Skip to contentSkip to conversation

x86: Fix static memory detection

PATCH Applied Guenter Roeck 2 participants 3 messages Jan 31, 2020 – Mar 19, 2020 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86: Fix static memory detection

When booting x86 images in qemu, the following warning is seen randomly if DEBUG_LOCKDEP is enabled.

WARNING: CPU: 0 PID: 1 at kernel/locking/lockdep.c:1119
	lockdep_register_key+0xc0/0x100

static_obj() returns true if an address is between _stext and _end. On x86, this includes the brk memory space. Problem is that this memory block is not static on x86; its unused portions are released after init and can be allocated. This results in the observed warning if a lockdep object is allocated from this memory.

Solve the problem by implementing arch_is_kernel_initmem_freed() for x86 and have it return true if an address is within the released memory range.

The same problem was solved for s390 with commit 7a5da02de8d6e ("locking/lockdep: check for freed initmem in static_obj()"), which introduced arch_is_kernel_initmem_freed().

---
 arch/x86/include/asm/sections.h | 20 ++++++++++++++++++++
 arch/x86/kernel/setup.c         |  1 -
 2 files changed, 20 insertions(+), 1 deletion(-)
arch/x86/include/asm/sections.h +20 -0
index 036c360910c5..a6e8373a5170 100644
@@ -2,6 +2,8 @@
 #ifndef _ASM_X86_SECTIONS_H
 #define _ASM_X86_SECTIONS_H
 
+#define arch_is_kernel_initmem_freed arch_is_kernel_initmem_freed
+
 #include <asm-generic/sections.h>
 #include <asm/extable.h>
 
@@ -14,4 +16,22 @@ extern char __end_rodata_hpage_align[];
 
 extern char __end_of_kernel_reserve[];
 
+extern unsigned long _brk_start, _brk_end;
+
+static inline bool arch_is_kernel_initmem_freed(unsigned long addr)
+{
+	/*
+	 * If _brk_start has not been cleared, brk allocation is incomplete,
+	 * and we can not make assumptions about its use.
+	 */
+	if (_brk_start)
+		return 0;
+
+	/*
+	 * After brk allocation is complete, space between _brk_end and _end
+	 * is available for allocation.
+	 */
+	return addr >= _brk_end && addr < (unsigned long)&_end;
+}
+
 #endif	/* _ASM_X86_SECTIONS_H */
arch/x86/kernel/setup.c +0 -1
index 1e4c20a1efec..08f5ceed70e0 100644
@@ -64,7 +64,6 @@ RESERVE_BRK(dmi_alloc, 65536);
  * at link time, with RESERVE_BRK*() facility reserving additional
  * chunks.
  */
-static __initdata
 unsigned long _brk_start = (unsigned long)__brk_base;
 unsigned long _brk_end   = (unsigned long)__brk_base;
 
Signature · 1 line
2.17.1
#1 lore raw

ping ... I still see this problem.

Guenter

Quoted 74 lines · Guenter Roeck

When booting x86 images in qemu, the following warning is seen randomly if DEBUG_LOCKDEP is enabled.

WARNING: CPU: 0 PID: 1 at kernel/locking/lockdep.c:1119
	lockdep_register_key+0xc0/0x100

static_obj() returns true if an address is between _stext and _end. On x86, this includes the brk memory space. Problem is that this memory block is not static on x86; its unused portions are released after init and can be allocated. This results in the observed warning if a lockdep object is allocated from this memory.

Solve the problem by implementing arch_is_kernel_initmem_freed() for x86 and have it return true if an address is within the released memory range.

The same problem was solved for s390 with commit 7a5da02de8d6e ("locking/lockdep: check for freed initmem in static_obj()"), which introduced arch_is_kernel_initmem_freed().

---
 arch/x86/include/asm/sections.h | 20 ++++++++++++++++++++
 arch/x86/kernel/setup.c         |  1 -
 2 files changed, 20 insertions(+), 1 deletion(-)
arch/x86/include/asm/sections.h +20 -0
index 036c360910c5..a6e8373a5170 100644
@@ -2,6 +2,8 @@
 #ifndef _ASM_X86_SECTIONS_H
 #define _ASM_X86_SECTIONS_H
 
+#define arch_is_kernel_initmem_freed arch_is_kernel_initmem_freed
+
 #include <asm-generic/sections.h>
 #include <asm/extable.h>
 
@@ -14,4 +16,22 @@ extern char __end_rodata_hpage_align[];
 
 extern char __end_of_kernel_reserve[];
 
