Re: [PATCH 1/2] mm: introduce MAP_FIXED_SAFE

From: Michael Ellerman
Date: Wed Dec 06 2017 - 00:15:46 EST


Hi Michal,

Some comments below.

Michal Hocko <mhocko@xxxxxxxxxx> writes:

> From: Michal Hocko <mhocko@xxxxxxxx>
>
> MAP_FIXED is used quite often to enforce mapping at the particular
> range. The main problem of this flag is, however, that it is inherently
> dangerous because it unmaps existing mappings covered by the requested
> range. This can cause silent memory corruptions. Some of them even with
> serious security implications. While the current semantic might be
> really desiderable in many cases there are others which would want to
> enforce the given range but rather see a failure than a silent memory
> corruption on a clashing range. Please note that there is no guarantee
> that a given range is obeyed by the mmap even when it is free - e.g.
> arch specific code is allowed to apply an alignment.

I don't think this last sentence is correct. Or maybe I don't understand
what you're referring to.

If you specifiy MAP_FIXED on a page boundary then the mapping must be
made at that address, I don't think arch code is allowed to add any
extra alignment.

> Introduce a new MAP_FIXED_SAFE flag for mmap to achieve this behavior.
> It has the same semantic as MAP_FIXED wrt. the given address request
> with a single exception that it fails with EEXIST if the requested
> address is already covered by an existing mapping. We still do rely on
> get_unmaped_area to handle all the arch specific MAP_FIXED treatment and
> check for a conflicting vma after it returns.
>
> [fail on clashing range with EEXIST as per Florian Weimer]
> [set MAP_FIXED before round_hint_to_min as per Khalid Aziz]
> Reviewed-by: Khalid Aziz <khalid.aziz@xxxxxxxxxx>
> Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
> ---
> arch/alpha/include/uapi/asm/mman.h | 2 ++
> arch/mips/include/uapi/asm/mman.h | 2 ++
> arch/parisc/include/uapi/asm/mman.h | 2 ++
> arch/powerpc/include/uapi/asm/mman.h | 1 +
> arch/sparc/include/uapi/asm/mman.h | 1 +
> arch/tile/include/uapi/asm/mman.h | 1 +
> arch/xtensa/include/uapi/asm/mman.h | 2 ++
> include/uapi/asm-generic/mman.h | 1 +
> mm/mmap.c | 11 +++++++++++
> 9 files changed, 23 insertions(+)
>
> diff --git a/arch/alpha/include/uapi/asm/mman.h b/arch/alpha/include/uapi/asm/mman.h
> index 6bf730063e3f..ef3770262925 100644
> --- a/arch/alpha/include/uapi/asm/mman.h
> +++ b/arch/alpha/include/uapi/asm/mman.h
> @@ -32,6 +32,8 @@
> #define MAP_STACK 0x80000 /* give out an address that is best suited for process/thread stacks */
> #define MAP_HUGETLB 0x100000 /* create a huge page mapping */
>
> +#define MAP_FIXED_SAFE 0x200000 /* MAP_FIXED which doesn't unmap underlying mapping */
> +

Why the new line before MAP_FIXED_SAFE? It should sit with the others.

You're using a different value to other arches here, but that's OK, and
alpha doesn't use asm-generic/mman.h or mman-common.h

> diff --git a/arch/powerpc/include/uapi/asm/mman.h b/arch/powerpc/include/uapi/asm/mman.h
> index e63bc37e33af..3ffd284e7160 100644
> --- a/arch/powerpc/include/uapi/asm/mman.h
> +++ b/arch/powerpc/include/uapi/asm/mman.h
> @@ -29,5 +29,6 @@
> #define MAP_NONBLOCK 0x10000 /* do not block on IO */
> #define MAP_STACK 0x20000 /* give out an address that is best suited for process/thread stacks */
> #define MAP_HUGETLB 0x40000 /* create a huge page mapping */
> +#define MAP_FIXED_SAFE 0x800000 /* MAP_FIXED which doesn't unmap underlying mapping */

Why did you pick 0x800000?

I don't see any reason you can't use 0x8000 on powerpc.


