mm/mmap/vma_merge: always check invariants

  We may still have inconsistent input parameters even if we choose not to
merge and the vma_merge() invariant checks are useful for checking this
with no production runtime cost (these are only relevant when
CONFIG_DEBUG_VM is specified).

Therefore, perform these checks regardless of whether we merge.

This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
Correctly update prev when policy is equal on mbind") in the mbind logic
was only picked up in the 6.2.y stable branch where these assertions are
performed prior to determining mergeability.

Had this remained the same in mainline this issue may have been picked up
faster, so moving forward let's always check them.

Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
---
 mm/mmap.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
  

On 4/30/23 22:19, Lorenzo Stoakes wrote:
> We may still have inconsistent input parameters even if we choose not to
> merge and the vma_merge() invariant checks are useful for checking this
> with no production runtime cost (these are only relevant when
> CONFIG_DEBUG_VM is specified).
> 
> Therefore, perform these checks regardless of whether we merge.
> 
> This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> Correctly update prev when policy is equal on mbind") in the mbind logic
> was only picked up in the 6.2.y stable branch where these assertions are
> performed prior to determining mergeability.
> 
> Had this remained the same in mainline this issue may have been picked up
> faster, so moving forward let's always check them.
> 
> Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>

Acked-by: Vlastimil Babka <vbabka@suse.cz>

> ---
>  mm/mmap.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 5522130ae606..13678edaa22c 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>  		merge_next = true;
>  	}
>  
> +	/* Verify some invariant that must be enforced by the caller. */
> +	VM_WARN_ON(prev && addr <= prev->vm_start);
> +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> +	VM_WARN_ON(addr >= end);
> +
>  	if (!merge_prev && !merge_next)
>  		return NULL; /* Not mergeable. */
>  
>  	res = vma = prev;
>  	remove = remove2 = adjust = NULL;
>  
> -	/* Verify some invariant that must be enforced by the caller. */
> -	VM_WARN_ON(prev && addr <= prev->vm_start);
> -	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> -	VM_WARN_ON(addr >= end);
> -
>  	/* Can we merge both the predecessor and the successor? */
>  	if (merge_prev && merge_next &&
>  	    is_mergeable_anon_vma(prev->anon_vma, next->anon_vma, NULL)) {
  

On 30.04.23 22:19, Lorenzo Stoakes wrote:
> We may still have inconsistent input parameters even if we choose not to
> merge and the vma_merge() invariant checks are useful for checking this
> with no production runtime cost (these are only relevant when
> CONFIG_DEBUG_VM is specified).
> 
> Therefore, perform these checks regardless of whether we merge.
> 
> This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> Correctly update prev when policy is equal on mbind") in the mbind logic
> was only picked up in the 6.2.y stable branch where these assertions are
> performed prior to determining mergeability.
> 
> Had this remained the same in mainline this issue may have been picked up
> faster, so moving forward let's always check them.
> 
> Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> ---
>   mm/mmap.c | 10 +++++-----
>   1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 5522130ae606..13678edaa22c 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>   		merge_next = true;
>   	}
>   
> +	/* Verify some invariant that must be enforced by the caller. */
> +	VM_WARN_ON(prev && addr <= prev->vm_start);
> +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> +	VM_WARN_ON(addr >= end);
> +
>   	if (!merge_prev && !merge_next)
>   		return NULL; /* Not mergeable. */
>   
>   	res = vma = prev;
>   	remove = remove2 = adjust = NULL;
>   
> -	/* Verify some invariant that must be enforced by the caller. */
> -	VM_WARN_ON(prev && addr <= prev->vm_start);
> -	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> -	VM_WARN_ON(addr >= end);
> -
>   	/* Can we merge both the predecessor and the successor? */
>   	if (merge_prev && merge_next &&
>   	    is_mergeable_anon_vma(prev->anon_vma, next->anon_vma, NULL)) {

Reviewed-by: David Hildenbrand <david@redhat.com>
  

* Lorenzo Stoakes <lstoakes@gmail.com> [230430 16:19]:
> We may still have inconsistent input parameters even if we choose not to
> merge and the vma_merge() invariant checks are useful for checking this
> with no production runtime cost (these are only relevant when
> CONFIG_DEBUG_VM is specified).
> 
> Therefore, perform these checks regardless of whether we merge.
> 
> This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> Correctly update prev when policy is equal on mbind") in the mbind logic
> was only picked up in the 6.2.y stable branch where these assertions are
> performed prior to determining mergeability.
> 
> Had this remained the same in mainline this issue may have been picked up
> faster, so moving forward let's always check them.
> 
> Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>

Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>

> ---
>  mm/mmap.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 5522130ae606..13678edaa22c 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>  		merge_next = true;
>  	}
>  
> +	/* Verify some invariant that must be enforced by the caller. */
> +	VM_WARN_ON(prev && addr <= prev->vm_start);
> +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> +	VM_WARN_ON(addr >= end);
> +
>  	if (!merge_prev && !merge_next)
>  		return NULL; /* Not mergeable. */
>  
>  	res = vma = prev;
>  	remove = remove2 = adjust = NULL;
>  
> -	/* Verify some invariant that must be enforced by the caller. */
> -	VM_WARN_ON(prev && addr <= prev->vm_start);
> -	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> -	VM_WARN_ON(addr >= end);
> -
>  	/* Can we merge both the predecessor and the successor? */
>  	if (merge_prev && merge_next &&
>  	    is_mergeable_anon_vma(prev->anon_vma, next->anon_vma, NULL)) {
> -- 
> 2.40.1
>
  

On Sun, 30 Apr 2023 21:19:17 +0100 Lorenzo Stoakes <lstoakes@gmail.com> wrote:

> We may still have inconsistent input parameters even if we choose not to
> merge and the vma_merge() invariant checks are useful for checking this
> with no production runtime cost (these are only relevant when
> CONFIG_DEBUG_VM is specified).
> 
> Therefore, perform these checks regardless of whether we merge.
> 
> This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> Correctly update prev when policy is equal on mbind") in the mbind logic
> was only picked up in the 6.2.y stable branch where these assertions are
> performed prior to determining mergeability.
> 
> Had this remained the same in mainline this issue may have been picked up
> faster, so moving forward let's always check them.
> 

I'll scoot this into 6.4-rc, given the recent problems in this area.

> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>  		merge_next = true;
>  	}
>  
> +	/* Verify some invariant that must be enforced by the caller. */
> +	VM_WARN_ON(prev && addr <= prev->vm_start);
> +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> +	VM_WARN_ON(addr >= end);

Maybe converting to VM_WARN_ON_ONCE() would be kinder.
  

