[PATCH v3 0/4] mm: convert to vma_is_initial_heap/stack()

[PATCH v3 0/4] mm: convert to vma_is_initial_heap/stack()

[Kefeng Wang]

Add vma_is_initial_stack() and vma_is_initial_heap() helper and use
them to simplify code.

v2:
- add comment for heap helper and remove one more goto cpy_name,
  per David Hildenbrand
- add RB
v2:
- address comments per David Hildenbrand and Christian Göttsche
- fix selinux build

Kefeng Wang (4):
  mm: factor out VMA stack and heap checks
  drm/amdkfd: use vma_is_initial_stack() and vma_is_initial_heap()
  selinux: use vma_is_initial_stack() and vma_is_initial_heap()
  perf/core: use vma_is_initial_stack() and vma_is_initial_heap()

 drivers/gpu/drm/amd/amdkfd/kfd_svm.c |  5 +----
 fs/proc/task_mmu.c                   | 24 ++++----------------
 fs/proc/task_nommu.c                 | 15 +------------
 include/linux/mm.h                   | 25 +++++++++++++++++++++
 kernel/events/core.c                 | 33 ++++++++++------------------
 security/selinux/hooks.c             |  7 ++----
 6 files changed, 44 insertions(+), 65 deletions(-)

-- 
2.41.0

[PATCH v3 1/4] mm: factor out VMA stack and heap checks

[Kefeng Wang]

Factor out VMA stack and heap checks and name them
vma_is_initial_stack() and vma_is_initial_heap() for
general use.

Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: David Hildenbrand <david@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
---
 fs/proc/task_mmu.c   | 24 ++++--------------------
 fs/proc/task_nommu.c | 15 +--------------
 include/linux/mm.h   | 25 +++++++++++++++++++++++++
 3 files changed, 30 insertions(+), 34 deletions(-)

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 2b434b3fa2ec..3f063201b1ec 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -236,21 +236,6 @@ static int do_maps_open(struct inode *inode, struct file *file,
 				sizeof(struct proc_maps_private));
 }
 
-/*
- * Indicate if the VMA is a stack for the given task; for
- * /proc/PID/maps that is the stack of the main task.
- */
-static int is_stack(struct vm_area_struct *vma)
-{
-	/*
-	 * We make no effort to guess what a given thread considers to be
-	 * its "stack".  It's not even well-defined for programs written
-	 * languages like Go.
-	 */
-	return vma->vm_start <= vma->vm_mm->start_stack &&
-		vma->vm_end >= vma->vm_mm->start_stack;
-}
-
 static void show_vma_header_prefix(struct seq_file *m,
 				   unsigned long start, unsigned long end,
 				   vm_flags_t flags, unsigned long long pgoff,
@@ -327,13 +312,12 @@ show_map_vma(struct seq_file *m, struct vm_area_struct *vma)
 			goto done;
 		}
 
-		if (vma->vm_start <= mm->brk &&
-		    vma->vm_end >= mm->start_brk) {
+		if (vma_is_initial_heap(vma)) {
 			name = "[heap]";
 			goto done;
 		}
 
-		if (is_stack(vma)) {
+		if (vma_is_initial_stack(vma)) {
 			name = "[stack]";
 			goto done;
 		}
@@ -1975,9 +1959,9 @@ static int show_numa_map(struct seq_file *m, void *v)
 	if (file) {
 		seq_puts(m, " file=");
 		seq_file_path(m, file, "\n\t= ");
-	} else if (vma->vm_start <= mm->brk && vma->vm_end >= mm->start_brk) {
+	} else if (vma_is_initial_heap(vma)) {
 		seq_puts(m, " heap");
-	} else if (is_stack(vma)) {
+	} else if (vma_is_initial_stack(vma)) {
 		seq_puts(m, " stack");
 	}
 
diff --git a/fs/proc/task_nommu.c b/fs/proc/task_nommu.c
index 2c8b62265981..a8ac0dd8041e 100644
--- a/fs/proc/task_nommu.c
+++ b/fs/proc/task_nommu.c
@@ -121,19 +121,6 @@ unsigned long task_statm(struct mm_struct *mm,
 	return size;
 }
 
-static int is_stack(struct vm_area_struct *vma)
-{
-	struct mm_struct *mm = vma->vm_mm;
-
-	/*
-	 * We make no effort to guess what a given thread considers to be
-	 * its "stack".  It's not even well-defined for programs written
-	 * languages like Go.
-	 */
-	return vma->vm_start <= mm->start_stack &&
-		vma->vm_end >= mm->start_stack;
-}
-
 /*
  * display a single VMA to a sequenced file
  */
@@ -171,7 +158,7 @@ static int nommu_vma_show(struct seq_file *m, struct vm_area_struct *vma)
 	if (file) {
 		seq_pad(m, ' ');
 		seq_file_path(m, file, "");
-	} else if (mm && is_stack(vma)) {
+	} else if (mm && vma_is_initial_stack(vma)) {
 		seq_pad(m, ' ');
 		seq_puts(m, "[stack]");
 	}
diff --git a/include/linux/mm.h b/include/linux/mm.h
index acaef8d106f0..2fbc6c631764 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -866,6 +866,31 @@ static inline bool vma_is_anonymous(struct vm_area_struct *vma)
 	return !vma->vm_ops;
 }
 
