Skip to contentSkip to conversation

common/qat: fix uninitialized variable bug

PATCH Applied Adam Dybkowski 4 participants 6 messages Jul 24, 2020 – Jul 29, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 common/qat: fix uninitialized variable bug

This patch fixes the uninitialized variable bug in QAT PMD.

Fixes 9f27a860dc16 ("crypto/qat: move generic qp function to qp file")
---
 drivers/common/qat/qat_qp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/common/qat/qat_qp.c +1 -1
index aacd4ab21..0ee713955 100644
@@ -582,7 +582,7 @@ qat_enqueue_op_burst(void *qp, void **ops, uint16_t nb_ops)
 	register struct qat_queue *queue;
 	struct qat_qp *tmp_qp = (struct qat_qp *)qp;
 	register uint32_t nb_ops_sent = 0;
-	register int ret;
+	register int ret = -1;
 	uint16_t nb_ops_possible = nb_ops;
 	register uint8_t *base_addr;
 	register uint32_t tail;
Signature · 1 line
2.25.1
#1 lore raw
Quoted 23 lines

-----Original Message-----
From: Dybkowski, AdamX <[email protected]>
Sent: Friday, July 24, 2020 10:40 AM
To: [email protected]; Trahe, Fiona <[email protected]>; [email protected]
Cc: Dybkowski, AdamX <[email protected]>
Subject: [PATCH] common/qat: fix uninitialized variable bug

This patch fixes the uninitialized variable bug in QAT PMD.

Fixes 9f27a860dc16 ("crypto/qat: move generic qp function to qp file")
---
 drivers/common/qat/qat_qp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/common/qat/qat_qp.c +0 -0
index aacd4ab21..0ee713955 100644

drivers/common/qat/qat_qp.c qat_enqueue_op_burst(void *qp, void **ops, uint16_t nb_ops) · line 585

Quoted 5 lines
drivers/common/qat/qat_qp.c +1 -1
@@ -582,7 +582,7 @@ qat_enqueue_op_burst(void *qp, void **ops, uint16_t nb_ops)
 	register struct qat_queue *queue;
 	struct qat_qp *tmp_qp = (struct qat_qp *)qp;
 	register uint32_t nb_ops_sent = 0;
-	register int ret;
+	register int ret = -1;

Nack - this fn returns an unsigned. So the correct option is to default to 0

#2 lore raw
Quoted 34 lines

-----Original Message-----
From: Trahe, Fiona <[email protected]>
Sent: Friday, 24 July, 2020 13:55
To: Dybkowski, AdamX <[email protected]>; [email protected]; [email protected]
Cc: Trahe, Fiona <[email protected]>
Subject: RE: [PATCH] common/qat: fix uninitialized variable bug

drivers/common/qat/qat_qp.c qat_enqueue_op_burst(void *qp, void **ops,

Quoted 23 lines

-----Original Message-----
From: Dybkowski, AdamX <[email protected]>
Sent: Friday, July 24, 2020 10:40 AM
To: [email protected]; Trahe, Fiona <[email protected]>; [email protected]
Cc: Dybkowski, AdamX <[email protected]>
Subject: [PATCH] common/qat: fix uninitialized variable bug

This patch fixes the uninitialized variable bug in QAT PMD.

Fixes 9f27a860dc16 ("crypto/qat: move generic qp function to qp

file")

---
 drivers/common/qat/qat_qp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/common/qat/qat_qp.c +0 -0
index aacd4ab21..0ee713955 100644
@@ -582,7 +582,7 @@ qat_enqueue_op_burst(void *qp, void **ops,

uint16_t nb_ops)

Quoted 6 lines

drivers/common/qat/qat_qp.c qat_enqueue_op_burst(void *qp, void **ops, uint16_t nb_ops) · line 585

Quoted 5 lines
drivers/common/qat/qat_qp.c +1 -1
 	register struct qat_queue *queue;
 	struct qat_qp *tmp_qp = (struct qat_qp *)qp;
 	register uint32_t nb_ops_sent = 0;
-	register int ret;
+	register int ret = -1;

Nack - this fn returns an unsigned. So the correct option is to default to 0

[Adam] The ret variable value (signed) is not returned directly, please check the rest of this function in src code. This is just checked to calculate how many ops were enqueued. And if all checks skip (meaning the op was not processed by sym crypto, asym crypto nor compression), we should note the user that the actual op was NOT enqueued. That's why ret is set to -1.

