[PATCH] x86/hugetlb: Use 32/64 mmap bases according to syscall

[PATCH] x86/hugetlb: Use 32/64 mmap bases according to syscall

[Dmitry Safonov]

Commit:
  1b028f784e8c ("x86/mm: Introduce mmap_compat_base() for 32-bit
mmap()")

introduced two mmap() bases for 32-bit syscalls and for 64-bit syscalls.
After that commit mm->mmap_base has address to base allocations for
64-bit syscalls, while mm->mmap_compat_base - for 32-bit syscalls.
mmap() code was changed accordingly, but hugetlb code was not changed,
which introduced bogus behavior: 32-bit application which mmaps
file on hugetlbfs uses mm->mmap_base and thou tries to allocate
space with 64-bit mmap() base.
Changed x86 hugetlbfs code to use two bases according to calling
syscall, which also will fix any problems with 32-bit syscalls
in 64-bit ELF and vice-versa.

Fixes: commit 1b028f784e8c ("x86/mm: Introduce mmap_compat_base() for
32-bit mmap()").
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Cc: 0x7f454c46@gmail.com
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: linux-mm@kvack.org
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: x86@kernel.org
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
---
Note: I've tested it on a simple hand-written test, will reply when
got libhugetlbfs tests running in my environment.

 arch/x86/include/asm/elf.h   |  1 +
 arch/x86/kernel/sys_x86_64.c | 12 ------------
 arch/x86/mm/hugetlbpage.c    |  9 ++++++---
 arch/x86/mm/mmap.c           | 13 +++++++++++++
 4 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/arch/x86/include/asm/elf.h b/arch/x86/include/asm/elf.h
index ac5be5ba8527..d4d3ed456cb7 100644
--- a/arch/x86/include/asm/elf.h
+++ b/arch/x86/include/asm/elf.h
@@ -305,6 +305,7 @@ static inline int mmap_is_ia32(void)
 
 extern unsigned long tasksize_32bit(void);
 extern unsigned long tasksize_64bit(void);
+extern unsigned long get_mmap_base(int is_legacy);
 
 #ifdef CONFIG_X86_32
 
diff --git a/arch/x86/kernel/sys_x86_64.c b/arch/x86/kernel/sys_x86_64.c
index 63e89dfc808a..207b8f2582c7 100644
--- a/arch/x86/kernel/sys_x86_64.c
+++ b/arch/x86/kernel/sys_x86_64.c
@@ -100,18 +100,6 @@ SYSCALL_DEFINE6(mmap, unsigned long, addr, unsigned long, len,
 	return error;
 }
 
-static unsigned long get_mmap_base(int is_legacy)
-{
-	struct mm_struct *mm = current->mm;
-
-#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
-	if (in_compat_syscall())
-		return is_legacy ? mm->mmap_compat_legacy_base
-				 : mm->mmap_compat_base;
-#endif
-	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
-}
-
 static void find_start_end(unsigned long flags, unsigned long *begin,
 			   unsigned long *end)
 {
diff --git a/arch/x86/mm/hugetlbpage.c b/arch/x86/mm/hugetlbpage.c
index c5066a260803..a50f4600a281 100644
--- a/arch/x86/mm/hugetlbpage.c
+++ b/arch/x86/mm/hugetlbpage.c
@@ -16,6 +16,8 @@
 #include <asm/tlb.h>
 #include <asm/tlbflush.h>
 #include <asm/pgalloc.h>
+#include <asm/elf.h>
+#include <asm/compat.h>
 
