Skip to contentSkip to conversation

doc: propose correction rte_bsf64 return type declaration

PATCH New Tyler Retzlaff 7 participants 26 messages Mar 10, 2021 – Oct 6, 2022

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 doc: propose correction rte_bsf64 return type declaration

It is clear from the change that introduced the rte_bsf64 inline function and an evaluation of usage that the initial revision was intended to return uint32_t and not int.

It is proposed the return type of int be changed to uint32_t.

-static inline int
+static inline uint32_t
rte_bsf64(uint64_t v)

 {
        return (uint32_t)__builtin_ctzll(v);

}

---
 doc/guides/rel_notes/deprecation.rst | 3 +++
 1 file changed, 3 insertions(+)
doc/guides/rel_notes/deprecation.rst +3 -0
index 64629e064..7ba86024a 100644
@@ -17,6 +17,9 @@ Deprecation Notices
 * eal: The function ``rte_eal_remote_launch`` will return new error codes
   after read or write error on the pipe, instead of calling ``rte_panic``.
 
+* eal: The inline function ``rte_bsf64`` will be changed to return ``uint32_t``
+  instead of ``int`` as originally intended.
+
 * rte_atomicNN_xxx: These APIs do not take memory order parameter. This does
   not allow for writing optimized code for all the CPU architectures supported
   in DPDK. DPDK will adopt C11 atomic operations semantics and provide wrappers
Signature · 1 line
2.30.0.vfs.0.2
#1 lore raw

PATCH v2 1/1 doc: propose correction rte_{bsf, fls} inline functions type use

The proposal has resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---
 doc/guides/rel_notes/deprecation.rst | 7 +++++++
 1 file changed, 7 insertions(+)
doc/guides/rel_notes/deprecation.rst +7 -0
index 64629e064..4934f4da4 100644
@@ -17,6 +17,13 @@ Deprecation Notices
 * eal: The function ``rte_eal_remote_launch`` will return new error codes
   after read or write error on the pipe, instead of calling ``rte_panic``.
 
+* eal: Fix inline function return and parameter types for rte_{bsf,fls}
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t`` to ``uint32_t``.
+  Change ``rte_bsf64`` return type to  ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u32`` return type to ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u64`` return type to ``uint32_t`` instead of ``int``.
+
 * rte_atomicNN_xxx: These APIs do not take memory order parameter. This does
   not allow for writing optimized code for all the CPU architectures supported
   in DPDK. DPDK will adopt C11 atomic operations semantics and provide wrappers
Signature · 1 line
2.30.0.vfs.0.2
#2 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

15/03/2021 20:34, Tyler Retzlaff:

Quoted 19 lines

The proposal has resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---

doc/guides/rel_notes/deprecation.rst Deprecation Notices · line 25

Quoted 6 lines
doc/guides/rel_notes/deprecation.rst +6 -0
+* eal: Fix inline function return and parameter types for rte_{bsf,fls}
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t`` to ``uint32_t``.
+  Change ``rte_bsf64`` return type to  ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u32`` return type to ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u64`` return type to ``uint32_t`` instead of ``int``.

It seems we completely forgot this.
How critical is it?

#3 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

Quoted 29 lines

From: dev [mailto:[email protected]] On Behalf Of Thomas Monjalon
Sent: Monday, 25 October 2021 21.14

15/03/2021 20:34, Tyler Retzlaff:

Quoted 19 lines

The proposal has resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---

doc/guides/rel_notes/deprecation.rst

Quoted 1 line
doc/guides/rel_notes/deprecation.rst +1 -0
+* eal: Fix inline function return and parameter types for

rte_{bsf,fls}

doc/guides/rel_notes/deprecation.rst

Quoted 2 lines
doc/guides/rel_notes/deprecation.rst +2 -0
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t`` to

``uint32_t``.

Quoted 1 line

+ Change ``rte_bsf64`` return type to ``uint32_t`` instead of

Quoted 8 lines

``int``.