Adam

Acked-by
#3 lore raw
Quoted 49 lines

-----Original Message-----
From: Dybkowski, AdamX <[email protected]>
Sent: Friday, July 24, 2020 12:58 PM
To: Trahe, Fiona <[email protected]>; [email protected]; [email protected]
Subject: RE: [PATCH] common/qat: fix uninitialized variable bug

Quoted 34 lines

-----Original Message-----
From: Trahe, Fiona <[email protected]>
Sent: Friday, 24 July, 2020 13:55
To: Dybkowski, AdamX <[email protected]>; [email protected]; [email protected]
Cc: Trahe, Fiona <[email protected]>
Subject: RE: [PATCH] common/qat: fix uninitialized variable bug

drivers/common/qat/qat_qp.c qat_enqueue_op_burst(void *qp, void **ops,

Quoted 23 lines

-----Original Message-----
From: Dybkowski, AdamX <[email protected]>
Sent: Friday, July 24, 2020 10:40 AM
To: [email protected]; Trahe, Fiona <[email protected]>; [email protected]
Cc: Dybkowski, AdamX <[email protected]>
Subject: [PATCH] common/qat: fix uninitialized variable bug

This patch fixes the uninitialized variable bug in QAT PMD.

Fixes 9f27a860dc16 ("crypto/qat: move generic qp function to qp

file")

---
 drivers/common/qat/qat_qp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/common/qat/qat_qp.c +0 -0
index aacd4ab21..0ee713955 100644
@@ -582,7 +582,7 @@ qat_enqueue_op_burst(void *qp, void **ops,

uint16_t nb_ops)

Quoted 6 lines

drivers/common/qat/qat_qp.c qat_enqueue_op_burst(void *qp, void **ops, uint16_t nb_ops) · line 585

Quoted 5 lines
drivers/common/qat/qat_qp.c +1 -1
 	register struct qat_queue *queue;
 	struct qat_qp *tmp_qp = (struct qat_qp *)qp;
 	register uint32_t nb_ops_sent = 0;
-	register int ret;
+	register int ret = -1;

Nack - this fn returns an unsigned. So the correct option is to default to 0

Quoted 4 lines

[Adam] The ret variable value (signed) is not returned directly, please check the rest of this function in src code. This is just checked to calculate how many ops were enqueued. And if all checks skip (meaning the op was not processed by sym crypto, asym crypto nor compression), we should note the user that the actual op was NOT enqueued. That's why ret is set to -1.

[Fiona] ok. makes sense thanks. In that case

Applied
#4 lore raw
Quoted 10 lines
Quoted 2 lines

[Adam] The ret variable value (signed) is not returned directly, please check the

rest of this function in src

Quoted 1 line

code. This is just checked to calculate how many ops were enqueued. And if all

checks skip (meaning the

Quoted 1 line

op was not processed by sym crypto, asym crypto nor compression), we

should note the user that the

Quoted 1 line

actual op was NOT enqueued. That's why ret is set to -1.

[Fiona] ok. makes sense thanks. In that case

Applied to dpdk-next-crypto

Thanks

#5 lore raw

26/07/2020 21:19, Akhil Goyal:

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

[Adam] The ret variable value (signed) is not returned directly, please check the

rest of this function in src

Quoted 1 line

code. This is just checked to calculate how many ops were enqueued. And if all

checks skip (meaning the

Quoted 1 line

op was not processed by sym crypto, asym crypto nor compression), we

Quoted 8 lines
Quoted 4 lines

should note the user that the

Quoted 1 line

actual op was NOT enqueued. That's why ret is set to -1.

[Fiona] ok. makes sense thanks. In that case

Applied to dpdk-next-crypto

This patch should not have been merged.
The explanation is missing.

This is the commit log:
"This patch fixes the uninitialized variable bug in QAT PMD."

We don't even know what is the consequence and the scope. That's not acceptable for a fix.