 #if 0	/* This is just for testing */
 struct page *
@@ -82,8 +84,9 @@ static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *file,
 
 	info.flags = 0;
 	info.length = len;
-	info.low_limit = current->mm->mmap_legacy_base;
-	info.high_limit = TASK_SIZE;
+	info.low_limit = get_mmap_base(1);
+	info.high_limit = in_compat_syscall() ?
+		tasksize_32bit() : tasksize_64bit();
 	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
 	info.align_offset = 0;
 	return vm_unmapped_area(&info);
@@ -100,7 +103,7 @@ static unsigned long hugetlb_get_unmapped_area_topdown(struct file *file,
 	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
 	info.length = len;
 	info.low_limit = PAGE_SIZE;
-	info.high_limit = current->mm->mmap_base;
+	info.high_limit = get_mmap_base(0);
 	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
 	info.align_offset = 0;
 	addr = vm_unmapped_area(&info);
diff --git a/arch/x86/mm/mmap.c b/arch/x86/mm/mmap.c
index 529ab79800af..0fbb5a71b826 100644
--- a/arch/x86/mm/mmap.c
+++ b/arch/x86/mm/mmap.c
@@ -31,6 +31,7 @@
 #include <linux/sched/signal.h>
 #include <linux/sched/mm.h>
 #include <asm/elf.h>
+#include <asm/compat.h>
 
 struct va_alignment __read_mostly va_align = {
 	.flags = -1,
@@ -153,6 +154,18 @@ void arch_pick_mmap_layout(struct mm_struct *mm)
 #endif
 }
 
+unsigned long get_mmap_base(int is_legacy)
+{
+	struct mm_struct *mm = current->mm;
+
+#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
+	if (in_compat_syscall())
+		return is_legacy ? mm->mmap_compat_legacy_base
+				 : mm->mmap_compat_base;
+#endif
+	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
+}
+
 const char *arch_vma_name(struct vm_area_struct *vma)
 {
 	if (vma->vm_flags & VM_MPX)
-- 
2.11.1

Re: [PATCH] x86/hugetlb: Use 32/64 mmap bases according to syscall

[Dmitry Safonov]

On 03/14/2017 02:41 PM, Dmitry Safonov wrote:
> Commit:
>   1b028f784e8c ("x86/mm: Introduce mmap_compat_base() for 32-bit
> mmap()")
>
> introduced two mmap() bases for 32-bit syscalls and for 64-bit syscalls.
> After that commit mm->mmap_base has address to base allocations for
> 64-bit syscalls, while mm->mmap_compat_base - for 32-bit syscalls.
> mmap() code was changed accordingly, but hugetlb code was not changed,
> which introduced bogus behavior: 32-bit application which mmaps
> file on hugetlbfs uses mm->mmap_base and thou tries to allocate
> space with 64-bit mmap() base.
> Changed x86 hugetlbfs code to use two bases according to calling
> syscall, which also will fix any problems with 32-bit syscalls
> in 64-bit ELF and vice-versa.
>
> Fixes: commit 1b028f784e8c ("x86/mm: Introduce mmap_compat_base() for
> 32-bit mmap()").
> Reported-by: kernel test robot <xiaolong.ye@intel.com>
> Cc: 0x7f454c46@gmail.com
> Cc: Cyrill Gorcunov <gorcunov@openvz.org>
> Cc: Borislav Petkov <bp@suse.de>
> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
> Cc: H. Peter Anvin <hpa@zytor.com>
> Cc: linux-mm@kvack.org
> Cc: Andy Lutomirski <luto@kernel.org>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: x86@kernel.org
> Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
> ---
> Note: I've tested it on a simple hand-written test, will reply when
> got libhugetlbfs tests running in my environment.

Can confirm: I've tested this on clean fedora-25 with libhugetlbfs
tests, it's fixed.
Sorry for the breaking.