Quoted 1 line

+ Change ``rte_fls_u32`` return type to ``uint32_t`` instead of

``int``.

Quoted 1 line

+ Change ``rte_fls_u64`` return type to ``uint32_t`` instead of

``int``.

It seems we completely forgot this.
How critical is it?

Not updating has near zero effect on bug probability: Incorrectly returning signed int instead of unsigned int is extremely unlikely to cause problems.

Updating has near zero performance improvement: The unnecessary expansion of a parameter value from 32 to 64 bits is cheap.

The update's only tangible benefit is API consistency. :-)

-Morten

#4 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

Quoted 29 lines · Morten Brørup
Quoted 29 lines

From: dev [mailto:[email protected]] On Behalf Of Thomas Monjalon
Sent: Monday, 25 October 2021 21.14

15/03/2021 20:34, Tyler Retzlaff:

Quoted 19 lines

The proposal has resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---

doc/guides/rel_notes/deprecation.rst

Quoted 1 line
doc/guides/rel_notes/deprecation.rst +1 -0
+* eal: Fix inline function return and parameter types for

rte_{bsf,fls}

doc/guides/rel_notes/deprecation.rst

Quoted 2 lines
doc/guides/rel_notes/deprecation.rst +2 -0
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t`` to

``uint32_t``.

Quoted 1 line

+ Change ``rte_bsf64`` return type to ``uint32_t`` instead of

Quoted 9 lines
Quoted 8 lines

``int``.

Quoted 1 line

+ Change ``rte_fls_u32`` return type to ``uint32_t`` instead of

``int``.

Quoted 1 line

+ Change ``rte_fls_u64`` return type to ``uint32_t`` instead of

``int``.

It seems we completely forgot this.
How critical is it?

our organization as a matter of internal security policy requires these sorts of things to be fixed. while i didn't see any bugs in the dpdk code there is an opportunity for users of these functions to accidentally write code that is prone to integer and buffer overflow class bugs.

there is no urgency, but why leave things sloppy? though i do wish this had been responded to in a more timely manner 7 months for something that should have almost been rubber stamped.

Quoted 7 lines

Not updating has near zero effect on bug probability: Incorrectly returning signed int instead of unsigned int is extremely unlikely to cause problems.

Updating has near zero performance improvement: The unnecessary expansion of a parameter value from 32 to 64 bits is cheap.

The update's only tangible benefit is API consistency. :-)

-Morten

#5 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

11/11/2021 05:15, Tyler Retzlaff:

Quoted 46 lines · Morten Brørup
Quoted 29 lines
Quoted 29 lines

From: dev [mailto:[email protected]] On Behalf Of Thomas Monjalon
Sent: Monday, 25 October 2021 21.14

15/03/2021 20:34, Tyler Retzlaff:

Quoted 19 lines

The proposal has resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---

doc/guides/rel_notes/deprecation.rst

Quoted 1 line
doc/guides/rel_notes/deprecation.rst +1 -0
+* eal: Fix inline function return and parameter types for

rte_{bsf,fls}

doc/guides/rel_notes/deprecation.rst

Quoted 2 lines
doc/guides/rel_notes/deprecation.rst +2 -0
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t`` to

``uint32_t``.

Quoted 1 line

+ Change ``rte_bsf64`` return type to ``uint32_t`` instead of

Quoted 9 lines
Quoted 8 lines

``int``.

Quoted 1 line

+ Change ``rte_fls_u32`` return type to ``uint32_t`` instead of

``int``.

Quoted 1 line

+ Change ``rte_fls_u64`` return type to ``uint32_t`` instead of

``int``.

It seems we completely forgot this.
How critical is it?

our organization as a matter of internal security policy requires these sorts of things to be fixed. while i didn't see any bugs in the dpdk code there is an opportunity for users of these functions to accidentally write code that is prone to integer and buffer overflow class bugs.

