ether: mark ethernet addresses as being 2-byte aligned
PATCH 1/1 ether: mark ethernet addresses as being 2-byte aligned
When including the rte_ether.h header in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.
.../include/rte_ether.h:149:2: warning: converting a packed ‘const
struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]
149 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea;
| ^~~~~Since ethernet addresses should always be aligned on a two-byte boundary, we can just inform the compiler of this assumption to remove the warnings and allow us to always access the addresses using 16-bit operations.
---
Although this is an ABI break, the network structures are all being renamed in this release, and a deprecation notice was previously posted for it.
--- lib/librte_net/rte_ether.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
lib/librte_net/rte_ether.h +2 -1
@@ -55,7 +55,8 @@ extern "C" {* See http://standards.ieee.org/regauth/groupmac/tutorial.html */ struct ether_addr { -uint8_t addr_bytes[ETHER_ADDR_LEN ];/**< Addr bytes in tx order */ +/** Addr bytes in tx order */ + uint8_t addr_bytes[ETHER_ADDR_LEN] __rte_aligned(2); } __attribute__((__packed__ ));#define ETHER_LOCAL_ADMIN_ADDR 0x02/**< Locally assigned Eth. address. */
Signature · 1 line
2.21.0
Quoted 11 lines · Bruce Richardson
When including the
rte_ether.hheader in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.
.../include/rte_ether.h:149:2: warning: converting a packed ‘const
struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]149 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea; | ^~~~~
Hi - There was a couple of these warnings in telemetry that were not squashed with the patch to disable the warning in 19.05. I was just about send a patch to address that when I saw this one.
As your patch is aiming for a better solution, I won't send my patch for master but I'll still send to stable.
Quoted 29 lines
Since ethernet addresses should always be aligned on a two-byte boundary, we can just inform the compiler of this assumption to remove the warnings and allow us to always access the addresses using 16-bit operations.
Signed-off-by Bruce Richardson <[email protected]>---
Although this is an ABI break, the network structures are all being renamed in this release, and a deprecation notice was previously posted for it.
--- lib/librte_net/rte_ether.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)lib/librte_net/rte_ether.h +2
-1@@ -55,7 +55,8 @@ extern "C" {* See http://standards.ieee.org/regauth/groupmac/tutorial.html */ struct ether_addr { -uint8_t addr_bytes[ETHER_ADDR_LEN ];/**< Addr bytes in tx order */ +/** Addr bytes in tx order */ + uint8_t addr_bytes[ETHER_ADDR_LEN] __rte_aligned(2); } __attribute__((__packed__ ));#define ETHER_LOCAL_ADMIN_ADDR 0x02/**< Locally assigned Eth. address. */
Quoted 13 lines · Kevin Traynor
Quoted 11 lines · Bruce Richardson
When including the
rte_ether.hheader in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.
.../include/rte_ether.h:149:2: warning: converting a packed ‘const
struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]149 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea; | ^~~~~
Quoted 7 lines
Hi - There was a couple of these warnings in telemetry that were not squashed with the patch to disable the warning in 19.05. I was just about send a patch to address that when I saw this one.
As your patch is aiming for a better solution, I won't send my patch for master but I'll still send to stable.
Good idea, thanks.
Hi Bruce,
Quoted 4 lines · Bruce Richardson
When including the
rte_ether.hheader in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.
Quoted 8 lines
.../include/rte_ether.h:149:2: warning: converting a packed ‘const
struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]149 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea; | ^~~~~Since ethernet addresses should always be aligned on a two-byte boundary,
I'm a bit reserved about this last assumption. The ethernet address structure may be used in a private structure, whose alignment is 1. Are we sure that there is no (funny) protocol that carries unaligned ethernet addresses?
Shouldn't we change the definition of unaligned_uint16_t instead? Or change the rte_is_broadcast_ether_addr() function?
Quoted 9 lines
we can just inform the compiler of this assumption to remove the warnings and allow us to always access the addresses using 16-bit operations.
Signed-off-by Bruce Richardson <[email protected]>---
Although this is an ABI break, the network structures are all being renamed in this release, and a deprecation notice was previously posted for it.
Yes, but the network renaming is identified in the release note as an API break, not an ABI break.
