Skip to contentSkip to conversation

Specify C-standard requirement for DPDK builds

RFC PATCH In review Bruce Richardson 4 participants 14 messages Jan 12, 2023 – Feb 23, 2023 Arm/Amlogic Meson SoC support

lore
Share & tools
lore
#0 lore raw

cover Specify C-standard requirement for DPDK builds

Traditionally, DPDK has never specified a minimum C standard used either in DPDK builds or for applications using DPDK. Following discussion on-list about C standards, this RFC attempts to start the process of codifying what our standards expectations are. No code changes are made by this RFC, instead only the build parameters are changed to explicitly specify:

* C99 standard is used to build DPDK itself. This is supported by all supported compiler versions of GCC and Clang.
* The headers are checked for compatibility with gcc89 standard, which was the default standard used by the oldest supported version of GCC. DPDK headers do not build with the official C89 standard, and, to the best of my knowledge, have never done so.

Bruce Richardson (1):
build: increase minimum C standard for DPDK builds

 buildtools/chkincs/meson.build | 1 +
 meson.build                    | 1 +
 2 files changed, 2 insertions(+)

--
2.37.2

#1 lore raw

RFC PATCH 1/1 build: increase minimum C standard for DPDK builds

Set the default C language standard to be used for DPDK builds to C99. This requires no actual code changes to build successfully.

To ensure compatibility is kept for external apps using DPDK headers, we explicitly set the build parameters for the chkincs binary to the old minimum standard of "gnu89". [NOTE: DPDK code does not compile and has previously not compiled for pure c89 standard, so that stricter requirement need not be checked.] By adding this additional check, we can separately manage C standards used internally in DPDK builds and that required in the build flags for external apps using DPDK.

---
 buildtools/chkincs/meson.build | 1 +
 meson.build                    | 1 +
 2 files changed, 2 insertions(+)
buildtools/chkincs/meson.build +1 -0
index 378c2f19ef..322ac775ce 100644
@@ -30,6 +30,7 @@ executable('chkincs', sources,
         c_args: cflags,
         include_directories: includes,
         dependencies: deps,
+        override_options: 'c_std=gnu89',
         install: false)
 
 # run tests for c++ builds also
meson.build +1 -0
index f91d652bc5..9a2963cc16 100644
@@ -9,6 +9,7 @@ project('DPDK', 'C',
         license: 'BSD',
         default_options: [
             'buildtype=release',
+            'c_std=c99',
             'default_library=static',
             'warning_level=2',
         ],
Signature · 1 line
2.37.2
Acked-by on v1 1/1
#2 lore raw

RE: [RFC PATCH 1/1] build: increase minimum C standard for DPDK builds

Quoted 5 lines

From: Bruce Richardson [mailto:[email protected]]
Sent: Thursday, 12 January 2023 12.36

Set the default C language standard to be used for DPDK builds to C99. This requires no actual code changes to build successfully.

Great!

Quoted 9 lines

To ensure compatibility is kept for external apps using DPDK headers, we
explicitly set the build parameters for the chkincs binary to the old minimum standard of "gnu89". [NOTE: DPDK code does not compile and has previously not compiled for pure c89 standard, so that stricter requirement need not be checked.] By adding this additional check, we can separately manage C standards used internally in DPDK builds and that required in the build flags for external apps using DPDK.

It seems I have to accept this techboard decision for now. ;-)

Quoted 3 lines

---

#3 lore raw

Re: [RFC PATCH 1/1] build: increase minimum C standard for DPDK builds

Quoted 18 lines · Morten Brørup
Quoted 5 lines

From: Bruce Richardson [mailto:[email protected]]
Sent: Thursday, 12 January 2023 12.36

Set the default C language standard to be used for DPDK builds to C99. This requires no actual code changes to build successfully.

Great!

Quoted 9 lines

To ensure compatibility is kept for external apps using DPDK headers, we
explicitly set the build parameters for the chkincs binary to the old minimum standard of "gnu89". [NOTE: DPDK code does not compile and has previously not compiled for pure c89 standard, so that stricter requirement need not be checked.] By adding this additional check, we can separately manage C standards used internally in DPDK builds and that required in the build flags for external apps using DPDK.

Quoted 2 lines

It seems I have to accept this techboard decision for now. ;-)

There is no techboard decision here - certainly not yet. There is plenty of discussion still to be had. However, for now this RFC is being very conservative - perhaps overly so.

