[dpdk-dev] [PATCH] eal: fix build with conflicting libc variable memory_order

[dpdk-dev] [PATCH] eal: fix build with conflicting libc variable memory_order

[Eli Britstein]

The cited commit introduced functions with 'int memory_order' argument.
The C11 standard section 7.17.1.4 defines 'memory_order' as the
"enumerated type whose enumerators identify memory ordering constraints".
Applications that use the standard enum (includes stdatomic.h), will
fail compilation with:
error: declaration of 'memory_order' shadows a global declaration
    [-Werror=shadow]
     rte_atomic_thread_fence(int memory_order)
Fix it by changing the argument name 'memory_order' to 'memorder'.

Fixes: 672a15056380 ("eal: add wrapper for C11 atomic thread fence")
Signed-off-by: Eli Britstein <elibr@nvidia.com>
---
 lib/librte_eal/arm/include/rte_atomic_32.h  | 4 ++--
 lib/librte_eal/arm/include/rte_atomic_64.h  | 4 ++--
 lib/librte_eal/include/generic/rte_atomic.h | 2 +-
 lib/librte_eal/ppc/include/rte_atomic.h     | 4 ++--
 lib/librte_eal/x86/include/rte_atomic.h     | 6 +++---
 5 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/lib/librte_eal/arm/include/rte_atomic_32.h b/lib/librte_eal/arm/include/rte_atomic_32.h
index 9d0568d497..fe48ab428e 100644
--- a/lib/librte_eal/arm/include/rte_atomic_32.h
+++ b/lib/librte_eal/arm/include/rte_atomic_32.h
@@ -34,9 +34,9 @@ extern "C" {
 #define rte_io_rmb() rte_rmb()
 
 static __rte_always_inline void
-rte_atomic_thread_fence(int memory_order)
+rte_atomic_thread_fence(int memorder)
 {
-	__atomic_thread_fence(memory_order);
+	__atomic_thread_fence(memorder);
 }
 
 #ifdef __cplusplus
diff --git a/lib/librte_eal/arm/include/rte_atomic_64.h b/lib/librte_eal/arm/include/rte_atomic_64.h
index c518559bc9..20dd6c75dd 100644
--- a/lib/librte_eal/arm/include/rte_atomic_64.h
+++ b/lib/librte_eal/arm/include/rte_atomic_64.h
@@ -38,9 +38,9 @@ extern "C" {
 #define rte_io_rmb() rte_rmb()
 
 static __rte_always_inline void
-rte_atomic_thread_fence(int memory_order)
+rte_atomic_thread_fence(int memorder)
 {
-	__atomic_thread_fence(memory_order);
+	__atomic_thread_fence(memorder);
 }
 
 /*------------------------ 128 bit atomic operations -------------------------*/
diff --git a/lib/librte_eal/include/generic/rte_atomic.h b/lib/librte_eal/include/generic/rte_atomic.h
index d1255b2d8c..276272f40b 100644
--- a/lib/librte_eal/include/generic/rte_atomic.h
+++ b/lib/librte_eal/include/generic/rte_atomic.h
@@ -122,7 +122,7 @@ static inline void rte_io_rmb(void);
 /**
  * Synchronization fence between threads based on the specified memory order.
  */
-static inline void rte_atomic_thread_fence(int memory_order);
+static inline void rte_atomic_thread_fence(int memorder);
 
 /*------------------------- 16 bit atomic operations -------------------------*/
 
diff --git a/lib/librte_eal/ppc/include/rte_atomic.h b/lib/librte_eal/ppc/include/rte_atomic.h
index a91989930b..6a7e65210c 100644
--- a/lib/librte_eal/ppc/include/rte_atomic.h
+++ b/lib/librte_eal/ppc/include/rte_atomic.h
@@ -37,9 +37,9 @@ extern "C" {
 #define rte_io_rmb() rte_rmb()
 
 static __rte_always_inline void
-rte_atomic_thread_fence(int memory_order)
+rte_atomic_thread_fence(int memorder)
 {
-	__atomic_thread_fence(memory_order);
+	__atomic_thread_fence(memorder);
 }
 