Quoted 3 lines

there is no urgency, but why leave things sloppy? though i do wish this had been responded to in a more timely manner 7 months for something that should have almost been rubber stamped.

It's difficult to be on all topics.
The best way to avoid such miss is to ping when you see no progress.

So what's next?
They are only inline functions, right? so no ABI breakage. Is it going to require any change on application-side? I guess no. Is it acceptable in 21.11-rc3? maybe too late?
Is it acceptable in 22.02?

#6 lore raw

RE: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

Quoted 63 lines

From: Thomas Monjalon [mailto:[email protected]]
Sent: Thursday, 11 November 2021 12.55

11/11/2021 05:15, Tyler Retzlaff:

Quoted 2 lines · Morten Brørup
Quoted 1 line
Quoted 1 line

From: dev [mailto:[email protected]] On Behalf Of Thomas

Monjalon

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Monday, 25 October 2021 21.14

15/03/2021 20:34, Tyler Retzlaff:

Quoted 1 line

The proposal has resulted from request to review [1] the

following

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

functions where there appeared to be inconsistency in return

type

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

or parameter type selections for the following inline

functions.

Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

doc/guides/rel_notes/deprecation.rst

Quoted 5 lines

---

doc/guides/rel_notes/deprecation.rst +1 -0
+* eal: Fix inline function return and parameter types for

rte_{bsf,fls}

Quoted 2 lines
doc/guides/rel_notes/deprecation.rst +2 -0
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t``

to

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

``uint32_t``.

Quoted 1 line

+ Change ``rte_bsf64`` return type to ``uint32_t`` instead of

``int``.

Quoted 1 line

+ Change ``rte_fls_u32`` return type to ``uint32_t`` instead

of

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

``int``.

Quoted 1 line

+ Change ``rte_fls_u64`` return type to ``uint32_t`` instead

of

Quoted 7 lines
Quoted 5 lines
Quoted 4 lines

``int``.

It seems we completely forgot this.
How critical is it?

our organization as a matter of internal security policy requires

these

Quoted 6 lines

sorts of things to be fixed. while i didn't see any bugs in the dpdk code there is an opportunity for users of these functions to accidentally write code that is prone to integer and buffer overflow class bugs.

there is no urgency, but why leave things sloppy? though i do wish

this

Quoted 2 lines

had been responded to in a more timely manner 7 months for something that should have almost been rubber stamped.

Quoted 8 lines

It's difficult to be on all topics.
The best way to avoid such miss is to ping when you see no progress.

So what's next?
They are only inline functions, right? so no ABI breakage. Is it going to require any change on application-side? I guess no. Is it acceptable in 21.11-rc3? maybe too late?
Is it acceptable in 22.02?

If Microsoft (represented by Tyler in this case) considers this a bug, I would prefer getting it into 21.11 - especially because it is an LTS release.

-Morten

Acked-by
#7 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

Quoted 27 lines · Morten Brørup
Quoted 27 lines

From: Thomas Monjalon [mailto:[email protected]]
Sent: Thursday, 11 November 2021 12.55

11/11/2021 05:15, Tyler Retzlaff:

Quoted 2 lines · Morten Brørup
Quoted 1 line
Quoted 1 line

From: dev [mailto:[email protected]] On Behalf Of Thomas

Monjalon

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Monday, 25 October 2021 21.14

15/03/2021 20:34, Tyler Retzlaff:

Quoted 1 line

The proposal has resulted from request to review [1] the

following

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

functions where there appeared to be inconsistency in return

type

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

or parameter type selections for the following inline

functions.

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

Quoted 47 lines
Quoted 34 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

doc/guides/rel_notes/deprecation.rst

Quoted 4 lines

---

doc/guides/rel_notes/deprecation.rst +1 -0
+* eal: Fix inline function return and parameter types for

rte_{bsf,fls}

Quoted 2 lines
doc/guides/rel_notes/deprecation.rst +2 -0
+  inline functions to be consistent.
+  Change ``rte_bsf32_safe`` parameter ``v`` from ``uint64_t``

to

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

``uint32_t``.