+/*
+ * Indicate if the VMA is a heap for the given task; for
+ * /proc/PID/maps that is the heap of the main task.
+ */
+static inline bool vma_is_initial_heap(const struct vm_area_struct *vma)
+{
+       return vma->vm_start <= vma->vm_mm->brk &&
+		vma->vm_end >= vma->vm_mm->start_brk;
+}
+
+/*
+ * Indicate if the VMA is a stack for the given task; for
+ * /proc/PID/maps that is the stack of the main task.
+ */
+static inline bool vma_is_initial_stack(const struct vm_area_struct *vma)
+{
+	/*
+	 * We make no effort to guess what a given thread considers to be
+	 * its "stack".  It's not even well-defined for programs written
+	 * languages like Go.
+	 */
+       return vma->vm_start <= vma->vm_mm->start_stack &&
+	       vma->vm_end >= vma->vm_mm->start_stack;
+}
+
 static inline bool vma_is_temporary_stack(struct vm_area_struct *vma)
 {
 	int maybe_stack = vma->vm_flags & (VM_GROWSDOWN | VM_GROWSUP);
-- 
2.41.0

[PATCH v3 2/4] drm/amdkfd: use vma_is_initial_stack() and vma_is_initial_heap()

[Kefeng Wang]

Use the helpers to simplify code.

Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Cc: David Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 01c7de2d6e19..308384dbc502 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -2624,10 +2624,7 @@ svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr,
 		return -EFAULT;
 	}
 
-	*is_heap_stack = (vma->vm_start <= vma->vm_mm->brk &&
-			  vma->vm_end >= vma->vm_mm->start_brk) ||
-			 (vma->vm_start <= vma->vm_mm->start_stack &&
-			  vma->vm_end >= vma->vm_mm->start_stack);
+	*is_heap_stack = vma_is_initial_heap(vma) || vma_is_initial_stack(vma);
 
 	start_limit = max(vma->vm_start >> PAGE_SHIFT,
 		      (unsigned long)ALIGN_DOWN(addr, 2UL << 8));
-- 
2.41.0

[PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Kefeng Wang]

Use the helpers to simplify code.

Cc: Paul Moore <paul@paul-moore.com>
Cc: Stephen Smalley <stephen.smalley.work@gmail.com>
Cc: Eric Paris <eparis@parisplace.org>
Acked-by: Paul Moore <paul@paul-moore.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
---
 security/selinux/hooks.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c
index c87b79a29fad..ac582c046c51 100644
--- a/security/selinux/hooks.c
+++ b/security/selinux/hooks.c
@@ -3800,13 +3800,10 @@ static int selinux_file_mprotect(struct vm_area_struct *vma,
 	if (default_noexec &&
 	    (prot & PROT_EXEC) && !(vma->vm_flags & VM_EXEC)) {
 		int rc = 0;
-		if (vma->vm_start >= vma->vm_mm->start_brk &&
-		    vma->vm_end <= vma->vm_mm->brk) {
+		if (vma_is_initial_heap(vma)) {
 			rc = avc_has_perm(sid, sid, SECCLASS_PROCESS,
 					  PROCESS__EXECHEAP, NULL);
-		} else if (!vma->vm_file &&
-			   ((vma->vm_start <= vma->vm_mm->start_stack &&
-			     vma->vm_end >= vma->vm_mm->start_stack) ||
+		} else if (!vma->vm_file && (vma_is_initial_stack(vma) ||
 			    vma_is_stack_for_current(vma))) {
 			rc = avc_has_perm(sid, sid, SECCLASS_PROCESS,
 					  PROCESS__EXECSTACK, NULL);
-- 
2.41.0

[PATCH v3 4/4] perf/core: use vma_is_initial_stack() and vma_is_initial_heap()

[Kefeng Wang]

Use the helpers to simplify code, also kill unneeded goto cpy_name.

Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
---
 kernel/events/core.c | 33 +++++++++++----------------------
 1 file changed, 11 insertions(+), 22 deletions(-)

diff --git a/kernel/events/core.c b/kernel/events/core.c
index f84e2640ea2f..b36d0823ff33 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -8631,7 +8631,7 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event)
 	unsigned int size;
 	char tmp[16];
 	char *buf = NULL;
-	char *name;
+	char *name = NULL;
 
 	if (vma->vm_flags & VM_READ)
 		prot |= PROT_READ;
@@ -8678,29 +8678,18 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event)
 
 		goto got_name;
 	} else {
-		if (vma->vm_ops && vma->vm_ops->name) {
+		if (vma->vm_ops && vma->vm_ops->name)
 			name = (char *) vma->vm_ops->name(vma);
-			if (name)
-				goto cpy_name;
+		if (!name)
+			name = (char *)arch_vma_name(vma);
+		if (!name) {
+			if (vma_is_initial_heap(vma))
+				name = "[heap]";
+			else if (vma_is_initial_stack(vma))
+				name = "[stack]";
+			else
+				name = "//anon";
 		}
-
-		name = (char *)arch_vma_name(vma);
-		if (name)
-			goto cpy_name;
-
-		if (vma->vm_start <= vma->vm_mm->start_brk &&
-				vma->vm_end >= vma->vm_mm->brk) {
-			name = "[heap]";
-			goto cpy_name;
-		}
-		if (vma->vm_start <= vma->vm_mm->start_stack &&
-				vma->vm_end >= vma->vm_mm->start_stack) {
-			name = "[stack]";
-			goto cpy_name;
-		}
-
-		name = "//anon";
-		goto cpy_name;
 	}
 
 cpy_name:
-- 
2.41.0

Re: [PATCH v3 0/4] mm: convert to vma_is_initial_heap/stack()