>
>  arch/x86/include/asm/elf.h   |  1 +
>  arch/x86/kernel/sys_x86_64.c | 12 ------------
>  arch/x86/mm/hugetlbpage.c    |  9 ++++++---
>  arch/x86/mm/mmap.c           | 13 +++++++++++++
>  4 files changed, 20 insertions(+), 15 deletions(-)
>
> diff --git a/arch/x86/include/asm/elf.h b/arch/x86/include/asm/elf.h
> index ac5be5ba8527..d4d3ed456cb7 100644
> --- a/arch/x86/include/asm/elf.h
> +++ b/arch/x86/include/asm/elf.h
> @@ -305,6 +305,7 @@ static inline int mmap_is_ia32(void)
>
>  extern unsigned long tasksize_32bit(void);
>  extern unsigned long tasksize_64bit(void);
> +extern unsigned long get_mmap_base(int is_legacy);
>
>  #ifdef CONFIG_X86_32
>
> diff --git a/arch/x86/kernel/sys_x86_64.c b/arch/x86/kernel/sys_x86_64.c
> index 63e89dfc808a..207b8f2582c7 100644
> --- a/arch/x86/kernel/sys_x86_64.c
> +++ b/arch/x86/kernel/sys_x86_64.c
> @@ -100,18 +100,6 @@ SYSCALL_DEFINE6(mmap, unsigned long, addr, unsigned long, len,
>  	return error;
>  }
>
> -static unsigned long get_mmap_base(int is_legacy)
> -{
> -	struct mm_struct *mm = current->mm;
> -
> -#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
> -	if (in_compat_syscall())
> -		return is_legacy ? mm->mmap_compat_legacy_base
> -				 : mm->mmap_compat_base;
> -#endif
> -	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
> -}
> -
>  static void find_start_end(unsigned long flags, unsigned long *begin,
>  			   unsigned long *end)
>  {
> diff --git a/arch/x86/mm/hugetlbpage.c b/arch/x86/mm/hugetlbpage.c
> index c5066a260803..a50f4600a281 100644
> --- a/arch/x86/mm/hugetlbpage.c
> +++ b/arch/x86/mm/hugetlbpage.c
> @@ -16,6 +16,8 @@
>  #include <asm/tlb.h>
>  #include <asm/tlbflush.h>
>  #include <asm/pgalloc.h>
> +#include <asm/elf.h>
> +#include <asm/compat.h>
>
>  #if 0	/* This is just for testing */
>  struct page *
> @@ -82,8 +84,9 @@ static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *file,
>
>  	info.flags = 0;
>  	info.length = len;
> -	info.low_limit = current->mm->mmap_legacy_base;
> -	info.high_limit = TASK_SIZE;
> +	info.low_limit = get_mmap_base(1);
> +	info.high_limit = in_compat_syscall() ?
> +		tasksize_32bit() : tasksize_64bit();
>  	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
>  	info.align_offset = 0;
>  	return vm_unmapped_area(&info);
> @@ -100,7 +103,7 @@ static unsigned long hugetlb_get_unmapped_area_topdown(struct file *file,
>  	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
>  	info.length = len;
>  	info.low_limit = PAGE_SIZE;
> -	info.high_limit = current->mm->mmap_base;
> +	info.high_limit = get_mmap_base(0);
>  	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
>  	info.align_offset = 0;
>  	addr = vm_unmapped_area(&info);
> diff --git a/arch/x86/mm/mmap.c b/arch/x86/mm/mmap.c
> index 529ab79800af..0fbb5a71b826 100644
> --- a/arch/x86/mm/mmap.c
> +++ b/arch/x86/mm/mmap.c
> @@ -31,6 +31,7 @@
>  #include <linux/sched/signal.h>
>  #include <linux/sched/mm.h>
>  #include <asm/elf.h>
> +#include <asm/compat.h>
>
>  struct va_alignment __read_mostly va_align = {
>  	.flags = -1,
> @@ -153,6 +154,18 @@ void arch_pick_mmap_layout(struct mm_struct *mm)
>  #endif
>  }
>
> +unsigned long get_mmap_base(int is_legacy)
> +{
> +	struct mm_struct *mm = current->mm;
> +
> +#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
> +	if (in_compat_syscall())
> +		return is_legacy ? mm->mmap_compat_legacy_base
> +				 : mm->mmap_compat_base;
> +#endif
> +	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
> +}
> +
>  const char *arch_vma_name(struct vm_area_struct *vma)
>  {
>  	if (vma->vm_flags & VM_MPX)
>


