Thread (14 messages) 14 messages, 3 authors, 2021-03-18

Re: [PATCH v2 4/6] qla2xxx: Suppress Coverity complaints about dseg_r*

From: Lee Duncan <lduncan@suse.com>
Date: 2021-03-18 16:01:35

On 3/17/21 8:28 PM, Bart Van Assche wrote:
quoted hunk ↗ jump to hunk
Change dseq_rq and dseg_rsp from scalar structure members into
single-element arrays such that Coverity does not complain about the
(*cur_dsd)++ statement in append_dsd64().

Cc: Quinn Tran <redacted>
Cc: Mike Christie <michael.christie@oracle.com>
Reviewed-by: Daniel Wagner <redacted>
Reviewed-by: Himanshu Madhani <redacted>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/scsi/qla2xxx/qla_mr.c | 12 ++++++------
 drivers/scsi/qla2xxx/qla_mr.h |  4 ++--
 2 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/drivers/scsi/qla2xxx/qla_mr.c b/drivers/scsi/qla2xxx/qla_mr.c
index ca7306685325..61eaf6c4ac47 100644
--- a/drivers/scsi/qla2xxx/qla_mr.c
+++ b/drivers/scsi/qla2xxx/qla_mr.c
@@ -3266,8 +3266,8 @@ qlafx00_fxdisc_iocb(srb_t *sp, struct fxdisc_entry_fx00 *pfxiocb)
 			fx_iocb.req_xfrcnt =
 			    cpu_to_le16(fxio->u.fxiocb.req_len);
 			put_unaligned_le64(fxio->u.fxiocb.req_dma_handle,
-					   &fx_iocb.dseg_rq.address);
-			fx_iocb.dseg_rq.length =
+					   &fx_iocb.dseg_rq[0].address);
+			fx_iocb.dseg_rq[0].length =
 			    cpu_to_le32(fxio->u.fxiocb.req_len);
 		}
 
@@ -3276,8 +3276,8 @@ qlafx00_fxdisc_iocb(srb_t *sp, struct fxdisc_entry_fx00 *pfxiocb)
 			fx_iocb.rsp_xfrcnt =
 			    cpu_to_le16(fxio->u.fxiocb.rsp_len);
 			put_unaligned_le64(fxio->u.fxiocb.rsp_dma_handle,
-					   &fx_iocb.dseg_rsp.address);
-			fx_iocb.dseg_rsp.length =
+					   &fx_iocb.dseg_rsp[0].address);
+			fx_iocb.dseg_rsp[0].length =
 			    cpu_to_le32(fxio->u.fxiocb.rsp_len);
 		}
 
@@ -3314,7 +3314,7 @@ qlafx00_fxdisc_iocb(srb_t *sp, struct fxdisc_entry_fx00 *pfxiocb)
 			    cpu_to_le16(bsg_job->request_payload.sg_cnt);
 			tot_dsds =
 			    bsg_job->request_payload.sg_cnt;
-			cur_dsd = &fx_iocb.dseg_rq;
+			cur_dsd = &fx_iocb.dseg_rq[0];
 			avail_dsds = 1;
 			for_each_sg(bsg_job->request_payload.sg_list, sg,
 			    tot_dsds, index) {
@@ -3369,7 +3369,7 @@ qlafx00_fxdisc_iocb(srb_t *sp, struct fxdisc_entry_fx00 *pfxiocb)
 			fx_iocb.rsp_dsdcnt =
 			   cpu_to_le16(bsg_job->reply_payload.sg_cnt);
 			tot_dsds = bsg_job->reply_payload.sg_cnt;
-			cur_dsd = &fx_iocb.dseg_rsp;
+			cur_dsd = &fx_iocb.dseg_rsp[0];
 			avail_dsds = 1;
 
 			for_each_sg(bsg_job->reply_payload.sg_list, sg,
diff --git a/drivers/scsi/qla2xxx/qla_mr.h b/drivers/scsi/qla2xxx/qla_mr.h
index 73be8348402a..eefbae9d7547 100644
--- a/drivers/scsi/qla2xxx/qla_mr.h
+++ b/drivers/scsi/qla2xxx/qla_mr.h
@@ -176,8 +176,8 @@ struct fxdisc_entry_fx00 {
 	uint8_t flags;-					   &
 	uint8_t reserved_1;
 
-	struct dsd64 dseg_rq;
-	struct dsd64 dseg_rsp;
+	struct dsd64 dseg_rq[1];
+	struct dsd64 dseg_rsp[1];
 
 	__le32 dataword;
 	__le32 adapid;
I dislike such an uncommented change just to keep a tool happy. Could
you add a comment in qla_mr.h saying why these are one-element-long
arrays, just so nobody optimizes those out later?

If you do that, please add my:

Reviewed-by: Lee Duncan <lduncan@suse.com>
-- 
Lee Duncan
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help