Skip to contentSkip to conversation

x86/vdso: small fixups for map_vdso

PATCH In review Dmitry Safonov 3 participants 6 messages Oct 27, 2016 – Oct 28, 2016

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

cover x86/vdso: small fixups for map_vdso

The first one is a fixup for arch_prctl constants uapi visability, the second is code simplification.

Dmitry Safonov (2):
x86/prctl/uapi: remove ifdef for CHECKPOINT_RESTORE x86/vdso: set vdso pointer only after success

 arch/x86/entry/vdso/vma.c         | 10 +++-------
 arch/x86/include/uapi/asm/prctl.h |  8 +++-----
 2 files changed, 6 insertions(+), 12 deletions(-)
Signature · 7 lines
2.10.1

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#1 lore raw

PATCH 1/2 x86/prctl/uapi: remove ifdef for CHECKPOINT_RESTORE

As userspace knows nothing about kernel config, this ifdefs will make prctl constants invisible to userspace. Let it be clean'n'simple: remove ifdefs.
If kernel has CONFIG_CHECKPOINT_RESTORE disabled, sys_prctl() will return -EINVAL for those prctls.

Fixes 2eefd8789698 ("x86/arch_prctl/vdso: Add ARCH_MAP_VDSO_*")
Cc Cyrill Gorcunov <[email protected]>
Cc Paul Bolle <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
 arch/x86/include/uapi/asm/prctl.h | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
arch/x86/include/uapi/asm/prctl.h +3 -5
index ae135de547f5..835aa51c7f6e 100644
@@ -6,10 +6,8 @@
 #define ARCH_GET_FS 0x1003
 #define ARCH_GET_GS 0x1004
 
-#ifdef CONFIG_CHECKPOINT_RESTORE
-# define ARCH_MAP_VDSO_X32	0x2001
-# define ARCH_MAP_VDSO_32	0x2002
-# define ARCH_MAP_VDSO_64	0x2003
-#endif
+#define ARCH_MAP_VDSO_X32	0x2001
+#define ARCH_MAP_VDSO_32	0x2002
+#define ARCH_MAP_VDSO_64	0x2003
 
 #endif /* _ASM_X86_PRCTL_H */
Signature · 7 lines
2.10.1

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#2 lore raw

[tip:x86/asm] x86/prctl/uapi: Remove #ifdef for CHECKPOINT_RESTORE

x86/prctl/uapi: Remove #ifdef for CHECKPOINT_RESTORE
Commit-ID:  a01aa6c9f40fe03c82032e7f8b3bcf1e6c93ac0e
Gitweb:     http://git.kernel.org/tip/a01aa6c9f40fe03c82032e7f8b3bcf1e6c93ac0e
Author:     Dmitry Safonov <[email protected]>
AuthorDate: Thu, 27 Oct 2016 17:15:15 +0300
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 28 Oct 2016 08:15:55 +0200

x86/prctl/uapi: Remove #ifdef for CHECKPOINT_RESTORE

As userspace knows nothing about kernel config, thus #ifdefs around ABI prctl constants makes them invisible to userspace.

Let it be clean'n'simple: remove #ifdefs.

If kernel has CONFIG_CHECKPOINT_RESTORE disabled, sys_prctl() will return -EINVAL for those prctls.

Cc Borislav Petkov <[email protected]>
Cc Brian Gerst <[email protected]>
Cc Cyrill Gorcunov <[email protected]>
Cc Denys Vlasenko <[email protected]>
Cc H. Peter Anvin <[email protected]>
Cc Josh Poimboeuf <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes 2eefd8789698 ("x86/arch_prctl/vdso: Add ARCH_MAP_VDSO_*")
---
 arch/x86/include/uapi/asm/prctl.h | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
arch/x86/include/uapi/asm/prctl.h +3 -5
index ae135de..835aa51 100644
@@ -6,10 +6,8 @@
 #define ARCH_GET_FS 0x1003
 #define ARCH_GET_GS 0x1004
 
-#ifdef CONFIG_CHECKPOINT_RESTORE
-# define ARCH_MAP_VDSO_X32	0x2001
-# define ARCH_MAP_VDSO_32	0x2002
-# define ARCH_MAP_VDSO_64	0x2003
-#endif
+#define ARCH_MAP_VDSO_X32	0x2001
+#define ARCH_MAP_VDSO_32	0x2002
+#define ARCH_MAP_VDSO_64	0x2003
 
 #endif /* _ASM_X86_PRCTL_H */
#3 lore raw

PATCH 2/2 x86/vdso: set vdso pointer only after success

In reply to Dmitry Safonov

Those pointers were initialized before call to _install_special_mapping after the commit f7b6eb3fa072 ("x86: Set context.vdso before installing the mapping"). This is not required anymore as special mappings have their vma name and don't use arch_vma_name() after commit a62c34bd2a8a ("x86, mm: Improve _install_special_mapping and fix x86 vdso naming"). So, this way to init looks less entangled.
I even belive, we can remove null initializers:
- on failure load_elf_binary() will not start a new thread;
- arch_prctl will have the same pointers as before syscall.