Quoted 21 lines
--- lib/librte_net/rte_ether.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)lib/librte_net/rte_ether.h +2
-1@@ -55,7 +55,8 @@ extern "C" {* See http://standards.ieee.org/regauth/groupmac/tutorial.html */ struct ether_addr { -uint8_t addr_bytes[ETHER_ADDR_LEN ];/**< Addr bytes in tx order */ +/** Addr bytes in tx order */ + uint8_t addr_bytes[ETHER_ADDR_LEN] __rte_aligned(2); } __attribute__((__packed__ ));#define ETHER_LOCAL_ADMIN_ADDR 0x02/**< Locally assigned Eth. address. */ Signature · 1 line
2.21.0
Quoted 16 lines · Olivier Matz
Hi Bruce,
Quoted 4 lines · Bruce Richardson
When including the
rte_ether.hheader in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.Quoted 8 lines
.../include/rte_ether.h:149:2: warning: converting a packed ‘const
struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]149 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea; | ^~~~~Since ethernet addresses should always be aligned on a two-byte boundary,
Quoted 8 lines
I'm a bit reserved about this last assumption. The ethernet address structure may be used in a private structure, whose alignment is 1. Are we sure that there is no (funny) protocol that carries unaligned ethernet addresses?
Shouldn't we change the definition of
unaligned_uint16_tinstead? Or change therte_is_broadcast_ether_addr()function?
We could, but I believe the correct behaviour is to make the addresses always 2-byte aligned, unless someone actually has a real-world case where there is a protocol that doesn't have the data 2-byte aligned.
/Bruce
Hi Bruce,
Quoted 22 lines · Bruce Richardson
Quoted 21 lines · Olivier Matz
Hi Bruce,
Quoted 4 lines · Bruce Richardson
When including the
rte_ether.hheader in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.Quoted 8 lines
.../include/rte_ether.h:149:2: warning: converting a packed ‘const
struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]149 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea; | ^~~~~Since ethernet addresses should always be aligned on a two-byte boundary,
I'm a bit reserved about this last assumption. The ethernet address structure may be used in a private structure, whose alignment is 1. Are we sure that there is no (funny) protocol that carries unaligned ethernet addresses?
Quoted 7 lines
Quoted 3 lines
Shouldn't we change the definition of
unaligned_uint16_tinstead? Or change therte_is_broadcast_ether_addr()function?We could, but I believe the correct behaviour is to make the addresses always 2-byte aligned, unless someone actually has a real-world case where there is a protocol that doesn't have the data 2-byte aligned.
Maybe you missed that part of my previous answer, I'm copy it again here:
Quoted 2 lines
Although this is an ABI break, the network structures are all being renamed in this release, and a deprecation notice was previously posted for it.
Yes, but the network renaming is identified in the release note as an API break, not an ABI break.
I thought we agreed to limit ABI breakages to cases where there is no other solution. Here, this is surely a "small" ABI breakage, but I suppose there is a way to do differently.
If we really want to do that way, it's better to announce it as an ABI break.
Olivier
Quoted 43 lines · Olivier Matz
Hi Bruce,
Quoted 23 lines · Bruce Richardson
Quoted 22 lines · Olivier Matz
Hi Bruce,
Quoted 11 lines · Bruce Richardson
When including the
rte_ether.hheader in applications with warnings enabled, a warning was given because of the assumption of 2-byte alignment of ethernet addresses when processing them.
.../include/rte_ether.h:149:2: warning: converting a packed ‘const struct ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member] 149 | constunaligned_uint16_t*ea_words= (constunaligned_uint16_t*)ea; | ^~~~~Quoted 2 lines
Since ethernet addresses should always be aligned on a two-byte boundary,
I'm a bit reserved about this last assumption. The ethernet address structure may be used in a private structure, whose alignment is 1. Are we sure that there is no (funny) protocol that carries unaligned ethernet addresses?
Quoted 7 lines
Quoted 3 lines
Shouldn't we change the definition of
unaligned_uint16_tinstead? Or change therte_is_broadcast_ether_addr()function?We could, but I believe the correct behaviour is to make the addresses always 2-byte aligned, unless someone actually has a real-world case where there is a protocol that doesn't have the data 2-byte aligned.
Maybe you missed that part of my previous answer, I'm copy it again here:
Quoted 3 lines
Although this is an ABI break, the network structures are all being renamed in this release, and a deprecation notice was previously posted for it.
Yes, but the network renaming is identified in the release note as an API break, not an ABI break.
Quoted 7 lines
I thought we agreed to limit ABI breakages to cases where there is no other solution. Here, this is surely a "small" ABI breakage, but I suppose there is a way to do differently.
