2013-08-13 04:49:38

by Nilanjan Roychowdhury

Subject: [PATCH 1/1] Resubmit mtd: mtdoops: fix for a potential memory leak in mtdoops_notify_remove

we are allocating cxt->oops_page_used using vmalloc in mtdoops_notify_add for
every mtd_info addition but not freeing it in mtdoops_notify_remove. Also care is
taken so that we do not free the same pointer again in module remove.

Signed-off-by: Nilanjan Roychowdhury <[email protected]>
---
drivers/mtd/mtdoops.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/mtd/mtdoops.c b/drivers/mtd/mtdoops.c
index 97bb8f6..4b8b621 100644
--- a/drivers/mtd/mtdoops.c
+++ b/drivers/mtd/mtdoops.c
@@ -386,6 +386,8 @@ static void mtdoops_notify_remove(struct mtd_info *mtd)
cxt->mtd = NULL;
flush_work(&cxt->work_erase);
flush_work(&cxt->work_write);
+ vfree(cxt->oops_page_used);
+ cxt->oops_page_used = NULL;
}


@@ -439,7 +441,8 @@ static void __exit mtdoops_exit(void)

unregister_mtd_user(&mtdoops_notifier);
vfree(cxt->oops_buf);
- vfree(cxt->oops_page_used);
+ if (cxt->oops_page_used)
+ vfree(cxt->oops_page_used);
}


--
1.7.9.5

2013-08-13 14:50:54

by Ezequiel Garcia

Subject: Re: [PATCH 1/1] Resubmit mtd: mtdoops: fix for a potential memory leak in mtdoops_notify_remove

Hi Nilanjan,

On Mon, Aug 12, 2013 at 09:48:35PM -0700, Nilanjan Roychowdhury wrote:
> we are allocating cxt->oops_page_used using vmalloc in mtdoops_notify_add for
> every mtd_info addition but not freeing it in mtdoops_notify_remove. Also care is
> taken so that we do not free the same pointer again in module remove.
>

Could you re-send this with a better commit message?

In particular we'd like to see why you're doing this, explaining
how the current implementation implies a possible leak, explaining under
what circumstance this leak.

I think it would be great if you could test the patch and also add
some information about the testing scenario.

Please note the subject should be "[PATCH v2] mtd: ...", with a
proper patch version number. The next patch will be v3 (and so on) which
should be stated in the subject.

Also...

> Signed-off-by: Nilanjan Roychowdhury <[email protected]>
> ---

In this place (between the "---" and the diffstat) you can put some
changelog information, like this:

Changes from v2:
* Did this and that.

Changes from v1:
* Added check to vfree() in remove function as suggested by Ezequiel.

> drivers/mtd/mtdoops.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mtd/mtdoops.c b/drivers/mtd/mtdoops.c
> index 97bb8f6..4b8b621 100644
> --- a/drivers/mtd/mtdoops.c
> +++ b/drivers/mtd/mtdoops.c
> @@ -386,6 +386,8 @@ static void mtdoops_notify_remove(struct mtd_info *mtd)
> cxt->mtd = NULL;
> flush_work(&cxt->work_erase);
> flush_work(&cxt->work_write);
> + vfree(cxt->oops_page_used);
> + cxt->oops_page_used = NULL;
> }
>
>
> @@ -439,7 +441,8 @@ static void __exit mtdoops_exit(void)
>
> unregister_mtd_user(&mtdoops_notifier);
> vfree(cxt->oops_buf);
> - vfree(cxt->oops_page_used);
> + if (cxt->oops_page_used)
> + vfree(cxt->oops_page_used);
> }
>
>
> --
> 1.7.9.5
>

Thanks!
--
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com