[Peter Zijlstra]

On Fri, Jul 28, 2023 at 01:00:39PM +0800, Kefeng Wang wrote:

> Kefeng Wang (4):
>   mm: factor out VMA stack and heap checks
>   drm/amdkfd: use vma_is_initial_stack() and vma_is_initial_heap()
>   selinux: use vma_is_initial_stack() and vma_is_initial_heap()
>   perf/core: use vma_is_initial_stack() and vma_is_initial_heap()
> 
>  drivers/gpu/drm/amd/amdkfd/kfd_svm.c |  5 +----
>  fs/proc/task_mmu.c                   | 24 ++++----------------
>  fs/proc/task_nommu.c                 | 15 +------------
>  include/linux/mm.h                   | 25 +++++++++++++++++++++
>  kernel/events/core.c                 | 33 ++++++++++------------------
>  security/selinux/hooks.c             |  7 ++----
>  6 files changed, 44 insertions(+), 65 deletions(-)

Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Ondrej Mosnacek]

(Restoring part of the Cc list to include more relevant lists &
people... If you are lost, the original email is here:
https://lore.kernel.org/selinux/20230728050043.59880-4-wangkefeng.wang@huawei.com/)

On Tue, Aug 1, 2023 at 1:08 AM Paul Moore <paul@paul-moore.com> wrote:
>
> On Mon, Jul 31, 2023 at 4:02 PM Stephen Smalley
> <stephen.smalley.work@gmail.com> wrote:
> > On Mon, Jul 31, 2023 at 12:19 PM Paul Moore <paul@paul-moore.com> wrote:
> > >
> > > On Mon, Jul 31, 2023 at 10:26 AM Stephen Smalley
> > > <stephen.smalley.work@gmail.com> wrote:
> > > > I believe this patch yields a semantic change in the SELinux execheap
> > > > permission check. That said, I think the change is for the better.
> > >
> > > Agreed.  I'm also in favor of using a helper which is maintained by
> > > the VM folks over open coded logic in the SELinux code.
> >
> > Yes, only caveat is in theory it could trigger new execheap denials
> > under existing policies.
> > Trying to construct an example based on the
> > selinux-testsuite/tests/mmap/mprot_heap.c example but coming up empty
> > so far on something that both works and yields different results
> > before and after this patch.
>
> My gut feeling is that this will not be an issue, but I could very
> well be wrong.  If it becomes a significant issue we can revert the
> SELinux portion of the patch.
>
> Of course, if you have any luck demonstrating this with reasonable
> code, that would be good input too.

So, it turns out this does affect actual code. Thus far, we know about
gcl [1] and wine [2]. The gcl case is easy to reproduce (just install
gcl on Fedora and run gcl without arguments), so I was able to dig a
bit deeper.

gcl has the following relevant memory mappings as captured by gdb:
Start Addr           End Addr       Size     Offset  Perms  objfile
  0x413000           0xf75000   0xb62000   0x3fa000  rw-p
/usr/lib/gcl-2.6.14/unixport/saved_ansi_gcl
  0xf75000           0xf79000     0x4000        0x0  rwxp   [heap]

It tries to call mprotect(0x883000, 7282688,
PROT_READ|PROT_WRITE|PROT_EXEC), i.e. it tries to make the region
0x883000 - 0xf75000 executable. Before this patch it was allowed,
whereas now it triggers an execheap SELinux denial. But this seems
wrong - the affected region is merely adjacent to the [heap] region,
it does not actually overlap with it. So even if we accept that the
correct semantics is to catch any region that overlaps with the heap
(before only subregions of the heap were subject to the execheap
check), this corner case doesn't seem to be handled correctly by the
new check (and the same bug seems to have been in fs/proc/task_mmu.c
before commit 11250fd12eb8 ("mm: factor out VMA stack and heap
checks")).

I didn't analyze the wine case ([2]), but it may be the same situation.

Unless I'm mistaken, those <= & >= in should in fact be just < & >.
And the expression in vma_is_initial_stack() is also suspicious (but
I'm not going to make any assumption on what is the intended semantics
there...)

[1] https://bugzilla.redhat.com/show_bug.cgi?id=2252391
[2] https://bugzilla.redhat.com/show_bug.cgi?id=2247299

--
Ondrej Mosnacek
Senior Software Engineer, Linux Security - SELinux kernel
Red Hat, Inc.