Cc Cyrill Gorcunov <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
 arch/x86/entry/vdso/vma.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
arch/x86/entry/vdso/vma.c +3 -7
index 23c881caabd1..e739002427ed 100644
@@ -161,8 +161,6 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
 	}
 
 	text_start = addr - image->sym_vvar_start;
-	current->mm->context.vdso = (void __user *)text_start;
-	current->mm->context.vdso_image = image;
 
 	/*
 	 * MAYWRITE to allow gdb to COW and set breakpoints
@@ -189,14 +187,12 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
 	if (IS_ERR(vma)) {
 		ret = PTR_ERR(vma);
 		do_munmap(mm, text_start, image->size);
+	} else {
+		current->mm->context.vdso = (void __user *)text_start;
+		current->mm->context.vdso_image = image;
 	}
 
 up_fail:
-	if (ret) {
-		current->mm->context.vdso = NULL;
-		current->mm->context.vdso_image = NULL;
-	}
-
 	up_write(&mm->mmap_sem);
 	return ret;
 }
Signature · 7 lines
2.10.1

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
Acked-by on v1 2/2
#4 lore raw

Re: [PATCH 2/2] x86/vdso: set vdso pointer only after success

Quoted 9 lines · Dmitry Safonov

Those pointers were initialized before call to _install_special_mapping after the commit f7b6eb3fa072 ("x86: Set context.vdso before installing the mapping"). This is not required anymore as special mappings have their vma name and don't use arch_vma_name() after commit a62c34bd2a8a ("x86, mm: Improve _install_special_mapping and fix x86 vdso naming"). So, this way to init looks less entangled.
I even belive, we can remove null initializers:
- on failure load_elf_binary() will not start a new thread;
- arch_prctl will have the same pointers as before syscall.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#5 lore raw

[tip:x86/asm] x86/vdso: Set vDSO pointer only after success

In reply to Dmitry Safonov
x86/vdso: Set vDSO pointer only after success
Commit-ID:  67dece7d4c5841e84a3c795e79bf0dcd5be54f55
Gitweb:     http://git.kernel.org/tip/67dece7d4c5841e84a3c795e79bf0dcd5be54f55
Author:     Dmitry Safonov <[email protected]>
AuthorDate: Thu, 27 Oct 2016 17:15:16 +0300
Committer:  Ingo Molnar <[email protected]>
CommitDate: Fri, 28 Oct 2016 08:15:55 +0200

x86/vdso: Set vDSO pointer only after success

Those pointers were initialized before call to _install_special_mapping() after the commit:

f7b6eb3fa072 ("x86: Set context.vdso before installing the mapping")

This is not required anymore as special mappings have their vma name and don't use arch_vma_name() after commit:

a62c34bd2a8a ("x86, mm: Improve _install_special_mapping and fix x86 vdso naming")

So, this way to init looks less entangled.

I even belive that we can remove NULL initializers:

- on failure load_elf_binary() will not start a new thread;
- arch_prctl will have the same pointers as before syscall.

---
 arch/x86/entry/vdso/vma.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
arch/x86/entry/vdso/vma.c +3 -7
index 23c881c..e739002 100644
@@ -161,8 +161,6 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
 	}
 
 	text_start = addr - image->sym_vvar_start;
-	current->mm->context.vdso = (void __user *)text_start;
-	current->mm->context.vdso_image = image;
 
 	/*
 	 * MAYWRITE to allow gdb to COW and set breakpoints
@@ -189,14 +187,12 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
 	if (IS_ERR(vma)) {
 		ret = PTR_ERR(vma);
 		do_munmap(mm, text_start, image->size);
+	} else {
+		current->mm->context.vdso = (void __user *)text_start;
+		current->mm->context.vdso_image = image;
 	}
 
 up_fail:
-	if (ret) {
-		current->mm->context.vdso = NULL;
-		current->mm->context.vdso_image = NULL;
-	}
-
 	up_write(&mm->mmap_sem);
 	return ret;
 }