Hi,

On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> We may still have inconsistent input parameters even if we choose not to
> merge and the vma_merge() invariant checks are useful for checking this
> with no production runtime cost (these are only relevant when
> CONFIG_DEBUG_VM is specified).
> 
> Therefore, perform these checks regardless of whether we merge.
> 
> This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> Correctly update prev when policy is equal on mbind") in the mbind logic
> was only picked up in the 6.2.y stable branch where these assertions are
> performed prior to determining mergeability.
> 
> Had this remained the same in mainline this issue may have been picked up
> faster, so moving forward let's always check them.
> 
> Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> ---
>  mm/mmap.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 5522130ae606..13678edaa22c 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
>  		merge_next = true;
>  	}
>  
> +	/* Verify some invariant that must be enforced by the caller. */
> +	VM_WARN_ON(prev && addr <= prev->vm_start);
> +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> +	VM_WARN_ON(addr >= end);
> +

I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.

The splat looks like:

| Syzkaller hit 'WARNING in vma_merge' bug.
|
| ------------[ cut here ]------------
| WARNING: CPU: 0 PID: 193 at mm/mmap.c:965 vma_merge+0x21c/0x1158 mm/mmap.c:965
| CPU: 0 PID: 193 Comm: syz-executor105 Not tainted 6.4.0-rc1-00001-g7d54d3135001 #1
| Hardware name: linux,dummy-virt (DT)
| pstate: 80400005 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
| pc : vma_merge+0x21c/0x1158 mm/mmap.c:965
| lr : vma_merge+0x21c/0x1158 mm/mmap.c:965
| sp : ffff800018ec7970
| x29: ffff800018ec7970 x28: 0000000020000000 x27: 0000000000000000
| x26: 0000000000000000 x25: 1ffff000031d8f42 x24: ffff000010d58000
| x23: 0000000000000000 x22: ffff000017acc9b0 x21: 0000000020ffd000
| x20: 0000000020ffb000 x19: ffff000017acc8b8 x18: 0000000000000005
| x17: 0000000000000000 x16: 0000000000000000 x15: 1fffe00002f27494
| x14: 0000000000000000 x13: 000000009a8feb3a x12: ffff700002ddc77d
| x11: 1ffff00002ddc77c x10: ffff700002ddc77c x9 : dfff800000000000
| x8 : ffff800016ee3be3 x7 : 0000000000000000 x6 : 0000000000000000
| x5 : ffff000017939b00 x4 : ffff800010c4a000 x3 : ffff800008000000
| x2 : 0000000000000000 x1 : ffff000017939b00 x0 : 0000000000000000
| Call trace:
|  vma_merge+0x21c/0x1158 mm/mmap.c:965
|  userfaultfd_register fs/userfaultfd.c:1485 [inline]
|  userfaultfd_ioctl+0x378c/0x4240 fs/userfaultfd.c:2050
|  vfs_ioctl fs/ioctl.c:51 [inline]
|  __do_sys_ioctl fs/ioctl.c:870 [inline]
|  __se_sys_ioctl fs/ioctl.c:856 [inline]
|  __arm64_sys_ioctl+0x184/0x218 fs/ioctl.c:856
|  __invoke_syscall arch/arm64/kernel/syscall.c:38 [inline]
|  invoke_syscall+0x8c/0x2d8 arch/arm64/kernel/syscall.c:52
|  el0_svc_common.constprop.0+0xf4/0x300 arch/arm64/kernel/syscall.c:142
|  do_el0_svc+0x6c/0x180 arch/arm64/kernel/syscall.c:193
|  el0_svc+0x4c/0x110 arch/arm64/kernel/entry-common.c:637
|  el0t_64_sync_handler+0xf4/0x120 arch/arm64/kernel/entry-common.c:655
|  el0t_64_sync+0x190/0x198 arch/arm64/kernel/entry.S:591
| irq event stamp: 2212
| hardirqs last  enabled at (2211): [<ffff80000805d03c>] local_daif_restore arch/arm64/include/asm/daifflags.h:75 [inline]
| hardirqs last  enabled at (2211): [<ffff80000805d03c>] el0_svc_common.constprop.0+0xac/0x300 arch/arm64/kernel/syscall.c:107
| hardirqs last disabled at (2212): [<ffff80000e67c40c>] el1_dbg+0x24/0xa0 arch/arm64/kernel/entry-common.c:405
| softirqs last  enabled at (2190): [<ffff800008021718>] softirq_handle_end kernel/softirq.c:414 [inline]
| softirqs last  enabled at (2190): [<ffff800008021718>] __do_softirq+0x8e8/0xe50 kernel/softirq.c:600
| softirqs last disabled at (2183): [<ffff80000802ad9c>] ____do_softirq+0x1c/0x30 arch/arm64/kernel/irq.c:80
| ---[ end trace 0000000000000000 ]---

I can reproduce that reliably with the below:

| #include <linux/userfaultfd.h>
| #include <sys/ioctl.h>
| #include <sys/mman.h>
| #include <sys/syscall.h>
| #include <sys/types.h>
| #include <unistd.h>
|   
| int main(int argc, char *argv[])
| {
|         int uffd;
|         void *addr;
| 
|         struct uffdio_api uffdio_api;
|         struct uffdio_register uffdio_register;
| 
|         uffd = syscall(__NR_userfaultfd, 0x801ul);
| 
|         uffdio_api.api = UFFD_API;
|         uffdio_api.features = 0;
|         ioctl(uffd, UFFDIO_API, &uffdio_api);
| 
|         addr = mmap(NULL, 0x1000000ul, PROT_READ | PROT_WRITE,
|                     MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
| 
|         uffdio_register.range.start = (unsigned long)addr + 0x10000;
|         uffdio_register.range.len = 0x2000;
|         uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING;
|         ioctl(uffd, UFFDIO_REGISTER, &uffdio_register);
| 
|         return 0;
| }

