE or other similar consts and macros.

4. Dynamic testing: adding BUG_ON(has_tag(addr)) to find_vma() and running a modified syzkaller version that passes tagged pointers to the kernel.

Based on the results of the testing the requried patches have been added to the patchset.

This patchset has been merged into the Pixel 2 kernel tree and is now being used to enable testing of Pixel 2 phones with HWASan.

This patchset is a prerequisite for ARM's memory tagging hardware feature support [4].

Thanks!

[1] https://lkml.org/lkml/2018/12/10/402

[2] http://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html

[3] https://github.com/lucvoo/sparse-dev/commit/5f960cb10f56ec2017c128ef9d16060e0145f292

[4] https://community.arm.com/processors/b/blog/posts/arm-a-profile-architecture-2018-developments-armv85a

Changes in v10:
- Added "mm, arm64: untag user pointers passed to memory syscalls" back.
- New patch "fs, arm64: untag user pointers in fs/userfaultfd.c".
- New patch "net, arm64: untag user pointers in tcp_zerocopy_receive".
- New patch "kernel, arm64: untag user pointers in prctl_set_mm*".
- New patch "tracing, arm64: untag user pointers in seq_print_user_ip".

Changes in v9:
- Rebased onto 4.20-rc6.
- Used u64 instead of __u64 in type casts in the untagged_addr macro for arm64.
- Added braces around (addr) in the untagged_addr macro for other arches.

Changes in v8:
- Rebased onto 65102238 (4.20-rc1).
- Added a note to the cover letter on why syscall wrappers/shims that untag user pointers won't work.
- Added a note to the cover letter that this patchset has been merged into the Pixel 2 kernel tree.
- Documentation fixes, in particular added a list of syscalls that don't support tagged user pointers.

Changes in v7:
- Rebased onto 17b57b18 (4.19-rc6).
- Dropped the "arm64: untag user address in __do_user_fault" patch, since the existing patches already handle user faults properly.
- Dropped the "usb, arm64: untag user addresses in devio" patch, since the passed pointer must come from a vma and therefore be untagged.
- Dropped the "arm64: annotate user pointers casts detected by sparse" patch (see the discussion to the replies of the v6 of this patchset).
- Added more context to the cover letter.
- Updated Documentation/arm64/tagged-pointers.txt.

Changes in v6:
- Added annotations for user pointer casts found by sparse.
- Rebased onto 050cdc6c (4.19-rc1+).

Changes in v5:
- Added 3 new patches that add untagging to places found with static analysis.
- Rebased onto 44c929e1 (4.18-rc8).

Changes in v4:
- Added a selftest for checking that passing tagged pointers to the kernel succeeds.
- Rebased onto 81e97f013 (4.18-rc1+).

Changes in v3:
- Rebased onto e5c51f30 (4.17-rc6+).
- Added linux-arch@ to the list of recipients.

Changes in v2:
- Rebased onto 2d618bdf (4.17-rc3+).
- Removed excessive untagging in gup.c.
- Removed untagging pointers returned from __uaccess_mask_ptr.

Changes in v1:
- Rebased onto 4.17-rc1.

Changes in RFC v2:
- Added "#ifndef untagged_addr..." fallback in linux/uaccess.h instead of defining it for each arch individually.
- Updated Documentation/arm64/tagged-pointers.txt.
- Dropped "mm, arm64: untag user addresses in memory syscalls".
- Rebased onto 3eb2ce82 (4.16-rc7).

Reviewed-by Luc Van Oostenryck <[email protected]>

Signed-off-by Andrey Konovalov <[email protected]>

Andrey Konovalov (12):
uaccess: add untagged_addr definition for other arches
arm64: untag user pointers in access_ok and __uaccess_mask_ptr lib, arm64: untag user pointers in strn*_user
mm, arm64: untag user pointers passed to memory syscalls mm, arm64: untag user pointers in mm/gup.c
fs, arm64: untag user pointers in copy_mount_options
fs, arm64: untag user pointers in fs/userfaultfd.c
net, arm64: untag user pointers in tcp_zerocopy_receive kernel, arm64: untag user pointers in prctl_set_mm* tracing, arm64: untag user pointers in seq_print_user_ip
arm64: update Documentation/arm64/tagged-pointers.txt selftests, arm64: add a selftest for passing tagged pointers to kernel