 /*------------------------- 16 bit atomic operations -------------------------*/
diff --git a/lib/librte_eal/x86/include/rte_atomic.h b/lib/librte_eal/x86/include/rte_atomic.h
index b7d6b06ddf..915afd9d27 100644
--- a/lib/librte_eal/x86/include/rte_atomic.h
+++ b/lib/librte_eal/x86/include/rte_atomic.h
@@ -87,12 +87,12 @@ rte_smp_mb(void)
  * used instead.
  */
 static __rte_always_inline void
-rte_atomic_thread_fence(int memory_order)
+rte_atomic_thread_fence(int memorder)
 {
-	if (memory_order == __ATOMIC_SEQ_CST)
+	if (memorder == __ATOMIC_SEQ_CST)
 		rte_smp_mb();
 	else
-		__atomic_thread_fence(memory_order);
+		__atomic_thread_fence(memorder);
 }
 
 /*------------------------- 16 bit atomic operations -------------------------*/
-- 
2.28.0.546.g385c171

Re: [dpdk-dev] [PATCH] eal: fix build with conflicting libc variable memory_order

[Asaf Penso]

>-----Original Message-----
>From: dev <dev-bounces@dpdk.org> On Behalf Of Eli Britstein
>Sent: Wednesday, October 14, 2020 9:19 AM
>To: dev@dpdk.org
>Cc: Eli Britstein <elibr@nvidia.com>
>Subject: [dpdk-dev] [PATCH] eal: fix build with conflicting libc variable
>memory_order
>
>The cited commit introduced functions with 'int memory_order' argument.
>The C11 standard section 7.17.1.4 defines 'memory_order' as the
>"enumerated type whose enumerators identify memory ordering
>constraints".
>Applications that use the standard enum (includes stdatomic.h), will fail
>compilation with:
>error: declaration of 'memory_order' shadows a global declaration
>    [-Werror=shadow]
>     rte_atomic_thread_fence(int memory_order) Fix it by changing the
>argument name 'memory_order' to 'memorder'.
>
>Fixes: 672a15056380 ("eal: add wrapper for C11 atomic thread fence")
>Signed-off-by: Eli Britstein <elibr@nvidia.com>
Reviewed-by: Asaf Penso <asafp@nvidia.com>

>---
> lib/librte_eal/arm/include/rte_atomic_32.h  | 4 ++--
>lib/librte_eal/arm/include/rte_atomic_64.h  | 4 ++--
>lib/librte_eal/include/generic/rte_atomic.h | 2 +-
> lib/librte_eal/ppc/include/rte_atomic.h     | 4 ++--
> lib/librte_eal/x86/include/rte_atomic.h     | 6 +++---
> 5 files changed, 10 insertions(+), 10 deletions(-)
>
>diff --git a/lib/librte_eal/arm/include/rte_atomic_32.h
>b/lib/librte_eal/arm/include/rte_atomic_32.h
>index 9d0568d497..fe48ab428e 100644
>--- a/lib/librte_eal/arm/include/rte_atomic_32.h
>+++ b/lib/librte_eal/arm/include/rte_atomic_32.h
>@@ -34,9 +34,9 @@ extern "C" {
> #define rte_io_rmb() rte_rmb()
>
> static __rte_always_inline void
>-rte_atomic_thread_fence(int memory_order)
>+rte_atomic_thread_fence(int memorder)
> {
>-	__atomic_thread_fence(memory_order);
>+	__atomic_thread_fence(memorder);
> }
>
> #ifdef __cplusplus
>diff --git a/lib/librte_eal/arm/include/rte_atomic_64.h
>b/lib/librte_eal/arm/include/rte_atomic_64.h
>index c518559bc9..20dd6c75dd 100644
>--- a/lib/librte_eal/arm/include/rte_atomic_64.h
>+++ b/lib/librte_eal/arm/include/rte_atomic_64.h
>@@ -38,9 +38,9 @@ extern "C" {
> #define rte_io_rmb() rte_rmb()
>
> static __rte_always_inline void
>-rte_atomic_thread_fence(int memory_order)
>+rte_atomic_thread_fence(int memorder)
> {
>-	__atomic_thread_fence(memory_order);
>+	__atomic_thread_fence(memorder);
> }
>
> /*------------------------ 128 bit atomic operations -------------------------*/ diff --
>git a/lib/librte_eal/include/generic/rte_atomic.h
>b/lib/librte_eal/include/generic/rte_atomic.h
>index d1255b2d8c..276272f40b 100644
>--- a/lib/librte_eal/include/generic/rte_atomic.h
>+++ b/lib/librte_eal/include/generic/rte_atomic.h
>@@ -122,7 +122,7 @@ static inline void rte_io_rmb(void);
> /**
>  * Synchronization fence between threads based on the specified memory
>order.
>  */
>-static inline void rte_atomic_thread_fence(int memory_order);
>+static inline void rte_atomic_thread_fence(int memorder);
>
> /*------------------------- 16 bit atomic operations -------------------------*/
>
>diff --git a/lib/librte_eal/ppc/include/rte_atomic.h
>b/lib/librte_eal/ppc/include/rte_atomic.h
>index a91989930b..6a7e65210c 100644
>--- a/lib/librte_eal/ppc/include/rte_atomic.h
>+++ b/lib/librte_eal/ppc/include/rte_atomic.h
>@@ -37,9 +37,9 @@ extern "C" {
> #define rte_io_rmb() rte_rmb()
>
> static __rte_always_inline void
>-rte_atomic_thread_fence(int memory_order)
>+rte_atomic_thread_fence(int memorder)
> {
>-	__atomic_thread_fence(memory_order);
>+	__atomic_thread_fence(memorder);
> }
>
> /*------------------------- 16 bit atomic operations -------------------------*/ diff --
>git a/lib/librte_eal/x86/include/rte_atomic.h
>b/lib/librte_eal/x86/include/rte_atomic.h
>index b7d6b06ddf..915afd9d27 100644
>--- a/lib/librte_eal/x86/include/rte_atomic.h
>+++ b/lib/librte_eal/x86/include/rte_atomic.h
>@@ -87,12 +87,12 @@ rte_smp_mb(void)
>  * used instead.
>  */
> static __rte_always_inline void
>-rte_atomic_thread_fence(int memory_order)
>+rte_atomic_thread_fence(int memorder)
> {
>-	if (memory_order == __ATOMIC_SEQ_CST)
>+	if (memorder == __ATOMIC_SEQ_CST)
> 		rte_smp_mb();
> 	else
>-		__atomic_thread_fence(memory_order);
>+		__atomic_thread_fence(memorder);
> }
>
> /*------------------------- 16 bit atomic operations -------------------------*/
>--
>2.28.0.546.g385c171

