fix potential integer overflows

Cast fi->interval.numerator to u64 in order to avoid a potential integer overflow. This variable is being used in a context that expects an expression of type u64.

Addresses-Coverity-ID: 1324146 ("Unintentional integer overflow")

Signed-off-by Gustavo A. R. Silva <[email protected]>

---
 drivers/media/i2c/ov9650.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

drivers/media/i2c/ov9650.c +0 -0

index e519f27..c674a49 100644

drivers/media/i2c/ov9650.c +0 -0

 	for (i = 0; i < ARRAY_SIZE(ov965x_intervals); i++) {

Signature · 1 line

2.7.4

Cast len to const u64 in order to avoid a potential integer overflow. This variable is being used in a context that expects an expression of type const u64.

Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

Signed-off-by Gustavo A. R. Silva <[email protected]>

---
 drivers/media/platform/vivid/vivid-cec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

drivers/media/platform/vivid/vivid-cec.c +0 -0

index b55d278..30240ab 100644

drivers/media/platform/vivid/vivid-cec.c +0 -0

 	cec_queue_pin_cec_event(adap, false, ts);
 	ts = ktime_add_us(ts, CEC_TIM_START_BIT_LOW);
 	cec_queue_pin_cec_event(adap, true, ts);

Hi Gustavo,

Quoted 10 lines · Gustavo A. R. Silva

Cast len to const u64 in order to avoid a potential integer overflow. This variable is being used in a context that expects an expression of type const u64.

Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

Signed-off-by Gustavo A. R. Silva <[email protected]>

---
 drivers/media/platform/vivid/vivid-cec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

drivers/media/platform/vivid/vivid-cec.c static void vivid_cec_pin_adap_events(struct · line 87

Quoted 11 lines

drivers/media/platform/vivid/vivid-cec.c +1 -1

index b55d278..30240ab 100644
@@ -83,7 +83,7 @@ static void vivid_cec_pin_adap_events(struct  
 cec_adapter *adap, ktime_t ts,
 	if (adap == NULL)
 		return;
 	ts = ktime_sub_us(ts, (CEC_TIM_START_BIT_TOTAL +
-			       len * 10 * CEC_TIM_DATA_BIT_TOTAL));
+			       (const u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL));

This makes no sense. Certainly the const part is pointless. And given that len is always <= 16 there definitely is no overflow.

I don't really want this cast in the code.

Sorry,

Hi Hans,

Quoting Hans Verkuil <[email protected]>:

Quoted 13 lines

Hi Gustavo,

Quoted 10 lines · Gustavo A. R. Silva

Cast len to const u64 in order to avoid a potential integer overflow. This variable is being used in a context that expects an expression of type const u64.

Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

Signed-off-by Gustavo A. R. Silva <[email protected]>

---
 drivers/media/platform/vivid/vivid-cec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Quoted 16 lines

drivers/media/platform/vivid/vivid-cec.c static void vivid_cec_pin_adap_events(struct · line 87

Quoted 12 lines

drivers/media/platform/vivid/vivid-cec.c +1 -1

index b55d278..30240ab 100644
@@ -83,7 +83,7 @@ static void vivid_cec_pin_adap_events(struct  
 cec_adapter *adap, ktime_t ts,
 	if (adap == NULL)
 		return;
 	ts = ktime_sub_us(ts, (CEC_TIM_START_BIT_TOTAL +
-			       len * 10 * CEC_TIM_DATA_BIT_TOTAL));
+			       (const u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL));

This makes no sense. Certainly the const part is pointless. And given that len is always <= 16 there definitely is no overflow.

Yeah, I understand your point and I know there is no chance of an overflow in this particular case.

Quoted 4 lines

I don't really want this cast in the code.

Sorry,

I'm working through all the Linux kernel Coverity reports, and I thought of sending a patch for this because IMHO it doesn't hurt to give the compiler complete information about the arithmetic in which an expression is intended to be evaluated.

I agree that the _const_ part is a bit odd. What do you think about the cast to u64 alone?

I appreciate your feedback.

Thanks
--
Gustavo

Quoted 47 lines

Hi Hans,

Quoting Hans Verkuil <[email protected]>:

Quoted 13 lines

Hi Gustavo,

Quoted 10 lines · Gustavo A. R. Silva

Cast len to const u64 in order to avoid a potential integer overflow. This variable is being used in a context that expects an expression of type const u64.

Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

Signed-off-by Gustavo A. R. Silva <[email protected]>

---
 drivers/media/platform/vivid/vivid-cec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Quoted 16 lines

drivers/media/platform/vivid/vivid-cec.c static void vivid_cec_pin_adap_events(struct · line 87

Quoted 12 lines

drivers/media/platform/vivid/vivid-cec.c +1 -1

index b55d278..30240ab 100644
@@ -83,7 +83,7 @@ static void vivid_cec_pin_adap_events(struct
 cec_adapter *adap, ktime_t ts,
 	if (adap == NULL)
 		return;
 	ts = ktime_sub_us(ts, (CEC_TIM_START_BIT_TOTAL +
-			       len * 10 * CEC_TIM_DATA_BIT_TOTAL));
+			       (const u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL));

This makes no sense. Certainly the const part is pointless. And given that len is always <= 16 there definitely is no overflow.

Yeah, I understand your point and I know there is no chance of an overflow in this particular case.

Quoted 4 lines

I don't really want this cast in the code.

Sorry,

I'm working through all the Linux kernel Coverity reports, and I thought of sending a patch for this because IMHO it doesn't hurt to give the compiler complete information about the arithmetic in which an expression is intended to be evaluated.

Quoted 2 lines

I agree that the _const_ part is a bit odd. What do you think about the cast to u64 alone?

What happens if you do: ((u64)CEC_TIM_START_BIT_TOTAL +

I think that forces everything else in the expression to be evaluated as u64.

It definitely needs a comment that this fixes a bogus Coverity report.

Quoting Hans Verkuil <[email protected]>:

Quoted 48 lines · Gustavo A. R. Silva

Quoted 47 lines

Hi Hans,

Quoting Hans Verkuil <[email protected]>:

Quoted 13 lines

Hi Gustavo,

Quoted 10 lines · Gustavo A. R. Silva

Cast len to const u64 in order to avoid a potential integer overflow. This variable is being used in a context that expects an expression of type const u64.

Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

Signed-off-by Gustavo A. R. Silva <[email protected]>

---
 drivers/media/platform/vivid/vivid-cec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Quoted 16 lines

drivers/media/platform/vivid/vivid-cec.c static void vivid_cec_pin_adap_events(struct · line 87

Quoted 12 lines

drivers/media/platform/vivid/vivid-cec.c +1 -1

index b55d278..30240ab 100644
@@ -83,7 +83,7 @@ static void vivid_cec_pin_adap_events(struct
 cec_adapter *adap, ktime_t ts,
 	if (adap == NULL)
 		return;
 	ts = ktime_sub_us(ts, (CEC_TIM_START_BIT_TOTAL +
-			       len * 10 * CEC_TIM_DATA_BIT_TOTAL));
+			       (const u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL));

This makes no sense. Certainly the const part is pointless. And given that len is always <= 16 there definitely is no overflow.

Yeah, I understand your point and I know there is no chance of an overflow in this particular case.

Quoted 4 lines

I don't really want this cast in the code.

Sorry,

I'm working through all the Linux kernel Coverity reports, and I thought of sending a patch for this because IMHO it doesn't hurt to give the compiler complete information about the arithmetic in which an expression is intended to be evaluated.

Quoted 8 lines

Quoted 2 lines

I agree that the _const_ part is a bit odd. What do you think about the cast to u64 alone?

What happens if you do: ((u64)CEC_TIM_START_BIT_TOTAL +

I think that forces everything else in the expression to be evaluated as u64.

Well, in this case the operator precedence takes place and the expression len * 10 * CEC_TIM_DATA_BIT_TOTAL is computed first. So the issue remains the same.

I can switch the expressions as follows:

(u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL + CEC_TIM_START_BIT_TOTAL

and avoid the cast in the middle.

What do you think?

Quoted 3 lines

It definitely needs a comment that this fixes a bogus Coverity report.

I actually added the following line to the message changelog:
Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

Certainly, I've run across multiple false positives as in this case, but I have also fixed many actual bugs thanks to the Coverity reports. So I think in general it is valuable to take a look into these reports, either if they spot actual bugs or promote code correctness.

