Re: [PATCH 2/2] loongarch/env: fix missing NULL check after kstrdup()

["Jackie Liu" <liu.yun@linux.dev>]

Hi Huacai.


2025年9月18日 15:15, "Huacai Chen" <chenhuacai@kernel.org mailto:chenhuacai@kernel.org?to=%22Huacai%20Chen%22%20%3Cchenhuacai%40kernel.org%3E > 写到:


> 
> Hi, Jackie,
> 
> Your partner has already sent a patch and be NAKed, don't you know?
> 
> https://lore.kernel.org/loongarch/5169d057-28c9-439c-a6b7-1138b4f79e65@xen0n.name/T/#t
> 
> Huacai

Yes, I was just reviewing the code and happened to check this while
investigating other crash issues. It looks like passing NULL to strsep
won’t cause a crash, so this can safely be ignored. Thanks for pointing
it out.

--- 
Jackie Liu

> 
> On Thu, Sep 18, 2025 at 2:45 PM Jackie Liu <liu.yun@linux.dev> wrote:
> 
> > 
> > From: Jackie Liu <liuyun01@kylinos.cn>
> > 
> >  The result of kstrdup() may be NULL when memory allocation fails.
> >  In init_cpu_fullname(), the return value was used directly by strsep()
> >  without validation, which could lead to NULL pointer dereference.
> > 
> >  Add a proper check for the kstrdup() result and return -ENOMEM when
> >  allocation fails.
> > 
> >  Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
> >  ---
> >  arch/loongarch/kernel/env.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> >  diff --git a/arch/loongarch/kernel/env.c b/arch/loongarch/kernel/env.c
> >  index c0a5dc9aeae2..b41a65ec45c3 100644
> >  --- a/arch/loongarch/kernel/env.c
> >  +++ b/arch/loongarch/kernel/env.c
> >  @@ -49,6 +49,8 @@ static int __init init_cpu_fullname(void)
> >  ret = of_property_read_string(root, "model", &model);
> >  if (ret == 0) {
> >  cpuname = kstrdup(model, GFP_KERNEL);
> >  + if (!cpuname)
> >  + return -ENOMEM;
> >  loongson_sysconf.cpuname = strsep(&cpuname, " ");
> >  }
> >  of_node_put(root);
> >  --
> >  2.51.0
> >
>