... which is cleaned up from the orginial Syzkaller reproducer:

| Syzkaller reproducer:
| # {Threaded:false Repeat:false RepeatTimes:0 Procs:1 Slowdown:1 Sandbox: SandboxArg:0 Leak:false NetInjection:false NetDevices:false NetReset:false Cgroups:false BinfmtMisc:false CloseFDs:false KCSAN:false DevlinkPCI:false NicVF:false USB:false VhciInjection:false Wifi:false IEEE802154:false Sysctl:false UseTmpDir:false HandleSegv:false Repro:false Trace:false LegacyOptions:{Collide:false Fault:false FaultCall:0 FaultNth:0}}
| r0 = userfaultfd(0x801)
| r1 = dup(r0)
| ioctl$UFFDIO_API(r1, 0xc018aa3f, &(0x7f0000000000))
| ioctl$UFFDIO_REGISTER(r1, 0xc020aa00, &(0x7f00000001c0)={{&(0x7f0000ffb000/0x2000)=nil, 0x2000}, 0x1})
| 
| 
| C reproducer:
| // autogenerated by syzkaller (https://github.com/google/syzkaller)
| 
| #define _GNU_SOURCE 
| 
| #include <endian.h>
| #include <stdint.h>
| #include <stdio.h>
| #include <stdlib.h>
| #include <string.h>
| #include <sys/syscall.h>
| #include <sys/types.h>
| #include <unistd.h>
| 
| #ifndef __NR_dup
| #define __NR_dup 23
| #endif
| #ifndef __NR_ioctl
| #define __NR_ioctl 29
| #endif
| #ifndef __NR_mmap
| #define __NR_mmap 222
| #endif
| #ifndef __NR_userfaultfd
| #define __NR_userfaultfd 282
| #endif
| 
| uint64_t r[2] = {0xffffffffffffffff, 0xffffffffffffffff};
| 
| int main(void)
| {
| 		syscall(__NR_mmap, 0x1ffff000ul, 0x1000ul, 0ul, 0x32ul, -1, 0ul);
| 	syscall(__NR_mmap, 0x20000000ul, 0x1000000ul, 7ul, 0x32ul, -1, 0ul);
| 	syscall(__NR_mmap, 0x21000000ul, 0x1000ul, 0ul, 0x32ul, -1, 0ul);
| 				intptr_t res = 0;
| 	res = syscall(__NR_userfaultfd, 0x801ul);
| 	if (res != -1)
| 		r[0] = res;
| 	res = syscall(__NR_dup, r[0]);
| 	if (res != -1)
| 		r[1] = res;
| *(uint64_t*)0x20000000 = 0xaa;
| *(uint64_t*)0x20000008 = 0;
| *(uint64_t*)0x20000010 = 0;
| 	syscall(__NR_ioctl, r[1], 0xc018aa3f, 0x20000000ul);
| *(uint64_t*)0x200001c0 = 0x20ffb000;
| *(uint64_t*)0x200001c8 = 0x2000;
| *(uint64_t*)0x200001d0 = 1;
| *(uint64_t*)0x200001d8 = 0;
| 	syscall(__NR_ioctl, r[1], 0xc020aa00, 0x200001c0ul);
| 	return 0;
| } 

Thanks,
Mark.
  

On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> Hi,
>
> On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > We may still have inconsistent input parameters even if we choose not to
> > merge and the vma_merge() invariant checks are useful for checking this
> > with no production runtime cost (these are only relevant when
> > CONFIG_DEBUG_VM is specified).
> >
> > Therefore, perform these checks regardless of whether we merge.
> >
> > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > Correctly update prev when policy is equal on mbind") in the mbind logic
> > was only picked up in the 6.2.y stable branch where these assertions are
> > performed prior to determining mergeability.
> >
> > Had this remained the same in mainline this issue may have been picked up
> > faster, so moving forward let's always check them.
> >
> > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > ---
> >  mm/mmap.c | 10 +++++-----
> >  1 file changed, 5 insertions(+), 5 deletions(-)
> >
> > diff --git a/mm/mmap.c b/mm/mmap.c
> > index 5522130ae606..13678edaa22c 100644
> > --- a/mm/mmap.c
> > +++ b/mm/mmap.c
> > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> >  		merge_next = true;
> >  	}
> >
> > +	/* Verify some invariant that must be enforced by the caller. */
> > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > +	VM_WARN_ON(addr >= end);
> > +
>
> I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
>

Thanks, from the line I suspect addr != curr->vm_start, but need to look
into the repro, at lsf/mm so a bit time lagged :)

Am looking into it!

