staging: most: video: use min_t() for comparison and assignment
Subsystems (2)
PATCH v2 1/1 staging: most: video: use min_t() for comparison and assignment
Simplify code by using min_t helper macro for logical evaluation and value assignment. Use the _t variant of min macro since the variable types are not same.
This issue is identified by coccicheck using the minmax.cocci file.
---
Changes in v2:
1. Revise patch description. No functional change.
drivers/staging/most/video/video.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/most/video/video.c +1 -1
@@ -173,7 +173,7 @@ static ssize_t comp_vdev_read(struct file *filp, char __user *buf,while (count >0 &&data_ready (mdev)) {struct mbo *const mbo =get_top_mbo (mdev);int const rem = mbo->processed_length - fh->offs; -int const cnt = rem < count ? rem : count; +int const cnt =min_t (int, rem, count);if (copy_to_user (buf, mbo->virt_address + fh->offs, cnt)) {v4l2_err (&mdev->v4l2_dev,"read: copy_to_user failed\n" );
--
2.34.1
Quoted 19 lines · Deepak R Varma
Simplify code by using
min_thelper macro for logical evaluation and value assignment. Use the _t variant of min macro since the variable types are not same.
This issue is identified by coccicheck using theminmax.coccifile.Signed-off-by Deepak R Varma <[email protected]>---
Changes in v2:
1. Revise patch description. No functional change.drivers/staging/most/video/video.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/staging/most/video/video.c +0
-0
drivers/staging/most/video/video.c static ssize_t comp_vdev_read(struct file *filp, char __user *buf, · line 176
Quoted 5 lines
drivers/staging/most/video/video.c +1
-1@@ -173,7 +173,7 @@ static ssize_t comp_vdev_read(struct file *filp, char __user *buf,while (count >0 &&data_ready (mdev)) {struct mbo *const mbo =get_top_mbo (mdev);int const rem = mbo->processed_length - fh->offs; -int const cnt = rem < count ? rem : count; +int const cnt =min_t (int, rem, count);
TL;DR use size_t instead of int.
Using "int" here is wrong. size_t is unsigned long meaning that it has 64 bits to use to represent positive values. (Let's ignore 32 bit arches). You have chopped it down to say that it now has 31 bits for positives and if BIT(31) is set then treat it as negative. Everything which is larger than INT_MAX will be broken.
Fortunately, in this code the value of count will never go higher than "INT_MAX - PAGE_SIZE" because Linus understands that it's easy to introduce bugs like this.
regards,
dan carpenter
Quoted 15 lines · Dan Carpenter
Quoted 14 lines · Deepak R Varma
Simplify code by using
min_thelper macro for logical evaluation and value assignment. Use the _t variant of min macro since the variable types are not same.
This issue is identified by coccicheck using theminmax.coccifile.Signed-off-by Deepak R Varma <[email protected]>---
Changes in v2:
1. Revise patch description. No functional change.drivers/staging/most/video/video.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 12 lines
drivers/staging/most/video/video.c static ssize_t comp_vdev_read(struct file *filp, char __user *buf, · line 176
Quoted 10 lines
drivers/staging/most/video/video.c +1
-1@@ -173,7 +173,7 @@ static ssize_t comp_vdev_read(struct file *filp, char __user *buf,while (count >0 &&data_ready (mdev)) {struct mbo *const mbo =get_top_mbo (mdev);int const rem = mbo->processed_length - fh->offs; -int const cnt = rem < count ? rem : count; +int const cnt =min_t (int, rem, count);TL;DR use
size_tinstead of int.
Hi Dan,
Thank you for reviewing the patch. Please see my queries inline.
Quoted 6 lines
Using "int" here is wrong.
size_tis unsigned long meaning that it has 64 bits to use to represent positive values. (Let's ignore 32 bit arches). You have chopped it down to say that it now has 31 bits for positives and if BIT(31) is set then treat it as negative. Everything which is larger thanINT_MAXwill be broken.
I did worry about the truncation int might cause to the size_t variable, however, as the result is being assigned to an int, I decided to go for int to be the typecast for min_t.
Also, won't size_t will force the int rem to be treated as unsigned value which will impact the comparison when rem indeed is negative. If rem will never be -ve, my worry will be void.
Can you please correct if my understanding is wrong?
Thanks,
./drv
Quoted 9 lines
Fortunately, in this code the value of count will never go higher than "
INT_MAX-PAGE_SIZE" because Linus understands that it's easy to introduce bugs like this.regards,
dan carpenter
Quoted 39 lines · Deepak R Varma
Quoted 15 lines · Dan Carpenter
Quoted 14 lines · Deepak R Varma
Simplify code by using
min_thelper macro for logical evaluation and value assignment. Use the _t variant of min macro since the variable types are not same.
