kernel: kallsyms: parameters checking, for EXPORT_SYMBOL_GPL functions
PATCH 1/1 kernel: kallsyms: parameters checking, for EXPORT_SYMBOL_GPL functions
for EXPORT_SYMBOL_GPL functions, necessary to check their parameters.
--- kernel/kallsyms.c | 12 ++++++++++++ 1 files changed, 12 insertions(+), 0 deletions(-)
kernel/kallsyms.c +12 -0
@@ -175,6 +175,9 @@ unsigned long kallsyms_lookup_name(const char *name)unsigned long i;unsigned int off; +if (!name || !name[0 ]) +return 0 ; +for (i =0 , off =0 ; i < kallsyms_num_syms; i++) { off =kallsyms_expand_symbol (off, namebuf);
@@ -194,6 +197,9 @@ int kallsyms_on_each_symbol(int (*fn)(void *, const char *, struct module *,unsigned int off;int ret; +if (!fn) +return -EINVAL ; +for (i =0 , off =0 ; i < kallsyms_num_syms; i++) { off =kallsyms_expand_symbol (off, namebuf); ret =fn (data, namebuf,NULL , kallsyms_addresses[i]);
@@ -382,6 +388,9 @@ static int __sprint_symbol(char *buffer, unsigned long address,*/ int sprint_symbol (char *buffer,unsigned long address) { +if (!buffer) +return 0 ; +return __sprint_symbol (buffer, address,0 ,1 ); }EXPORT_SYMBOL_GPL (sprint_symbol);
@@ -399,6 +408,9 @@ EXPORT_SYMBOL_GPL(sprint_symbol);*/ int sprint_symbol_no_offset (char *buffer,unsigned long address) { +if (!buffer) +return 0 ; +return __sprint_symbol (buffer, address,0 ,0 ); }EXPORT_SYMBOL_GPL (sprint_symbol_no_offset);
Signature · 1 line
1.7.7.6
Quoted 3 lines · Chen Gang
for
EXPORT_SYMBOL_GPLfunctions, necessary to check their parameters.Signed-off-by Chen Gang <[email protected]>
Why?
If someone misuses these functions, they crash and thus indicate that the caller shouldn't do that.
Or is someone already doing this?
Confused,
Rusty.
Quoted 50 lines
--- kernel/kallsyms.c | 12 ++++++++++++ 1 files changed, 12 insertions(+), 0 deletions(-)kernel/kallsyms.c +12
-0@@ -175,6 +175,9 @@ unsigned long kallsyms_lookup_name(const char *name)unsigned long i;unsigned int off; +if (!name || !name[0 ]) +return 0 ; +for (i =0 , off =0 ; i < kallsyms_num_syms; i++) { off =kallsyms_expand_symbol (off, namebuf);@@ -194,6 +197,9 @@ int kallsyms_on_each_symbol(int (*fn)(void *, const char *, struct module *,unsigned int off;int ret; +if (!fn) +return -EINVAL ; +for (i =0 , off =0 ; i < kallsyms_num_syms; i++) { off =kallsyms_expand_symbol (off, namebuf); ret =fn (data, namebuf,NULL , kallsyms_addresses[i]);@@ -382,6 +388,9 @@ static int __sprint_symbol(char *buffer, unsigned long address,*/ int sprint_symbol (char *buffer,unsigned long address) { +if (!buffer) +return 0 ; +return __sprint_symbol (buffer, address,0 ,1 ); }EXPORT_SYMBOL_GPL (sprint_symbol);@@ -399,6 +408,9 @@ EXPORT_SYMBOL_GPL(sprint_symbol);*/ int sprint_symbol_no_offset (char *buffer,unsigned long address) { +if (!buffer) +return 0 ; +return __sprint_symbol (buffer, address,0 ,0 ); }EXPORT_SYMBOL_GPL (sprint_symbol_no_offset);Signature · 1 line
1.7.7.6
Quoted 9 lines · Rusty Russell
Quoted 3 lines · Chen Gang
Quoted 3 lines
for
EXPORT_SYMBOL_GPLfunctions, necessary to check their parameters.Signed-off-by Chen Gang <[email protected]>Why?
If someone misuses these functions, they crash and thus indicate that the caller shouldn't do that.
for me, I think:
if it is used by self (such as static functions):
I prefer to crash immediatly.
it will help us to find issue, quickly.
if it can be used by others (such as EXPORT_SYMBOL_GPL):
I prefer to return fail and tell caller that parameter is invalid. it is more polite to callers, and still indicate it may be an issue.
:-)
Quoted 2 lines
Or is someone already doing this?
really has:
kernel: __wake_up_sync_key in kernel/sched/core.c.
lib: *printf.
mm: kfree.
Quoted 3 lines
Confused,
Rusty.
Signature · 3 lines
Chen Gang Asianux Corporation
Quoted 17 lines · Chen Gang
Quoted 9 lines · Rusty Russell
Quoted 3 lines · Chen Gang
Quoted 3 lines
for
EXPORT_SYMBOL_GPLfunctions, necessary to check their parameters.Signed-off-by Chen Gang <[email protected]>Why?
If someone misuses these functions, they crash and thus indicate that the caller shouldn't do that.
for me, I think:
if it is used by self (such as static functions):
I prefer to crash immediatly.
it will help us to find issue, quickly.
Quoted 5 lines
if it can be used by others (such as
EXPORT_SYMBOL_GPL):
I prefer to return fail and tell caller that parameter is invalid. it is more polite to callers, and still indicate it may be an issue.:-)
I disagree. Calling with invalid parameters is a bug. You've just covered up some cases of invalid use and made it less likely to be found. Because the caller won't notice they screwed up.
We could sprinkle WARN_ON() everywhere, but I prefer the crash. Even harder to ignore.
There's no limit to how many of these checks we could put in, and we can *never* take them out. I don't want to code that way.
Quoted 8 lines
Quoted 2 lines
Or is someone already doing this?
really has:
kernel:
__wake_up_sync_keyinkernel/sched/core.c.
lib: *printf.
mm: kfree.
No, I mean "is someone calling these functions with NULL".
Cheers,
Rusty.
Quoted 5 lines · Rusty Russell
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Or is someone already doing this?
really has:
Quoted 9 lines
Quoted 3 lines
Quoted 3 lines
kernel:
__wake_up_sync_keyinkernel/sched/core.c.
lib: *printf.
mm: kfree.No, I mean "is someone calling these functions with NULL".
Cheers,
Rusty.
often "kfree(NULL)", that is ok. it will let caller easier using it. also often give 0 size to snprintf, it still let caller easy using.
if we treat EXPORT functions of kallsyms as commonly used (or we want to)
I suggest to give parameter check for them.
thanks.
:-)
Signature · 3 lines
Chen Gang Asianux Corporation