Skip to contentSkip to conversation

eal: fix argument to rte_bsf32_safe

PATCH Superseded Stephen Hemminger 3 participants 8 messages Jul 13, 2021 – Jul 24, 2021

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 eal: fix argument to rte_bsf32_safe

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. This function only correctly handles on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

The bug passed silently under the radar until some other code was built with -Wall and -Wextra in C++ and C++ complains about the missing cast.

Yes, this is a API signature change, but the original code was wrong. It is an inline so not an ABI change.

Fixes 4e261f551986 ("eal: add 64-bit bsf and 32-bit safe bsf functions")
---
 lib/eal/include/rte_common.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/eal/include/rte_common.h +1 -1
index d5a32c66a5fe..99eb5f1820ae 100644
@@ -623,7 +623,7 @@ rte_bsf32(uint32_t v)
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
 static inline int
-rte_bsf32_safe(uint64_t v, uint32_t *pos)
+rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
 	if (v == 0)
 		return 0;
Signature · 1 line
2.30.2
Acked-by on v1 1/1
#1 lore raw
Quoted 5 lines · Stephen Hemminger

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. This function only correctly handles on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

there are multiple errors in this family of functions [1] both in usage and signatures. we previously discussed rolling all fixes up into a single patch and announcing an api break.

a doc patch was submitted as per the process documented for breaking api but received no replies [2]

i have a full patch that corrects the whole family if you would like to take it instead. contact me offline if you are interested.

1. http://mails.dpdk.org/archives/dev/2021-March/201590.html
2. http://mails.dpdk.org/archives/dev/2021-March/201868.html

the change stand-alone is correct so

Quoted 30 lines

The bug passed silently under the radar until some other code was built with -Wall and -Wextra in C++ and C++ complains about the missing cast.

Yes, this is a API signature change, but the original code was wrong. It is an inline so not an ABI change.

Fixes 4e261f551986 ("eal: add 64-bit bsf and 32-bit safe bsf functions")
---
 lib/eal/include/rte_common.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/eal/include/rte_common.h +1 -1
index d5a32c66a5fe..99eb5f1820ae 100644
@@ -623,7 +623,7 @@ rte_bsf32(uint32_t v)
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
 static inline int
-rte_bsf32_safe(uint64_t v, uint32_t *pos)
+rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
 	if (v == 0)
 		return 0;
Signature · 1 line
2.30.2
#2 lore raw
Quoted 17 lines · On Mon, 19 Jul 2021 10:15:34 -0700 Tyler Retzlaff…
Quoted 5 lines · Stephen Hemminger

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. This function only correctly handles on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

there are multiple errors in this family of functions [1] both in usage and signatures. we previously discussed rolling all fixes up into a single patch and announcing an api break.

a doc patch was submitted as per the process documented for breaking api but received no replies [2]

i have a full patch that corrects the whole family if you would like to take it instead. contact me offline if you are interested.

Quoted 6 lines

1. http://mails.dpdk.org/archives/dev/2021-March/201590.html
2. http://mails.dpdk.org/archives/dev/2021-March/201868.html

the change stand-alone is correct so

Thanks, I think the larger set should go into 21.11 where API/ABI break would be ok. My bit was all about fixing the bug where current code breaks C++ users.

#3 lore raw

20/07/2021 00:00, Stephen Hemminger:

Quoted 23 lines
Quoted 20 lines · On Mon, 19 Jul 2021 10:15:34 -0700 Tyler Retzlaff…
Quoted 5 lines · Stephen Hemminger

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. This function only correctly handles on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

there are multiple errors in this family of functions [1] both in usage and signatures. we previously discussed rolling all fixes up into a single patch and announcing an api break.

a doc patch was submitted as per the process documented for breaking api but received no replies [2]

i have a full patch that corrects the whole family if you would like to take it instead. contact me offline if you are interested.

1. http://mails.dpdk.org/archives/dev/2021-March/201590.html
2. http://mails.dpdk.org/archives/dev/2021-March/201868.html

Quoted 7 lines
Quoted 3 lines

the change stand-alone is correct so

Thanks, I think the larger set should go into 21.11 where API/ABI break would be ok. My bit was all about fixing the bug where current code breaks C++ users.

Shouldn't we have a note in the API changes section of the release notes?

#4 lore raw

PATCH v2 1/1 eal: fix argument to rte_bsf32_safe

In reply to Stephen Hemminger

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. The code only works on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was a mistake introduced when the safe version was added and probaly cause by copy/paste from the 64 bit version.

