[1/2] net/i40e: compilation fix for GCC-12
Checks
Commit Message
GCC 12 raises the following warning:
meson --werror --buildtype=debugoptimized
--cross-file config/x86/cross-mingw -Dexamples=helloworld build
ninja -C build
In function 'i40e_hash_get_pattern_type',
inlined from 'i40e_hash_get_pattern_pctypes' at
../drivers/net/i40e/i40e_hash.c:520:8,
inlined from 'i40e_hash_parse_pattern_act' at
../drivers/net/i40e/i40e_hash.c:1147:9,
inlined from 'i40e_hash_parse' at
../drivers/net/i40e/i40e_hash.c:1181:9:
../drivers/net/i40e/i40e_hash.c:389:47:
error: array subscript 53 is above array
bounds of 'const uint64_t[53]'
{aka 'const long long unsigned int[53]'} [-Werror=array-bounds]
389 | item_hdr = pattern_item_header[last_item_type];
| ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
../drivers/net/i40e/i40e_hash.c: In function 'i40e_hash_parse':
../drivers/net/i40e/i40e_hash.c:182:23: note: while referencing
'pattern_item_header'
182 | static const uint64_t pattern_item_header[] = {
| ^~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
Fixes: ef4c16fd9148 (net/i40e: refactor RSS flow)
Cc: stable@dpdk.org
Signed-off-by: Amit Prakash Shukla <amitprakashs@marvell.com>
---
v2:
- Removed "examples/ipsec-secgw" patch from this series and posted it as
seperate patch.
drivers/net/i40e/i40e_hash.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
Comments
> From: Amit Prakash Shukla [mailto:amitprakashs@marvell.com]
> Sent: Wednesday, 24 August 2022 16.04
>
> GCC 12 raises the following warning:
>
> meson --werror --buildtype=debugoptimized
> --cross-file config/x86/cross-mingw -Dexamples=helloworld build
> ninja -C build
>
> In function 'i40e_hash_get_pattern_type',
> inlined from 'i40e_hash_get_pattern_pctypes' at
> ../drivers/net/i40e/i40e_hash.c:520:8,
> inlined from 'i40e_hash_parse_pattern_act' at
> ../drivers/net/i40e/i40e_hash.c:1147:9,
> inlined from 'i40e_hash_parse' at
> ../drivers/net/i40e/i40e_hash.c:1181:9:
> ../drivers/net/i40e/i40e_hash.c:389:47:
> error: array subscript 53 is above array
> bounds of 'const uint64_t[53]'
> {aka 'const long long unsigned int[53]'} [-Werror=array-bounds]
> 389 | item_hdr = pattern_item_header[last_item_type];
> | ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
> ../drivers/net/i40e/i40e_hash.c: In function 'i40e_hash_parse':
> ../drivers/net/i40e/i40e_hash.c:182:23: note: while referencing
> 'pattern_item_header'
> 182 | static const uint64_t pattern_item_header[] = {
> | ^~~~~~~~~~~~~~~~~~~
> cc1: all warnings being treated as errors
>
> Fixes: ef4c16fd9148 (net/i40e: refactor RSS flow)
> Cc: stable@dpdk.org
>
> Signed-off-by: Amit Prakash Shukla <amitprakashs@marvell.com>
> ---
> v2:
> - Removed "examples/ipsec-secgw" patch from this series and posted it
> as
> seperate patch.
>
> drivers/net/i40e/i40e_hash.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/i40e/i40e_hash.c
> b/drivers/net/i40e/i40e_hash.c
> index 8962e9d97a..a1ff85fceb 100644
> --- a/drivers/net/i40e/i40e_hash.c
> +++ b/drivers/net/i40e/i40e_hash.c
> @@ -384,8 +384,10 @@ i40e_hash_get_pattern_type(const struct
> rte_flow_item pattern[],
> }
>
> prev_item_type = last_item_type;
> - assert(last_item_type < (enum rte_flow_item_type)
> - RTE_DIM(pattern_item_header));
> + if (last_item_type >= (enum rte_flow_item_type)
> + RTE_DIM(pattern_item_header))
Does this compile with the correct static branch prediction? If not, please add unlikely() to error checks like this.
> + goto not_sup;
> +
> item_hdr = pattern_item_header[last_item_type];
> assert(item_hdr);
>
> --
> 2.25.1
>
Thanks Morten for the review comment.
As the change is not in datapath, branch prediction optimization was not done.