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Paul Moore]

On Wed, Dec 6, 2023 at 9:22 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
>
> (Restoring part of the Cc list to include more relevant lists &
> people... If you are lost, the original email is here:
> https://lore.kernel.org/selinux/20230728050043.59880-4-wangkefeng.wang@huawei.com/)
>
> On Tue, Aug 1, 2023 at 1:08 AM Paul Moore <paul@paul-moore.com> wrote:
> > On Mon, Jul 31, 2023 at 4:02 PM Stephen Smalley
> > <stephen.smalley.work@gmail.com> wrote:
> > > On Mon, Jul 31, 2023 at 12:19 PM Paul Moore <paul@paul-moore.com> wrote:
> > > >
> > > > On Mon, Jul 31, 2023 at 10:26 AM Stephen Smalley
> > > > <stephen.smalley.work@gmail.com> wrote:
> > > > > I believe this patch yields a semantic change in the SELinux execheap
> > > > > permission check. That said, I think the change is for the better.
> > > >
> > > > Agreed.  I'm also in favor of using a helper which is maintained by
> > > > the VM folks over open coded logic in the SELinux code.
> > >
> > > Yes, only caveat is in theory it could trigger new execheap denials
> > > under existing policies.
> > > Trying to construct an example based on the
> > > selinux-testsuite/tests/mmap/mprot_heap.c example but coming up empty
> > > so far on something that both works and yields different results
> > > before and after this patch.
> >
> > My gut feeling is that this will not be an issue, but I could very
> > well be wrong.  If it becomes a significant issue we can revert the
> > SELinux portion of the patch.
> >
> > Of course, if you have any luck demonstrating this with reasonable
> > code, that would be good input too.
>
> So, it turns out this does affect actual code. Thus far, we know about
> gcl [1] and wine [2]. The gcl case is easy to reproduce (just install
> gcl on Fedora and run gcl without arguments), so I was able to dig a
> bit deeper.
>
> gcl has the following relevant memory mappings as captured by gdb:
> Start Addr           End Addr       Size     Offset  Perms  objfile
>   0x413000           0xf75000   0xb62000   0x3fa000  rw-p
> /usr/lib/gcl-2.6.14/unixport/saved_ansi_gcl
>   0xf75000           0xf79000     0x4000        0x0  rwxp   [heap]
>
> It tries to call mprotect(0x883000, 7282688,
> PROT_READ|PROT_WRITE|PROT_EXEC), i.e. it tries to make the region
> 0x883000 - 0xf75000 executable. Before this patch it was allowed,
> whereas now it triggers an execheap SELinux denial. But this seems
> wrong - the affected region is merely adjacent to the [heap] region,
> it does not actually overlap with it. So even if we accept that the
> correct semantics is to catch any region that overlaps with the heap
> (before only subregions of the heap were subject to the execheap
> check), this corner case doesn't seem to be handled correctly by the
> new check (and the same bug seems to have been in fs/proc/task_mmu.c
> before commit 11250fd12eb8 ("mm: factor out VMA stack and heap
> checks")).
>
> I didn't analyze the wine case ([2]), but it may be the same situation.
>
> Unless I'm mistaken, those <= & >= in should in fact be just < & >.
> And the expression in vma_is_initial_stack() is also suspicious (but
> I'm not going to make any assumption on what is the intended semantics
> there...)
>
> [1] https://bugzilla.redhat.com/show_bug.cgi?id=2252391
> [2] https://bugzilla.redhat.com/show_bug.cgi?id=2247299

Thanks Ondrej.

I'm hoping the mm folks will comment on this as it looks like this is
an issue with the helper functions, but just in case I'm going to prep
a revert for just the SELinux changes.  If we don't hear anything in
the next couple of days I'll send the revert up to Linus with the idea
that we can eventually shift back to the helpers when this is sorted.

-- 
paul-moore.com

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Kefeng Wang]

On 2023/12/7 4:47, Paul Moore wrote:
> On Wed, Dec 6, 2023 at 9:22 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
>>
>> (Restoring part of the Cc list to include more relevant lists &
>> people... If you are lost, the original email is here:
>> https://lore.kernel.org/selinux/20230728050043.59880-4-wangkefeng.wang@huawei.com/)
>>
>> On Tue, Aug 1, 2023 at 1:08 AM Paul Moore <paul@paul-moore.com> wrote:
>>> On Mon, Jul 31, 2023 at 4:02 PM Stephen Smalley
>>> <stephen.smalley.work@gmail.com> wrote:
>>>> On Mon, Jul 31, 2023 at 12:19 PM Paul Moore <paul@paul-moore.com> wrote:
>>>>>
>>>>> On Mon, Jul 31, 2023 at 10:26 AM Stephen Smalley
>>>>> <stephen.smalley.work@gmail.com> wrote:
>>>>>> I believe this patch yields a semantic change in the SELinux execheap
>>>>>> permission check. That said, I think the change is for the better.
>>>>>
>>>>> Agreed.  I'm also in favor of using a helper which is maintained by
>>>>> the VM folks over open coded logic in the SELinux code.
>>>>
>>>> Yes, only caveat is in theory it could trigger new execheap denials
>>>> under existing policies.
>>>> Trying to construct an example based on the
>>>> selinux-testsuite/tests/mmap/mprot_heap.c example but coming up empty
>>>> so far on something that both works and yields different results
>>>> before and after this patch.
>>>
>>> My gut feeling is that this will not be an issue, but I could very
>>> well be wrong.  If it becomes a significant issue we can revert the
>>> SELinux portion of the patch.
>>>
>>> Of course, if you have any luck demonstrating this with reasonable
>>> code, that would be good input too.
>>
>> So, it turns out this does affect actual code. Thus far, we know about
>> gcl [1] and wine [2]. The gcl case is easy to reproduce (just install
>> gcl on Fedora and run gcl without arguments), so I was able to dig a
>> bit deeper.
>>
>> gcl has the following relevant memory mappings as captured by gdb:
>> Start Addr           End Addr       Size     Offset  Perms  objfile
>>    0x413000           0xf75000   0xb62000   0x3fa000  rw-p
>> /usr/lib/gcl-2.6.14/unixport/saved_ansi_gcl
>>    0xf75000           0xf79000     0x4000        0x0  rwxp   [heap]
>>
>> It tries to call mprotect(0x883000, 7282688,
>> PROT_READ|PROT_WRITE|PROT_EXEC), i.e. it tries to make the region
>> 0x883000 - 0xf75000 executable. Before this patch it was allowed,
>> whereas now it triggers an execheap SELinux denial. But this seems
>> wrong - the affected region is merely adjacent to the [heap] region,
>> it does not actually overlap with it. So even if we accept that the
>> correct semantics is to catch any region that overlaps with the heap
>> (before only subregions of the heap were subject to the execheap
>> check), this corner case doesn't seem to be handled correctly by the
>> new check (and the same bug seems to have been in fs/proc/task_mmu.c
>> before commit 11250fd12eb8 ("mm: factor out VMA stack and heap
>> checks")).