-- 
              Dmitry

[tip:x86/mm] x86/hugetlb: Adjust to the new native/compat mmap bases

[tip-bot for Dmitry Safonov]

Commit-ID:  7d5c038a03acfbedeb3ffef9da1814a47191f6f1
Gitweb:     http://git.kernel.org/tip/7d5c038a03acfbedeb3ffef9da1814a47191f6f1
Author:     Dmitry Safonov <dsafonov@virtuozzo.com>
AuthorDate: Tue, 14 Mar 2017 14:41:26 +0300
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Tue, 14 Mar 2017 16:11:47 +0100

x86/hugetlb: Adjust to the new native/compat mmap bases

Commit 1b028f784e8c introduced two mmap() bases for 32-bit syscalls and for
64-bit syscalls. The mmap() code in x86 was modified to handle the
separation, but the patch series missed to update the hugetlb code.

As a consequence a 32bit application mapping a file on hugetlbfs uses the
64-bit mmap base for address space allocation, which fails.

Adjust the hugetlb mapping code to use the proper bases depending on the
syscall invocation mode (64-bit or compat).

[ tglx: Massaged changelog ]

Fixes: commit 1b028f784e8c ("x86/mm: Introduce mmap_compat_base() for 32-bit mmap()")
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: 0x7f454c46@gmail.com
Cc: linux-mm@kvack.org
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Link: http://lkml.kernel.org/r/20170314114126.9280-1-dsafonov@virtuozzo.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

---
 arch/x86/include/asm/elf.h   |  1 +
 arch/x86/kernel/sys_x86_64.c | 12 ------------
 arch/x86/mm/hugetlbpage.c    |  9 ++++++---
 arch/x86/mm/mmap.c           | 14 ++++++++++++++
 4 files changed, 21 insertions(+), 15 deletions(-)

diff --git a/arch/x86/include/asm/elf.h b/arch/x86/include/asm/elf.h
index ac5be5b..d4d3ed4 100644
--- a/arch/x86/include/asm/elf.h
+++ b/arch/x86/include/asm/elf.h
@@ -305,6 +305,7 @@ static inline int mmap_is_ia32(void)
 
 extern unsigned long tasksize_32bit(void);
 extern unsigned long tasksize_64bit(void);
+extern unsigned long get_mmap_base(int is_legacy);
 
 #ifdef CONFIG_X86_32
 
diff --git a/arch/x86/kernel/sys_x86_64.c b/arch/x86/kernel/sys_x86_64.c
index 63e89df..207b8f2 100644
--- a/arch/x86/kernel/sys_x86_64.c
+++ b/arch/x86/kernel/sys_x86_64.c
@@ -100,18 +100,6 @@ out:
 	return error;
 }
 
-static unsigned long get_mmap_base(int is_legacy)
-{
-	struct mm_struct *mm = current->mm;
-
-#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
-	if (in_compat_syscall())
-		return is_legacy ? mm->mmap_compat_legacy_base
-				 : mm->mmap_compat_base;
-#endif
-	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
-}
-
 static void find_start_end(unsigned long flags, unsigned long *begin,
 			   unsigned long *end)
 {
diff --git a/arch/x86/mm/hugetlbpage.c b/arch/x86/mm/hugetlbpage.c
index c5066a2..a50f460 100644
--- a/arch/x86/mm/hugetlbpage.c
+++ b/arch/x86/mm/hugetlbpage.c
@@ -16,6 +16,8 @@
 #include <asm/tlb.h>
 #include <asm/tlbflush.h>
 #include <asm/pgalloc.h>
+#include <asm/elf.h>
+#include <asm/compat.h>
 
