[PATCH v2 1/4] mm: Create a new system state and fix core_kernel_text()

core_kernel_text() considers that until system_state in at least
SYSTEM_RUNNING, init memory is valid.

But init memory is freed a few lines before setting SYSTEM_RUNNING,
so we have a small period of time when core_kernel_text() is wrong.

Create an intermediate system state called SYSTEM_FREEING_INIT that
is set before starting freeing init memory, and use it in
core_kernel_text() to report init memory invalid earlier.

Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Christophe Leroy <redacted>
---
v2: New
---
 include/linux/kernel.h | 1 +
 init/main.c            | 2 ++
 kernel/extable.c       | 2 +-
 3 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/include/linux/kernel.h b/include/linux/kernel.h
index 2776423a587e..471bc0593679 100644
--- a/include/linux/kernel.h
+++ b/include/linux/kernel.h

@@ -248,6 +248,7 @@ extern bool early_boot_irqs_disabled;
 extern enum system_states {
 	SYSTEM_BOOTING,
 	SYSTEM_SCHEDULING,
+	SYSTEM_FREEING_INITMEM,
 	SYSTEM_RUNNING,
 	SYSTEM_HALT,
 	SYSTEM_POWER_OFF,

diff --git a/init/main.c b/init/main.c
index 3f7216934441..c457d393fdd4 100644
--- a/init/main.c
+++ b/init/main.c

@@ -1505,6 +1505,8 @@ static int __ref kernel_init(void *unused)
 	kernel_init_freeable();
 	/* need to finish all async __init code before freeing the memory */
 	async_synchronize_full();
+
+	system_state = SYSTEM_FREEING_INITMEM;
 	kprobe_free_init_mem();
 	ftrace_free_init_mem();
 	kgdb_free_init_mem();

diff --git a/kernel/extable.c b/kernel/extable.c
index b0ea5eb0c3b4..290661f68e6b 100644
--- a/kernel/extable.c
+++ b/kernel/extable.c

@@ -76,7 +76,7 @@ int notrace core_kernel_text(unsigned long addr)
 	    addr < (unsigned long)_etext)
 		return 1;
 
-	if (system_state < SYSTEM_RUNNING &&
+	if (system_state < SYSTEM_FREEING_INITMEM &&
 	    init_kernel_text(addr))
 		return 1;
 	return 0;

-- 
2.31.1

Generic version of arch_is_kernel_initmem_freed() now does the same
as s390 version.

Remove the s390 version.

Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Christophe Leroy <redacted>
---
v2: No change
---
 arch/s390/include/asm/sections.h | 12 ------------
 arch/s390/mm/init.c              |  3 ---
 2 files changed, 15 deletions(-)

diff --git a/arch/s390/include/asm/sections.h b/arch/s390/include/asm/sections.h
index 85881dd48022..3fecaa4e8b74 100644
--- a/arch/s390/include/asm/sections.h
+++ b/arch/s390/include/asm/sections.h

@@ -2,20 +2,8 @@
 #ifndef _S390_SECTIONS_H
 #define _S390_SECTIONS_H
 
-#define arch_is_kernel_initmem_freed arch_is_kernel_initmem_freed
-
 #include <asm-generic/sections.h>
 
-extern bool initmem_freed;
-
-static inline int arch_is_kernel_initmem_freed(unsigned long addr)
-{
-	if (!initmem_freed)
-		return 0;
-	return addr >= (unsigned long)__init_begin &&
-	       addr < (unsigned long)__init_end;
-}
-
 /*
  * .boot.data section contains variables "shared" between the decompressor and
  * the decompressed kernel. The decompressor will store values in them, and

diff --git a/arch/s390/mm/init.c b/arch/s390/mm/init.c
index a04faf49001a..8c6f258a6183 100644
--- a/arch/s390/mm/init.c
+++ b/arch/s390/mm/init.c

@@ -58,8 +58,6 @@ unsigned long empty_zero_page, zero_page_mask;
 EXPORT_SYMBOL(empty_zero_page);
 EXPORT_SYMBOL(zero_page_mask);
 
-bool initmem_freed;
-
 static void __init setup_zero_pages(void)
 {
 	unsigned int order;

@@ -214,7 +212,6 @@ void __init mem_init(void)
 
 void free_initmem(void)
 {
-	initmem_freed = true;
 	__set_memory((unsigned long)_sinittext,
 		     (unsigned long)(_einittext - _sinittext) >> PAGE_SHIFT,
 		     SET_MEMORY_RW | SET_MEMORY_NX);

-- 
2.31.1

Generic version of arch_is_kernel_initmem_freed() now does the same
as powerpc version.

Remove the powerpc version.

Signed-off-by: Christophe Leroy <redacted>
---
v2: No change
---
 arch/powerpc/include/asm/sections.h | 13 -------------
 1 file changed, 13 deletions(-)

diff --git a/arch/powerpc/include/asm/sections.h b/arch/powerpc/include/asm/sections.h
index 6e4af4492a14..79cb7a25a5fb 100644
--- a/arch/powerpc/include/asm/sections.h
+++ b/arch/powerpc/include/asm/sections.h

@@ -6,21 +6,8 @@
 #include <linux/elf.h>
 #include <linux/uaccess.h>
 
-#define arch_is_kernel_initmem_freed arch_is_kernel_initmem_freed
-
 #include <asm-generic/sections.h>
 
-extern bool init_mem_is_free;
-
-static inline int arch_is_kernel_initmem_freed(unsigned long addr)
-{
-	if (!init_mem_is_free)
-		return 0;
-
-	return addr >= (unsigned long)__init_begin &&
-		addr < (unsigned long)__init_end;
-}
-
 extern char __head_end[];
 
 #ifdef __powerpc64__

-- 
2.31.1

Commit 7a5da02de8d6 ("locking/lockdep: check for freed initmem in
static_obj()") added arch_is_kernel_initmem_freed() which is supposed
to report whether an object is part of already freed init memory.