 Documentation/arm64/tagged-pointers.txt       | 25 +++++++++++--------
 arch/arm64/include/asm/uaccess.h              | 10 +++++---
 fs/namespace.c                                |  2 +-
 fs/userfaultfd.c                              |  5 ++++
 include/linux/memory.h                        |  4 +++
 ipc/shm.c                                     |  2 ++
 kernel/sys.c                                  | 14 +++++++++++
 kernel/trace/trace_output.c                   |  2 +-
 lib/strncpy_from_user.c                       |  2 ++
 lib/strnlen_user.c                            |  2 ++
 mm/gup.c                                      |  4 +++
 mm/madvise.c                                  |  2 ++
 mm/mempolicy.c                                |  5 ++++
 mm/migrate.c                                  |  1 +
 mm/mincore.c                                  |  2 ++
 mm/mlock.c                                    |  5 ++++
 mm/mmap.c                                     |  7 ++++++
 mm/mprotect.c                                 |  2 ++
 mm/mremap.c                                   |  2 ++
 mm/msync.c                                    |  2 ++
 net/ipv4/tcp.c                                |  2 ++
 tools/testing/selftests/arm64/.gitignore      |  1 +
 tools/testing/selftests/arm64/Makefile        | 11 ++++++++
 .../testing/selftests/arm64/run_tags_test.sh  | 12 +++++++++
 tools/testing/selftests/arm64/tags_test.c     | 19 ++++++++++++++
 25 files changed, 129 insertions(+), 16 deletions(-)
 create mode 100644 tools/testing/selftests/arm64/.gitignore
 create mode 100644 tools/testing/selftests/arm64/Makefile
 create mode 100755 tools/testing/selftests/arm64/run_tags_test.sh
 create mode 100644 tools/testing/selftests/arm64/tags_test.c

Quoted 25 lines · Andrey Konovalov

This patchset is meant to be merged together with "arm64 relaxed ABI" [1].

arm64 has a feature called Top Byte Ignore, which allows to embed pointer tags into the top byte of each pointer. Userspace programs (such as HWASan, a memory debugging tool [2]) might use this feature and pass tagged user pointers to the kernel through syscalls or other interfaces.

Right now the kernel is already able to handle user faults with tagged pointers, due to these patches:

1. 81cddd65 ("arm64: traps: fix userspace cache maintenance emulation on a tagged pointer")
2. 7dcd9dd8 ("arm64: hw_breakpoint: fix watchpoint matching for tagged pointers")
3. 276e9327 ("arm64: entry: improve data abort handling of tagged pointers")

This patchset extends tagged pointer support to syscall arguments.

For non-memory syscalls this is done by untaging user pointers when the kernel performs pointer checking to find out whether the pointer comes from userspace (most notably in access_ok). The untagging is done only when the pointer is being checked, the tag is preserved as the pointer makes its way through the kernel.

Quoted 4 lines

Since memory syscalls (mmap, mprotect, etc.) don't do memory accesses but rather deal with memory ranges, untagged pointers are better suited to describe memory ranges internally. Thus for memory syscalls we untag pointers completely when they enter the kernel.

i think the same is true when user pointers are compared.

e.g. i suspect there may be issues with tagged robust mutex list pointers because the kernel does

futex.c:3541: while (entry != &head->list) {

where entry is a user pointer that may be tagged, and &head->list is probably not tagged.

Quoted 157 lines

One of the alternative approaches to untagging that was considered is to completely strip the pointer tag as the pointer enters the kernel with some kind of a syscall wrapper, but that won't work with the countless number of different ioctl calls. With this approach we would need a custom wrapper for each ioctl variation, which doesn't seem practical.

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

2. Static testing with grep to find parts of the kernel that call find_vma() (and other similar functions) or directly compare against vm_start/vm_end fields of vma.

3. Static testing with grep to find parts of the kernel that compare user pointers with TASK_SIZE or other similar consts and macros.

4. Dynamic testing: adding BUG_ON(has_tag(addr)) to find_vma() and running a modified syzkaller version that passes tagged pointers to the kernel.