Quoted 1 line

+ Change ``rte_bsf64`` return type to ``uint32_t`` instead of

``int``.

Quoted 1 line

+ Change ``rte_fls_u32`` return type to ``uint32_t`` instead

of

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

``int``.

Quoted 1 line

+ Change ``rte_fls_u64`` return type to ``uint32_t`` instead

of

Quoted 7 lines
Quoted 5 lines
Quoted 4 lines

``int``.

It seems we completely forgot this.
How critical is it?

our organization as a matter of internal security policy requires

these

Quoted 6 lines

sorts of things to be fixed. while i didn't see any bugs in the dpdk code there is an opportunity for users of these functions to accidentally write code that is prone to integer and buffer overflow class bugs.

there is no urgency, but why leave things sloppy? though i do wish

this

Quoted 2 lines

had been responded to in a more timely manner 7 months for something that should have almost been rubber stamped.

Quoted 8 lines

It's difficult to be on all topics.
The best way to avoid such miss is to ping when you see no progress.

So what's next?
They are only inline functions, right? so no ABI breakage. Is it going to require any change on application-side? I guess no. Is it acceptable in 21.11-rc3? maybe too late?
Is it acceptable in 22.02?

If Microsoft (represented by Tyler in this case) considers this a bug, I would prefer getting it into 21.11 - especially because it is an LTS release.

-Morten

#8 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

In reply to Morten Brørup

Tyler, there was no progress on this topic during the past year. Please could you send the patch to fix API inconsistencies, so we will merge it in 22.11?

#9 lore raw

Re: [dpdk-dev] [PATCH v2] doc: propose correction rte_{bsf, fls} inline functions type use

Hi Thomas,

I will see if i can carve out some time budget for it, it's a pretty old patch so it could take some time to review.

Quoted 4 lines · Thomas Monjalon

Tyler, there was no progress on this topic during the past year. Please could you send the patch to fix API inconsistencies, so we will merge it in 22.11?

#10 lore raw

cover cleanup bsf and fls inline function return types

In reply to Tyler Retzlaff

The cleanup resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

Since the original request some instances have been fixed so this series completes the outstanding return value type updates.

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

Tyler Retzlaff (3):
doc: announce cleanup of rte_{bsf, fls} inline functions type use
eal: change rte_fls and rte_bsf to return uint32_t
test: fix sign compare warning for rte_bsf64 return type change

 app/test/test_mbuf.c                 | 2 +-
 doc/guides/rel_notes/deprecation.rst | 6 ++++++
 lib/eal/include/rte_common.h         | 6 +++---
 3 files changed, 10 insertions(+), 4 deletions(-)
Signature · 1 line
1.8.3.1
#11 lore raw

PATCH 1/3 doc: announce cleanup of rte_{bsf, fls} inline functions type use

From: Tyler Retzlaff <[email protected]>

The cleanup resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()
rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---
 doc/guides/rel_notes/deprecation.rst | 6 ++++++
 1 file changed, 6 insertions(+)
doc/guides/rel_notes/deprecation.rst +6 -0
index e7583ca..58f4c24 100644
@@ -17,6 +17,12 @@ Deprecation Notices
 * eal: The function ``rte_eal_remote_launch`` will return new error codes
   after read or write error on the pipe, instead of calling ``rte_panic``.
 
+* eal: Fix inline function return and parameter types for rte_{bsf,fls}
+  inline functions to be consistent in DPDK 22.11.
+  Change ``rte_bsf64`` return type to  ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u32`` return type to ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u64`` return type to ``uint32_t`` instead of ``int``.
+
 * rte_atomicNN_xxx: These APIs do not take memory order parameter. This does
   not allow for writing optimized code for all the CPU architectures supported
   in DPDK. DPDK has adopted the atomic operations from
Signature · 1 line
1.8.3.1
#12 lore raw

Re: [PATCH 1/3] doc: announce cleanup of rte_{bsf, fls} inline functions type use

08/08/2022 23:21, Tyler Retzlaff:

Quoted 10 lines

From: Tyler Retzlaff <[email protected]>

The cleanup resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

rte_bsf32()
rte_bsf32_safe()
rte_bsf64()
rte_bsf64_safe()

Why _safe functions are not changed?

