diff mbox series

mm: fix BUG_ON() in vmf_insert_pfn_pud() from VM_MIXEDMAP removal

Message ID 153565957352.35524.1005746906902065126.stgit@djiang5-desk3.ch.intel.com (mailing list archive)
State New, archived
Headers show
Series mm: fix BUG_ON() in vmf_insert_pfn_pud() from VM_MIXEDMAP removal | expand

Commit Message

Dave Jiang Aug. 30, 2018, 8:06 p.m. UTC
It looks like I missed the PUD path when doing VM_MIXEDMAP removal.
This can be triggered by:
1. Boot with memmap=4G!8G
2. build ndctl with destructive flag on
3. make TESTS=device-dax check

[  +0.000675] kernel BUG at mm/huge_memory.c:824!

Applying the same change that was applied to vmf_insert_pfn_pmd() in the
original patch.

Fixes: e1fb4a08649 ("dax: remove VM_MIXEDMAP for fsdax and device dax")

Reported-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
---
 mm/huge_memory.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Jeff Moyer Aug. 30, 2018, 8:12 p.m. UTC | #1
Dave Jiang <dave.jiang@intel.com> writes:

> It looks like I missed the PUD path when doing VM_MIXEDMAP removal.
> This can be triggered by:
> 1. Boot with memmap=4G!8G
> 2. build ndctl with destructive flag on
> 3. make TESTS=device-dax check
>
> [  +0.000675] kernel BUG at mm/huge_memory.c:824!
>
> Applying the same change that was applied to vmf_insert_pfn_pmd() in the
> original patch.
>
> Fixes: e1fb4a08649 ("dax: remove VM_MIXEDMAP for fsdax and device dax")
>
> Reported-by: Vishal Verma <vishal.l.verma@intel.com>
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>

Acked-by: Jeff Moyer <jmoyer@redhat.com>


> ---
>  mm/huge_memory.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index c3bc7e9c9a2a..533f9b00147d 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -821,11 +821,11 @@ vm_fault_t vmf_insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
>  	 * but we need to be consistent with PTEs and architectures that
>  	 * can't support a 'special' bit.
>  	 */
> -	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)));
> +	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) &&
> +			!pfn_t_devmap(pfn));
>  	BUG_ON((vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) ==
>  						(VM_PFNMAP|VM_MIXEDMAP));
>  	BUG_ON((vma->vm_flags & VM_PFNMAP) && is_cow_mapping(vma->vm_flags));
> -	BUG_ON(!pfn_t_devmap(pfn));
>  
>  	if (addr < vma->vm_start || addr >= vma->vm_end)
>  		return VM_FAULT_SIGBUS;
>
> _______________________________________________
> Linux-nvdimm mailing list
> Linux-nvdimm@lists.01.org
> https://lists.01.org/mailman/listinfo/linux-nvdimm
Verma, Vishal L Aug. 30, 2018, 8:28 p.m. UTC | #2
On Thu, 2018-08-30 at 13:06 -0700, Dave Jiang wrote:
> It looks like I missed the PUD path when doing VM_MIXEDMAP removal.
> This can be triggered by:
> 1. Boot with memmap=4G!8G
> 2. build ndctl with destructive flag on
> 3. make TESTS=device-dax check
> 
> [  +0.000675] kernel BUG at mm/huge_memory.c:824!
> 
> Applying the same change that was applied to vmf_insert_pfn_pmd() in
> the
> original patch.
> 
> Fixes: e1fb4a08649 ("dax: remove VM_MIXEDMAP for fsdax and device
> dax")
> 
> Reported-by: Vishal Verma <vishal.l.verma@intel.com>
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>
> ---
>  mm/huge_memory.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

This fixes the unit test failure, feel free to add:
Tested-by: Vishal Verma <vishal.l.verma@intel.com>

> 
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index c3bc7e9c9a2a..533f9b00147d 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -821,11 +821,11 @@ vm_fault_t vmf_insert_pfn_pud(struct
> vm_area_struct *vma, unsigned long addr,
>  	 * but we need to be consistent with PTEs and architectures
> that
>  	 * can't support a 'special' bit.
>  	 */
> -	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)));
> +	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) &&
> +			!pfn_t_devmap(pfn));
>  	BUG_ON((vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) ==
>  						(VM_PFNMAP|VM_MIXEDM
> AP));
>  	BUG_ON((vma->vm_flags & VM_PFNMAP) && is_cow_mapping(vma-
> >vm_flags));
> -	BUG_ON(!pfn_t_devmap(pfn));
>  
>  	if (addr < vma->vm_start || addr >= vma->vm_end)
>  		return VM_FAULT_SIGBUS;
>
Jan Kara Aug. 31, 2018, 10:22 a.m. UTC | #3
On Thu 30-08-18 13:06:13, Dave Jiang wrote:
> It looks like I missed the PUD path when doing VM_MIXEDMAP removal.
> This can be triggered by:
> 1. Boot with memmap=4G!8G
> 2. build ndctl with destructive flag on
> 3. make TESTS=device-dax check
> 
> [  +0.000675] kernel BUG at mm/huge_memory.c:824!
> 
> Applying the same change that was applied to vmf_insert_pfn_pmd() in the
> original patch.
> 
> Fixes: e1fb4a08649 ("dax: remove VM_MIXEDMAP for fsdax and device dax")
> 
> Reported-by: Vishal Verma <vishal.l.verma@intel.com>
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>

Looks good. You can add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
>  mm/huge_memory.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index c3bc7e9c9a2a..533f9b00147d 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -821,11 +821,11 @@ vm_fault_t vmf_insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
>  	 * but we need to be consistent with PTEs and architectures that
>  	 * can't support a 'special' bit.
>  	 */
> -	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)));
> +	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) &&
> +			!pfn_t_devmap(pfn));
>  	BUG_ON((vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) ==
>  						(VM_PFNMAP|VM_MIXEDMAP));
>  	BUG_ON((vma->vm_flags & VM_PFNMAP) && is_cow_mapping(vma->vm_flags));
> -	BUG_ON(!pfn_t_devmap(pfn));
>  
>  	if (addr < vma->vm_start || addr >= vma->vm_end)
>  		return VM_FAULT_SIGBUS;
> 
>
diff mbox series

Patch

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index c3bc7e9c9a2a..533f9b00147d 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -821,11 +821,11 @@  vm_fault_t vmf_insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
 	 * but we need to be consistent with PTEs and architectures that
 	 * can't support a 'special' bit.
 	 */
-	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)));
+	BUG_ON(!(vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) &&
+			!pfn_t_devmap(pfn));
 	BUG_ON((vma->vm_flags & (VM_PFNMAP|VM_MIXEDMAP)) ==
 						(VM_PFNMAP|VM_MIXEDMAP));
 	BUG_ON((vma->vm_flags & VM_PFNMAP) && is_cow_mapping(vma->vm_flags));
-	BUG_ON(!pfn_t_devmap(pfn));
 
 	if (addr < vma->vm_start || addr >= vma->vm_end)
 		return VM_FAULT_SIGBUS;