/Bruce

#4 lore raw

RE: [RFC PATCH 1/1] build: increase minimum C standard for DPDK builds

Quoted 26 lines

From: Bruce Richardson [mailto:[email protected]]
Sent: Thursday, 12 January 2023 13.47

Quoted 4 lines · Morten Brørup
Quoted 4 lines

From: Bruce Richardson [mailto:[email protected]]
Sent: Thursday, 12 January 2023 12.36

Set the default C language standard to be used for DPDK builds to

C99.

Quoted 6 lines
Quoted 1 line

This requires no actual code changes to build successfully.

Great!

Quoted 2 lines

To ensure compatibility is kept for external apps using DPDK

headers,

Quoted 2 lines
Quoted 2 lines

we
explicitly set the build parameters for the chkincs binary to the

old

Quoted 1 line
Quoted 1 line

minimum standard of "gnu89". [NOTE: DPDK code does not compile and

has

Quoted 2 lines
Quoted 2 lines

previously not compiled for pure c89 standard, so that stricter requirement need not be checked.] By adding this additional check,

we

Quoted 1 line
Quoted 1 line

can separately manage C standards used internally in DPDK builds

and

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

that required in the build flags for external apps using DPDK.

It seems I have to accept this techboard decision for now. ;-)

There is no techboard decision here - certainly not yet. There is plenty of
discussion still to be had. However, for now this RFC is being very conservative - perhaps overly so.

Yeah, perhaps I was phrasing that somewhat emotionally loaded. :-)

I didn't mean to cause confusion on the mailing list. Thanks for clarifying, Bruce.

Acked-by on v1 1/1
#5 lore raw

Re: [RFC PATCH 1/1] build: increase minimum C standard for DPDK builds

In reply to Bruce Richardson
Quoted 34 lines · Bruce Richardson

Set the default C language standard to be used for DPDK builds to C99. This requires no actual code changes to build successfully.

To ensure compatibility is kept for external apps using DPDK headers, we explicitly set the build parameters for the chkincs binary to the old minimum standard of "gnu89". [NOTE: DPDK code does not compile and has previously not compiled for pure c89 standard, so that stricter requirement need not be checked.] By adding this additional check, we can separately manage C standards used internally in DPDK builds and that required in the build flags for external apps using DPDK.

---
 buildtools/chkincs/meson.build | 1 +
 meson.build                    | 1 +
 2 files changed, 2 insertions(+)
buildtools/chkincs/meson.build +1 -0
index 378c2f19ef..322ac775ce 100644
@@ -30,6 +30,7 @@ executable('chkincs', sources,
         c_args: cflags,
         include_directories: includes,
         dependencies: deps,
+        override_options: 'c_std=gnu89',
         install: false)
 
 # run tests for c++ builds also
meson.build +0 -0
index f91d652bc5..9a2963cc16 100644

meson.build project('DPDK', 'C', · line 15

Quoted 7 lines
meson.build +1 -0
@@ -9,6 +9,7 @@ project('DPDK', 'C',
         license: 'BSD',
         default_options: [
             'buildtype=release',
+            'c_std=c99',
             'default_library=static',
             'warning_level=2',
         ],

subject to the atomics abstraction proposal where a meson option is provided enable_stdatomics=true we'll want to be able to upgrade to -std=c11 / c_std=c11 (as a minimum).

is there a meson mechanism that will let us evaluate that condition and specify the higher required standard version in default_options or is it a matter of some post project() or is this just done by adding -std=c11 using add_project_arguments() after project()?

second, i think you will run into a build break with this change on some platform / compiler combinations. somewhere we are using strerror_r where the required posix versions are not specified in the translation unit worked around by using c_std=gnu99 or by adding appropriate undef GNUC etc.. in the place where strerror_r is used. (just something i noticed when testing a similar change during prototyping i think ubuntu 22.04?)

going one step further i'd just ask that the default options be "portable" i.e. they work with ! gcc and ! clang to help me with future work. though i know that contradicts with my advice in the previous paragraph, maybe we can set c_std=xxx subject to toolchain/platform?

#6 lore raw
In reply to Bruce Richardson

Since this topic keeps coming up in other threads I'll chime in with my $0.01 here. We've been using CentOS 7 for awhile (and working on migrating off) but have had to leverage devtoolset/llvmtoolset for various reasons. I remember a discussion of installing a different compiler coming up but don't remember which thread that was in/what the outcome was. While I'd like to just brush over C7 and say there is a compatible compiler available so just make the change I also realize that making that change could be quite disruptive to existing code bases.