Quoted 16 lines

rte_fls_u32()
rte_fls_u64()
rte_log2_u32()
rte_log2_u64()

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

---
 doc/guides/rel_notes/deprecation.rst | 6 ++++++
 1 file changed, 6 insertions(+)
doc/guides/rel_notes/deprecation.rst +0 -0
index e7583ca..58f4c24 100644

doc/guides/rel_notes/deprecation.rst Deprecation Notices · line 24

Quoted 5 lines
doc/guides/rel_notes/deprecation.rst +5 -0
+* eal: Fix inline function return and parameter types for rte_{bsf,fls}
+  inline functions to be consistent in DPDK 22.11.
+  Change ``rte_bsf64`` return type to  ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u32`` return type to ``uint32_t`` instead of ``int``.
+  Change ``rte_fls_u64`` return type to ``uint32_t`` instead of ``int``.

It is too late for the announce,
but the release notes requires to be updated.

#13 lore raw

PATCH 2/3 eal: change rte_fls and rte_bsf to return uint32_t

In reply to Tyler Retzlaff

From: Tyler Retzlaff <[email protected]>

return fixed width uint32_t to be consistent with what appears to be the original authors intent. it doesn't make much sense to return signed integers for these functions.

---
 lib/eal/include/rte_common.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
lib/eal/include/rte_common.h +3 -3
index a96cc2a..bd4184d 100644
@@ -707,7 +707,7 @@ static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void)
  * @return
  *     The last (most-significant) bit set, or 0 if the input is 0.
  */
-static inline int
+static inline uint32_t
 rte_fls_u32(uint32_t x)
 {
 	return (x == 0) ? 0 : 32 - __builtin_clz(x);
@@ -724,7 +724,7 @@ static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void)
  * @return
  *     least significant set bit in the input parameter.
  */
-static inline int
+static inline uint32_t
 rte_bsf64(uint64_t v)
 {
 	return (uint32_t)__builtin_ctzll(v);
@@ -766,7 +766,7 @@ static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void)
  * @return
  *     The last (most-significant) bit set, or 0 if the input is 0.
  */
-static inline int
+static inline uint32_t
 rte_fls_u64(uint64_t x)
 {
 	return (x == 0) ? 0 : 64 - __builtin_clzll(x);
Signature · 1 line
1.8.3.1
#14 lore raw

Re: [PATCH 2/3] eal: change rte_fls and rte_bsf to return uint32_t

08/08/2022 23:21, Tyler Retzlaff:

Quoted 35 lines

From: Tyler Retzlaff <[email protected]>

return fixed width uint32_t to be consistent with what appears to be the original authors intent. it doesn't make much sense to return signed integers for these functions.

---
 lib/eal/include/rte_common.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
lib/eal/include/rte_common.h +2 -2
index a96cc2a..bd4184d 100644
@@ -707,7 +707,7 @@ static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void)
  * @return
  *     The last (most-significant) bit set, or 0 if the input is 0.
  */
-static inline int
+static inline uint32_t
 rte_fls_u32(uint32_t x)
 {
 	return (x == 0) ? 0 : 32 - __builtin_clz(x);
@@ -724,7 +724,7 @@ static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void)
  * @return
  *     least significant set bit in the input parameter.
  */
-static inline int
+static inline uint32_t
 rte_bsf64(uint64_t v)
 {
 	return (uint32_t)__builtin_ctzll(v);

lib/eal/include/rte_common.h static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void) · line 772