Quoted 10 lines

Quoted 5 lines

What happens if you do: ((u64)CEC_TIM_START_BIT_TOTAL +

I think that forces everything else in the expression to be evaluated as u64.

Well, in this case the operator precedence takes place and the expression len * 10 * CEC_TIM_DATA_BIT_TOTAL is computed first. So the issue remains the same.

Quoted 3 lines

I can switch the expressions as follows:

(u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL + CEC_TIM_START_BIT_TOTAL

What about:

10ULL * len * ...

Quoted 4 lines

and avoid the cast in the middle.

What do you think?

My problem is that (u64)len suggests that there is some problem with len specifically, which isn't true.

Quoted 6 lines

Quoted 3 lines

It definitely needs a comment that this fixes a bogus Coverity report.

I actually added the following line to the message changelog:
Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

That needs to be in the source, otherwise someone will remove the cast (or ULL) at some time in the future since it isn't clear why it is done. And nobody reads commit logs from X years back :-)

Quoting Hans Verkuil <[email protected]>:

[...]

Quoted 10 lines

Quoted 10 lines

Quoted 5 lines

What happens if you do: ((u64)CEC_TIM_START_BIT_TOTAL +

I think that forces everything else in the expression to be evaluated as u64.

Well, in this case the operator precedence takes place and the expression len * 10 * CEC_TIM_DATA_BIT_TOTAL is computed first. So the issue remains the same.

Quoted 8 lines

Quoted 3 lines

I can switch the expressions as follows:

(u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL + CEC_TIM_START_BIT_TOTAL

What about:

10ULL * len * ...

Yeah, I like it.

Quoted 8 lines

Quoted 4 lines

and avoid the cast in the middle.

What do you think?

My problem is that (u64)len suggests that there is some problem with len specifically, which isn't true.

That's a good point. Actually, I think the same applies for the rest of the patch series. Maybe it is a good idea to send a v2 of the whole patchset with that update.

Quoted 11 lines

Quoted 6 lines

Quoted 3 lines

It definitely needs a comment that this fixes a bogus Coverity report.

I actually added the following line to the message changelog:
Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

That needs to be in the source, otherwise someone will remove the cast (or ULL) at some time in the future since it isn't clear why it is done. And nobody reads commit logs from X years back :-)

You're right. I thought you were talking about the changelog.

And unless you think otherwise, I think there is no need for any additional code comment if the update you suggest is applied:

len * 10ULL * CEC_TIM_DATA_BIT_TOTAL

Quoted 53 lines

Quoting Hans Verkuil <[email protected]>:

[...]

Quoted 10 lines

Quoted 10 lines

Quoted 5 lines

What happens if you do: ((u64)CEC_TIM_START_BIT_TOTAL +

I think that forces everything else in the expression to be evaluated as u64.

Well, in this case the operator precedence takes place and the expression len * 10 * CEC_TIM_DATA_BIT_TOTAL is computed first. So the issue remains the same.

Quoted 8 lines

Quoted 3 lines

I can switch the expressions as follows:

(u64)len * 10 * CEC_TIM_DATA_BIT_TOTAL + CEC_TIM_START_BIT_TOTAL

What about:

10ULL * len * ...

Yeah, I like it.

Quoted 8 lines

Quoted 4 lines

and avoid the cast in the middle.

What do you think?

My problem is that (u64)len suggests that there is some problem with len specifically, which isn't true.

That's a good point. Actually, I think the same applies for the rest of the patch series. Maybe it is a good idea to send a v2 of the whole patchset with that update.

Quoted 11 lines

Quoted 6 lines

Quoted 3 lines

It definitely needs a comment that this fixes a bogus Coverity report.

I actually added the following line to the message changelog:
Addresses-Coverity-ID: 1454996 ("Unintentional integer overflow")

That needs to be in the source, otherwise someone will remove the cast (or ULL) at some time in the future since it isn't clear why it is done. And nobody reads commit logs from X years back :-)

You're right. I thought you were talking about the changelog.

Quoted 4 lines

And unless you think otherwise, I think there is no need for any additional code comment if the update you suggest is applied:

len * 10ULL * CEC_TIM_DATA_BIT_TOTAL

I still think I'd like to see a comment. It is still not obvious why you would want to use ULL here.

Please use "10ULL * len", it's actually a bit better to have it in that order (it's 10 bits per character, so '10 * len' is a more logical order).