[PATCH skeleton v5 07/19] fix info memory leak

[Vishwanatha Subbanna]

I remember reviewing a very similar code ( or perhaps the same ) in the
past. Is it a new patch now ?

Here is what I had sent before

a close(fd) on open failure may lead to unexpected behavior since the file
is not even opened.

Also, the pattern seems to be multiple error paths with same set of
instructions. Does it make sense to put them in a macro -or- better yet
have fd auto cleaned using the cleanup attribute ?

https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html

----------------------------------------------

"openbmc" <openbmc-bounces+vishwanath=in.ibm.com at lists.ozlabs.org> wrote on
11/04/2016 06:50:36 am:

> From: OpenBMC Patches <openbmc-patches at stwcx.xyz>
> To: openbmc at lists.ozlabs.org
> Cc: Ken <ken1029 at gmail.com>
> Date: 11/04/2016 06:51 am
> Subject: [PATCH skeleton v5 07/19] fix info memory leak
> Sent by: "openbmc" <openbmc-bounces
+vishwanath=in.ibm.com at lists.ozlabs.org>
>
> From: Ken <ken1029 at gmail.com>
>
> ---
>  objects/info.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/objects/info.c b/objects/info.c
> index ab2d98c..d872c3f 100644
> --- a/objects/info.c
> +++ b/objects/info.c
> @@ -42,6 +42,7 @@ static int i2c_open() {
>    fd = open(fn, O_RDWR);
>    if (fd == -1) {
>      LOG_ERR(errno, "Failed to open i2c device %s", fn);
> +    close(fd);

I remember giving this same comment back the.. There is nothing to close if
open itself fails.

>      return -1;
>    }
>
> @@ -108,6 +109,7 @@ static int i2c_io(int fd) {
>    rc = ioctl(fd, I2C_RDWR, &data);
>    if (rc < 0) {
>      LOG_ERR(errno, "Failed to do raw io");
> +    close(fd);
>      return -1;
>    }
>
> @@ -126,6 +128,7 @@ int get_hdd_status(void)
>    }
>
>    if (i2c_io(fd) < 0) {
> +    close(fd);
>      return -1;
>    }
>
> @@ -145,7 +148,7 @@ int get_hdd_status(void)
>
>      g_read_tmp[2]=g_read_bytes[2];
>      g_read_tmp[3]=g_read_bytes[3];
> -
> +    close(fd);
>  }
>
>  int send_esel_to_dbus(const char *desc, const char *sev, const char
> *details, uint8_t *debug, size_t debuglen) {
> --
> 2.7.1
>
>
> _______________________________________________
> openbmc mailing list
> openbmc at lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/openbmc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/openbmc/attachments/20160411/0e17cb80/attachment.html>