Based on the results of the testing the requried patches have been added to the patchset.

This patchset has been merged into the Pixel 2 kernel tree and is now being used to enable testing of Pixel 2 phones with HWASan.

This patchset is a prerequisite for ARM's memory tagging hardware feature support [4].

Thanks!

[1] https://lkml.org/lkml/2018/12/10/402

[2] http://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html

[3] https://github.com/lucvoo/sparse-dev/commit/5f960cb10f56ec2017c128ef9d16060e0145f292

[4] https://community.arm.com/processors/b/blog/posts/arm-a-profile-architecture-2018-developments-armv85a

Changes in v10:
- Added "mm, arm64: untag user pointers passed to memory syscalls" back.
- New patch "fs, arm64: untag user pointers in fs/userfaultfd.c".
- New patch "net, arm64: untag user pointers in tcp_zerocopy_receive".
- New patch "kernel, arm64: untag user pointers in prctl_set_mm*".
- New patch "tracing, arm64: untag user pointers in seq_print_user_ip".

Changes in v9:
- Rebased onto 4.20-rc6.
- Used u64 instead of __u64 in type casts in the untagged_addr macro for arm64.
- Added braces around (addr) in the untagged_addr macro for other arches.

Changes in v8:
- Rebased onto 65102238 (4.20-rc1).
- Added a note to the cover letter on why syscall wrappers/shims that untag user pointers won't work.
- Added a note to the cover letter that this patchset has been merged into the Pixel 2 kernel tree.
- Documentation fixes, in particular added a list of syscalls that don't support tagged user pointers.

Changes in v7:
- Rebased onto 17b57b18 (4.19-rc6).
- Dropped the "arm64: untag user address in __do_user_fault" patch, since the existing patches already handle user faults properly.
- Dropped the "usb, arm64: untag user addresses in devio" patch, since the passed pointer must come from a vma and therefore be untagged.
- Dropped the "arm64: annotate user pointers casts detected by sparse" patch (see the discussion to the replies of the v6 of this patchset).
- Added more context to the cover letter.
- Updated Documentation/arm64/tagged-pointers.txt.

Changes in v6:
- Added annotations for user pointer casts found by sparse.
- Rebased onto 050cdc6c (4.19-rc1+).

Changes in v5:
- Added 3 new patches that add untagging to places found with static analysis.
- Rebased onto 44c929e1 (4.18-rc8).

Changes in v4:
- Added a selftest for checking that passing tagged pointers to the kernel succeeds.
- Rebased onto 81e97f013 (4.18-rc1+).

Changes in v3:
- Rebased onto e5c51f30 (4.17-rc6+).
- Added linux-arch@ to the list of recipients.

Changes in v2:
- Rebased onto 2d618bdf (4.17-rc3+).
- Removed excessive untagging in gup.c.
- Removed untagging pointers returned from __uaccess_mask_ptr.

Changes in v1:
- Rebased onto 4.17-rc1.

Changes in RFC v2:
- Added "#ifndef untagged_addr..." fallback in linux/uaccess.h instead of defining it for each arch individually.
- Updated Documentation/arm64/tagged-pointers.txt.
- Dropped "mm, arm64: untag user addresses in memory syscalls".
- Rebased onto 3eb2ce82 (4.16-rc7).

Reviewed-by Luc Van Oostenryck <[email protected]>

Signed-off-by Andrey Konovalov <[email protected]>

Andrey Konovalov (12):
uaccess: add untagged_addr definition for other arches
arm64: untag user pointers in access_ok and __uaccess_mask_ptr lib, arm64: untag user pointers in strn*_user
mm, arm64: untag user pointers passed to memory syscalls mm, arm64: untag user pointers in mm/gup.c
fs, arm64: untag user pointers in copy_mount_options
fs, arm64: untag user pointers in fs/userfaultfd.c
net, arm64: untag user pointers in tcp_zerocopy_receive kernel, arm64: untag user pointers in prctl_set_mm* tracing, arm64: untag user pointers in seq_print_user_ip
arm64: update Documentation/arm64/tagged-pointers.txt selftests, arm64: add a selftest for passing tagged pointers to kernel

