Skip to contentSkip to conversation

Update num_symtab, symtab and strtab right after setup of core_num_syms, core_symtab and core_strtab.

PATCH Superseded Hu Tao 2 participants 5 messages Sep 27, 2010 – Sep 28, 2010

lore
Share & tools
lore

Hu Tao

author
#0 lore raw

PATCH 1/1 Update num_symtab, symtab and strtab right after setup of core_num_syms, core_symtab and core_strtab.

---
 kernel/module.c |    9 ++++-----
 1 files changed, 4 insertions(+), 5 deletions(-)
kernel/module.c +4 -5
index d0b5f8d..d5dc98b 100644
@@ -2029,6 +2029,10 @@ static void add_kallsyms(struct module *mod, const struct load_info *info)
 	for (*s = 0, i = 1; i < info->sechdrs[info->index.str].sh_size; ++i)
 		if (test_bit(i, info->strmap))
 			*++s = mod->strtab[i];
+
+	mod->num_symtab = mod->core_num_syms;
+	mod->symtab = mod->core_symtab;
+	mod->strtab = mod->core_strtab;
 }
 #else
 static inline void layout_symtab(struct module *mod, struct load_info *info)
@@ -2738,11 +2742,6 @@ SYSCALL_DEFINE3(init_module, void __user *, umod,
 	/* Drop initial reference. */
 	module_put(mod);
 	trim_init_extable(mod);
-#ifdef CONFIG_KALLSYMS
-	mod->num_symtab = mod->core_num_syms;
-	mod->symtab = mod->core_symtab;
-	mod->strtab = mod->core_strtab;
-#endif
 	module_free(mod, mod->module_init);
 	mod->module_init = NULL;
 	mod->init_size = 0;
Signature · 1 line
1.7.3
#1 lore raw
Quoted 1 line · Hu Tao

Changelog is empty.
What bug did you observe?

Quoted 21 lines

---
kernel/module.c | 9 ++++-----

1 files changed, 4 insertions(+), 5 deletions(-)
kernel/module.c +4 -0
index d0b5f8d..d5dc98b 100644
@@ -2029,6 +2029,10 @@ static void add_kallsyms(struct module *mod, const struct load_info *info)
 	for (*s = 0, i = 1; i < info->sechdrs[info->index.str].sh_size; ++i)
 		if (test_bit(i, info->strmap))
 			*++s = mod->strtab[i];
+
+	mod->num_symtab = mod->core_num_syms;
+	mod->symtab = mod->core_symtab;
+	mod->strtab = mod->core_strtab;
 }
 #else
 static inline void layout_symtab(struct module *mod, struct load_info *info)
@@ -2738,11 +2742,6 @@ SYSCALL_DEFINE3(init_module, void __user *, umod,
 	/* Drop initial reference. */

kernel/module.c SYSCALL_DEFINE3(init_module, void __user *, umod, · line 2747

Quoted 10 lines
kernel/module.c +0 -5
 	module_put(mod);
 	trim_init_extable(mod);
-#ifdef CONFIG_KALLSYMS
-	mod->num_symtab = mod->core_num_syms;
-	mod->symtab = mod->core_symtab;
-	mod->strtab = mod->core_strtab;
-#endif
 	module_free(mod, mod->module_init);
 	mod->module_init = NULL;
 	mod->init_size = 0;

This patch is not correct, symbols that are not core symbols will be dropped right after this piece of code. So NACK.

author
#2 lore raw
Quoted 6 lines · Américo Wang
Quoted 1 line · Hu Tao

Changelog is empty.
What bug did you observe?

symtab and strtab will point to vfreed memory after free_copy(&info) near end of load_module().

Quoted 28 lines
Quoted 28 lines

---
kernel/module.c | 9 ++++-----

1 files changed, 4 insertions(+), 5 deletions(-)
kernel/module.c +4 -4
index d0b5f8d..d5dc98b 100644
@@ -2029,6 +2029,10 @@ static void add_kallsyms(struct module *mod, const struct load_info *info)
 	for (*s = 0, i = 1; i < info->sechdrs[info->index.str].sh_size; ++i)
 		if (test_bit(i, info->strmap))
 			*++s = mod->strtab[i];
+
+	mod->num_symtab = mod->core_num_syms;
+	mod->symtab = mod->core_symtab;
+	mod->strtab = mod->core_strtab;
 }
 #else
 static inline void layout_symtab(struct module *mod, struct load_info *info)
@@ -2738,11 +2742,6 @@ SYSCALL_DEFINE3(init_module, void __user *, umod,
 	/* Drop initial reference. */
 	module_put(mod);
 	trim_init_extable(mod);
-#ifdef CONFIG_KALLSYMS
-	mod->num_symtab = mod->core_num_syms;
-	mod->symtab = mod->core_symtab;
-	mod->strtab = mod->core_strtab;
Quoted 8 lines

kernel/module.c SYSCALL_DEFINE3(init_module, void __user *, umod, · line 2747

Quoted 4 lines
kernel/module.c +0 -1
-#endif
 	module_free(mod, mod->module_init);
 	mod->module_init = NULL;
 	mod->init_size = 0;

This patch is not correct, symbols that are not core symbols will be dropped right after this piece of code. So NACK.

There is the reason why we should update num_symtab, symtab and strtab before non-core-symbols are dropped.

Reviewed-by
#3 lore raw
Quoted 11 lines · Hu Tao
Quoted 6 lines · Américo Wang
Quoted 1 line · Hu Tao

Changelog is empty.
What bug did you observe?

symtab and strtab will point to vfreed memory after free_copy(&info) near end of load_module().

I think you must mean the error path in load_module(), yeah, probably.

But you need to check if this patch misses other thing, like
the other functions which are called after add_kallsyms() inside load_module(). According to my eyes, this seems fine (at least on x86), so

Quoted 7 lines
Quoted 4 lines

This patch is not correct, symbols that are not core symbols will be dropped right after this piece of code. So NACK.

There is the reason why we should update num_symtab, symtab and strtab before non-core-symbols are dropped.

init section is dropped *after* these value are update, which is correct.

Again, you need to describe what you are trying to fix in your changelog. Please resend it with a proper changelog.

Thanks.

author
#4 lore raw
Quoted 6 lines · Américo Wang
Quoted 5 lines · Hu Tao
Quoted 4 lines · Américo Wang
Quoted 1 line · Hu Tao
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

Changelog is empty.
What bug did you observe?

symtab and strtab will point to vfreed memory after free_copy(&info) near end of load_module().

I think you must mean the error path in load_module(), yeah, probably.

No. There is a free_copy(&info) at the end of normal path of load_module().

Quoted 6 lines

But you need to check if this patch misses other thing, like
the other functions which are called after add_kallsyms() inside load_module(). According to my eyes, this seems fine (at least on x86), so

Thanks for review.

Quoted 15 lines
Quoted 8 lines
Quoted 5 lines

This patch is not correct, symbols that are not core symbols will be dropped right after this piece of code. So NACK.

There is the reason why we should update num_symtab, symtab and strtab before non-core-symbols are dropped.

init section is dropped *after* these value are update, which is correct.

Again, you need to describe what you are trying to fix in your changelog. Please resend it with a proper changelog.

Thanks.