> The splat looks like:
>
> | Syzkaller hit 'WARNING in vma_merge' bug.
> |
> | ------------[ cut here ]------------
> | WARNING: CPU: 0 PID: 193 at mm/mmap.c:965 vma_merge+0x21c/0x1158 mm/mmap.c:965
> | CPU: 0 PID: 193 Comm: syz-executor105 Not tainted 6.4.0-rc1-00001-g7d54d3135001 #1
> | Hardware name: linux,dummy-virt (DT)
> | pstate: 80400005 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> | pc : vma_merge+0x21c/0x1158 mm/mmap.c:965
> | lr : vma_merge+0x21c/0x1158 mm/mmap.c:965
> | sp : ffff800018ec7970
> | x29: ffff800018ec7970 x28: 0000000020000000 x27: 0000000000000000
> | x26: 0000000000000000 x25: 1ffff000031d8f42 x24: ffff000010d58000
> | x23: 0000000000000000 x22: ffff000017acc9b0 x21: 0000000020ffd000
> | x20: 0000000020ffb000 x19: ffff000017acc8b8 x18: 0000000000000005
> | x17: 0000000000000000 x16: 0000000000000000 x15: 1fffe00002f27494
> | x14: 0000000000000000 x13: 000000009a8feb3a x12: ffff700002ddc77d
> | x11: 1ffff00002ddc77c x10: ffff700002ddc77c x9 : dfff800000000000
> | x8 : ffff800016ee3be3 x7 : 0000000000000000 x6 : 0000000000000000
> | x5 : ffff000017939b00 x4 : ffff800010c4a000 x3 : ffff800008000000
> | x2 : 0000000000000000 x1 : ffff000017939b00 x0 : 0000000000000000
> | Call trace:
> |  vma_merge+0x21c/0x1158 mm/mmap.c:965
> |  userfaultfd_register fs/userfaultfd.c:1485 [inline]
> |  userfaultfd_ioctl+0x378c/0x4240 fs/userfaultfd.c:2050
> |  vfs_ioctl fs/ioctl.c:51 [inline]
> |  __do_sys_ioctl fs/ioctl.c:870 [inline]
> |  __se_sys_ioctl fs/ioctl.c:856 [inline]
> |  __arm64_sys_ioctl+0x184/0x218 fs/ioctl.c:856
> |  __invoke_syscall arch/arm64/kernel/syscall.c:38 [inline]
> |  invoke_syscall+0x8c/0x2d8 arch/arm64/kernel/syscall.c:52
> |  el0_svc_common.constprop.0+0xf4/0x300 arch/arm64/kernel/syscall.c:142
> |  do_el0_svc+0x6c/0x180 arch/arm64/kernel/syscall.c:193
> |  el0_svc+0x4c/0x110 arch/arm64/kernel/entry-common.c:637
> |  el0t_64_sync_handler+0xf4/0x120 arch/arm64/kernel/entry-common.c:655
> |  el0t_64_sync+0x190/0x198 arch/arm64/kernel/entry.S:591
> | irq event stamp: 2212
> | hardirqs last  enabled at (2211): [<ffff80000805d03c>] local_daif_restore arch/arm64/include/asm/daifflags.h:75 [inline]
> | hardirqs last  enabled at (2211): [<ffff80000805d03c>] el0_svc_common.constprop.0+0xac/0x300 arch/arm64/kernel/syscall.c:107
> | hardirqs last disabled at (2212): [<ffff80000e67c40c>] el1_dbg+0x24/0xa0 arch/arm64/kernel/entry-common.c:405
> | softirqs last  enabled at (2190): [<ffff800008021718>] softirq_handle_end kernel/softirq.c:414 [inline]
> | softirqs last  enabled at (2190): [<ffff800008021718>] __do_softirq+0x8e8/0xe50 kernel/softirq.c:600
> | softirqs last disabled at (2183): [<ffff80000802ad9c>] ____do_softirq+0x1c/0x30 arch/arm64/kernel/irq.c:80
> | ---[ end trace 0000000000000000 ]---
>
> I can reproduce that reliably with the below:
>
> | #include <linux/userfaultfd.h>
> | #include <sys/ioctl.h>
> | #include <sys/mman.h>
> | #include <sys/syscall.h>
> | #include <sys/types.h>
> | #include <unistd.h>
> |
> | int main(int argc, char *argv[])
> | {
> |         int uffd;
> |         void *addr;
> |
> |         struct uffdio_api uffdio_api;
> |         struct uffdio_register uffdio_register;
> |
> |         uffd = syscall(__NR_userfaultfd, 0x801ul);
> |
> |         uffdio_api.api = UFFD_API;
> |         uffdio_api.features = 0;
> |         ioctl(uffd, UFFDIO_API, &uffdio_api);
> |
> |         addr = mmap(NULL, 0x1000000ul, PROT_READ | PROT_WRITE,
> |                     MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
> |
> |         uffdio_register.range.start = (unsigned long)addr + 0x10000;
> |         uffdio_register.range.len = 0x2000;
> |         uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING;
> |         ioctl(uffd, UFFDIO_REGISTER, &uffdio_register);
> |
> |         return 0;
> | }
>
> ... which is cleaned up from the orginial Syzkaller reproducer:
>
> | Syzkaller reproducer:
> | # {Threaded:false Repeat:false RepeatTimes:0 Procs:1 Slowdown:1 Sandbox: SandboxArg:0 Leak:false NetInjection:false NetDevices:false NetReset:false Cgroups:false BinfmtMisc:false CloseFDs:false KCSAN:false DevlinkPCI:false NicVF:false USB:false VhciInjection:false Wifi:false IEEE802154:false Sysctl:false UseTmpDir:false HandleSegv:false Repro:false Trace:false LegacyOptions:{Collide:false Fault:false FaultCall:0 FaultNth:0}}
> | r0 = userfaultfd(0x801)
> | r1 = dup(r0)
> | ioctl$UFFDIO_API(r1, 0xc018aa3f, &(0x7f0000000000))
> | ioctl$UFFDIO_REGISTER(r1, 0xc020aa00, &(0x7f00000001c0)={{&(0x7f0000ffb000/0x2000)=nil, 0x2000}, 0x1})
> |
> |
> | C reproducer:
> | // autogenerated by syzkaller (https://github.com/google/syzkaller)
> |
> | #define _GNU_SOURCE
> |
> | #include <endian.h>
> | #include <stdint.h>
> | #include <stdio.h>
> | #include <stdlib.h>
> | #include <string.h>
> | #include <sys/syscall.h>
> | #include <sys/types.h>
> | #include <unistd.h>
> |
> | #ifndef __NR_dup
> | #define __NR_dup 23
> | #endif
> | #ifndef __NR_ioctl
> | #define __NR_ioctl 29
> | #endif
> | #ifndef __NR_mmap
> | #define __NR_mmap 222
> | #endif
> | #ifndef __NR_userfaultfd
> | #define __NR_userfaultfd 282
> | #endif
> |
> | uint64_t r[2] = {0xffffffffffffffff, 0xffffffffffffffff};
> |
> | int main(void)
> | {
> | 		syscall(__NR_mmap, 0x1ffff000ul, 0x1000ul, 0ul, 0x32ul, -1, 0ul);
> | 	syscall(__NR_mmap, 0x20000000ul, 0x1000000ul, 7ul, 0x32ul, -1, 0ul);
> | 	syscall(__NR_mmap, 0x21000000ul, 0x1000ul, 0ul, 0x32ul, -1, 0ul);
> | 				intptr_t res = 0;
> | 	res = syscall(__NR_userfaultfd, 0x801ul);
> | 	if (res != -1)
> | 		r[0] = res;
> | 	res = syscall(__NR_dup, r[0]);
> | 	if (res != -1)
> | 		r[1] = res;
> | *(uint64_t*)0x20000000 = 0xaa;
> | *(uint64_t*)0x20000008 = 0;
> | *(uint64_t*)0x20000010 = 0;
> | 	syscall(__NR_ioctl, r[1], 0xc018aa3f, 0x20000000ul);
> | *(uint64_t*)0x200001c0 = 0x20ffb000;
> | *(uint64_t*)0x200001c8 = 0x2000;
> | *(uint64_t*)0x200001d0 = 1;
> | *(uint64_t*)0x200001d8 = 0;
> | 	syscall(__NR_ioctl, r[1], 0xc020aa00, 0x200001c0ul);
> | 	return 0;
> | }
>
> Thanks,
> Mark.
  