> -----Original Message-----
> From: Amit Prakash Shukla <amitprakashs@marvell.com>
> Sent: Wednesday, August 24, 2022 7:34 PM
> To: Yuying Zhang <Yuying.Zhang@intel.com>; Beilei Xing
> <beilei.xing@intel.com>
> Cc: dev@dpdk.org; Jerin Jacob Kollanukkaran <jerinj@marvell.com>;
> stable@dpdk.org; Amit Prakash Shukla <amitprakashs@marvell.com>
> Subject: [PATCH 1/2] net/i40e: compilation fix for GCC-12
>
> GCC 12 raises the following warning:
>
> meson --werror --buildtype=debugoptimized
> --cross-file config/x86/cross-mingw -Dexamples=helloworld build
> ninja -C build
>
> In function 'i40e_hash_get_pattern_type',
> inlined from 'i40e_hash_get_pattern_pctypes' at
> ../drivers/net/i40e/i40e_hash.c:520:8,
> inlined from 'i40e_hash_parse_pattern_act' at
> ../drivers/net/i40e/i40e_hash.c:1147:9,
> inlined from 'i40e_hash_parse' at
> ../drivers/net/i40e/i40e_hash.c:1181:9:
> ../drivers/net/i40e/i40e_hash.c:389:47:
> error: array subscript 53 is above array
> bounds of 'const uint64_t[53]'
> {aka 'const long long unsigned int[53]'} [-Werror=array-bounds]
> 389 | item_hdr = pattern_item_header[last_item_type];
> | ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
> ../drivers/net/i40e/i40e_hash.c: In function 'i40e_hash_parse':
> ../drivers/net/i40e/i40e_hash.c:182:23: note: while referencing
> 'pattern_item_header'
> 182 | static const uint64_t pattern_item_header[] = {
> | ^~~~~~~~~~~~~~~~~~~
> cc1: all warnings being treated as errors
>
> Fixes: ef4c16fd9148 (net/i40e: refactor RSS flow)
> Cc: stable@dpdk.org
>
> Signed-off-by: Amit Prakash Shukla <amitprakashs@marvell.com>
> ---
> v2:
> - Removed "examples/ipsec-secgw" patch from this series and posted it as
> seperate patch.
>
> drivers/net/i40e/i40e_hash.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/i40e/i40e_hash.c b/drivers/net/i40e/i40e_hash.c
> index 8962e9d97a..a1ff85fceb 100644
> --- a/drivers/net/i40e/i40e_hash.c
> +++ b/drivers/net/i40e/i40e_hash.c
> @@ -384,8 +384,10 @@ i40e_hash_get_pattern_type(const struct
> rte_flow_item pattern[],
> }
>
> prev_item_type = last_item_type;
> - assert(last_item_type < (enum rte_flow_item_type)
> - RTE_DIM(pattern_item_header));
> + if (last_item_type >= (enum rte_flow_item_type)
> + RTE_DIM(pattern_item_header))
> + goto not_sup;
> +
> item_hdr = pattern_item_header[last_item_type];
> assert(item_hdr);
>
> --
> 2.25.1
> From: Amit Prakash Shukla [mailto:amitprakashs@marvell.com]
> Sent: Friday, 26 August 2022 11.45
>
> Thanks Morten for the review comment.
>
> As the change is not in datapath, branch prediction optimization was
> not done.
OK. Then I agree that it is not required.
>
> > -----Original Message-----
> > From: Amit Prakash Shukla <amitprakashs@marvell.com>
> > Sent: Wednesday, August 24, 2022 7:34 PM
> >
> > GCC 12 raises the following warning:
> >
> > meson --werror --buildtype=debugoptimized
> > --cross-file config/x86/cross-mingw -Dexamples=helloworld build
> > ninja -C build
> >
> > In function 'i40e_hash_get_pattern_type',
> > inlined from 'i40e_hash_get_pattern_pctypes' at
> > ../drivers/net/i40e/i40e_hash.c:520:8,
> > inlined from 'i40e_hash_parse_pattern_act' at
> > ../drivers/net/i40e/i40e_hash.c:1147:9,
> > inlined from 'i40e_hash_parse' at
> > ../drivers/net/i40e/i40e_hash.c:1181:9:
> > ../drivers/net/i40e/i40e_hash.c:389:47:
> > error: array subscript 53 is above array
> > bounds of 'const uint64_t[53]'
> > {aka 'const long long unsigned int[53]'} [-Werror=array-bounds]
> > 389 | item_hdr =
> pattern_item_header[last_item_type];
> > |
> ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
> > ../drivers/net/i40e/i40e_hash.c: In function 'i40e_hash_parse':
> > ../drivers/net/i40e/i40e_hash.c:182:23: note: while referencing
> > 'pattern_item_header'
> > 182 | static const uint64_t pattern_item_header[] = {
> > | ^~~~~~~~~~~~~~~~~~~
> > cc1: all warnings being treated as errors
> >
> > Fixes: ef4c16fd9148 (net/i40e: refactor RSS flow)
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Amit Prakash Shukla <amitprakashs@marvell.com>
> > ---
> > v2:
> > - Removed "examples/ipsec-secgw" patch from this series and posted it
> as
> > seperate patch.