This issue is identified by coccicheck using theminmax.coccifile.Signed-off-by Deepak R Varma <[email protected]>---
Changes in v2:
1. Revise patch description. No functional change.drivers/staging/most/video/video.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)Quoted 12 lines
drivers/staging/most/video/video.c static ssize_t comp_vdev_read(struct file *filp, char __user *buf, · line 176
Quoted 10 lines
drivers/staging/most/video/video.c +1
-1@@ -173,7 +173,7 @@ static ssize_t comp_vdev_read(struct file *filp, char __user *buf,while (count >0 &&data_ready (mdev)) {struct mbo *const mbo =get_top_mbo (mdev);int const rem = mbo->processed_length - fh->offs; -int const cnt = rem < count ? rem : count; +int const cnt =min_t (int, rem, count);TL;DR use
size_tinstead of int.Hi Dan,
Thank you for reviewing the patch. Please see my queries inline.Quoted 6 lines
Using "int" here is wrong.
size_tis unsigned long meaning that it has 64 bits to use to represent positive values. (Let's ignore 32 bit arches). You have chopped it down to say that it now has 31 bits for positives and if BIT(31) is set then treat it as negative. Everything which is larger thanINT_MAXwill be broken.
Quoted 3 lines
I did worry about the truncation int might cause to the
size_tvariable, however, as the result is being assigned to an int, I decided to go for int to be the typecast formin_t.
Let's ignore that other layers prevent "count" from being greater than INT_MAX. mbo->processed_length is a u16. Also if "fh->offs" is more than mbo->processed_length that's a separate bug and we are already screwed.
So that means rem is a relatively small number. A small number can easily fit in "int cnt". So we are eating a big pie ("count") but we are taking small bites ("cnt"). Everything works fine.
But if we chop the pie in half or treat it as negative pie then the math breaks.
Quoted 4 lines
Also, won't
size_twill force the int rem to be treated as unsigned value which will impact the comparison when rem indeed is negative. If rem will never be -ve, my worry will be void.
Is "-ve" the TikTok way of abbreviating negative? Am I old?
The small bites are always positive. But if we are eating negative pie then we take negative size bites. min_t() should almost always use unsigned types. Everything else is a headache. I have often wondered why people do it but I think it's because of the 80 character rule and the word "int" is shorter than "unsigned long".
regards,
dan carpenter
Quoted 40 lines · Dan Carpenter
Quoted 39 lines · Deepak R Varma
Quoted 15 lines · Dan Carpenter
Quoted 14 lines · Deepak R Varma
Simplify code by using
min_thelper macro for logical evaluation and value assignment. Use the _t variant of min macro since the variable types are not same.
This issue is identified by coccicheck using theminmax.coccifile.Signed-off-by Deepak R Varma <[email protected]>---
Changes in v2:
1. Revise patch description. No functional change.drivers/staging/most/video/video.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)Quoted 12 lines
drivers/staging/most/video/video.c static ssize_t comp_vdev_read(struct file *filp, char __user *buf, · line 176
Quoted 10 lines
drivers/staging/most/video/video.c +1
-1@@ -173,7 +173,7 @@ static ssize_t comp_vdev_read(struct file *filp, char __user *buf,while (count >0 &&data_ready (mdev)) {struct mbo *const mbo =get_top_mbo (mdev);int const rem = mbo->processed_length - fh->offs; -int const cnt = rem < count ? rem : count; +int const cnt =min_t (int, rem, count);TL;DR use
size_tinstead of int.Hi Dan,
Thank you for reviewing the patch. Please see my queries inline.Quoted 6 lines
Using "int" here is wrong.
size_tis unsigned long meaning that it has 64 bits to use to represent positive values. (Let's ignore 32 bit arches). You have chopped it down to say that it now has 31 bits for positives and if BIT(31) is set then treat it as negative. Everything which is larger thanINT_MAXwill be broken.
Quoted 8 lines
Quoted 3 lines
I did worry about the truncation int might cause to the
size_tvariable, however, as the result is being assigned to an int, I decided to go for int to be the typecast formin_t.Let's ignore that other layers prevent "count" from being greater than
INT_MAX.mbo->processed_lengthis a u16. Also if "fh->offs" is more thanmbo->processed_lengththat's a separate bug and we are already screwed.
Yes, "u16 - u32" looks wrong on the face, but it should always evaluat positive, else we would have had other issues down the line.
Quoted 7 lines
So that means rem is a relatively small number. A small number can easily fit in "int cnt". So we are eating a big pie ("count") but we are taking small bites ("cnt"). Everything works fine.
But if we chop the pie in half or treat it as negative pie then the math breaks.
That makes perfect sense. Thank you so much Dan for the explanation.
Quoted 7 lines
Quoted 5 lines
Also, won't
size_twill force the int rem to be treated as unsigned value which will impact the comparison when rem indeed is negative. If rem will never be -ve, my worry will be void.Is "-ve" the TikTok way of abbreviating negative? Am I old?
No no... sorry. My fingers were catching up with the thoughts. :)
Quoted 6 lines
The small bites are always positive. But if we are eating negative pie then we take negative size bites.
min_t()should almost always use unsigned types. Everything else is a headache. I have often wondered why people do it but I think it's because of the 80 character rule and the word "int" is shorter than "unsigned long".
Understood. I will send in a revision with size_t type for min_t. Thank you once again for the detailed answer. Much appreciate.
./drv
Quoted 5 lines
regards,
dan carpenter