linux-kernel - Re: [PATCH] lightnvm: fix memory leak in pblk_luns

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Fri, 23 Feb 2018 09:56:18 +0100
From:   Javier González <jg@...htnvm.io>
To:     Matias Bjørling <mb@...htnvm.io>
Cc:     Huaicheng Li <huaicheng@...uchicago.edu>,
        linux-block@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] lightnvm: fix memory leak in pblk_luns_init

> On 23 Feb 2018, at 07.12, Matias Bjørling <mb@...htnvm.io> wrote:
> 
> On 02/23/2018 12:03 AM, Huaicheng Li wrote:
>> Please ignore my previous email as I found the memory is free'ed at
>> pblk_init()'s error handling logic.
>> Sorry for the interruption.
>> On Thu, Feb 22, 2018 at 3:01 PM, Huaicheng Li <huaicheng@...uchicago.edu>
>> wrote:
>>> Signed-off-by: Huaicheng Li <huaicheng@...uchicago.edu>
>>> ---
>>> drivers/lightnvm/pblk-init.c | 1 +
>>> 1 file changed, 1 insertion(+)
>>> 
>>> diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
>>> index 93d671ca518e..330665d91d8d 100644
>>> --- a/drivers/lightnvm/pblk-init.c
>>> +++ b/drivers/lightnvm/pblk-init.c
>>> @@ -510,6 +510,7 @@ static int pblk_luns_init(struct pblk *pblk, struct
>>> ppa_addr *luns)
>>>                 if (ret) {
>>>                         while (--i >= 0)
>>>                                 kfree(pblk->luns[i].bb_list);
>>> +                       kfree(pblk->luns);
>>>                         return ret;
>>>                 }
>>>         }
>>> --
>>> 2.13.6
> 
> No worries. The initialization part is pretty complex due to all the
> data structures being set up. I would love to have a clean
> initialization function, which cleans up after it self. But being able
> to share the "fail" bringup and tear-down code is very helpful and
> makes other parts easier.

As part of the 2.0 patches, I took a first try at refactoring the
init/exit sequences to maintain the same fail paths but minimizing
dependencies across structures. One of the big parts have been the bad
block initialization, which now is much cleaner, but still needs more
work to be as I would like it to. Patches are very welcome :)

Javier

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)