 Documentation/arm64/tagged-pointers.txt       | 25 +++++++++++--------
 arch/arm64/include/asm/uaccess.h              | 10 +++++---
 fs/namespace.c                                |  2 +-
 fs/userfaultfd.c                              |  5 ++++
 include/linux/memory.h                        |  4 +++
 ipc/shm.c                                     |  2 ++
 kernel/sys.c                                  | 14 +++++++++++
 kernel/trace/trace_output.c                   |  2 +-
 lib/strncpy_from_user.c                       |  2 ++
 lib/strnlen_user.c                            |  2 ++
 mm/gup.c                                      |  4 +++
 mm/madvise.c                                  |  2 ++
 mm/mempolicy.c                                |  5 ++++
 mm/migrate.c                                  |  1 +
 mm/mincore.c                                  |  2 ++
 mm/mlock.c                                    |  5 ++++
 mm/mmap.c                                     |  7 ++++++
 mm/mprotect.c                                 |  2 ++
 mm/mremap.c                                   |  2 ++
 mm/msync.c                                    |  2 ++
 net/ipv4/tcp.c                                |  2 ++
 tools/testing/selftests/arm64/.gitignore      |  1 +
 tools/testing/selftests/arm64/Makefile        | 11 ++++++++
 .../testing/selftests/arm64/run_tags_test.sh  | 12 +++++++++
 tools/testing/selftests/arm64/tags_test.c     | 19 ++++++++++++++
 25 files changed, 129 insertions(+), 16 deletions(-)
 create mode 100644 tools/testing/selftests/arm64/.gitignore
 create mode 100644 tools/testing/selftests/arm64/Makefile
 create mode 100755 tools/testing/selftests/arm64/run_tags_test.sh
 create mode 100644 tools/testing/selftests/arm64/tags_test.c

Quoted 39 lines · Szabolcs Nagy

Quoted 25 lines · Andrey Konovalov

This patchset is meant to be merged together with "arm64 relaxed ABI" [1].

arm64 has a feature called Top Byte Ignore, which allows to embed pointer tags into the top byte of each pointer. Userspace programs (such as HWASan, a memory debugging tool [2]) might use this feature and pass tagged user pointers to the kernel through syscalls or other interfaces.

Right now the kernel is already able to handle user faults with tagged pointers, due to these patches:

1. 81cddd65 ("arm64: traps: fix userspace cache maintenance emulation on a tagged pointer")
2. 7dcd9dd8 ("arm64: hw_breakpoint: fix watchpoint matching for tagged pointers")
3. 276e9327 ("arm64: entry: improve data abort handling of tagged pointers")

This patchset extends tagged pointer support to syscall arguments.

For non-memory syscalls this is done by untaging user pointers when the kernel performs pointer checking to find out whether the pointer comes from userspace (most notably in access_ok). The untagging is done only when the pointer is being checked, the tag is preserved as the pointer makes its way through the kernel.

Quoted 4 lines

Since memory syscalls (mmap, mprotect, etc.) don't do memory accesses but rather deal with memory ranges, untagged pointers are better suited to describe memory ranges internally. Thus for memory syscalls we untag pointers completely when they enter the kernel.

i think the same is true when user pointers are compared.

e.g. i suspect there may be issues with tagged robust mutex list pointers because the kernel does

futex.c:3541: while (entry != &head->list) {

Quoted 2 lines

where entry is a user pointer that may be tagged, and &head->list is probably not tagged.

You're right. I'll expand the cover letter in the next version to describe this more accurately. The patchset however contains "mm,
arm64: untag user pointers in mm/gup.c" that should take care of futex pointers.

Quoted 158 lines

Quoted 157 lines

One of the alternative approaches to untagging that was considered is to completely strip the pointer tag as the pointer enters the kernel with some kind of a syscall wrapper, but that won't work with the countless number of different ioctl calls. With this approach we would need a custom wrapper for each ioctl variation, which doesn't seem practical.

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

2. Static testing with grep to find parts of the kernel that call find_vma() (and other similar functions) or directly compare against vm_start/vm_end fields of vma.

3. Static testing with grep to find parts of the kernel that compare user pointers with TASK_SIZE or other similar consts and macros.