 #if 0	/* This is just for testing */
 struct page *
@@ -82,8 +84,9 @@ static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *file,
 
 	info.flags = 0;
 	info.length = len;
-	info.low_limit = current->mm->mmap_legacy_base;
-	info.high_limit = TASK_SIZE;
+	info.low_limit = get_mmap_base(1);
+	info.high_limit = in_compat_syscall() ?
+		tasksize_32bit() : tasksize_64bit();
 	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
 	info.align_offset = 0;
 	return vm_unmapped_area(&info);
@@ -100,7 +103,7 @@ static unsigned long hugetlb_get_unmapped_area_topdown(struct file *file,
 	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
 	info.length = len;
 	info.low_limit = PAGE_SIZE;
-	info.high_limit = current->mm->mmap_base;
+	info.high_limit = get_mmap_base(0);
 	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
 	info.align_offset = 0;
 	addr = vm_unmapped_area(&info);
diff --git a/arch/x86/mm/mmap.c b/arch/x86/mm/mmap.c
index 529ab79..c3ea70f 100644
--- a/arch/x86/mm/mmap.c
+++ b/arch/x86/mm/mmap.c
@@ -31,6 +31,7 @@
 #include <linux/sched/signal.h>
 #include <linux/sched/mm.h>
 #include <asm/elf.h>
+#include <asm/compat.h>
 
 struct va_alignment __read_mostly va_align = {
 	.flags = -1,
@@ -153,6 +154,19 @@ void arch_pick_mmap_layout(struct mm_struct *mm)
 #endif
 }
 
+unsigned long get_mmap_base(int is_legacy)
+{
+	struct mm_struct *mm = current->mm;
+
+#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
+	if (in_compat_syscall()) {
+		return is_legacy ? mm->mmap_compat_legacy_base
+				 : mm->mmap_compat_base;
+	}
+#endif
+	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
+}
+
 const char *arch_vma_name(struct vm_area_struct *vma)
 {
 	if (vma->vm_flags & VM_MPX)

[tip:x86/mm] x86/hugetlb: Adjust to the new native/compat mmap bases

[tip-bot for Dmitry Safonov]

Commit-ID:  e13b73dd9c8003993b171173ba803363faf74238
Gitweb:     http://git.kernel.org/tip/e13b73dd9c8003993b171173ba803363faf74238
Author:     Dmitry Safonov <dsafonov@virtuozzo.com>
AuthorDate: Tue, 14 Mar 2017 14:41:26 +0300
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Tue, 14 Mar 2017 16:29:16 +0100

x86/hugetlb: Adjust to the new native/compat mmap bases

Commit 1b028f784e8c introduced two mmap() bases for 32-bit syscalls and for
64-bit syscalls. The mmap() code in x86 was modified to handle the
separation, but the patch series missed to update the hugetlb code.

As a consequence a 32bit application mapping a file on hugetlbfs uses the
64-bit mmap base for address space allocation, which fails.

Adjust the hugetlb mapping code to use the proper bases depending on the
syscall invocation mode (64-bit or compat).

[ tglx: Massaged changelog and switched from asm/compat.h to linux/compat.h ]

Fixes: commit 1b028f784e8c ("x86/mm: Introduce mmap_compat_base() for 32-bit mmap()")
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Dmitry Safonov <dsafonov@virtuozzo.com>
Cc: 0x7f454c46@gmail.com
Cc: linux-mm@kvack.org
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Link: http://lkml.kernel.org/r/20170314114126.9280-1-dsafonov@virtuozzo.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
 arch/x86/include/asm/elf.h   |  1 +
 arch/x86/kernel/sys_x86_64.c | 12 ------------
 arch/x86/mm/hugetlbpage.c    |  9 ++++++---
 arch/x86/mm/mmap.c           | 14 ++++++++++++++
 4 files changed, 21 insertions(+), 15 deletions(-)

diff --git a/arch/x86/include/asm/elf.h b/arch/x86/include/asm/elf.h
index ac5be5b..d4d3ed4 100644
--- a/arch/x86/include/asm/elf.h
+++ b/arch/x86/include/asm/elf.h
@@ -305,6 +305,7 @@ static inline int mmap_is_ia32(void)
 
 extern unsigned long tasksize_32bit(void);
 extern unsigned long tasksize_64bit(void);
+extern unsigned long get_mmap_base(int is_legacy);
 
