Re: memory leak in do_epoll_create

Re: memory leak in do_epoll_create

[Eric Biggers]

On Fri, May 19, 2023 at 04:30:26PM +0800, 范俊杰 wrote:
> Hi Kernel maintainers,
> 
> Our tool found a new bug memory leak in do_epoll_create in Kernel commit
> v5.14.
> 

v5.14 is almost 2 years old.  Why are you testing such an old kernel version?
This bug could have already been fixed almost 2 years ago.

Also, if you think this is a bug in eventpoll, this report should be sent to
linux-fsdevel, as per './scripts/get_maintainer.pl fs/eventpoll.c'.  It's
unclear why you are sending this report to linux-scsi.

> The report is as below and this bug don't have a repro C program until
> now. Please inform me if you confirm this is a reproducible bug.

I think you answered your own question.  It doesn't have a reproducer;
therefore, it's not reproducible.

- Eric

Re: memory leak in do_epoll_create

[范俊杰]

Thank you for your response. This is my first time submitting crashes to kernel developers, so forgive me if there are any shortcomings. In my opinion, some of the code crashes in the old version may also be present in the new version. That’s why I want to report these crash to you. I will take note of the issues you mentioned and make a meaningful contribution by submitting valid kernel errors next time.!
Sincerely!
> 2023年5月20日 02:57，Eric Biggers <ebiggers@kernel.org> 写道：
> 
> On Fri, May 19, 2023 at 04:30:26PM +0800, 范俊杰 wrote:
>> Hi Kernel maintainers,
>> 
>> Our tool found a new bug memory leak in do_epoll_create in Kernel commit
>> v5.14.
>> 
> 
> v5.14 is almost 2 years old.  Why are you testing such an old kernel version?
> This bug could have already been fixed almost 2 years ago.
> 
> Also, if you think this is a bug in eventpoll, this report should be sent to
> linux-fsdevel, as per './scripts/get_maintainer.pl fs/eventpoll.c'.  It's
> unclear why you are sending this report to linux-scsi.
> 
>> The report is as below and this bug don't have a repro C program until
>> now. Please inform me if you confirm this is a reproducible bug.
> 
> I think you answered your own question.  It doesn't have a reproducer;
> therefore, it's not reproducible.
> 
> - Eric