Re: [dpdk-dev] [PATCH] eal: fix build with conflicting libc variable memory_order

[Thomas Monjalon]

14/10/2020 08:19, Eli Britstein:
> The cited commit introduced functions with 'int memory_order' argument.
> The C11 standard section 7.17.1.4 defines 'memory_order' as the
> "enumerated type whose enumerators identify memory ordering constraints".
> Applications that use the standard enum (includes stdatomic.h), will
> fail compilation with:
> error: declaration of 'memory_order' shadows a global declaration
>     [-Werror=shadow]
>      rte_atomic_thread_fence(int memory_order)
> Fix it by changing the argument name 'memory_order' to 'memorder'.

Not clear why it builds fine with most compilers,
but the fix does not hurt.

> Fixes: 672a15056380 ("eal: add wrapper for C11 atomic thread fence")

A blank line should be inserted here.

> Signed-off-by: Eli Britstein <elibr@nvidia.com>

Acked-by: Thomas Monjalon <thomas@monjalon.net>

Re: [dpdk-dev] [PATCH] eal: fix build with conflicting libc variable memory_order

[David Marchand]

On Thu, Oct 15, 2020 at 1:16 AM Thomas Monjalon <thomas@monjalon.net> wrote:
>
> 14/10/2020 08:19, Eli Britstein:
> > The cited commit introduced functions with 'int memory_order' argument.
> > The C11 standard section 7.17.1.4 defines 'memory_order' as the
> > "enumerated type whose enumerators identify memory ordering constraints".
> > Applications that use the standard enum (includes stdatomic.h), will
> > fail compilation with:

Simply including stdatomic.h does not trigger the problem.
Can you rework this commitlog with below info?

> > error: declaration of 'memory_order' shadows a global declaration
> >     [-Werror=shadow]
> >      rte_atomic_thread_fence(int memory_order)
> > Fix it by changing the argument name 'memory_order' to 'memorder'.
>
> Not clear why it builds fine with most compilers,

I can reproduce in two cases:
- with a gcc that provides a stdatomic.h header + passing
-Wsystem-headers in the CFLAGS,
- with a compiler that does not provide stdatomic.h and you redefine
memory_order in your code (like OVS does),


> but the fix does not hurt.

Otherwise, yes, lgtm.


> > Fixes: 672a15056380 ("eal: add wrapper for C11 atomic thread fence")
>
> A blank line should be inserted here.
>
> > Signed-off-by: Eli Britstein <elibr@nvidia.com>
>
> Acked-by: Thomas Monjalon <thomas@monjalon.net>

Acked-by: David Marchand <david.marchand@redhat.com>


-- 
David Marchand