For the time being, the generic version of arch_is_kernel_initmem_freed()
always reports 'false', allthough free_initmem() is generically called
on all architectures.

Therefore, change the generic version of arch_is_kernel_initmem_freed()
to check whether free_initmem() has been called. If so, then check
if a given address falls into init memory.

In order to use function init_section_contains(), the fonction is
moved at the end of asm-generic/section.h

Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Christophe Leroy <redacted>
---
v2: Change to using the new SYSTEM_FREEING_INITMEM state
---
 include/asm-generic/sections.h | 31 +++++++++++++++++--------------
 1 file changed, 17 insertions(+), 14 deletions(-)

diff --git a/include/asm-generic/sections.h b/include/asm-generic/sections.h
index d16302d3eb59..13f301239007 100644
--- a/include/asm-generic/sections.h
+++ b/include/asm-generic/sections.h

@@ -80,20 +80,6 @@ static inline int arch_is_kernel_data(unsigned long addr)
 }
 #endif
 
-/*
- * Check if an address is part of freed initmem. This is needed on architectures
- * with virt == phys kernel mapping, for code that wants to check if an address
- * is part of a static object within [_stext, _end]. After initmem is freed,
- * memory can be allocated from it, and such allocations would then have
- * addresses within the range [_stext, _end].
- */
-#ifndef arch_is_kernel_initmem_freed
-static inline int arch_is_kernel_initmem_freed(unsigned long addr)
-{
-	return 0;
-}
-#endif
-
 /**
  * memory_contains - checks if an object is contained within a memory region
  * @begin: virtual address of the beginning of the memory region

@@ -172,4 +158,21 @@ static inline bool is_kernel_rodata(unsigned long addr)
 	       addr < (unsigned long)__end_rodata;
 }
 
+/*
+ * Check if an address is part of freed initmem. This is needed on architectures
+ * with virt == phys kernel mapping, for code that wants to check if an address
+ * is part of a static object within [_stext, _end]. After initmem is freed,
+ * memory can be allocated from it, and such allocations would then have
+ * addresses within the range [_stext, _end].
+ */
+#ifndef arch_is_kernel_initmem_freed
+static inline int arch_is_kernel_initmem_freed(unsigned long addr)
+{
+	if (system_state < SYSTEM_FREEING_INITMEM)
+		return 0;
+
+	return init_section_contains((void *)addr, 1);
+}
+#endif
+
 #endif /* _ASM_GENERIC_SECTIONS_H_ */

-- 
2.31.1

On Tue, Sep 28, 2021 at 09:15:37AM +0200, Christophe Leroy wrote:

Generic version of arch_is_kernel_initmem_freed() now does the same
as s390 version.

Remove the s390 version.

Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Christophe Leroy <redacted>
---
v2: No change
---
 arch/s390/include/asm/sections.h | 12 ------------
 arch/s390/mm/init.c              |  3 ---
 2 files changed, 15 deletions(-)

Looks good. Thanks for cleaning this up!

Acked-by: Heiko Carstens <hca@linux.ibm.com>

On Tue, 28 Sep 2021 09:15:35 +0200 Christophe Leroy [off-list ref] wrote:

Commit 7a5da02de8d6 ("locking/lockdep: check for freed initmem in
static_obj()") added arch_is_kernel_initmem_freed() which is supposed
to report whether an object is part of already freed init memory.

For the time being, the generic version of arch_is_kernel_initmem_freed()
always reports 'false', allthough free_initmem() is generically called
on all architectures.

Therefore, change the generic version of arch_is_kernel_initmem_freed()
to check whether free_initmem() has been called. If so, then check
if a given address falls into init memory.

In order to use function init_section_contains(), the fonction is
moved at the end of asm-generic/section.h

i386 allmodconfig:

In file included from arch/x86/platform/intel-quark/imr.c:28:
./include/asm-generic/sections.h: In function 'arch_is_kernel_initmem_freed':
./include/asm-generic/sections.h:171:6: error: 'system_state' undeclared (first use in this function)
  171 |  if (system_state < SYSTEM_FREEING_INITMEM)
      |      ^~~~~~~~~~~~
./include/asm-generic/sections.h:171:6: note: each undeclared identifier is reported only once for each function it appears in
./include/asm-generic/sections.h:171:21: error: 'SYSTEM_FREEING_INITMEM' undeclared (first use in this function)
  171 |  if (system_state < SYSTEM_FREEING_INITMEM)
      |                     ^~~~~~~~~~~~~~~~~~~~~~


I don't think it would be a good idea to include kernel.h from
sections.h - it's unclear to me which is the "innermost" of those two. 
It would be better to uninline arch_is_kernel_initmem_freed().  Surely
there's no real reason for inlining it?

Anyway, I'll drop the series for now.