netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] bnx2fc: reduce stack usage in __bnx2fc_enable
@ 2015-10-07 13:11 Arnd Bergmann
  2015-10-07 13:28 ` Maurizio Lombardi
  0 siblings, 1 reply; 3+ messages in thread
From: Arnd Bergmann @ 2015-10-07 13:11 UTC (permalink / raw)
  To: linux-scsi
  Cc: QLogic-Storage-Upstream, James E.J. Bottomley, linux-kernel,
	linux-arm-kernel, Joe Carnuccio, Chad Dupuis, Yuval Mintz,
	David S. Miller, netdev

When the bnx2fc driver was changed to read the npiv table from
nvram, the stack of the __bnx2fc_enable function gained an
additional 1028 byte structure that gcc rightfully warns about:

drivers/scsi/bnx2fc/bnx2fc_fcoe.c: In function '__bnx2fc_enable':
drivers/scsi/bnx2fc/bnx2fc_fcoe.c:2134:1: warning: the frame size of 1128 bytes is larger than 1024 bytes [-Wframe-larger-than=]

In order to avoid a possible kernel stack overflow and to get rid
of the warning, this changes the function to use a dynamic allocation
of the structure using kzalloc.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 2971ff67bd3 ("bnx2fc: Read npiv table from nvram and create vports.")
---
The original patch was merged through netdev along with the respective
ethernet driver changes, but this should probably go through the scsi
tree, as no ethernet changes are involved.

diff --git a/drivers/scsi/bnx2fc/bnx2fc_fcoe.c b/drivers/scsi/bnx2fc/bnx2fc_fcoe.c
index d5cdc4776707..72ba0dfae3e0 100644
--- a/drivers/scsi/bnx2fc/bnx2fc_fcoe.c
+++ b/drivers/scsi/bnx2fc/bnx2fc_fcoe.c
@@ -2092,7 +2092,7 @@ static int __bnx2fc_enable(struct fcoe_ctlr *ctlr)
 {
 	struct bnx2fc_interface *interface = fcoe_ctlr_priv(ctlr);
 	struct bnx2fc_hba *hba;
-	struct cnic_fc_npiv_tbl npiv_tbl;
+	struct cnic_fc_npiv_tbl *npiv_tbl;
 	struct fc_lport *lport;
 
 	if (interface->enabled == false) {
@@ -2124,11 +2124,16 @@ static int __bnx2fc_enable(struct fcoe_ctlr *ctlr)
 	if (!hba->cnic->get_fc_npiv_tbl)
 		goto done;
 
-	memset(&npiv_tbl, 0, sizeof(npiv_tbl));
-	if (hba->cnic->get_fc_npiv_tbl(hba->cnic, &npiv_tbl))
+	npiv_tbl = kzalloc(sizeof(struct cnic_fc_npiv_tbl), GFP_KERNEL);
+	if (!npiv_tbl)
 		goto done;
 
-	bnx2fc_npiv_create_vports(lport, &npiv_tbl);
+	if (hba->cnic->get_fc_npiv_tbl(hba->cnic, npiv_tbl))
+		goto done_free;
+
+	bnx2fc_npiv_create_vports(lport, npiv_tbl);
+done_free:
+	kfree(npiv_tbl);
 done:
 	return 0;
 }

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH] bnx2fc: reduce stack usage in __bnx2fc_enable
  2015-10-07 13:11 [PATCH] bnx2fc: reduce stack usage in __bnx2fc_enable Arnd Bergmann
@ 2015-10-07 13:28 ` Maurizio Lombardi
  2015-10-07 18:49   ` Chad Dupuis
  0 siblings, 1 reply; 3+ messages in thread
From: Maurizio Lombardi @ 2015-10-07 13:28 UTC (permalink / raw)
  To: Arnd Bergmann, linux-scsi
  Cc: QLogic-Storage-Upstream, James E.J. Bottomley, linux-kernel,
	linux-arm-kernel, Joe Carnuccio, Chad Dupuis, Yuval Mintz,
	David S. Miller, netdev

Hi,

On 10/07/2015 03:11 PM, Arnd Bergmann wrote:
> -	memset(&npiv_tbl, 0, sizeof(npiv_tbl));
> -	if (hba->cnic->get_fc_npiv_tbl(hba->cnic, &npiv_tbl))
> +	npiv_tbl = kzalloc(sizeof(struct cnic_fc_npiv_tbl), GFP_KERNEL);
> +	if (!npiv_tbl)
>  		goto done;
>  

If kzalloc() fails perhaps the function should return -ENOMEM, not zero.


Regards,
Maurizio Lombardi

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] bnx2fc: reduce stack usage in __bnx2fc_enable
  2015-10-07 13:28 ` Maurizio Lombardi
@ 2015-10-07 18:49   ` Chad Dupuis
  0 siblings, 0 replies; 3+ messages in thread
From: Chad Dupuis @ 2015-10-07 18:49 UTC (permalink / raw)
  To: Maurizio Lombardi
  Cc: Arnd Bergmann, linux-scsi, QLogic-Storage-Upstream,
	James E.J. Bottomley, linux-kernel, linux-arm-kernel,
	Joe Carnuccio, Yuval Mintz, David S. Miller, netdev



On Wed, 7 Oct 2015, Maurizio Lombardi wrote:

> Hi,
>
> On 10/07/2015 03:11 PM, Arnd Bergmann wrote:
>> -	memset(&npiv_tbl, 0, sizeof(npiv_tbl));
>> -	if (hba->cnic->get_fc_npiv_tbl(hba->cnic, &npiv_tbl))
>> +	npiv_tbl = kzalloc(sizeof(struct cnic_fc_npiv_tbl), GFP_KERNEL);
>> +	if (!npiv_tbl)
>>  		goto done;
>>
>
> If kzalloc() fails perhaps the function should return -ENOMEM, not zero.
>

The enablement of the fcoe interface can still proceed if this particular 
allocation fails so returning 0 would be appropriate here.

This patch looks good to me.

>
> Regards,
> Maurizio Lombardi
>

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2015-10-07 18:50 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-10-07 13:11 [PATCH] bnx2fc: reduce stack usage in __bnx2fc_enable Arnd Bergmann
2015-10-07 13:28 ` Maurizio Lombardi
2015-10-07 18:49   ` Chad Dupuis

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).