However, the 22.11 LTS will be EOL in Nov 2024. CentOS 7 is EOL Jun 2024. For the 23.x series and going forward I don't think starting with a C11 requirement is an unreasonable ask.

Quoted 25 lines · Bruce Richardson

Traditionally, DPDK has never specified a minimum C standard used either in DPDK builds or for applications using DPDK. Following discussion on-list about C standards, this RFC attempts to start the process of codifying what our standards expectations are. No code changes are made by this RFC, instead only the build parameters are changed to explicitly specify:

* C99 standard is used to build DPDK itself. This is supported by all supported compiler versions of GCC and Clang.
* The headers are checked for compatibility with gcc89 standard, which was the default standard used by the oldest supported version of GCC. DPDK headers do not build with the official C89 standard, and, to the best of my knowledge, have never done so.

Bruce Richardson (1):
build: increase minimum C standard for DPDK builds

 buildtools/chkincs/meson.build | 1 +
 meson.build                    | 1 +
 2 files changed, 2 insertions(+)

--
2.37.2

#7 lore raw
Quoted 4 lines · Ben Magistro

Since this topic keeps coming up in other threads I'll chime in with my $0.01 here. We've been using CentOS 7 for awhile (and working on migrating off) but have had to leverage devtoolset/llvmtoolset for various reasons. I remember a discussion of installing a different

Quoted 9 lines

compiler coming up but don't remember which thread that was in/what the outcome was. While I'd like to just brush over C7 and say there is a compatible compiler available so just make the change I also realize that making that change could be quite disruptive to existing code bases.
However, the 22.11 LTS will be EOL in Nov 2024. CentOS 7 is EOL Jun 2024. For the 23.x series and going forward I don't think starting with a C11 requirement is an unreasonable ask.

Thanks for that input. If we drop support for Centos/RHEL 7, I think we should be ok to pass -std=c11 for the build of DPDK.

Have you any thoughts on the second part of the c11 move - where our headers require c11 support and therefore may require that the end user builds their own code using -std=c11? This latter part is the bit that concerns me a little, as I feel it may be problematic for some with older codebases.

/Bruce

#8 lore raw

In our case we have other libraries that we are using that have required us to specify a minimum c++ version (14/17 most recently for one) so it doesn't feel like a big ask/issue to us (provided things don't start conflicting...hah; not anticipating any issue). Our software is also used internally so we have a fair bit of control over how fast we can adopt changes.

This got me wondering what some other projects in the DPDK ecosystem are saying/doing around language standards/gcc versions. So some quick checking of the projects I am aware of/looked at/using...
* trex: cannot find an obvious minimum gcc requirement
* tldk: we are running our own public folk with several fixes, need to find time to solve the build sys change aspect to continue providing patches upstream; I know I have hit some places where it was easier to say the new minimum DPDK version is x at which point you just adopt the minimum requirements of DPDK
* ovs: looks to be comfortable with an older gcc still
* seastar: seems to be the most aggressive with adopting language standards/compilers I've seen [1] and are asking for gcc 9+ and cpp17+
* ans: based on release 19.02 (2019), they are on gcc >= 5.4 [2] and is the same on the main README file

I do understand the concern, but if no one is voicing an opinion/objection does that mean they agree with/will not be affected by the change....

1) https://docs.seastar.io/master/md_compatibility.html
2) https://github.com/ansyun/dpdk-ans/releases

Cheers

Quoted 27 lines · Bruce Richardson
Quoted 1 line · Ben Magistro

Since this topic keeps coming up in other threads I'll chime in with

my

Quoted 4 lines

$0.01 here. We've been using CentOS 7 for awhile (and working on migrating off) but have had to leverage devtoolset/llvmtoolset for various reasons. I remember a discussion of installing a different compiler coming up but don't remember which thread that was in/what

the

Quoted 8 lines

outcome was. While I'd like to just brush over C7 and say there is a compatible compiler available so just make the change I also realize that making that change could be quite disruptive to existing code bases.
However, the 22.11 LTS will be EOL in Nov 2024. CentOS 7 is EOL Jun 2024. For the 23.x series and going forward I don't think starting with a C11 requirement is an unreasonable ask.

Thanks for that input. If we drop support for Centos/RHEL 7, I think we should be ok to pass -std=c11 for the build of DPDK.

Have you any thoughts on the second part of the c11 move - where our headers require c11 support and therefore may require that the end user builds their own code using -std=c11? This latter part is the bit that concerns me a little, as I feel it may be problematic for some with older codebases.