> >
> > drivers/net/i40e/i40e_hash.c | 6 ++++--
> > 1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/i40e/i40e_hash.c
> b/drivers/net/i40e/i40e_hash.c
> > index 8962e9d97a..a1ff85fceb 100644
> > --- a/drivers/net/i40e/i40e_hash.c
> > +++ b/drivers/net/i40e/i40e_hash.c
> > @@ -384,8 +384,10 @@ i40e_hash_get_pattern_type(const struct
> > rte_flow_item pattern[],
> > }
> >
> > prev_item_type = last_item_type;
> > - assert(last_item_type < (enum rte_flow_item_type)
> > - RTE_DIM(pattern_item_header));
> > + if (last_item_type >= (enum rte_flow_item_type)
> > + RTE_DIM(pattern_item_header))
> > + goto not_sup;
> > +
> > item_hdr = pattern_item_header[last_item_type];
> > assert(item_hdr);
> >
> > --
> > 2.25.1
>
24/08/2022 16:03, Amit Prakash Shukla:
> GCC 12 raises the following warning:
>
> meson --werror --buildtype=debugoptimized
> --cross-file config/x86/cross-mingw -Dexamples=helloworld build
> ninja -C build
>
> In function 'i40e_hash_get_pattern_type',
> inlined from 'i40e_hash_get_pattern_pctypes' at
> ../drivers/net/i40e/i40e_hash.c:520:8,
> inlined from 'i40e_hash_parse_pattern_act' at
> ../drivers/net/i40e/i40e_hash.c:1147:9,
> inlined from 'i40e_hash_parse' at
> ../drivers/net/i40e/i40e_hash.c:1181:9:
> ../drivers/net/i40e/i40e_hash.c:389:47:
> error: array subscript 53 is above array
> bounds of 'const uint64_t[53]'
> {aka 'const long long unsigned int[53]'} [-Werror=array-bounds]
> 389 | item_hdr = pattern_item_header[last_item_type];
> | ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
> ../drivers/net/i40e/i40e_hash.c: In function 'i40e_hash_parse':
> ../drivers/net/i40e/i40e_hash.c:182:23: note: while referencing
> 'pattern_item_header'
> 182 | static const uint64_t pattern_item_header[] = {
> | ^~~~~~~~~~~~~~~~~~~
> cc1: all warnings being treated as errors
>
> Fixes: ef4c16fd9148 (net/i40e: refactor RSS flow)
> Cc: stable@dpdk.org
>
> Signed-off-by: Amit Prakash Shukla <amitprakashs@marvell.com>
Sorry I did not notice this patch and I recently submitted one
which I will abandon.
It seems there was no reply from i40e maintainers after 6 weeks,
but I will apply anyway.
I will use my commit message:
net/i40e: fix build with MinGW GCC 12
When compiling with MinGW GCC 12,
the rte_flow_item array is seen as read out of bound:
net/i40e/i40e_hash.c:389:47: error:
array subscript 50 is above array bounds of ‘const uint64_t[50]’
{aka ‘const long long unsigned int[50]’} [-Werror=array-bounds]
389 | item_hdr = pattern_item_header[last_item_type];
| ~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
It seems the assert check done above this line has no impact.
A real check is added to make the compiler happy.
Fixes: ef4c16fd9148 ("net/i40e: refactor RSS flow")
Cc: stable@dpdk.org
@@ -384,8 +384,10 @@ i40e_hash_get_pattern_type(const struct rte_flow_item pattern[],
}
prev_item_type = last_item_type;
- assert(last_item_type < (enum rte_flow_item_type)
- RTE_DIM(pattern_item_header));
+ if (last_item_type >= (enum rte_flow_item_type)
+ RTE_DIM(pattern_item_header))
+ goto not_sup;
+
item_hdr = pattern_item_header[last_item_type];
assert(item_hdr);