4. Dynamic testing: adding BUG_ON(has_tag(addr)) to find_vma() and running a modified syzkaller version that passes tagged pointers to the kernel.

Based on the results of the testing the requried patches have been added to the patchset.

This patchset has been merged into the Pixel 2 kernel tree and is now being used to enable testing of Pixel 2 phones with HWASan.

This patchset is a prerequisite for ARM's memory tagging hardware feature support [4].

Thanks!

[1] https://lkml.org/lkml/2018/12/10/402

[2] http://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html

[3] https://github.com/lucvoo/sparse-dev/commit/5f960cb10f56ec2017c128ef9d16060e0145f292

[4] https://community.arm.com/processors/b/blog/posts/arm-a-profile-architecture-2018-developments-armv85a

Changes in v10:
- Added "mm, arm64: untag user pointers passed to memory syscalls" back.
- New patch "fs, arm64: untag user pointers in fs/userfaultfd.c".
- New patch "net, arm64: untag user pointers in tcp_zerocopy_receive".
- New patch "kernel, arm64: untag user pointers in prctl_set_mm*".
- New patch "tracing, arm64: untag user pointers in seq_print_user_ip".

Changes in v9:
- Rebased onto 4.20-rc6.
- Used u64 instead of __u64 in type casts in the untagged_addr macro for arm64.
- Added braces around (addr) in the untagged_addr macro for other arches.

Changes in v8:
- Rebased onto 65102238 (4.20-rc1).
- Added a note to the cover letter on why syscall wrappers/shims that untag user pointers won't work.
- Added a note to the cover letter that this patchset has been merged into the Pixel 2 kernel tree.
- Documentation fixes, in particular added a list of syscalls that don't support tagged user pointers.

Changes in v7:
- Rebased onto 17b57b18 (4.19-rc6).
- Dropped the "arm64: untag user address in __do_user_fault" patch, since the existing patches already handle user faults properly.
- Dropped the "usb, arm64: untag user addresses in devio" patch, since the passed pointer must come from a vma and therefore be untagged.
- Dropped the "arm64: annotate user pointers casts detected by sparse" patch (see the discussion to the replies of the v6 of this patchset).
- Added more context to the cover letter.
- Updated Documentation/arm64/tagged-pointers.txt.

Changes in v6:
- Added annotations for user pointer casts found by sparse.
- Rebased onto 050cdc6c (4.19-rc1+).

Changes in v5:
- Added 3 new patches that add untagging to places found with static analysis.
- Rebased onto 44c929e1 (4.18-rc8).

Changes in v4:
- Added a selftest for checking that passing tagged pointers to the kernel succeeds.
- Rebased onto 81e97f013 (4.18-rc1+).

Changes in v3:
- Rebased onto e5c51f30 (4.17-rc6+).
- Added linux-arch@ to the list of recipients.

Changes in v2:
- Rebased onto 2d618bdf (4.17-rc3+).
- Removed excessive untagging in gup.c.
- Removed untagging pointers returned from __uaccess_mask_ptr.

Changes in v1:
- Rebased onto 4.17-rc1.

Changes in RFC v2:
- Added "#ifndef untagged_addr..." fallback in linux/uaccess.h instead of defining it for each arch individually.
- Updated Documentation/arm64/tagged-pointers.txt.
- Dropped "mm, arm64: untag user addresses in memory syscalls".
- Rebased onto 3eb2ce82 (4.16-rc7).

Reviewed-by Luc Van Oostenryck <[email protected]>

Signed-off-by Andrey Konovalov <[email protected]>

Andrey Konovalov (12):
uaccess: add untagged_addr definition for other arches
arm64: untag user pointers in access_ok and __uaccess_mask_ptr lib, arm64: untag user pointers in strn*_user
mm, arm64: untag user pointers passed to memory syscalls mm, arm64: untag user pointers in mm/gup.c
fs, arm64: untag user pointers in copy_mount_options
fs, arm64: untag user pointers in fs/userfaultfd.c
net, arm64: untag user pointers in tcp_zerocopy_receive kernel, arm64: untag user pointers in prctl_set_mm* tracing, arm64: untag user pointers in seq_print_user_ip
arm64: update Documentation/arm64/tagged-pointers.txt selftests, arm64: add a selftest for passing tagged pointers to kernel