On Wed, May 10, 2023 at 09:04:44AM -0700, Lorenzo Stoakes wrote:
> On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> > Hi,
> >
> > On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > > We may still have inconsistent input parameters even if we choose not to
> > > merge and the vma_merge() invariant checks are useful for checking this
> > > with no production runtime cost (these are only relevant when
> > > CONFIG_DEBUG_VM is specified).
> > >
> > > Therefore, perform these checks regardless of whether we merge.
> > >
> > > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > > Correctly update prev when policy is equal on mbind") in the mbind logic
> > > was only picked up in the 6.2.y stable branch where these assertions are
> > > performed prior to determining mergeability.
> > >
> > > Had this remained the same in mainline this issue may have been picked up
> > > faster, so moving forward let's always check them.
> > >
> > > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > > ---
> > >  mm/mmap.c | 10 +++++-----
> > >  1 file changed, 5 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/mm/mmap.c b/mm/mmap.c
> > > index 5522130ae606..13678edaa22c 100644
> > > --- a/mm/mmap.c
> > > +++ b/mm/mmap.c
> > > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > >  		merge_next = true;
> > >  	}
> > >
> > > +	/* Verify some invariant that must be enforced by the caller. */
> > > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > +	VM_WARN_ON(addr >= end);
> > > +
> >
> > I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
> >
> 
> Thanks, from the line I suspect addr != curr->vm_start, but need to look
> into the repro, at lsf/mm so a bit time lagged :)

No problem; FWIW I can confirm your theory, the reproducer is causing:

	addr > curr->vm_start

... confirmed the the following hack, log below.

| diff --git a/mm/mmap.c b/mm/mmap.c
| index 13678edaa22c..2cdebba15719 100644
| --- a/mm/mmap.c
| +++ b/mm/mmap.c
| @@ -961,9 +961,21 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
|         }
|  
|         /* Verify some invariant that must be enforced by the caller. */
| -       VM_WARN_ON(prev && addr <= prev->vm_start);
| -       VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
| -       VM_WARN_ON(addr >= end);
| +       VM_WARN(prev && addr <= prev->vm_start,
| +               "addr = 0x%016lx, prev->vm_start = 0x%016lx\n",
| +               addr, prev->vm_start);
| +
| +       VM_WARN(curr && addr != curr->vm_start,
| +               "addr = 0x%016lx, curr->vm_start = 0x%016lx\n",
| +               addr, curr->vm_start);
| +
| +       VM_WARN(curr && addr > curr->vm_end,
| +               "addr = 0x%016lx, curr->vm_end = 0x%016lx\n",
| +               addr, curr->vm_end);
| +
| +       VM_WARN(addr >= end,
| +               "addr = 0x%016lx, end = 0x%016lx\n",
| +               addr, end);
|  
|         if (!merge_prev && !merge_next)
|                 return NULL; /* Not mergeable. */

... with that applied, running the reproducer results in:

| addr = 0x0000ffff99dc2000, curr->vm_start = 0x0000ffff99db2000
| WARNING: CPU: 0 PID: 163 at mm/mmap.c:968 vma_merge+0x3d4/0x1260

... i.e. addr > curr->vm_start

Thanks,
Mark.
  

On Wed, May 10, 2023 at 05:17:49PM +0100, Mark Rutland wrote:
> On Wed, May 10, 2023 at 09:04:44AM -0700, Lorenzo Stoakes wrote:
> > On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> > > Hi,
> > >
> > > On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > > > We may still have inconsistent input parameters even if we choose not to
> > > > merge and the vma_merge() invariant checks are useful for checking this
> > > > with no production runtime cost (these are only relevant when
> > > > CONFIG_DEBUG_VM is specified).
> > > >
> > > > Therefore, perform these checks regardless of whether we merge.
> > > >
> > > > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > > > Correctly update prev when policy is equal on mbind") in the mbind logic
> > > > was only picked up in the 6.2.y stable branch where these assertions are
> > > > performed prior to determining mergeability.
> > > >
> > > > Had this remained the same in mainline this issue may have been picked up
> > > > faster, so moving forward let's always check them.
> > > >
> > > > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > > > ---
> > > >  mm/mmap.c | 10 +++++-----
> > > >  1 file changed, 5 insertions(+), 5 deletions(-)
> > > >
> > > > diff --git a/mm/mmap.c b/mm/mmap.c
> > > > index 5522130ae606..13678edaa22c 100644
> > > > --- a/mm/mmap.c
> > > > +++ b/mm/mmap.c
> > > > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > > >  		merge_next = true;
> > > >  	}
> > > >
> > > > +	/* Verify some invariant that must be enforced by the caller. */
> > > > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > > > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > > +	VM_WARN_ON(addr >= end);
> > > > +
> > >
> > > I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
> > >
> >
> > Thanks, from the line I suspect addr != curr->vm_start, but need to look
> > into the repro, at lsf/mm so a bit time lagged :)
>
> No problem; FWIW I can confirm your theory, the reproducer is causing:
>
> 	addr > curr->vm_start
>
> ... confirmed the the following hack, log below.

Awesome thanks for that! Just been firing up qemu to do this.

Cases 5-8 should really have addr == curr->vm_start, I wonder if it's
another case but curr is being set incorrectly, it should in theory not be
the case.

(See [1] for a visualisation of merge cases as a handy reference)

Of course userfaultfd might be the offender here and might be relying on no
merge case arising but passing dodgy parameters.

[1]:https://ljs.io/merge_cases.png

>
> | diff --git a/mm/mmap.c b/mm/mmap.c
> | index 13678edaa22c..2cdebba15719 100644
> | --- a/mm/mmap.c
> | +++ b/mm/mmap.c
> | @@ -961,9 +961,21 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> |         }
> |
> |         /* Verify some invariant that must be enforced by the caller. */
> | -       VM_WARN_ON(prev && addr <= prev->vm_start);
> | -       VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> | -       VM_WARN_ON(addr >= end);
> | +       VM_WARN(prev && addr <= prev->vm_start,
> | +               "addr = 0x%016lx, prev->vm_start = 0x%016lx\n",
> | +               addr, prev->vm_start);
> | +
> | +       VM_WARN(curr && addr != curr->vm_start,
> | +               "addr = 0x%016lx, curr->vm_start = 0x%016lx\n",
> | +               addr, curr->vm_start);
> | +
> | +       VM_WARN(curr && addr > curr->vm_end,
> | +               "addr = 0x%016lx, curr->vm_end = 0x%016lx\n",
> | +               addr, curr->vm_end);
> | +
> | +       VM_WARN(addr >= end,
> | +               "addr = 0x%016lx, end = 0x%016lx\n",
> | +               addr, end);
> |
> |         if (!merge_prev && !merge_next)
> |                 return NULL; /* Not mergeable. */
>
> ... with that applied, running the reproducer results in:
>
> | addr = 0x0000ffff99dc2000, curr->vm_start = 0x0000ffff99db2000
> | WARNING: CPU: 0 PID: 163 at mm/mmap.c:968 vma_merge+0x3d4/0x1260
>
> ... i.e. addr > curr->vm_start
>
> Thanks,
> Mark.
  