/Bruce

#9 lore raw
Quoted 18 lines · Ben Magistro

In our case we have other libraries that we are using that have required us to specify a minimum c++ version (14/17 most recently for one) so it doesn't feel like a big ask/issue to us (provided things don't start conflicting...hah; not anticipating any issue). Our software is also used internally so we have a fair bit of control over how fast we can adopt changes.
This got me wondering what some other projects in the DPDK ecosystem are saying/doing around language standards/gcc versions. So some quick checking of the projects I am aware of/looked at/using...
* trex: cannot find an obvious minimum gcc requirement
* tldk: we are running our own public folk with several fixes, need to find time to solve the build sys change aspect to continue providing patches upstream; I know I have hit some places where it was easier to say the new minimum DPDK version is x at which point you just adopt the minimum requirements of DPDK
* ovs: looks to be comfortable with an older gcc still
* seastar: seems to be the most aggressive with adopting language standards/compilers I've seen [1] and are asking for gcc 9+ and cpp17+

Quoted 9 lines

* ans: based on release 19.02 (2019), they are on gcc >= 5.4 [2] and is the same on the main README file
I do understand the concern, but if no one is voicing an opinion/objection does that mean they agree with/will not be affected by the change....
1) [1]https://docs.seastar.io/master/md_compatibility.html
2) [2]https://github.com/ansyun/dpdk-ans/releases
Cheers

Thanks for the info.
I also notice that since gcc 5, the default language version used - if none is explicitly specified - is gnu11 (or higher for later versions). Clang seems to do something similar, but not sure at what point it started defaulting to a standard >=c11.

/Bruce

#10 lore raw

Adding Tyler

Sort of following along on the RFC: introduce atomics [1] it seems like the decision to use 99 vs 11 here could make an impact on the approach taken in that thread.

1) http://mails.dpdk.org/archives/dev/2023-February/262042.html

Quoted 39 lines · Bruce Richardson
Quoted 8 lines · Ben Magistro

In our case we have other libraries that we are using that have required us to specify a minimum c++ version (14/17 most recently for one) so it doesn't feel like a big ask/issue to us (provided things don't start conflicting...hah; not anticipating any issue). Our software is also used internally so we have a fair bit of control over how fast we can adopt changes.
This got me wondering what some other projects in the DPDK ecosystem are saying/doing around language standards/gcc versions. So some

quick

Quoted 6 lines

checking of the projects I am aware of/looked at/using...
* trex: cannot find an obvious minimum gcc requirement
* tldk: we are running our own public folk with several fixes, need to find time to solve the build sys change aspect to continue providing patches upstream; I know I have hit some places where it was easier to say the new minimum DPDK version is x at which point you just adopt

the

Quoted 5 lines

minimum requirements of DPDK
* ovs: looks to be comfortable with an older gcc still
* seastar: seems to be the most aggressive with adopting language standards/compilers I've seen [1] and are asking for gcc 9+ and cpp17+
* ans: based on release 19.02 (2019), they are on gcc >= 5.4 [2] and

is

Quoted 8 lines

the same on the main README file
I do understand the concern, but if no one is voicing an opinion/objection does that mean they agree with/will not be affected by the change....
1) [1]https://docs.seastar.io/master/md_compatibility.html
2) [2]https://github.com/ansyun/dpdk-ans/releases
Cheers

Thanks for the info.
I also notice that since gcc 5, the default language version used - if none is explicitly specified - is gnu11 (or higher for later versions). Clang seems to do something similar, but not sure at what point it started defaulting to a standard >=c11.

/Bruce

#11 lore raw
Quoted 5 lines · Ben Magistro

Adding Tyler

Sort of following along on the RFC: introduce atomics [1] it seems like the decision to use 99 vs 11 here could make an impact on the approach taken in that thread.

hey Ben thanks for keeping an eye across threads on the topic. the atomics thread is fairly long but somewhere in it i did provide a rationale for why we can't just go straight to using C11 even if we declared that dpdk on supports compilers >= C11.

i wish we could it would certainly make my life way easier if i could just -std=c11 and cut & paste my way to completion. the reason why we can't (aside from not requiring C11 compiler as a minimum) is that there is potential issue with abi compatibility for existing applications using non-atomic types currently passed to ABI suddenly requiring standard atomic types. this is because _Atomic type and type are not guaranteed to have the same size, alignment, representation etc..

anyway, i welcome us establishing c99 as a minimum for all toolchain/platform combinations.