Quoted 8 lines
lib/eal/include/rte_common.h +1 -1
@@ -766,7 +766,7 @@ static void __attribute__((destructor(RTE_PRIO(prio)), used)) func(void)
  * @return
  *     The last (most-significant) bit set, or 0 if the input is 0.
  */
-static inline int
+static inline uint32_t
 rte_fls_u64(uint64_t x)
 {
 	return (x == 0) ? 0 : 64 - __builtin_clzll(x);

You forgot the _safe versions:

lib/eal/include/rte_common.h +2 -2
@@ -660,7 +660,7 @@ rte_bsf32(uint32_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline int
+static inline uint32_t
 rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
        if (v == 0)
@@ -739,7 +739,7 @@ rte_bsf64(uint64_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline int
+static inline uint32_t
 rte_bsf64_safe(uint64_t v, uint32_t *pos)
 {
        if (v == 0)
#15 lore raw

Re: [PATCH 2/3] eal: change rte_fls and rte_bsf to return uint32_t

Quoted 23 lines · Thomas Monjalon

08/08/2022 23:21, Tyler Retzlaff:

Quoted 2 lines

From: Tyler Retzlaff <[email protected]>

You forgot the _safe versions:

lib/eal/include/rte_common.h +1 -1
@@ -660,7 +660,7 @@ rte_bsf32(uint32_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline int
+static inline uint32_t
 rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
        if (v == 0)

lib/eal/include/rte_common.h rte_bsf64(uint64_t v) · line 745

Quoted 8 lines
lib/eal/include/rte_common.h +1 -1
@@ -739,7 +739,7 @@ rte_bsf64(uint64_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline int
+static inline uint32_t
 rte_bsf64_safe(uint64_t v, uint32_t *pos)
 {
        if (v == 0)

the return values from the _safe versions are `int' treated like a bool type. they have been left as is to be consistent with the rest of dpdk return value types.

the non-safe version returns were returning actual values and not an indication of success or failure.

they could certainly be changed to C99 fixed width types but if they are changed at all perhaps they should be changed to _Bool or bool from stdbool.h?

it looks like this change has been merged already but if you would like to make any further changes let me know i'll take care of it.

thanks

#16 lore raw

Re: [PATCH 2/3] eal: change rte_fls and rte_bsf to return uint32_t

05/10/2022 17:15, Tyler Retzlaff:

Quoted 40 lines · Thomas Monjalon
Quoted 23 lines

08/08/2022 23:21, Tyler Retzlaff:

Quoted 2 lines

From: Tyler Retzlaff <[email protected]>

You forgot the _safe versions:

lib/eal/include/rte_common.h +1 -1
@@ -660,7 +660,7 @@ rte_bsf32(uint32_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline int
+static inline uint32_t
 rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
        if (v == 0)

lib/eal/include/rte_common.h rte_bsf64(uint64_t v) · line 745

Quoted 8 lines
lib/eal/include/rte_common.h +1 -1
@@ -739,7 +739,7 @@ rte_bsf64(uint64_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline int
+static inline uint32_t
 rte_bsf64_safe(uint64_t v, uint32_t *pos)
 {
        if (v == 0)

the return values from the _safe versions are `int' treated like a bool type. they have been left as is to be consistent with the rest of dpdk return value types.

the non-safe version returns were returning actual values and not an indication of success or failure.

Quoted 7 lines

they could certainly be changed to C99 fixed width types but if they are changed at all perhaps they should be changed to _Bool or bool from stdbool.h?

it looks like this change has been merged already but if you would like to make any further changes let me know i'll take care of it.

Sorry, it's my mistake, I went too fast.
I'll revert them to int.

#17 lore raw

PATCH 1/1 eal: fix return type of bsf safe functions

In a recent commit, changing return type from int to uint32_t, I did a last minute change to functions rte_bsf32_safe and rte_bsf64_safe, because thought they were forgotten.
Actually these functions are returning 0 or 1, so it should be int. The return type is reverted to the original type for these 2 functions.