On Wed, May 10, 2023 at 09:26:10AM -0700, Lorenzo Stoakes wrote:
> On Wed, May 10, 2023 at 05:17:49PM +0100, Mark Rutland wrote:
> > On Wed, May 10, 2023 at 09:04:44AM -0700, Lorenzo Stoakes wrote:
> > > On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> > > > Hi,
> > > >
> > > > On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > > > > We may still have inconsistent input parameters even if we choose not to
> > > > > merge and the vma_merge() invariant checks are useful for checking this
> > > > > with no production runtime cost (these are only relevant when
> > > > > CONFIG_DEBUG_VM is specified).
> > > > >
> > > > > Therefore, perform these checks regardless of whether we merge.
> > > > >
> > > > > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > > > > Correctly update prev when policy is equal on mbind") in the mbind logic
> > > > > was only picked up in the 6.2.y stable branch where these assertions are
> > > > > performed prior to determining mergeability.
> > > > >
> > > > > Had this remained the same in mainline this issue may have been picked up
> > > > > faster, so moving forward let's always check them.
> > > > >
> > > > > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > > > > ---
> > > > >  mm/mmap.c | 10 +++++-----
> > > > >  1 file changed, 5 insertions(+), 5 deletions(-)
> > > > >
> > > > > diff --git a/mm/mmap.c b/mm/mmap.c
> > > > > index 5522130ae606..13678edaa22c 100644
> > > > > --- a/mm/mmap.c
> > > > > +++ b/mm/mmap.c
> > > > > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > > > >  		merge_next = true;
> > > > >  	}
> > > > >
> > > > > +	/* Verify some invariant that must be enforced by the caller. */
> > > > > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > > > > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > > > +	VM_WARN_ON(addr >= end);
> > > > > +
> > > >
> > > > I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
> > > >
> > >
> > > Thanks, from the line I suspect addr != curr->vm_start, but need to look
> > > into the repro, at lsf/mm so a bit time lagged :)
> >
> > No problem; FWIW I can confirm your theory, the reproducer is causing:
> >
> > 	addr > curr->vm_start
> >
> > ... confirmed the the following hack, log below.
>
> Awesome thanks for that! Just been firing up qemu to do this.
>
> Cases 5-8 should really have addr == curr->vm_start, I wonder if it's
> another case but curr is being set incorrectly, it should in theory not be
> the case.
>
> (See [1] for a visualisation of merge cases as a handy reference)
>
> Of course userfaultfd might be the offender here and might be relying on no
> merge case arising but passing dodgy parameters.
>
> [1]:https://ljs.io/merge_cases.png
>
> >
> > | diff --git a/mm/mmap.c b/mm/mmap.c
> > | index 13678edaa22c..2cdebba15719 100644
> > | --- a/mm/mmap.c
> > | +++ b/mm/mmap.c
> > | @@ -961,9 +961,21 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > |         }
> > |
> > |         /* Verify some invariant that must be enforced by the caller. */
> > | -       VM_WARN_ON(prev && addr <= prev->vm_start);
> > | -       VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > | -       VM_WARN_ON(addr >= end);
> > | +       VM_WARN(prev && addr <= prev->vm_start,
> > | +               "addr = 0x%016lx, prev->vm_start = 0x%016lx\n",
> > | +               addr, prev->vm_start);
> > | +
> > | +       VM_WARN(curr && addr != curr->vm_start,
> > | +               "addr = 0x%016lx, curr->vm_start = 0x%016lx\n",
> > | +               addr, curr->vm_start);
> > | +
> > | +       VM_WARN(curr && addr > curr->vm_end,
> > | +               "addr = 0x%016lx, curr->vm_end = 0x%016lx\n",
> > | +               addr, curr->vm_end);
> > | +
> > | +       VM_WARN(addr >= end,
> > | +               "addr = 0x%016lx, end = 0x%016lx\n",
> > | +               addr, end);
> > |
> > |         if (!merge_prev && !merge_next)
> > |                 return NULL; /* Not mergeable. */
> >
> > ... with that applied, running the reproducer results in:
> >
> > | addr = 0x0000ffff99dc2000, curr->vm_start = 0x0000ffff99db2000
> > | WARNING: CPU: 0 PID: 163 at mm/mmap.c:968 vma_merge+0x3d4/0x1260
> >
> > ... i.e. addr > curr->vm_start
> >
> > Thanks,
> > Mark.

It looks like userfaultfd_register() is indeed using vma_merge() to
determine whether potentially broken input is mergeable, e.g.:-

		if (vma->vm_start < start) {
			ret = split_vma(&vmi, vma, start, 1);
			if (ret)
				break;
		}


So it is proactively using this to determine whether the VMA needs
splitting which is... interesting :)

I need to look at this in detail, but I do suspect uffd could do this check
_before_ the vma_merge() (I'd need to double-check this though), or perhaps
vma_merge could un-warn it.

What's worrying here is that having addr > curr->vm_start but end ==
next->vm_start could result in a broken, overlapped merged.

In any case this needs some post-jet lag analysis (happy for you or Liam or
whoever to step in too if you fancy ;)

THe most proximal solution is to just drop the addr != curr->vm_start check
or at least return NULL if that check is failed, but I think it'd be better
to take a step back and examing the uffd code in detail because the
approach seems like it might have a broken edge case.

I think this check has more so brought out an issue rather than caused one,
but until I can give a post-lsf/mm, post-jet lag analysis I can't be
conclusively coherent on this.

Instinct says we should probably change uffd but might be wrong!
  

