Re: [PATCH v2] staging: dgnc: check if kzalloc fails in dgnc_tty_init()

[Dan Carpenter <dan.carpenter@oracle.com>]

On Tue, Apr 07, 2015 at 03:40:17PM +0300, Giedrius Statkevičius wrote:
> If one of the allocations of memory for storing a channel information struct
> fails then free all the successful allocations and return -ENOMEM that gets
> propogated to the pci layer.  Also, remove a bogus skipping in the next part of
> the initiation if a previous memory allocation failed because we won't execute
> that if any of the allocations failed.
> 
> Signed-off-by: Giedrius Statkevičius <giedrius.statkevicius@gmail.com>
> ---
> v2: Only returning -ENOMEM if an allocation failed isn't enough as it was
> spotted by Sudip so create a new label that frees all successfully allocated
> stuff and only then returns -ENOMEM. Also, remove a unnecessary check in the
> next loop.
> 
>  drivers/staging/dgnc/dgnc_tty.c | 11 +++++++----
>  1 file changed, 7 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/staging/dgnc/dgnc_tty.c b/drivers/staging/dgnc/dgnc_tty.c
> index ce4187f..60d7e49 100644
> --- a/drivers/staging/dgnc/dgnc_tty.c
> +++ b/drivers/staging/dgnc/dgnc_tty.c
> @@ -316,6 +316,8 @@ int dgnc_tty_init(struct dgnc_board *brd)
>  			 * interrupt context, and there are no locks held.
>  			 */
>  			brd->channels[i] = kzalloc(sizeof(*brd->channels[i]), GFP_KERNEL);
> +			if (!brd->channels[i])
> +				goto err_free_channels;

The comments here say that sometimes brd->channels[] are allocated
earlier.  If that's true then the error handling is not correct.  But
I don't think it is true...  Could you investigate and delete the
comments and unnecessary "if (!brd->channels[i])" NULL check.

regards,
dan carpenter