 Documentation/arm64/tagged-pointers.txt       | 25 +++++++++++--------
 arch/arm64/include/asm/uaccess.h              | 10 +++++---
 fs/namespace.c                                |  2 +-
 fs/userfaultfd.c                              |  5 ++++
 include/linux/memory.h                        |  4 +++
 ipc/shm.c                                     |  2 ++
 kernel/sys.c                                  | 14 +++++++++++
 kernel/trace/trace_output.c                   |  2 +-
 lib/strncpy_from_user.c                       |  2 ++
 lib/strnlen_user.c                            |  2 ++
 mm/gup.c                                      |  4 +++
 mm/madvise.c                                  |  2 ++
 mm/mempolicy.c                                |  5 ++++
 mm/migrate.c                                  |  1 +
 mm/mincore.c                                  |  2 ++
 mm/mlock.c                                    |  5 ++++
 mm/mmap.c                                     |  7 ++++++
 mm/mprotect.c                                 |  2 ++
 mm/mremap.c                                   |  2 ++
 mm/msync.c                                    |  2 ++
 net/ipv4/tcp.c                                |  2 ++
 tools/testing/selftests/arm64/.gitignore      |  1 +
 tools/testing/selftests/arm64/Makefile        | 11 ++++++++
 .../testing/selftests/arm64/run_tags_test.sh  | 12 +++++++++
 tools/testing/selftests/arm64/tags_test.c     | 19 ++++++++++++++
 25 files changed, 129 insertions(+), 16 deletions(-)
 create mode 100644 tools/testing/selftests/arm64/.gitignore
 create mode 100644 tools/testing/selftests/arm64/Makefile
 create mode 100755 tools/testing/selftests/arm64/run_tags_test.sh
 create mode 100644 tools/testing/selftests/arm64/tags_test.c

Quoted 43 lines · Andrey Konovalov

Quoted 41 lines · Szabolcs Nagy

Quoted 25 lines · Andrey Konovalov

This patchset is meant to be merged together with "arm64 relaxed ABI" [1].

arm64 has a feature called Top Byte Ignore, which allows to embed pointer tags into the top byte of each pointer. Userspace programs (such as HWASan, a memory debugging tool [2]) might use this feature and pass tagged user pointers to the kernel through syscalls or other interfaces.

Right now the kernel is already able to handle user faults with tagged pointers, due to these patches:

1. 81cddd65 ("arm64: traps: fix userspace cache maintenance emulation on a tagged pointer")
2. 7dcd9dd8 ("arm64: hw_breakpoint: fix watchpoint matching for tagged pointers")
3. 276e9327 ("arm64: entry: improve data abort handling of tagged pointers")

This patchset extends tagged pointer support to syscall arguments.

For non-memory syscalls this is done by untaging user pointers when the kernel performs pointer checking to find out whether the pointer comes from userspace (most notably in access_ok). The untagging is done only when the pointer is being checked, the tag is preserved as the pointer makes its way through the kernel.

Quoted 4 lines

Since memory syscalls (mmap, mprotect, etc.) don't do memory accesses but rather deal with memory ranges, untagged pointers are better suited to describe memory ranges internally. Thus for memory syscalls we untag pointers completely when they enter the kernel.

i think the same is true when user pointers are compared.

e.g. i suspect there may be issues with tagged robust mutex list pointers because the kernel does

futex.c:3541: while (entry != &head->list) {

where entry is a user pointer that may be tagged, and &head->list is probably not tagged.

Quoted 4 lines

You're right. I'll expand the cover letter in the next version to describe this more accurately. The patchset however contains "mm,
arm64: untag user pointers in mm/gup.c" that should take care of futex pointers.

the robust mutex list pointer is not a futex pointer, i'm not sure how the mm/gup.c patch helps.

Quoted 1 line

Quoted 1 line

futex.c:3541: while (entry != &head->list) {

Quoted 159 lines

Quoted 159 lines

Quoted 158 lines

One of the alternative approaches to untagging that was considered is to completely strip the pointer tag as the pointer enters the kernel with some kind of a syscall wrapper, but that won't work with the countless number of different ioctl calls. With this approach we would need a custom wrapper for each ioctl variation, which doesn't seem practical.