(adding Peter)

On Wed, May 10, 2023 at 09:26:10AM -0700, Lorenzo Stoakes wrote:
> On Wed, May 10, 2023 at 05:17:49PM +0100, Mark Rutland wrote:
> > On Wed, May 10, 2023 at 09:04:44AM -0700, Lorenzo Stoakes wrote:
> > > On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> > > > Hi,
> > > >
> > > > On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > > > > We may still have inconsistent input parameters even if we choose not to
> > > > > merge and the vma_merge() invariant checks are useful for checking this
> > > > > with no production runtime cost (these are only relevant when
> > > > > CONFIG_DEBUG_VM is specified).
> > > > >
> > > > > Therefore, perform these checks regardless of whether we merge.
> > > > >
> > > > > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > > > > Correctly update prev when policy is equal on mbind") in the mbind logic
> > > > > was only picked up in the 6.2.y stable branch where these assertions are
> > > > > performed prior to determining mergeability.
> > > > >
> > > > > Had this remained the same in mainline this issue may have been picked up
> > > > > faster, so moving forward let's always check them.
> > > > >
> > > > > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > > > > ---
> > > > >  mm/mmap.c | 10 +++++-----
> > > > >  1 file changed, 5 insertions(+), 5 deletions(-)
> > > > >
> > > > > diff --git a/mm/mmap.c b/mm/mmap.c
> > > > > index 5522130ae606..13678edaa22c 100644
> > > > > --- a/mm/mmap.c
> > > > > +++ b/mm/mmap.c
> > > > > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > > > >  		merge_next = true;
> > > > >  	}
> > > > >
> > > > > +	/* Verify some invariant that must be enforced by the caller. */
> > > > > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > > > > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > > > +	VM_WARN_ON(addr >= end);
> > > > > +
> > > >
> > > > I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
> > > >
> > >
> > > Thanks, from the line I suspect addr != curr->vm_start, but need to look
> > > into the repro, at lsf/mm so a bit time lagged :)
> >
> > No problem; FWIW I can confirm your theory, the reproducer is causing:
> >
> > 	addr > curr->vm_start
> >
> > ... confirmed the the following hack, log below.
> 
> Awesome thanks for that! Just been firing up qemu to do this.
> 
> Cases 5-8 should really have addr == curr->vm_start, I wonder if it's
> another case but curr is being set incorrectly, it should in theory not be
> the case.

AFAIU, it's a case of "adjust vma, but don't merge, because prev is not
compatible". Looks like uffd first attempts to merge compatible the newly
registered range with adjacent vmas relying on that there won't be no merge
when addr != curr->vm_start and only after the merge attempt it splits the
edges.

I think that moving the split in fs/userfaultfd.c:1495 (as of v6.4-rc1)
before vma_merge() will be the right fix.
 
> (See [1] for a visualisation of merge cases as a handy reference)
> 
> Of course userfaultfd might be the offender here and might be relying on no
> merge case arising but passing dodgy parameters.
> 
> [1]:https://ljs.io/merge_cases.png

You really should put it into Documentation/mm ;-)
 
> >
> > | diff --git a/mm/mmap.c b/mm/mmap.c
> > | index 13678edaa22c..2cdebba15719 100644
> > | --- a/mm/mmap.c
> > | +++ b/mm/mmap.c
> > | @@ -961,9 +961,21 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > |         }
> > |
> > |         /* Verify some invariant that must be enforced by the caller. */
> > | -       VM_WARN_ON(prev && addr <= prev->vm_start);
> > | -       VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > | -       VM_WARN_ON(addr >= end);
> > | +       VM_WARN(prev && addr <= prev->vm_start,
> > | +               "addr = 0x%016lx, prev->vm_start = 0x%016lx\n",
> > | +               addr, prev->vm_start);
> > | +
> > | +       VM_WARN(curr && addr != curr->vm_start,
> > | +               "addr = 0x%016lx, curr->vm_start = 0x%016lx\n",
> > | +               addr, curr->vm_start);
> > | +
> > | +       VM_WARN(curr && addr > curr->vm_end,
> > | +               "addr = 0x%016lx, curr->vm_end = 0x%016lx\n",
> > | +               addr, curr->vm_end);
> > | +
> > | +       VM_WARN(addr >= end,
> > | +               "addr = 0x%016lx, end = 0x%016lx\n",
> > | +               addr, end);
> > |
> > |         if (!merge_prev && !merge_next)
> > |                 return NULL; /* Not mergeable. */
> >
> > ... with that applied, running the reproducer results in:
> >
> > | addr = 0x0000ffff99dc2000, curr->vm_start = 0x0000ffff99db2000
> > | WARNING: CPU: 0 PID: 163 at mm/mmap.c:968 vma_merge+0x3d4/0x1260
> >
> > ... i.e. addr > curr->vm_start
> >
> > Thanks,
> > Mark.
>
  

On Thu, May 11, 2023 at 11:08:41AM -0700, Mike Rapoport wrote:
> (adding Peter)
>
> On Wed, May 10, 2023 at 09:26:10AM -0700, Lorenzo Stoakes wrote:
> > On Wed, May 10, 2023 at 05:17:49PM +0100, Mark Rutland wrote:
> > > On Wed, May 10, 2023 at 09:04:44AM -0700, Lorenzo Stoakes wrote:
> > > > On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> > > > > Hi,
> > > > >
> > > > > On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > > > > > We may still have inconsistent input parameters even if we choose not to
> > > > > > merge and the vma_merge() invariant checks are useful for checking this
> > > > > > with no production runtime cost (these are only relevant when
> > > > > > CONFIG_DEBUG_VM is specified).
> > > > > >
> > > > > > Therefore, perform these checks regardless of whether we merge.
> > > > > >
> > > > > > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > > > > > Correctly update prev when policy is equal on mbind") in the mbind logic
> > > > > > was only picked up in the 6.2.y stable branch where these assertions are
> > > > > > performed prior to determining mergeability.
> > > > > >
> > > > > > Had this remained the same in mainline this issue may have been picked up
> > > > > > faster, so moving forward let's always check them.
> > > > > >
> > > > > > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > > > > > ---
> > > > > >  mm/mmap.c | 10 +++++-----
> > > > > >  1 file changed, 5 insertions(+), 5 deletions(-)
> > > > > >
> > > > > > diff --git a/mm/mmap.c b/mm/mmap.c
> > > > > > index 5522130ae606..13678edaa22c 100644
> > > > > > --- a/mm/mmap.c
> > > > > > +++ b/mm/mmap.c
> > > > > > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > > > > >  		merge_next = true;
> > > > > >  	}
> > > > > >
> > > > > > +	/* Verify some invariant that must be enforced by the caller. */
> > > > > > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > > > > > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > > > > +	VM_WARN_ON(addr >= end);
> > > > > > +
> > > > >
> > > > > I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
> > > > >
> > > >
> > > > Thanks, from the line I suspect addr != curr->vm_start, but need to look
> > > > into the repro, at lsf/mm so a bit time lagged :)
> > >
> > > No problem; FWIW I can confirm your theory, the reproducer is causing:
> > >
> > > 	addr > curr->vm_start
> > >
> > > ... confirmed the the following hack, log below.
> >
> > Awesome thanks for that! Just been firing up qemu to do this.
> >
> > Cases 5-8 should really have addr == curr->vm_start, I wonder if it's
> > another case but curr is being set incorrectly, it should in theory not be
> > the case.
>
> AFAIU, it's a case of "adjust vma, but don't merge, because prev is not
> compatible". Looks like uffd first attempts to merge compatible the newly
> registered range with adjacent vmas relying on that there won't be no merge
> when addr != curr->vm_start and only after the merge attempt it splits the
> edges.
>
> I think that moving the split in fs/userfaultfd.c:1495 (as of v6.4-rc1)
> before vma_merge() will be the right fix.
>

Ack this was my strong suspicion, just want to get back to the UK and
de-lagged before I dig in properly.