Yes, the heap check exists for a long time,

                          [start_brk        brk]
case1:                     [vm_start,vm_end]
case2:             [vm_start,vm_end]
case3:                               [vm_start,vm_end]
case4:         [vm_start,                      vm_end]

case5:   [vm_start,vm_end]
case6:                                          [vm_start,vm_end]

old check:
vma->vm_start >= vma->vm_mm->start_brk && vma->vm_end <= vma->vm_mm->brk

Only include case1， vma range must be within heap

new check:
vma->vm_start <= vma->vm_mm->brk && vma->vm_end >= vma->vm_mm->start_brk

Include case1~case6, but case5(vm_end=start_brk) and case6(vm_start=brk)
are the corner cases, gcl issue matchs the case5.

>>
>> I didn't analyze the wine case ([2]), but it may be the same situation.
>>
>> Unless I'm mistaken, those <= & >= in should in fact be just < & >.

I support this change.
  vma->vm_start < vma->vm_mm->brk && vma->vm_end > vma->vm_mm->start_brk

>> And the expression in vma_is_initial_stack() is also suspicious (but
>> I'm not going to make any assumption on what is the intended semantics
>> there...)
>>
>> [1] https://bugzilla.redhat.com/show_bug.cgi?id=2252391
>> [2] https://bugzilla.redhat.com/show_bug.cgi?id=2247299

Could you quickly verify after the above change?
> 
> Thanks Ondrej.
> 
> I'm hoping the mm folks will comment on this as it looks like this is
> an issue with the helper functions, but just in case I'm going to prep
> a revert for just the SELinux changes.  If we don't hear anything in
> the next couple of days I'll send the revert up to Linus with the idea
> that we can eventually shift back to the helpers when this is sorted.
> 

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Ondrej Mosnacek]