+extern unsigned long _brk_start, _brk_end;
+
+static inline bool arch_is_kernel_initmem_freed(unsigned long addr)
+{
+	/*
+	 * If _brk_start has not been cleared, brk allocation is incomplete,
+	 * and we can not make assumptions about its use.
+	 */
+	if (_brk_start)
+		return 0;
+
+	/*
+	 * After brk allocation is complete, space between _brk_end and _end
+	 * is available for allocation.
+	 */
+	return addr >= _brk_end && addr < (unsigned long)&_end;
+}
+
 #endif	/* _ASM_X86_SECTIONS_H */
arch/x86/kernel/setup.c +0 -1
index 1e4c20a1efec..08f5ceed70e0 100644
@@ -64,7 +64,6 @@ RESERVE_BRK(dmi_alloc, 65536);
  * at link time, with RESERVE_BRK*() facility reserving additional
  * chunks.
  */
-static __initdata
 unsigned long _brk_start = (unsigned long)__brk_base;
 unsigned long _brk_end   = (unsigned long)__brk_base;
 
merge bot Applied
#2 lore raw

[tip: x86/boot] x86/setup: Fix static memory detection

In reply to Guenter Roeck
The following commit has been merged into the x86/boot branch of tip:

The following commit has been merged into the x86/boot branch of tip:

Commit-ID:     bac59d18c7018a2fd5e800a1e72a8271bf404977
Gitweb:        https://git.kernel.org/tip/bac59d18c7018a2fd5e800a1e72a8271bf404977
Author:        Guenter Roeck <[email protected]>
AuthorDate:    Thu, 30 Jan 2020 18:11:59 -08:00
Committer:     Borislav Petkov <[email protected]>
CommitterDate: Thu, 19 Mar 2020 11:58:13 +01:00

x86/setup: Fix static memory detection

When booting x86 images in qemu, the following warning is seen randomly if DEBUG_LOCKDEP is enabled.

  WARNING: CPU: 0 PID: 1 at kernel/locking/lockdep.c:1119
	  lockdep_register_key+0xc0/0x100

static_obj() returns true if an address is between _stext and _end.

On x86, this includes the brk memory space. Problem is that this memory block is not static on x86; its unused portions are released after init and can be allocated. This results in the observed warning if a lockdep object is allocated from this memory.

Solve the problem by implementing arch_is_kernel_initmem_freed() for x86 and have it return true if an address is within the released memory range.

The same problem was solved for s390 with commit

7a5da02de8d6e ("locking/lockdep: check for freed initmem in static_obj()"),

which introduced arch_is_kernel_initmem_freed().

---
 arch/x86/include/asm/sections.h | 20 ++++++++++++++++++++
 arch/x86/kernel/setup.c         |  1 -
 2 files changed, 20 insertions(+), 1 deletion(-)
arch/x86/include/asm/sections.h +20 -0
index 036c360..a6e8373 100644
@@ -2,6 +2,8 @@
 #ifndef _ASM_X86_SECTIONS_H
 #define _ASM_X86_SECTIONS_H
 
+#define arch_is_kernel_initmem_freed arch_is_kernel_initmem_freed
+
 #include <asm-generic/sections.h>
 #include <asm/extable.h>
 
@@ -14,4 +16,22 @@ extern char __end_rodata_hpage_align[];
 
 extern char __end_of_kernel_reserve[];
 
+extern unsigned long _brk_start, _brk_end;
+
+static inline bool arch_is_kernel_initmem_freed(unsigned long addr)
+{
+	/*
+	 * If _brk_start has not been cleared, brk allocation is incomplete,
+	 * and we can not make assumptions about its use.
+	 */
+	if (_brk_start)
+		return 0;
+
+	/*
+	 * After brk allocation is complete, space between _brk_end and _end
+	 * is available for allocation.
+	 */
+	return addr >= _brk_end && addr < (unsigned long)&_end;
+}
+
 #endif	/* _ASM_X86_SECTIONS_H */
arch/x86/kernel/setup.c +0 -1
index a74262c..e6b5450 100644
@@ -64,7 +64,6 @@ RESERVE_BRK(dmi_alloc, 65536);
  * at link time, with RESERVE_BRK*() facility reserving additional
  * chunks.
  */
-static __initdata
 unsigned long _brk_start = (unsigned long)__brk_base;
 unsigned long _brk_end   = (unsigned long)__brk_base;