Return-path: Received: from esa4.microchip.iphmx.com ([68.232.154.123]:18410 "EHLO esa4.microchip.iphmx.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750878AbeC0FLC (ORCPT ); Tue, 27 Mar 2018 01:11:02 -0400 Date: Tue, 27 Mar 2018 10:40:54 +0530 From: Ajay Singh To: Colin King CC: Aditya Shankar , Ganesh Krishna , Greg Kroah-Hartman , , , , Subject: Re: [PATCH] staging: wilc1000: replace kmalloc + memcpy with kmemdup Message-ID: <20180327104054.69479b47@ajaysk-VirtualBox> (sfid-20180327_071119_391830_C72D3B0F) In-Reply-To: <20180326171629.28700-1-colin.king@canonical.com> References: <20180326171629.28700-1-colin.king@canonical.com> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Sender: linux-wireless-owner@vger.kernel.org List-ID: On Mon, 26 Mar 2018 18:16:29 +0100 Colin King wrote: > From: Colin Ian King > > Replace several allocation and memcpys with kmemdup and add in some > missing memory allocation failure checks. Also fix an incorrect > -EFAULT return with -ENOMEM. > > Signed-off-by: Colin Ian King > --- > drivers/staging/wilc1000/host_interface.c | 75 +++++++++++++++++++------------ > 1 file changed, 46 insertions(+), 29 deletions(-) > > diff --git a/drivers/staging/wilc1000/host_interface.c b/drivers/staging/wilc1000/host_interface.c > index 9b9b86654958..8fd367f87fa5 100644 > --- a/drivers/staging/wilc1000/host_interface.c > +++ b/drivers/staging/wilc1000/host_interface.c > @@ -797,6 +797,10 @@ static s32 handle_scan(struct wilc_vif *vif, struct scan_attr *scan_info) > for (i = 0; i < hidden_net->n_ssids; i++) > valuesize += ((hidden_net->net_info[i].ssid_len) + 1); > hdn_ntwk_wid_val = kmalloc(valuesize + 1, GFP_KERNEL); > + if (!hdn_ntwk_wid_val) { > + result = -ENOMEM; > + goto error; > + } Please do not apply this changes. It will change the code flow differently. Check for NULl value in '(wid_list[index].val)' is already presented. It has to proceed with the below flow instead of returning from there. > wid_list[index].val = hdn_ntwk_wid_val; > if (wid_list[index].val) { > buffer = wid_list[index].val; > @@ -943,39 +947,35 @@ static s32 handle_connect(struct wilc_vif *vif, > } > > if (conn_attr->bssid) { > - hif_drv->usr_conn_req.bssid = kmalloc(6, GFP_KERNEL); > + hif_drv->usr_conn_req.bssid = kmemdup(conn_attr->bssid, 6, > + GFP_KERNEL); > if (!hif_drv->usr_conn_req.bssid) { > result = -ENOMEM; > goto error; > } > - memcpy(hif_drv->usr_conn_req.bssid, conn_attr->bssid, 6); > } > > hif_drv->usr_conn_req.ssid_len = conn_attr->ssid_len; > if (conn_attr->ssid) { > - hif_drv->usr_conn_req.ssid = kmalloc(conn_attr->ssid_len + 1, > + hif_drv->usr_conn_req.ssid = kmemdup(conn_attr->ssid, > + conn_attr->ssid_len + 1, > GFP_KERNEL); Sorry, I too missed to see that scenario. As suggested, kmemdup can not be used directly to replace kmalloc & memcpy in this case. The size used for kmalloc is not equal to size of data copy in memcpy i.e kmalloc is done for 1 byte extra to keep the NULL character. The direct replacement of kmalloc with kmemdup is not applicable here. > if (!hif_drv->usr_conn_req.ssid) { > result = -ENOMEM; > goto error; > } > - memcpy(hif_drv->usr_conn_req.ssid, > - conn_attr->ssid, > - conn_attr->ssid_len); > hif_drv->usr_conn_req.ssid[conn_attr->ssid_len] = '\0'; > } > > hif_drv->usr_conn_req.ies_len = conn_attr->ies_len; > if (conn_attr->ies) { > - hif_drv->usr_conn_req.ies = kmalloc(conn_attr->ies_len, > + hif_drv->usr_conn_req.ies = kmemdup(conn_attr->ies, > + conn_attr->ies_len, > GFP_KERNEL); > if (!hif_drv->usr_conn_req.ies) { > result = -ENOMEM; > goto error; > } > - memcpy(hif_drv->usr_conn_req.ies, > - conn_attr->ies, > - conn_attr->ies_len); > } > > hif_drv->usr_conn_req.security = conn_attr->security; > @@ -1009,9 +1009,12 @@ static s32 handle_connect(struct wilc_vif *vif, > > if (memcmp("DIRECT-", conn_attr->ssid, 7)) { > info_element_size = hif_drv->usr_conn_req.ies_len; > - info_element = kmalloc(info_element_size, GFP_KERNEL); > - memcpy(info_element, hif_drv->usr_conn_req.ies, > - info_element_size); > + info_element = kmemdup(hif_drv->usr_conn_req.ies, > + info_element_size, GFP_KERNEL); > + if (!info_element) { > + result = -ENOMEM; > + goto error; > + } > } "info_element" variable was removed in my previous submitted patchset. Those changes are still not included in Greg's staging repo. Few changes in this patch are already included in previous patchset,which might give conflict. But few changes are not present which can be applied like returning -ENOMEM in case of allocation failure. Regards, Ajay