On Thu, Dec 7, 2023 at 5:50 AM Kefeng Wang <wangkefeng.wang@huawei.com> wrote:
>
>
>
> On 2023/12/7 4:47, Paul Moore wrote:
> > On Wed, Dec 6, 2023 at 9:22 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
> >>
> >> (Restoring part of the Cc list to include more relevant lists &
> >> people... If you are lost, the original email is here:
> >> https://lore.kernel.org/selinux/20230728050043.59880-4-wangkefeng.wang@huawei.com/)
> >>
> >> On Tue, Aug 1, 2023 at 1:08 AM Paul Moore <paul@paul-moore.com> wrote:
> >>> On Mon, Jul 31, 2023 at 4:02 PM Stephen Smalley
> >>> <stephen.smalley.work@gmail.com> wrote:
> >>>> On Mon, Jul 31, 2023 at 12:19 PM Paul Moore <paul@paul-moore.com> wrote:
> >>>>>
> >>>>> On Mon, Jul 31, 2023 at 10:26 AM Stephen Smalley
> >>>>> <stephen.smalley.work@gmail.com> wrote:
> >>>>>> I believe this patch yields a semantic change in the SELinux execheap
> >>>>>> permission check. That said, I think the change is for the better.
> >>>>>
> >>>>> Agreed.  I'm also in favor of using a helper which is maintained by
> >>>>> the VM folks over open coded logic in the SELinux code.
> >>>>
> >>>> Yes, only caveat is in theory it could trigger new execheap denials
> >>>> under existing policies.
> >>>> Trying to construct an example based on the
> >>>> selinux-testsuite/tests/mmap/mprot_heap.c example but coming up empty
> >>>> so far on something that both works and yields different results
> >>>> before and after this patch.
> >>>
> >>> My gut feeling is that this will not be an issue, but I could very
> >>> well be wrong.  If it becomes a significant issue we can revert the
> >>> SELinux portion of the patch.
> >>>
> >>> Of course, if you have any luck demonstrating this with reasonable
> >>> code, that would be good input too.
> >>
> >> So, it turns out this does affect actual code. Thus far, we know about
> >> gcl [1] and wine [2]. The gcl case is easy to reproduce (just install
> >> gcl on Fedora and run gcl without arguments), so I was able to dig a
> >> bit deeper.
> >>
> >> gcl has the following relevant memory mappings as captured by gdb:
> >> Start Addr           End Addr       Size     Offset  Perms  objfile
> >>    0x413000           0xf75000   0xb62000   0x3fa000  rw-p
> >> /usr/lib/gcl-2.6.14/unixport/saved_ansi_gcl
> >>    0xf75000           0xf79000     0x4000        0x0  rwxp   [heap]
> >>
> >> It tries to call mprotect(0x883000, 7282688,
> >> PROT_READ|PROT_WRITE|PROT_EXEC), i.e. it tries to make the region
> >> 0x883000 - 0xf75000 executable. Before this patch it was allowed,
> >> whereas now it triggers an execheap SELinux denial. But this seems
> >> wrong - the affected region is merely adjacent to the [heap] region,
> >> it does not actually overlap with it. So even if we accept that the
> >> correct semantics is to catch any region that overlaps with the heap
> >> (before only subregions of the heap were subject to the execheap
> >> check), this corner case doesn't seem to be handled correctly by the
> >> new check (and the same bug seems to have been in fs/proc/task_mmu.c
> >> before commit 11250fd12eb8 ("mm: factor out VMA stack and heap
> >> checks")).
>
> Yes, the heap check exists for a long time,
>
>                           [start_brk        brk]
> case1:                     [vm_start,vm_end]
> case2:             [vm_start,vm_end]
> case3:                               [vm_start,vm_end]
> case4:         [vm_start,                      vm_end]
>
> case5:   [vm_start,vm_end]
> case6:                                          [vm_start,vm_end]
>
> old check:
> vma->vm_start >= vma->vm_mm->start_brk && vma->vm_end <= vma->vm_mm->brk
>
> Only include case1， vma range must be within heap
>
> new check:
> vma->vm_start <= vma->vm_mm->brk && vma->vm_end >= vma->vm_mm->start_brk
>
> Include case1~case6, but case5(vm_end=start_brk) and case6(vm_start=brk)
> are the corner cases, gcl issue matchs the case5.
>
> >>
> >> I didn't analyze the wine case ([2]), but it may be the same situation.
> >>
> >> Unless I'm mistaken, those <= & >= in should in fact be just < & >.
>
> I support this change.
>   vma->vm_start < vma->vm_mm->brk && vma->vm_end > vma->vm_mm->start_brk
>
> >> And the expression in vma_is_initial_stack() is also suspicious (but
> >> I'm not going to make any assumption on what is the intended semantics
> >> there...)
> >>
> >> [1] https://bugzilla.redhat.com/show_bug.cgi?id=2252391
> >> [2] https://bugzilla.redhat.com/show_bug.cgi?id=2247299
>
> Could you quickly verify after the above change?

Yes, changing the operators as suggested fixes the gcl case.

BTW, the vma_is_*() helpers introduced by your patch also have wrong
indentation - the "return" lines are indented by 7 spaces instead of a
tab. If you are going to submit a fixing patch you might want to fix
that, too.

-- 
Ondrej Mosnacek
Senior Software Engineer, Linux Security - SELinux kernel
Red Hat, Inc.

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Kefeng Wang]