Fixes 4b81c145ae37 ("eal: change return type of bsf/fls functions")
---
 lib/eal/include/rte_common.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
lib/eal/include/rte_common.h +2 -2
index 23d9e05cbb..15765b408d 100644
@@ -660,7 +660,7 @@ rte_bsf32(uint32_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline uint32_t
+static inline int
 rte_bsf32_safe(uint32_t v, uint32_t *pos)
 {
 	if (v == 0)
@@ -739,7 +739,7 @@ rte_bsf64(uint64_t v)
  * @return
  *     Returns 0 if ``v`` was 0, otherwise returns 1.
  */
-static inline uint32_t
+static inline int
 rte_bsf64_safe(uint64_t v, uint32_t *pos)
 {
 	if (v == 0)
Signature · 1 line
2.36.1
Reviewed-by on v1 1/1
#18 lore raw

Re: [PATCH] eal: fix return type of bsf safe functions

Quoted 10 lines · Thomas Monjalon

In a recent commit, changing return type from int to uint32_t, I did a last minute change to functions rte_bsf32_safe and rte_bsf64_safe, because thought they were forgotten.
Actually these functions are returning 0 or 1, so it should be int. The return type is reverted to the original type for these 2 functions.

Fixes 4b81c145ae37 ("eal: change return type of bsf/fls functions")
Signature · 1 line
David Marchand
author Acked-by
#19 lore raw

Re: [PATCH] eal: fix return type of bsf safe functions

Quoted 8 lines · David Marchand
Quoted 7 lines · Thomas Monjalon

In a recent commit, changing return type from int to uint32_t, I did a last minute change to functions rte_bsf32_safe and rte_bsf64_safe, because thought they were forgotten.
Actually these functions are returning 0 or 1, so it should be int. The return type is reverted to the original type for these 2 functions.

Quoted 5 lines
Quoted 3 lines
Fixes 4b81c145ae37 ("eal: change return type of bsf/fls functions")
#20 lore raw

Re: [PATCH] eal: fix return type of bsf safe functions

06/10/2022 00:20, Tyler Retzlaff:

Quoted 9 lines · David Marchand
Quoted 8 lines · Thomas Monjalon
Quoted 7 lines

In a recent commit, changing return type from int to uint32_t, I did a last minute change to functions rte_bsf32_safe and rte_bsf64_safe, because thought they were forgotten.
Actually these functions are returning 0 or 1, so it should be int. The return type is reverted to the original type for these 2 functions.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines
Fixes 4b81c145ae37 ("eal: change return type of bsf/fls functions")

Applied, sorry for the noise.

#21 lore raw

Re: [PATCH] eal: fix return type of bsf safe functions

In reply to Thomas Monjalon
Quoted 4 lines · Thomas Monjalon

In a recent commit, changing return type from int to uint32_t, I did a last minute change to functions rte_bsf32_safe and rte_bsf64_safe, because thought they were forgotten.
Actually these functions are returning 0 or 1, so it should be int.

Wouldn't bool be a better choice?

Quoted 31 lines

The return type is reverted to the original type for these 2 functions.

Fixes 4b81c145ae37 ("eal: change return type of bsf/fls functions")
---
  lib/eal/include/rte_common.h | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
lib/eal/include/rte_common.h +2 -2
index 23d9e05cbb..15765b408d 100644
@@ -660,7 +660,7 @@ rte_bsf32(uint32_t v)
   * @return
   *     Returns 0 if ``v`` was 0, otherwise returns 1.
   */
-static inline uint32_t
+static inline int
  rte_bsf32_safe(uint32_t v, uint32_t *pos)
  {
  	if (v == 0)
@@ -739,7 +739,7 @@ rte_bsf64(uint64_t v)
   * @return
   *     Returns 0 if ``v`` was 0, otherwise returns 1.
   */
-static inline uint32_t
+static inline int
  rte_bsf64_safe(uint64_t v, uint32_t *pos)
  {
  	if (v == 0)
#22 lore raw

PATCH 3/3 test: fix sign compare warning for rte_bsf64 return type change

In reply to Tyler Retzlaff

From: Tyler Retzlaff <[email protected]>

rte_bsf64 return type has been changed insert cast to suppress sign-comapre warning with int.