> > (See [1] for a visualisation of merge cases as a handy reference)
> >
> > Of course userfaultfd might be the offender here and might be relying on no
> > merge case arising but passing dodgy parameters.
> >
> > [1]:https://ljs.io/merge_cases.png
>
> You really should put it into Documentation/mm ;-)

Well... will reply to your lsf/mm docs thread on this topic ;)

>
> > >
> > > | diff --git a/mm/mmap.c b/mm/mmap.c
> > > | index 13678edaa22c..2cdebba15719 100644
> > > | --- a/mm/mmap.c
> > > | +++ b/mm/mmap.c
> > > | @@ -961,9 +961,21 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > > |         }
> > > |
> > > |         /* Verify some invariant that must be enforced by the caller. */
> > > | -       VM_WARN_ON(prev && addr <= prev->vm_start);
> > > | -       VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > | -       VM_WARN_ON(addr >= end);
> > > | +       VM_WARN(prev && addr <= prev->vm_start,
> > > | +               "addr = 0x%016lx, prev->vm_start = 0x%016lx\n",
> > > | +               addr, prev->vm_start);
> > > | +
> > > | +       VM_WARN(curr && addr != curr->vm_start,
> > > | +               "addr = 0x%016lx, curr->vm_start = 0x%016lx\n",
> > > | +               addr, curr->vm_start);
> > > | +
> > > | +       VM_WARN(curr && addr > curr->vm_end,
> > > | +               "addr = 0x%016lx, curr->vm_end = 0x%016lx\n",
> > > | +               addr, curr->vm_end);
> > > | +
> > > | +       VM_WARN(addr >= end,
> > > | +               "addr = 0x%016lx, end = 0x%016lx\n",
> > > | +               addr, end);
> > > |
> > > |         if (!merge_prev && !merge_next)
> > > |                 return NULL; /* Not mergeable. */
> > >
> > > ... with that applied, running the reproducer results in:
> > >
> > > | addr = 0x0000ffff99dc2000, curr->vm_start = 0x0000ffff99db2000
> > > | WARNING: CPU: 0 PID: 163 at mm/mmap.c:968 vma_merge+0x3d4/0x1260
> > >
> > > ... i.e. addr > curr->vm_start
> > >
> > > Thanks,
> > > Mark.
> >
  

On Thu, May 11, 2023 at 11:08:41AM -0700, Mike Rapoport wrote:
> (adding Peter)
>
> On Wed, May 10, 2023 at 09:26:10AM -0700, Lorenzo Stoakes wrote:
> > On Wed, May 10, 2023 at 05:17:49PM +0100, Mark Rutland wrote:
> > > On Wed, May 10, 2023 at 09:04:44AM -0700, Lorenzo Stoakes wrote:
> > > > On Wed, May 10, 2023 at 03:15:51PM +0100, Mark Rutland wrote:
> > > > > Hi,
> > > > >
> > > > > On Sun, Apr 30, 2023 at 09:19:17PM +0100, Lorenzo Stoakes wrote:
> > > > > > We may still have inconsistent input parameters even if we choose not to
> > > > > > merge and the vma_merge() invariant checks are useful for checking this
> > > > > > with no production runtime cost (these are only relevant when
> > > > > > CONFIG_DEBUG_VM is specified).
> > > > > >
> > > > > > Therefore, perform these checks regardless of whether we merge.
> > > > > >
> > > > > > This is relevant, as a recent issue (addressed in commit "mm/mempolicy:
> > > > > > Correctly update prev when policy is equal on mbind") in the mbind logic
> > > > > > was only picked up in the 6.2.y stable branch where these assertions are
> > > > > > performed prior to determining mergeability.
> > > > > >
> > > > > > Had this remained the same in mainline this issue may have been picked up
> > > > > > faster, so moving forward let's always check them.
> > > > > >
> > > > > > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > > > > > ---
> > > > > >  mm/mmap.c | 10 +++++-----
> > > > > >  1 file changed, 5 insertions(+), 5 deletions(-)
> > > > > >
> > > > > > diff --git a/mm/mmap.c b/mm/mmap.c
> > > > > > index 5522130ae606..13678edaa22c 100644
> > > > > > --- a/mm/mmap.c
> > > > > > +++ b/mm/mmap.c
> > > > > > @@ -960,17 +960,17 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
> > > > > >  		merge_next = true;
> > > > > >  	}
> > > > > >
> > > > > > +	/* Verify some invariant that must be enforced by the caller. */
> > > > > > +	VM_WARN_ON(prev && addr <= prev->vm_start);
> > > > > > +	VM_WARN_ON(curr && (addr != curr->vm_start || end > curr->vm_end));
> > > > > > +	VM_WARN_ON(addr >= end);
> > > > > > +
> > > > >
> > > > > I'm seeing this fire a lot when fuzzing v6.4-rc1 on arm64 using Syzkaller.
> > > > >
> > > >
> > > > Thanks, from the line I suspect addr != curr->vm_start, but need to look
> > > > into the repro, at lsf/mm so a bit time lagged :)
> > >
> > > No problem; FWIW I can confirm your theory, the reproducer is causing:
> > >
> > > 	addr > curr->vm_start
> > >
> > > ... confirmed the the following hack, log below.
> >
> > Awesome thanks for that! Just been firing up qemu to do this.
> >
> > Cases 5-8 should really have addr == curr->vm_start, I wonder if it's
> > another case but curr is being set incorrectly, it should in theory not be
> > the case.
>
> AFAIU, it's a case of "adjust vma, but don't merge, because prev is not
> compatible". Looks like uffd first attempts to merge compatible the newly
> registered range with adjacent vmas relying on that there won't be no merge
> when addr != curr->vm_start and only after the merge attempt it splits the
> edges.
>
> I think that moving the split in fs/userfaultfd.c:1495 (as of v6.4-rc1)
> before vma_merge() will be the right fix.
>

Indeed it was, patch at
https://lore.kernel.org/all/20230514172731.134188-1-lstoakes@gmail.com/

[snip]