Re: [PATCH -next] tracing/osnoise: fix null-ptr-deref in bitmap_parselist

[Wang Liang <wangliang74@huawei.com>]

在 2025/9/6 6:44, Steven Rostedt 写道:
> On Fri, 5 Sep 2025 11:25:44 +0800
> Wang Liang <wangliang74@huawei.com> wrote:
>
> Hi Wang,
>
> FYI, the tracincg subsystem requires capital subject:
>
>    tracing/osnoise: Fix null-ptr-deref in bitmap_parselist()
>
> And parenthesis for function names.


Thanks for your guide, it is clear and helpful!

>> A crash was observed with the following output:
>>
>> BUG: kernel NULL pointer dereference, address: 0000000000000010
>> Oops: Oops: 0000 [#1] SMP NOPTI
>> CPU: 2 UID: 0 PID: 92 Comm: osnoise_cpus Not tainted 6.17.0-rc4-00201-gd69eb204c255 #138 PREEMPT(voluntary)
>> RIP: 0010:bitmap_parselist+0x53/0x3e0
>> Call Trace:
>>   <TASK>
>>   osnoise_cpus_write+0x7a/0x190
>>   vfs_write+0xf8/0x410
>>   ? do_sys_openat2+0x88/0xd0
>>   ksys_write+0x60/0xd0
>>   do_syscall_64+0xa4/0x260
>>   entry_SYSCALL_64_after_hwframe+0x77/0x7f
>>   </TASK>
>>
>> This issue can be reproduced by below code:
>>
>> fd=open("/sys/kernel/debug/tracing/osnoise/cpus", O_WRONLY);
>> write(fd, "0-2", 0);
>>
>> When user pass 'count=0' to osnoise_cpus_write(), kmalloc() will return
>> ZERO_SIZE_PTR (16) and cpulist_parse() treat it as a normal value, which
>> trigger the null pointer dereference. Add check for the parameter 'count'.
>>
>> Fixes: 17f89102fe23 ("tracing/osnoise: Allow arbitrarily long CPU string")
>> Signed-off-by: Wang Liang <wangliang74@huawei.com>
>> ---
>>   kernel/trace/trace_osnoise.c | 3 +++
>>   1 file changed, 3 insertions(+)
>>
>> diff --git a/kernel/trace/trace_osnoise.c b/kernel/trace/trace_osnoise.c
>> index fd259da0aa64..edf5178d0317 100644
>> --- a/kernel/trace/trace_osnoise.c
>> +++ b/kernel/trace/trace_osnoise.c
>> @@ -2322,6 +2322,9 @@ osnoise_cpus_write(struct file *filp, const char __user *ubuf, size_t count,
>>   	int running, err;
>>   	char *buf __free(kfree) = NULL;
>>   
>> +	if (count < 1)
>> +		return -EINVAL;
> Sending a count of 0 is not invalid. This should simply return 0;
>
> -- Steve


Ok. I will correct it and send a v2 patch later.

------
Best regards
Wang Liang

>> +
>>   	buf = kmalloc(count, GFP_KERNEL);
>>   	if (!buf)
>>   		return -ENOMEM;
>