On 2023/12/7 16:37, Ondrej Mosnacek wrote:
> On Thu, Dec 7, 2023 at 5:50 AM Kefeng Wang <wangkefeng.wang@huawei.com> wrote:
>>
>>
>>
>> On 2023/12/7 4:47, Paul Moore wrote:
>>> On Wed, Dec 6, 2023 at 9:22 AM Ondrej Mosnacek <omosnace@redhat.com> wrote:
>>>>
>>>> (Restoring part of the Cc list to include more relevant lists &
>>>> people... If you are lost, the original email is here:
>>>> https://lore.kernel.org/selinux/20230728050043.59880-4-wangkefeng.wang@huawei.com/)
>>>>
>>>> On Tue, Aug 1, 2023 at 1:08 AM Paul Moore <paul@paul-moore.com> wrote:
>>>>> On Mon, Jul 31, 2023 at 4:02 PM Stephen Smalley
>>>>> <stephen.smalley.work@gmail.com> wrote:
>>>>>> On Mon, Jul 31, 2023 at 12:19 PM Paul Moore <paul@paul-moore.com> wrote:
>>>>>>>
>>>>>>> On Mon, Jul 31, 2023 at 10:26 AM Stephen Smalley
>>>>>>> <stephen.smalley.work@gmail.com> wrote:
>>>>>>>> I believe this patch yields a semantic change in the SELinux execheap
>>>>>>>> permission check. That said, I think the change is for the better.
>>>>>>>
>>>>>>> Agreed.  I'm also in favor of using a helper which is maintained by
>>>>>>> the VM folks over open coded logic in the SELinux code.
>>>>>>
>>>>>> Yes, only caveat is in theory it could trigger new execheap denials
>>>>>> under existing policies.
>>>>>> Trying to construct an example based on the
>>>>>> selinux-testsuite/tests/mmap/mprot_heap.c example but coming up empty
>>>>>> so far on something that both works and yields different results
>>>>>> before and after this patch.
>>>>>
>>>>> My gut feeling is that this will not be an issue, but I could very
>>>>> well be wrong.  If it becomes a significant issue we can revert the
>>>>> SELinux portion of the patch.
>>>>>
>>>>> Of course, if you have any luck demonstrating this with reasonable
>>>>> code, that would be good input too.
>>>>
>>>> So, it turns out this does affect actual code. Thus far, we know about
>>>> gcl [1] and wine [2]. The gcl case is easy to reproduce (just install
>>>> gcl on Fedora and run gcl without arguments), so I was able to dig a
>>>> bit deeper.
>>>>
>>>> gcl has the following relevant memory mappings as captured by gdb:
>>>> Start Addr           End Addr       Size     Offset  Perms  objfile
>>>>     0x413000           0xf75000   0xb62000   0x3fa000  rw-p
>>>> /usr/lib/gcl-2.6.14/unixport/saved_ansi_gcl
>>>>     0xf75000           0xf79000     0x4000        0x0  rwxp   [heap]
>>>>
>>>> It tries to call mprotect(0x883000, 7282688,
>>>> PROT_READ|PROT_WRITE|PROT_EXEC), i.e. it tries to make the region
>>>> 0x883000 - 0xf75000 executable. Before this patch it was allowed,
>>>> whereas now it triggers an execheap SELinux denial. But this seems
>>>> wrong - the affected region is merely adjacent to the [heap] region,
>>>> it does not actually overlap with it. So even if we accept that the
>>>> correct semantics is to catch any region that overlaps with the heap
>>>> (before only subregions of the heap were subject to the execheap
>>>> check), this corner case doesn't seem to be handled correctly by the
>>>> new check (and the same bug seems to have been in fs/proc/task_mmu.c
>>>> before commit 11250fd12eb8 ("mm: factor out VMA stack and heap
>>>> checks")).
>>
>> Yes, the heap check exists for a long time,
>>
>>                            [start_brk        brk]
>> case1:                     [vm_start,vm_end]
>> case2:             [vm_start,vm_end]
>> case3:                               [vm_start,vm_end]
>> case4:         [vm_start,                      vm_end]
>>
>> case5:   [vm_start,vm_end]
>> case6:                                          [vm_start,vm_end]
>>
>> old check:
>> vma->vm_start >= vma->vm_mm->start_brk && vma->vm_end <= vma->vm_mm->brk
>>
>> Only include case1， vma range must be within heap
>>
>> new check:
>> vma->vm_start <= vma->vm_mm->brk && vma->vm_end >= vma->vm_mm->start_brk
>>
>> Include case1~case6, but case5(vm_end=start_brk) and case6(vm_start=brk)
>> are the corner cases, gcl issue matchs the case5.
>>
>>>>
>>>> I didn't analyze the wine case ([2]), but it may be the same situation.
>>>>
>>>> Unless I'm mistaken, those <= & >= in should in fact be just < & >.
>>
>> I support this change.
>>    vma->vm_start < vma->vm_mm->brk && vma->vm_end > vma->vm_mm->start_brk
>>
>>>> And the expression in vma_is_initial_stack() is also suspicious (but
>>>> I'm not going to make any assumption on what is the intended semantics
>>>> there...)
>>>>
>>>> [1] https://bugzilla.redhat.com/show_bug.cgi?id=2252391
>>>> [2] https://bugzilla.redhat.com/show_bug.cgi?id=2247299
>>
>> Could you quickly verify after the above change?
> 
> Yes, changing the operators as suggested fixes the gcl case.

Thanks for your confirm，win[2] fixed too, right?
> 
> BTW, the vma_is_*() helpers introduced by your patch also have wrong
> indentation - the "return" lines are indented by 7 spaces instead of a
> tab. If you are going to submit a fixing patch you might want to fix
> that, too.

Sure. will fix.

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Liam R. Howlett]

* Marc Reisner <reisner.marc@gmail.com> [240808 11:03]:
> On Thu, Aug 08, 2024 at 09:12:59PM +0800, Kefeng Wang wrote:
> >
> > OK，revert patch is sent, but I am also curious about it.
> >
> > https://lore.kernel.org/all/20240808130909.1027860-1-wangkefeng.wang@huawei.com/
> 
> I am also curious. It seems like the "real" fix would be in mmap - my
> understanding is that it should not intersect with heap, even when heap
> is empty (start_brk == brk).
> 
> It looks like start_brk is fixed in place when the ELF is
> loaded in fs/binfmt_elf.c:load_elf_binary (line 1288).
> 
>         if ((current->flags & PF_RANDOMIZE) && (snapshot_randomize_va_space > 1)) {
>                 /*
>                  * For architectures with ELF randomization, when executing
>                  * a loader directly (i.e. no interpreter listed in ELF
>                  * headers), move the brk area out of the mmap region
>                  * (since it grows up, and may collide early with the stack
>                  * growing down), and into the unused ELF_ET_DYN_BASE region.
>                  */
>                 if (IS_ENABLED(CONFIG_ARCH_HAS_ELF_RANDOMIZE) &&
>                     elf_ex->e_type == ET_DYN && !interpreter) {
>                         mm->brk = mm->start_brk = ELF_ET_DYN_BASE;
>                 } else {
>                         /* Otherwise leave a gap between .bss and brk. */
>                         mm->brk = mm->start_brk = mm->brk + PAGE_SIZE;
>                 }
> 
>                 mm->brk = mm->start_brk = arch_randomize_brk(mm);
> #ifdef compat_brk_randomized
>                 current->brk_randomized = 1;
> #endif
>         }


