2021-09-02 01:55:18

by Jiasheng Jiang

Subject: [PATCH 9/9] x86/microcode/amd: Add __list_del_entry_valid() in front of __list_del() in free_cache()

We have found that in the complied files __list_del()
appear more than 100 times, and under at least 90% circumstances
that __list_del_entry_valid() and __list_del() appear in pairs.
For example, they appear together in the __list_del_entry()
of the header file, 'include/linux/list.h'.
But we have found that in the free_cache(), there is only
__list_del() instead of the pair.
Therefore, we consider that the __list_del_entry_valid()
might be forgotten.

Signed-off-by: jiasheng <[email protected]>
---
arch/x86/kernel/cpu/microcode/amd.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/arch/x86/kernel/cpu/microcode/amd.c b/arch/x86/kernel/cpu/microcode/amd.c
index 3d4a483..1987ee1 100644
--- a/arch/x86/kernel/cpu/microcode/amd.c
+++ b/arch/x86/kernel/cpu/microcode/amd.c
@@ -626,6 +626,8 @@ static void free_cache(void)
struct ucode_patch *p, *tmp;

list_for_each_entry_safe(p, tmp, &microcode_cache, plist) {
+ if (!__list_del_entry_valid(p->plist))
+ continue;
__list_del(p->plist.prev, p->plist.next);
kfree(p->data);
kfree(p);
--
2.7.4

2021-09-02 10:00:24

by Borislav Petkov

Subject: Re: [PATCH 9/9] x86/microcode/amd: Add __list_del_entry_valid() in front of __list_del() in free_cache()

Hello,

On Thu, Sep 02, 2021 at 01:45:22AM +0000, jiasheng wrote:
> We have found that in the complied files __list_del()
> appear more than 100 times, and under at least 90% circumstances
> that __list_del_entry_valid() and __list_del() appear in pairs.
> For example, they appear together in the __list_del_entry()
> of the header file, 'include/linux/list.h'.
> But we have found that in the free_cache(), there is only
> __list_del() instead of the pair.
> Therefore, we consider that the __list_del_entry_valid()
> might be forgotten.

if this is how you're going to "analyze" whether something is amiss
in the code - by looking at how some other code does something, i.e.,
patterns - and not involve grey matter and actually *think* what you're
doing before doing it and what those functions do, your patches will
simply get ingored.

> Signed-off-by: jiasheng <[email protected]>

When you sign off on a patch, please give your full name.

> ---
> arch/x86/kernel/cpu/microcode/amd.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/arch/x86/kernel/cpu/microcode/amd.c b/arch/x86/kernel/cpu/microcode/amd.c
> index 3d4a483..1987ee1 100644
> --- a/arch/x86/kernel/cpu/microcode/amd.c
> +++ b/arch/x86/kernel/cpu/microcode/amd.c
> @@ -626,6 +626,8 @@ static void free_cache(void)
> struct ucode_patch *p, *tmp;
>
> list_for_each_entry_safe(p, tmp, &microcode_cache, plist) {
> + if (!__list_del_entry_valid(p->plist))
> + continue;
> __list_del(p->plist.prev, p->plist.next);
> kfree(p->data);
> kfree(p);
> --

Also, always CC lkml when submitting patches. CCed now.

Ok, it looks like you're a newbie to this - I'd suggest you read all of
this here:

https://www.kernel.org/doc/html/latest/process/index.html

to get acquainted with how this work is usually done and then try to fix
real bugs first.

HTH.

--
Regards/Gruss,
Boris.

https://people.kernel.org/tglx/notes-about-netiquette