---
 app/test/test_mbuf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
app/test/test_mbuf.c +1 -1
index e09b254..be4c3ff 100644
@@ -2677,7 +2677,7 @@ struct test_case {
 
 	flag3 = rte_mbuf_dynflag_register_bitnum(&dynflag3,
 						rte_bsf64(RTE_MBUF_F_LAST_FREE));
-	if (flag3 != rte_bsf64(RTE_MBUF_F_LAST_FREE))
+	if ((uint32_t)flag3 != rte_bsf64(RTE_MBUF_F_LAST_FREE))
 		GOTO_FAIL("failed to register dynamic flag 3, flag3=%d: %s",
 			flag3, strerror(errno));
 
Signature · 1 line
1.8.3.1
#23 lore raw

Re: [PATCH 0/3] cleanup bsf and fls inline function return types

In reply to Tyler Retzlaff
Quoted 14 lines · On Mon, 8 Aug 2022 14:21:29 -0700 Tyler Retzlaff…

The cleanup resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

Since the original request some instances have been fixed so this series completes the outstanding return value type updates.

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

Tyler Retzlaff (3):
doc: announce cleanup of rte_{bsf, fls} inline functions type use
eal: change rte_fls and rte_bsf to return uint32_t
test: fix sign compare warning for rte_bsf64 return type change

Quoted 5 lines
 app/test/test_mbuf.c                 | 2 +-
 doc/guides/rel_notes/deprecation.rst | 6 ++++++
 lib/eal/include/rte_common.h         | 6 +++---
 3 files changed, 10 insertions(+), 4 deletions(-)
#24 lore raw

RE: [PATCH 0/3] cleanup bsf and fls inline function return types

In reply to Tyler Retzlaff
Quoted 17 lines

From: Tyler Retzlaff [mailto:[email protected]]
Sent: Monday, 8 August 2022 23.21

The cleanup resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

Since the original request some instances have been fixed so this series completes the outstanding return value type updates.

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

Tyler Retzlaff (3):
doc: announce cleanup of rte_{bsf, fls} inline functions type use
eal: change rte_fls and rte_bsf to return uint32_t
test: fix sign compare warning for rte_bsf64 return type change

Quoted 8 lines
 app/test/test_mbuf.c                 | 2 +-
 doc/guides/rel_notes/deprecation.rst | 6 ++++++
 lib/eal/include/rte_common.h         | 6 +++---
 3 files changed, 10 insertions(+), 4 deletions(-)

--
1.8.3.1

#25 lore raw

Re: [PATCH 0/3] cleanup bsf and fls inline function return types

09/08/2022 10:26, Morten Brørup:

Quoted 17 lines
Quoted 17 lines

From: Tyler Retzlaff [mailto:[email protected]]
Sent: Monday, 8 August 2022 23.21

The cleanup resulted from request to review [1] the following functions where there appeared to be inconsistency in return type or parameter type selections for the following inline functions.

Since the original request some instances have been fixed so this series completes the outstanding return value type updates.

[1] http://mails.dpdk.org/archives/dev/2021-March/201590.html

Tyler Retzlaff (3):
doc: announce cleanup of rte_{bsf, fls} inline functions type use
eal: change rte_fls and rte_bsf to return uint32_t
test: fix sign compare warning for rte_bsf64 return type change

Quoted 6 lines
Quoted 4 lines
 app/test/test_mbuf.c                 | 2 +-
 doc/guides/rel_notes/deprecation.rst | 6 ++++++
 lib/eal/include/rte_common.h         | 6 +++---
 3 files changed, 10 insertions(+), 4 deletions(-)

Applied and completed.
Release notes added.
All squashed.