 #ifdef CONFIG_X86_32
 
diff --git a/arch/x86/kernel/sys_x86_64.c b/arch/x86/kernel/sys_x86_64.c
index 63e89df..207b8f2 100644
--- a/arch/x86/kernel/sys_x86_64.c
+++ b/arch/x86/kernel/sys_x86_64.c
@@ -100,18 +100,6 @@ out:
 	return error;
 }
 
-static unsigned long get_mmap_base(int is_legacy)
-{
-	struct mm_struct *mm = current->mm;
-
-#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
-	if (in_compat_syscall())
-		return is_legacy ? mm->mmap_compat_legacy_base
-				 : mm->mmap_compat_base;
-#endif
-	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
-}
-
 static void find_start_end(unsigned long flags, unsigned long *begin,
 			   unsigned long *end)
 {
diff --git a/arch/x86/mm/hugetlbpage.c b/arch/x86/mm/hugetlbpage.c
index c5066a2..302f43f 100644
--- a/arch/x86/mm/hugetlbpage.c
+++ b/arch/x86/mm/hugetlbpage.c
@@ -12,10 +12,12 @@
 #include <linux/pagemap.h>
 #include <linux/err.h>
 #include <linux/sysctl.h>
+#include <linux/compat.h>
 #include <asm/mman.h>
 #include <asm/tlb.h>
 #include <asm/tlbflush.h>
 #include <asm/pgalloc.h>
+#include <asm/elf.h>
 
 #if 0	/* This is just for testing */
 struct page *
@@ -82,8 +84,9 @@ static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *file,
 
 	info.flags = 0;
 	info.length = len;
-	info.low_limit = current->mm->mmap_legacy_base;
-	info.high_limit = TASK_SIZE;
+	info.low_limit = get_mmap_base(1);
+	info.high_limit = in_compat_syscall() ?
+		tasksize_32bit() : tasksize_64bit();
 	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
 	info.align_offset = 0;
 	return vm_unmapped_area(&info);
@@ -100,7 +103,7 @@ static unsigned long hugetlb_get_unmapped_area_topdown(struct file *file,
 	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
 	info.length = len;
 	info.low_limit = PAGE_SIZE;
-	info.high_limit = current->mm->mmap_base;
+	info.high_limit = get_mmap_base(0);
 	info.align_mask = PAGE_MASK & ~huge_page_mask(h);
 	info.align_offset = 0;
 	addr = vm_unmapped_area(&info);
diff --git a/arch/x86/mm/mmap.c b/arch/x86/mm/mmap.c
index 529ab79..19ad095 100644
--- a/arch/x86/mm/mmap.c
+++ b/arch/x86/mm/mmap.c
@@ -30,6 +30,7 @@
 #include <linux/limits.h>
 #include <linux/sched/signal.h>
 #include <linux/sched/mm.h>
+#include <linux/compat.h>
 #include <asm/elf.h>
 
 struct va_alignment __read_mostly va_align = {
@@ -153,6 +154,19 @@ void arch_pick_mmap_layout(struct mm_struct *mm)
 #endif
 }
 
+unsigned long get_mmap_base(int is_legacy)
+{
+	struct mm_struct *mm = current->mm;
+
+#ifdef CONFIG_HAVE_ARCH_COMPAT_MMAP_BASES
+	if (in_compat_syscall()) {
+		return is_legacy ? mm->mmap_compat_legacy_base
+				 : mm->mmap_compat_base;
+	}
+#endif
+	return is_legacy ? mm->mmap_legacy_base : mm->mmap_base;
+}
+
 const char *arch_vma_name(struct vm_area_struct *vma)
 {
 	if (vma->vm_flags & VM_MPX)