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

2. Static testing with grep to find parts of the kernel that call find_vma() (and other similar functions) or directly compare against vm_start/vm_end fields of vma.

3. Static testing with grep to find parts of the kernel that compare user pointers with TASK_SIZE or other similar consts and macros.

4. Dynamic testing: adding BUG_ON(has_tag(addr)) to find_vma() and running a modified syzkaller version that passes tagged pointers to the kernel.

Based on the results of the testing the requried patches have been added to the patchset.

This patchset has been merged into the Pixel 2 kernel tree and is now being used to enable testing of Pixel 2 phones with HWASan.

This patchset is a prerequisite for ARM's memory tagging hardware feature support [4].

Thanks!

[1] https://lkml.org/lkml/2018/12/10/402

[2] http://clang.llvm.org/docs/HardwareAssistedAddressSanitizerDesign.html

[3] https://github.com/lucvoo/sparse-dev/commit/5f960cb10f56ec2017c128ef9d16060e0145f292

[4] https://community.arm.com/processors/b/blog/posts/arm-a-profile-architecture-2018-developments-armv85a

Changes in v10:
- Added "mm, arm64: untag user pointers passed to memory syscalls" back.
- New patch "fs, arm64: untag user pointers in fs/userfaultfd.c".
- New patch "net, arm64: untag user pointers in tcp_zerocopy_receive".
- New patch "kernel, arm64: untag user pointers in prctl_set_mm*".
- New patch "tracing, arm64: untag user pointers in seq_print_user_ip".

Changes in v9:
- Rebased onto 4.20-rc6.
- Used u64 instead of __u64 in type casts in the untagged_addr macro for arm64.
- Added braces around (addr) in the untagged_addr macro for other arches.

Changes in v8:
- Rebased onto 65102238 (4.20-rc1).
- Added a note to the cover letter on why syscall wrappers/shims that untag user pointers won't work.
- Added a note to the cover letter that this patchset has been merged into the Pixel 2 kernel tree.
- Documentation fixes, in particular added a list of syscalls that don't support tagged user pointers.

Changes in v7:
- Rebased onto 17b57b18 (4.19-rc6).
- Dropped the "arm64: untag user address in __do_user_fault" patch, since the existing patches already handle user faults properly.
- Dropped the "usb, arm64: untag user addresses in devio" patch, since the passed pointer must come from a vma and therefore be untagged.
- Dropped the "arm64: annotate user pointers casts detected by sparse" patch (see the discussion to the replies of the v6 of this patchset).
- Added more context to the cover letter.
- Updated Documentation/arm64/tagged-pointers.txt.

Changes in v6:
- Added annotations for user pointer casts found by sparse.
- Rebased onto 050cdc6c (4.19-rc1+).

Changes in v5:
- Added 3 new patches that add untagging to places found with static analysis.
- Rebased onto 44c929e1 (4.18-rc8).

Changes in v4:
- Added a selftest for checking that passing tagged pointers to the kernel succeeds.
- Rebased onto 81e97f013 (4.18-rc1+).

Changes in v3:
- Rebased onto e5c51f30 (4.17-rc6+).
- Added linux-arch@ to the list of recipients.

Changes in v2:
- Rebased onto 2d618bdf (4.17-rc3+).
- Removed excessive untagging in gup.c.
- Removed untagging pointers returned from __uaccess_mask_ptr.

Changes in v1:
- Rebased onto 4.17-rc1.

Changes in RFC v2:
- Added "#ifndef untagged_addr..." fallback in linux/uaccess.h instead of defining it for each arch individually.
- Updated Documentation/arm64/tagged-pointers.txt.
- Dropped "mm, arm64: untag user addresses in memory syscalls".
- Rebased onto 3eb2ce82 (4.16-rc7).