If we really want to do that way, it's better to announce it as an ABI break.
At this stage, I'm ok with pushing this to 19.11 to follow the official deprecation process.
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 26 lines · Martins Eglitis
Dear Bruce and Kevin,
I tried building an application (NFF-GO) which has DPDK as a dependency. I am still getting the same warnings you and Kevin were discussing. My current DPDK version is 19.11-1.
Do you know if this issue has been resolved?
This is the output:
# github.com/intel-go/nff-go/internal/low In file included from /home/zebra/Projects/nff-go/dpdk/dpdk/x86_64-native-linuxapp-gcc-install/usr/local/share/dpdk/x86_64-native-linuxapp-gcc/include/rte_ethdev.h:160, from ../../go/pkg/mod/github.com/intel-go/[email protected]/internal/low/low.h:11, from ../../go/pkg/mod/github.com/intel-go/[email protected]/internal/low/low.go:16: /home/zebra/Projects/nff-go/dpdk/dpdk/x86_64-native-linuxapp-gcc-install/usr/local/share/dpdk/x86_64-native-linuxapp-gcc/include/rte_ether.h: In function ‘rte_is_same_ether_addr’: /home/zebra/Projects/nff-go/dpdk/dpdk/x86_64-native-linuxapp-gcc-install/usr/local/share/dpdk/x86_64-native-linuxapp-gcc/include/rte_ether.h:84:2: warning: converting a packed ‘const struct rte_ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member] 84 | const unaligned_uint16_t *w1 = (const uint16_t *)ea1; | ^~~~~
<snip>
Quoted 1 line
Hi,
looking at the code in DPDK for 19.11, rte_ether.h no longer has the unaligned_uint16_t type in rte_ether.h. For example, line 84 of rte_ether.h should read as below:
81 static inline int rte_is_same_ether_addr(const struct rte_ether_addr *ea1,
82 const struct rte_ether_addr *ea2)
83 {
84 const uint16_t *w1 = (const uint16_t *)ea1;
85 const uint16_t *w2 = (const uint16_t *)ea2;
86
Have you got mixed header files from two different DPDK releases?
/Bruce
Hi,
Thank you for the help. I think it has something to do how the NFF-Go is being built. Will address the question to NFF-Go devs.
Thank you,
Martins
Quoted 44 lines · Bruce Richardson
Quoted 10 lines · Martins Eglitis
Dear Bruce and Kevin,
I tried building an application (NFF-GO) which has DPDK as a dependency. I am still getting the same warnings you and Kevin were discussing. My current DPDK version is 19.11-1.
Do you know if this issue has been resolved?
This is the output:
Quoted 16 lines
# github.com/intel-go/nff-go/internal/low In file included from /home/zebra/Projects/nff-go/dpdk/dpdk/x86_64-native-linuxapp-gcc-install/usr/local/share/dpdk/x86_64-native-linuxapp-gcc/include/rte_ethdev.h:160, from ../../go/pkg/mod/github.com/intel-go/[email protected]/internal/low/low.h:11, from ../../go/pkg/mod/github.com/intel-go/[email protected]/internal/low/low.go:16: /home/zebra/Projects/nff-go/dpdk/dpdk/x86_64-native-linuxapp-gcc-install/usr/local/share/dpdk/x86_64-native-linuxapp-gcc/include/rte_ether.h: In function ‘rte_is_same_ether_addr’: /home/zebra/Projects/nff-go/dpdk/dpdk/x86_64-native-linuxapp-gcc-install/usr/local/share/dpdk/x86_64-native-linuxapp-gcc/include/rte_ether.h:84:2: warning: converting a packed ‘const struct rte_ether_addr’ pointer (alignment 1) to a ‘unaligned_uint16_t’ {aka ‘const short unsigned int’} pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member] 84 | const unaligned_uint16_t *w1 = (const uint16_t *)ea1; | ^~~~~<snip>
Hi,looking at the code in DPDK for 19.11,
rte_ether.hno longer has theunaligned_uint16_ttype inrte_ether.h. For example, line 84 ofrte_ether.hshould read as below:81 static inline int rte_is_same_ether_addr(const struct rte_ether_addr *ea1, 82 const struct rte_ether_addr *ea2) 83 { 84 const uint16_t *w1 = (const uint16_t *)ea1; 85 const uint16_t *w2 = (const uint16_t *)ea2; 86Have you got mixed header files from two different DPDK releases?
/Bruce
Signature · 2 lines
Best regards, Martins Eglitis