Have a look at the mmapstress 3 test in ltp [1].  The tests pokes holes
and mmaps into those holes throughout the brk range.

[1]. https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/mem/mmapstress/mmapstress03.c

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Marc Reisner]

On Thu, Aug 08, 2024 at 02:00:09PM -0400, Liam R. Howlett wrote:
> Have a look at the mmapstress 3 test in ltp [1].  The tests pokes holes
> and mmaps into those holes throughout the brk range.
> 
> [1]. https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/mem/mmapstress/mmapstress03.c

In investigating this further, with additional reproducers, I believe
that the whole bug is in vma_is_initial_heap(). Here is what I have
tested so far:

1. Use only sbrk to allocate heap memory - sbrk(0) returns the start_brk
before calling sbrk(increment). Afterwards, sbrk(0) returns start_brk +
increment.
2. Use sbrk(0) to obtain start_brk, then request 512 MB of address space
from mmap starting at start_brk. mmap allocates 512 MB of address space
starting at start_brk and ending at start_brk + 0x20000000. sbrk(0)
still returns start_brk. However, /proc/PID/maps flags the mmapped
address space with "[heap]"
3. Use sbrk(0) to obtain start_brk, then request 512 MB of address space
from mmap starting at start_brk + _SC_PAGESIZE. mmap allocates 512 MB of
address space starting at start_brk + _SC_PAGESIZE and ending at
start_brk + _SC_PAGESIZE + 0x20000000. sbrk(0) still returns start_brk,
and /proc/PID/maps does NOT flag the mmapped address space with
"[heap]".

I believe that the entire bug may reside in vma_is_initial_heap because
/proc/PID/maps also uses vma_is_initial_heap to flag entries with
"[heap]" [1]. Also, sbrk(0) is not actually getting updated after a call
to mmap, so mmap is not actually allocating heap memory.

What do you all think about this patch? If it doesn't have any obvious
flaws I can submit it (along with a revert for the revert).

diff --git a/include/linux/mm.h b/include/linux/mm.h
index c4b238a20b76..1dd588833af8 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -918,7 +918,8 @@ static inline bool vma_is_anonymous(struct vm_area_struct *vma)
  */
 static inline bool vma_is_initial_heap(const struct vm_area_struct *vma)
 {
-       return vma->vm_start < vma->vm_mm->brk &&
+       return vma->vm_mm->brk != vma->vm_mm->start_brk &&
+               vma->vm_start < vma->vm_mm->brk &&
                vma->vm_end > vma->vm_mm->start_brk;
 }

--


[1]. https://github.com/torvalds/linux/blob/6a0e38264012809afa24113ee2162dc07f4ed22b/fs/proc/task_mmu.c#L287

Re: [PATCH v3 3/4] selinux: use vma_is_initial_stack() and vma_is_initial_heap()

[Paul Moore]

On Thu, Aug 8, 2024 at 3:35 PM Marc Reisner <reisner.marc@gmail.com> wrote:
> On Thu, Aug 08, 2024 at 02:00:09PM -0400, Liam R. Howlett wrote:
> > Have a look at the mmapstress 3 test in ltp [1].  The tests pokes holes
> > and mmaps into those holes throughout the brk range.
> >
> > [1]. https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/mem/mmapstress/mmapstress03.c
>
> In investigating this further, with additional reproducers, I believe
> that the whole bug is in vma_is_initial_heap().

That's my feeling at this point too.  Unfortunately, there are a few
callers other than SELinux so I don't want to change the helper
function without an explicit ACK from the mm folks and I think now
that we understand the problem we want to get this fixed ASAP in
Linus' tree (and get it marked for -stable).

I just posted a patch that reverts just our use of
vma_is_initial_heap() in favor of our old logic and adds a few lines
of comments about the problem with vma_is_initial_heap().  I'm okay
with moving back to vma_is_initial_heap() when it's fixed, but I'd
prefer it to be fixed for a while before we transition back to it.
We've gotten burned twice now with vma_is_initial_heap() so I'm going
to be a little extra cautious here.

https://lore.kernel.org/selinux/20240808203353.202352-2-paul@paul-moore.com

> What do you all think about this patch? If it doesn't have any obvious
> flaws I can submit it (along with a revert for the revert).

I'll leave the mm folks to weigh in on the fix to
vma_is_initial_heap(), but as I said above, please don't submit a
patch to SELinux right now, I want the fixed version of
vma_is_initial_heap() to "soak" for a bit before we go back to it.

-- 
paul-moore.com