> diff --git a/arch/sparc/include/uapi/asm/mman.h b/arch/sparc/include/uapi/asm/mman.h
> index 715a2c927e79..0c282c09fae8 100644
> --- a/arch/sparc/include/uapi/asm/mman.h
> +++ b/arch/sparc/include/uapi/asm/mman.h
> @@ -24,6 +24,7 @@
> #define MAP_NONBLOCK 0x10000 /* do not block on IO */
> #define MAP_STACK 0x20000 /* give out an address that is best suited for process/thread stacks */
> #define MAP_HUGETLB 0x40000 /* create a huge page mapping */
> +#define MAP_FIXED_SAFE 0x80000 /* MAP_FIXED which doesn't unmap underlying mapping */

Using 0x80000 on sparc, sparc uses mman-common.h.

> diff --git a/arch/tile/include/uapi/asm/mman.h b/arch/tile/include/uapi/asm/mman.h
> index 9b7add95926b..b212f5fd5345 100644
> --- a/arch/tile/include/uapi/asm/mman.h
> +++ b/arch/tile/include/uapi/asm/mman.h
> @@ -30,6 +30,7 @@
> #define MAP_DENYWRITE 0x0800 /* ETXTBSY */
> #define MAP_EXECUTABLE 0x1000 /* mark it as an executable */
> #define MAP_HUGETLB 0x4000 /* create a huge page mapping */
> +#define MAP_FIXED_SAFE 0x8000 /* MAP_FIXED which doesn't unmap underlying mapping */

That is the next free flag, but you could also use 0x80000 on tile.

tile uses mman-common.h.

> diff --git a/arch/xtensa/include/uapi/asm/mman.h b/arch/xtensa/include/uapi/asm/mman.h
> index 2bfe590694fc..0daf199caa57 100644
> --- a/arch/xtensa/include/uapi/asm/mman.h
> +++ b/arch/xtensa/include/uapi/asm/mman.h
> @@ -56,6 +56,7 @@
> #define MAP_NONBLOCK 0x20000 /* do not block on IO */
> #define MAP_STACK 0x40000 /* give out an address that is best suited for process/thread stacks */
> #define MAP_HUGETLB 0x80000 /* create a huge page mapping */
> +#define MAP_FIXED_SAFE 0x100000 /* MAP_FIXED which doesn't unmap underlying mapping */

xtensa doesn't use asm-generic/mman.h or mman-common.h

> #ifdef CONFIG_MMAP_ALLOW_UNINITIALIZED
> # define MAP_UNINITIALIZED 0x4000000 /* For anonymous mmap, memory could be
> * uninitialized */
> @@ -63,6 +64,7 @@
> # define MAP_UNINITIALIZED 0x0 /* Don't support this flag */
> #endif
>
> +

Stray new line.

> /*
> * Flags for msync
> */
> diff --git a/include/uapi/asm-generic/mman.h b/include/uapi/asm-generic/mman.h
> index 2dffcbf705b3..56cde132a80a 100644
> --- a/include/uapi/asm-generic/mman.h
> +++ b/include/uapi/asm-generic/mman.h
> @@ -13,6 +13,7 @@
> #define MAP_NONBLOCK 0x10000 /* do not block on IO */
> #define MAP_STACK 0x20000 /* give out an address that is best suited for process/thread stacks */
> #define MAP_HUGETLB 0x40000 /* create a huge page mapping */
> +#define MAP_FIXED_SAFE 0x80000 /* MAP_FIXED which doesn't unmap underlying mapping */

So I think I proved above that all the arches that are using 0x80000 are
also using mman-common.h, and vice-versa.

So you can put this in mman-common.h can't you?

> diff --git a/mm/mmap.c b/mm/mmap.c
> index 476e810cf100..e84339842bb8 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1342,6 +1342,10 @@ unsigned long do_mmap(struct file *file, unsigned long addr,
> if (!(file && path_noexec(&file->f_path)))
> prot |= PROT_EXEC;
>
> + /* force arch specific MAP_FIXED handling in get_unmapped_area */
> + if (flags & MAP_FIXED_SAFE)
> + flags |= MAP_FIXED;
> +

The comment is misleading, because literally on the next line below we
check MAP_FIXED and change the behaviour, but not in the arch code.

> if (!(flags & MAP_FIXED))
> addr = round_hint_to_min(addr);

So it would be more accurate to say something like:

/*
* Internal to the kernel MAP_FIXED_SAFE is a superset of
* MAP_FIXED, so set MAP_FIXED in flags if MAP_FIXED_SAFE was
* set by the caller. This avoids all the arch code having to
* check for MAP_FIXED and MAP_FIXED_SAFE.
*/
if (flags & MAP_FIXED_SAFE)
flags |= MAP_FIXED;


cheers