Quoted 45 lines

1) http://mails.dpdk.org/archives/dev/2023-February/262042.html

Quoted 39 lines · Bruce Richardson
Quoted 8 lines · Ben Magistro

In our case we have other libraries that we are using that have required us to specify a minimum c++ version (14/17 most recently for one) so it doesn't feel like a big ask/issue to us (provided things don't start conflicting...hah; not anticipating any issue). Our software is also used internally so we have a fair bit of control over how fast we can adopt changes.
This got me wondering what some other projects in the DPDK ecosystem are saying/doing around language standards/gcc versions. So some

quick

Quoted 6 lines

checking of the projects I am aware of/looked at/using...
* trex: cannot find an obvious minimum gcc requirement
* tldk: we are running our own public folk with several fixes, need to find time to solve the build sys change aspect to continue providing patches upstream; I know I have hit some places where it was easier to say the new minimum DPDK version is x at which point you just adopt

the

Quoted 5 lines

minimum requirements of DPDK
* ovs: looks to be comfortable with an older gcc still
* seastar: seems to be the most aggressive with adopting language standards/compilers I've seen [1] and are asking for gcc 9+ and cpp17+
* ans: based on release 19.02 (2019), they are on gcc >= 5.4 [2] and

is

Quoted 8 lines

the same on the main README file
I do understand the concern, but if no one is voicing an opinion/objection does that mean they agree with/will not be affected by the change....
1) [1]https://docs.seastar.io/master/md_compatibility.html
2) [2]https://github.com/ansyun/dpdk-ans/releases
Cheers

Thanks for the info.
I also notice that since gcc 5, the default language version used - if none is explicitly specified - is gnu11 (or higher for later versions). Clang seems to do something similar, but not sure at what point it started defaulting to a standard >=c11.

/Bruce

#12 lore raw
In reply to Bruce Richardson
Quoted 7 lines · Bruce Richardson

Traditionally, DPDK has never specified a minimum C standard used either in DPDK builds or for applications using DPDK. Following discussion on-list about C standards, this RFC attempts to start the process of codifying what our standards expectations are. No code changes are made by this RFC, instead only the build parameters are changed to explicitly specify:

Quoted 6 lines

* C99 standard is used to build DPDK itself. This is supported by all supported compiler versions of GCC and Clang.
* The headers are checked for compatibility with gcc89 standard, which was the default standard used by the oldest supported version of GCC. DPDK headers do not build with the official C89 standard, and, to the best of my knowledge, have never done so.

subject to the technical board meeting 2023/02/22 in relation to atomics and adoption of C11 starting in 23.11 does anything stop us from conditionally enabling/defaulting -std=C11 for all platforms immediately except for RHEL/CentOS 7?

so long as we don't actually start using C11 features we should be able to do this? or would we be worried that C11 feature use would creep in?

just curious.

Quoted 10 lines

Bruce Richardson (1):
build: increase minimum C standard for DPDK builds

 buildtools/chkincs/meson.build | 1 +
 meson.build                    | 1 +
 2 files changed, 2 insertions(+)

--
2.37.2

#13 lore raw
Quoted 20 lines · Tyler Retzlaff
Quoted 7 lines · Bruce Richardson

Traditionally, DPDK has never specified a minimum C standard used either in DPDK builds or for applications using DPDK. Following discussion on-list about C standards, this RFC attempts to start the process of codifying what our standards expectations are. No code changes are made by this RFC, instead only the build parameters are changed to explicitly specify:

Quoted 6 lines

* C99 standard is used to build DPDK itself. This is supported by all supported compiler versions of GCC and Clang.
* The headers are checked for compatibility with gcc89 standard, which was the default standard used by the oldest supported version of GCC. DPDK headers do not build with the official C89 standard, and, to the best of my knowledge, have never done so.

subject to the technical board meeting 2023/02/22 in relation to atomics and adoption of C11 starting in 23.11 does anything stop us from conditionally enabling/defaulting -std=C11 for all platforms immediately except for RHEL/CentOS 7?

Quoted 5 lines

so long as we don't actually start using C11 features we should be able to do this? or would we be worried that C11 feature use would creep in?

just curious.

Actually, if we don't do anything, the versions of gcc (and clang AFAIK) already default to C11 or later from GCC 5 onwards. If we were to specify a version, I think it would have to be gnu11 as we may still be using some GCC extensions. However, feel free to do up a patch for c11 if it works. The change to the header checks probably don't need to be included, only a change to the default options in the top-level meson.build file.

Incidentally, even though it is missing support for the c11 atomics, gcc 4.8.5 on RHEL 7 does have the -std=c11 flag that can be used, so adding that shouldn't break anything.

/Bruce