Reviewed-by Luc Van Oostenryck <[email protected]>

Signed-off-by Andrey Konovalov <[email protected]>

Andrey Konovalov (12):
uaccess: add untagged_addr definition for other arches
arm64: untag user pointers in access_ok and __uaccess_mask_ptr lib, arm64: untag user pointers in strn*_user
mm, arm64: untag user pointers passed to memory syscalls mm, arm64: untag user pointers in mm/gup.c
fs, arm64: untag user pointers in copy_mount_options
fs, arm64: untag user pointers in fs/userfaultfd.c
net, arm64: untag user pointers in tcp_zerocopy_receive kernel, arm64: untag user pointers in prctl_set_mm* tracing, arm64: untag user pointers in seq_print_user_ip
arm64: update Documentation/arm64/tagged-pointers.txt selftests, arm64: add a selftest for passing tagged pointers to kernel

 Documentation/arm64/tagged-pointers.txt       | 25 +++++++++++--------
 arch/arm64/include/asm/uaccess.h              | 10 +++++---
 fs/namespace.c                                |  2 +-
 fs/userfaultfd.c                              |  5 ++++
 include/linux/memory.h                        |  4 +++
 ipc/shm.c                                     |  2 ++
 kernel/sys.c                                  | 14 +++++++++++
 kernel/trace/trace_output.c                   |  2 +-
 lib/strncpy_from_user.c                       |  2 ++
 lib/strnlen_user.c                            |  2 ++
 mm/gup.c                                      |  4 +++
 mm/madvise.c                                  |  2 ++
 mm/mempolicy.c                                |  5 ++++
 mm/migrate.c                                  |  1 +
 mm/mincore.c                                  |  2 ++
 mm/mlock.c                                    |  5 ++++
 mm/mmap.c                                     |  7 ++++++
 mm/mprotect.c                                 |  2 ++
 mm/mremap.c                                   |  2 ++
 mm/msync.c                                    |  2 ++
 net/ipv4/tcp.c                                |  2 ++
 tools/testing/selftests/arm64/.gitignore      |  1 +
 tools/testing/selftests/arm64/Makefile        | 11 ++++++++
 .../testing/selftests/arm64/run_tags_test.sh  | 12 +++++++++
 tools/testing/selftests/arm64/tags_test.c     | 19 ++++++++++++++
 25 files changed, 129 insertions(+), 16 deletions(-)
 create mode 100644 tools/testing/selftests/arm64/.gitignore
 create mode 100644 tools/testing/selftests/arm64/Makefile
 create mode 100755 tools/testing/selftests/arm64/run_tags_test.sh
 create mode 100644 tools/testing/selftests/arm64/tags_test.c

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

Quoted 6 lines · Andrey Konovalov

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

First of all, it's really cool that you took this approach. Sounds like there was a lot of systematic work to fix up the sites in the existing codebase.

But, isn't this a _bit_ fragile going forward? Folks can't just "make sparse" to find issues with missing untags.

This seems like something
where we would ideally add an __tagged annotation (or something) to the source tree and then have sparse rules that can look for missed untags.

Quoted 3 lines

This seems like something
where we would ideally add an __tagged annotation (or something) to the source tree and then have sparse rules that can look for missed untags.

This has been suggested before, search for __untagged here [1]. However there are many places in the kernel where a __user pointer is casted into unsigned long and passed further. I'm not sure if it's possible apply a __tagged/__untagged kind of attribute to non-pointer types, is it?

[1] https://patchwork.kernel.org/patch/10581535/

Quoted 9 lines · Dave Hansen

Quoted 6 lines · Andrey Konovalov

The following testing approaches has been taken to find potential issues with user pointer untagging:

1. Static testing (with sparse [3] and separately with a custom static analyzer based on Clang) to track casts of __user pointers to integer types to find places where untagging needs to be done.

Quoted 6 lines

First of all, it's really cool that you took this approach. Sounds like there was a lot of systematic work to fix up the sites in the existing codebase.

But, isn't this a _bit_ fragile going forward? Folks can't just "make sparse" to find issues with missing untags.

Yes, this static approach can only be used as a hint to find some places where untagging is needed, but certainly not all.

Quoted 3 lines

This seems like something
where we would ideally add an __tagged annotation (or something) to the source tree and then have sparse rules that can look for missed untags.

This has been suggested before, search for __untagged here [1]. However there are many places in the kernel where a __user pointer is casted into unsigned long and passed further. I'm not sure if it's possible apply a __tagged/__untagged kind of attribute to non-pointer types, is it?

[1] https://patchwork.kernel.org/patch/10581535/