The bug passed silently under the radar until some other code was built with -Wall and -Wextra in C++ and C++ complains about the missing cast.

Yes, this is a API signature change, but the original code was wrong. It is an inline so not an ABI change.

Fixes 4e261f551986 ("eal: add 64-bit bsf and 32-bit safe bsf functions")

---
v2 - add suggested release note

 doc/guides/rel_notes/release_21_08.rst | 4 ++++
 lib/eal/include/rte_common.h           | 2 +-
 2 files changed, 5 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_08.rst +4 -0
index e2c5ccbf7d90..148405891fcb 100644
@@ -196,6 +196,10 @@ API Changes
   to be thread safe; all Rx queues affected by the API will now need to be
   stopped before making any changes to the power management scheme.
 
+* eal: ``rte_bsf32_safe`` now takes a 32 bit value for its first
+  argument. This fixes warnings about loss of precision when used
+  with some compilers settings.
+
 
 ABI Changes
 -----------
lib/eal/include/rte_common.h +1 -1
index d5a32c66a5fe..99eb5f1820ae 100644
@@ -623,7 +623,7 @@ rte_bsf32(uint32_t v)
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
 static inline int
-rte_bsf32_safe(uint64_t v, uint32_t *pos)
+rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
 	if (v == 0)
 		return 0;
Signature · 1 line
2.30.2
#5 lore raw

PATCH v3 1/1 eal: fix argument to rte_bsf32_safe

In reply to Stephen Hemminger

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. The code only works on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was a mistake introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

The bug passed silently under the radar until some other code was built with -Wall and -Wextra in C++ and C++ complains about the missing cast.

Yes, this is a API signature change, but the original code was wrong. It is an inline so not an ABI change.

Fixes 4e261f551986 ("eal: add 64-bit bsf and 32-bit safe bsf functions")

---
v3 - reword commit description for checkpatch

 doc/guides/rel_notes/release_21_08.rst | 4 ++++
 lib/eal/include/rte_common.h           | 2 +-
 2 files changed, 5 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_08.rst +4 -0
index e2c5ccbf7d90..148405891fcb 100644
@@ -196,6 +196,10 @@ API Changes
   to be thread safe; all Rx queues affected by the API will now need to be
   stopped before making any changes to the power management scheme.
 
+* eal: ``rte_bsf32_safe`` now takes a 32 bit value for its first
+  argument. This fixes warnings about loss of precision when used
+  with some compilers settings.
+
 
 ABI Changes
 -----------
lib/eal/include/rte_common.h +1 -1
index d5a32c66a5fe..99eb5f1820ae 100644
@@ -623,7 +623,7 @@ rte_bsf32(uint32_t v)
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
 static inline int
-rte_bsf32_safe(uint64_t v, uint32_t *pos)
+rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
 	if (v == 0)
 		return 0;
Signature · 1 line
2.30.2
#6 lore raw

23/07/2021 17:45, Stephen Hemminger:

Quoted 10 lines

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. The code only works on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was a mistake introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

The bug passed silently under the radar until some other code was built with -Wall and -Wextra in C++ and C++ complains about the missing cast.

Quoted 7 lines

Yes, this is a API signature change, but the original code was wrong. It is an inline so not an ABI change.

Fixes 4e261f551986 ("eal: add 64-bit bsf and 32-bit safe bsf functions")

+Cc: [email protected]

Applied, thanks.

I think these functions lack a reference to the name Bit Scan Forward.

#7 lore raw
Quoted 19 lines · On Sat, 24 Jul 2021 09:58:44 +0200 Thomas Monjalon <[email protected]> wrote:

23/07/2021 17:45, Stephen Hemminger:

Quoted 10 lines

The first argument to rte_bsf32_safe was incorrectly declared as a 64 bit value. The code only works on 32 bit values and the underlying function rte_bsf32 only accepts 32 bit values. This was a mistake introduced when the safe version was added and probably cause by copy/paste from the 64 bit version.

The bug passed silently under the radar until some other code was built with -Wall and -Wextra in C++ and C++ complains about the missing cast.

Quoted 7 lines

Yes, this is a API signature change, but the original code was wrong. It is an inline so not an ABI change.

Fixes 4e261f551986 ("eal: add 64-bit bsf and 32-bit safe bsf functions")
Quoted 9 lines

+Cc: [email protected]

Applied, thanks.

I think these functions lack a reference to the name Bit Scan Forward.

Tyler wanted to fix a bunch more stuff in these for 21.11 where it will be a bigger API change.