From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jia-Ju Bai Date: Thu, 29 Mar 2018 08:00:42 +0000 Subject: Re: [PATCH v2] staging: vt6655: check for memory allocation failures Message-Id: <798fbe60-52aa-4fb0-0cd3-e2c067bd6c04@gmail.com> List-Id: References: <1522308157-26463-1-git-send-email-ji_hun.kim@samsung.com> In-Reply-To: <1522308157-26463-1-git-send-email-ji_hun.kim@samsung.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Ji-Hun Kim , gregkh@linuxfoundation.org, forest@alittletooquiet.net Cc: devel@driverdev.osuosl.org, y.k.oh@samsung.com, kernel-janitors@vger.kernel.org, linux-kernel@vger.kernel.org, julia.lawall@lip6.fr, santhameena13@gmail.com On 2018/3/29 15:22, Ji-Hun Kim wrote: > There are no null pointer checking on rd_info and td_info values which > are allocated by kzalloc. It has potential null pointer dereferencing > issues. Add return when allocation is failed. > > Signed-off-by: Ji-Hun Kim > --- > > Change: since v1: > > - Delete WARN_ON which can makes crashes on some machines. > - Instead of return directly, goto freeing function for freeing previously > allocated memory in the for loop after kzalloc() failed. > - In the freeing function, if td_info and rd_info are not allocated, no > needs to free. > > drivers/staging/vt6655/device_main.c | 64 +++++++++++++++++++++++++----------- > 1 file changed, 44 insertions(+), 20 deletions(-) > > diff --git a/drivers/staging/vt6655/device_main.c b/drivers/staging/vt6655/device_main.c > index fbc4bc6..ecbba43 100644 > --- a/drivers/staging/vt6655/device_main.c > +++ b/drivers/staging/vt6655/device_main.c > @@ -539,7 +539,8 @@ static void device_init_rd0_ring(struct vnt_private *priv) > i ++, curr += sizeof(struct vnt_rx_desc)) { > desc = &priv->aRD0Ring[i]; > desc->rd_info = kzalloc(sizeof(*desc->rd_info), GFP_KERNEL); > - > + if (!desc->rd_info) > + goto error; > if (!device_alloc_rx_buf(priv, desc)) > dev_err(&priv->pcid->dev, "can not alloc rx bufs\n"); > > @@ -550,6 +551,10 @@ static void device_init_rd0_ring(struct vnt_private *priv) > if (i > 0) > priv->aRD0Ring[i-1].next_desc = cpu_to_le32(priv->rd0_pool_dma); > priv->pCurrRD[0] = &priv->aRD0Ring[0]; > + > + return; > +error: > + device_free_rd0_ring(priv); > } > I think you should return an error number here, because device_init_rd0_ring() is called by vnt_start(). You should also implement error handling code in vnt_start(), and let vnt_start() returns an error number too. The same for device_init_rd1_ring(), device_init_td0_ring() and device_init_td1_ring(). Best wishes, Jia-Ju Bai