SCSI: aacraid, fix memory leak
Subsystems (4)
PATCH 1/5 SCSI: aacraid, fix memory leak
Stanse found a memory leak on one fail path in aac_send_raw_srb. Add kfree there.
--- drivers/scsi/aacraid/commctrl.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/aacraid/commctrl.c +1 -0
@@ -649,6 +649,7 @@ static int aac_send_raw_srb(struct aac_dev* dev, void __user * arg) AAC_OPT_NEW_COMM) ? (dev->scsi_host_ptr->max_sectors << 9) : 65536) { +kfree (usg); rcode = -EINVAL ; goto cleanup; }
Signature · 1 line
1.6.4.2
PATCH 2/5 SCSI: remove unnecessary NULL test
Stanse found that c3cn is poked many times around in cxgb3i_conn_pdu_ready, there is no need to check if it is NULL.
Remove the test.
--- drivers/scsi/cxgb3i/cxgb3i_pdu.c | 6 ++---- 1 files changed, 2 insertions(+), 4 deletions(-)
drivers/scsi/cxgb3i/cxgb3i_pdu.c +2 -4
@@ -461,10 +461,8 @@ void cxgb3i_conn_pdu_ready(struct s3_conn *c3cn) skb =skb_peek (&c3cn->receive_queue); }read_unlock (&c3cn->callback_lock); -if (c3cn) { - c3cn->copied_seq += read; -cxgb3i_c3cn_rx_credits (c3cn, read); - } + c3cn->copied_seq += read; +cxgb3i_c3cn_rx_credits (c3cn, read); conn->rxdata_octets += read;if (err) {
Signature · 1 line
1.6.4.2
PATCH 3/5 SCSI: lpfc, fix memory leak
In reply to Jiri SlabyStanse found a memory leak on one fail path in lpfc_sli4_read_rev.
Add there kfree.
--- drivers/scsi/lpfc/lpfc_sli.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/lpfc/lpfc_sli.c +1 -0
@@ -4211,6 +4211,7 @@ lpfc_sli4_read_rev(struct lpfc_hba *phba, LPFC_MBOXQ_t *mboxq,if (rc) {dma_free_coherent (&phba->pcidev->dev, dma_size, dmabuf->virt, dmabuf->phys); +kfree (dmabuf);return -EIO ; }
Signature · 1 line
1.6.4.2
PATCH 4/5 SCSI: qla2xxx, fix NULL-ptr dereference
In reply to Jiri SlabyStanse found a NULL-ptr dereference on one fail path in qla25xx_create_rsp_que. It jumps to que_failed label and qla25xx_free_rsp_que there dereferences rsp which is NULL.
Return with ENOMEM directly instead.
--- drivers/scsi/qla2xxx/qla_mid.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/scsi/qla2xxx/qla_mid.c +1 -1
@@ -659,7 +659,7 @@ qla25xx_create_rsp_que(struct qla_hw_data *ha, uint16_t options,if (rsp ==NULL ) {qla_printk (KERN_WARNING , ha,"could not allocate memory for" " response que\n" ); - goto que_failed; +return -ENOMEM ; } rsp->length =RESPONSE_ENTRY_CNT_MQ ;
Signature · 1 line
1.6.4.2
Re: [PATCH 4/5] SCSI: qla2xxx, fix NULL-ptr dereference
Quoted 6 lines · Jiri Slaby
Stanse found a NULL-ptr dereference on one fail path in
qla25xx_create_rsp_que. It jumps toque_failedlabel andqla25xx_free_rsp_quethere dereferences rsp which is NULL.Return with ENOMEM directly instead.
Thanks for pointing it out. There is another code path (create_req_que) where the same bug exists. I will send a patch that takes care of the both.
-Anirban
PATCH 5/5 SCSI: scsi_lib, fix potential NULL dereference
In reply to Jiri SlabyStanse found a potential NULL dereference in scsi_kill_request.
Instead of triggering BUG() in 'if (unlikely(cmd == NULL))' branch, the kernel will Oops earlier on cmd dereference.
Move the dereferences after the if.
--- drivers/scsi/scsi_lib.c | 9 ++++++--- 1 files changed, 6 insertions(+), 3 deletions(-)
drivers/scsi/scsi_lib.c +6 -3
@@ -1359,9 +1359,9 @@ static int scsi_lld_busy(struct request_queue *q)static void scsi_kill_request (struct request *req,struct request_queue *q) {struct scsi_cmnd *cmd = req->special; -struct scsi_device *sdev = cmd->device; -struct scsi_target *starget =scsi_target (sdev); -struct Scsi_Host *shost = sdev->host; +struct scsi_device *sdev; +struct scsi_target *starget; +struct Scsi_Host *shost;blk_start_request (req);
@@ -1371,6 +1371,9 @@ static void scsi_kill_request(struct request *req, struct request_queue *q)BUG (); } + sdev = cmd->device; + starget =scsi_target (sdev); + shost = sdev->host;scsi_init_cmd_errh (cmd); cmd->result =DID_NO_CONNECT <<16 ;atomic_inc (&cmd->device->iorequest_cnt);
Signature · 1 line
1.6.4.2