* [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload
@ 2022-01-25 10:26 Baowen Zheng
2022-01-25 15:22 ` Victor Nogueira
0 siblings, 1 reply; 5+ messages in thread
From: Baowen Zheng @ 2022-01-25 10:26 UTC (permalink / raw)
To: David Ahern; +Cc: netdev, oss-drivers, jhs, baowen zheng, Simon Horman
Add skip_hw and skip_sw flags for user to control whether
offload action to hardware.
Also we add hw_count to show how many hardwares accept to offload
the action.
Change man page to describe the usage of skip_sw and skip_hw flag.
An example to add and query action as below.
$ tc actions add action police rate 1mbit burst 100k index 100 skip_sw
$ tc -s -d actions list action police
total acts 1
action order 0: police 0x64 rate 1Mbit burst 100Kb mtu 2Kb action reclassify overhead 0b linklayer ethernet
ref 1 bind 0 installed 2 sec used 2 sec
Action statistics:
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
skip_sw in_hw in_hw_count 1
used_hw_stats delayed
Signed-off-by: baowen zheng <baowen.zheng@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
man/man8/tc-actions.8 | 24 ++++++++++++++++++++
tc/m_action.c | 63 +++++++++++++++++++++++++++++++++++++++++++--------
2 files changed, 77 insertions(+), 10 deletions(-)
diff --git a/man/man8/tc-actions.8 b/man/man8/tc-actions.8
index 6f1c201..5c399cd 100644
--- a/man/man8/tc-actions.8
+++ b/man/man8/tc-actions.8
@@ -52,6 +52,8 @@ actions \- independently defined actions in tc
.I HWSTATSSPEC
] [
.I CONTROL
+] [
+.I SKIPSPEC
]
.I ACTISPEC
@@ -99,6 +101,11 @@ Time since last update.
.IR reclassify " | " pipe " | " drop " | " continue " | " ok
}
+.I SKIPSPEC
+:= {
+.IR skip_sw " | " skip_hw
+}
+
.I TC_OPTIONS
These are the options that are specific to
.B tc
@@ -270,6 +277,23 @@ Return to the calling qdisc for packet processing, and end classification of
this packet.
.RE
+.TP
+.I SKIPSPEC
+The
+.I SKIPSPEC
+indicates how
+.B tc
+should proceed when executing the action. Any of the following are valid:
+.RS
+.TP
+.B skip_sw
+Do not process action by software. If hardware has no offload support for this
+action, operation will fail.
+.TP
+.B skip_hw
+Do not process action by hardware.
+.RE
+
.SH SEE ALSO
.BR tc (8),
.BR tc-bpf (8),
diff --git a/tc/m_action.c b/tc/m_action.c
index b16882a..b9fed6f 100644
--- a/tc/m_action.c
+++ b/tc/m_action.c
@@ -51,9 +51,10 @@ static void act_usage(void)
" FL := ls | list | flush | <ACTNAMESPEC>\n"
" ACTNAMESPEC := action <ACTNAME>\n"
" ACTISPEC := <ACTNAMESPEC> <INDEXSPEC>\n"
- " ACTSPEC := action <ACTDETAIL> [INDEXSPEC] [HWSTATSSPEC]\n"
+ " ACTSPEC := action <ACTDETAIL> [INDEXSPEC] [HWSTATSSPEC] [SKIPSPEC]\n"
" INDEXSPEC := index <32 bit indexvalue>\n"
" HWSTATSSPEC := hw_stats [ immediate | delayed | disabled ]\n"
+ " SKIPSPEC := [ skip_sw | skip_hw ]\n"
" ACTDETAIL := <ACTNAME> <ACTPARAMS>\n"
" Example ACTNAME is gact, mirred, bpf, etc\n"
" Each action has its own parameters (ACTPARAMS)\n"
@@ -210,12 +211,14 @@ int parse_action(int *argc_p, char ***argv_p, int tca_id, struct nlmsghdr *n)
struct rtattr *tail, *tail2;
char k[FILTER_NAMESZ];
int act_ck_len = 0;
+ __u32 flag = 0;
int ok = 0;
int eap = 0; /* expect action parameters */
int ret = 0;
int prio = 0;
unsigned char act_ck[TC_COOKIE_MAX_SIZE];
+ int skip_loop = 2;
if (argc <= 0)
return -1;
@@ -314,13 +317,27 @@ done0:
}
if (*argv && strcmp(*argv, "no_percpu") == 0) {
+ flag |= TCA_ACT_FLAGS_NO_PERCPU_STATS;
+ NEXT_ARG_FWD();
+ }
+
+ /* we need to parse twice to fix skip flag out of order */
+ while (skip_loop--) {
+ if (*argv && strcmp(*argv, "skip_sw") == 0) {
+ flag |= TCA_ACT_FLAGS_SKIP_SW;
+ NEXT_ARG_FWD();
+ } else if (*argv && strcmp(*argv, "skip_hw") == 0) {
+ flag |= TCA_ACT_FLAGS_SKIP_HW;
+ NEXT_ARG_FWD();
+ }
+ }
+
+ if (flag) {
struct nla_bitfield32 flags =
- { TCA_ACT_FLAGS_NO_PERCPU_STATS,
- TCA_ACT_FLAGS_NO_PERCPU_STATS };
+ { flag, flag };
addattr_l(n, MAX_MSG, TCA_ACT_FLAGS, &flags,
sizeof(struct nla_bitfield32));
- NEXT_ARG_FWD();
}
addattr_nest_end(n, tail);
@@ -396,13 +413,39 @@ static int tc_print_one_action(FILE *f, struct rtattr *arg)
strsz, b1, sizeof(b1)));
print_nl();
}
- if (tb[TCA_ACT_FLAGS]) {
- struct nla_bitfield32 *flags = RTA_DATA(tb[TCA_ACT_FLAGS]);
+ if (tb[TCA_ACT_FLAGS] || tb[TCA_ACT_IN_HW_COUNT]) {
+ bool skip_hw = false;
+ if (tb[TCA_ACT_FLAGS]) {
+ struct nla_bitfield32 *flags = RTA_DATA(tb[TCA_ACT_FLAGS]);
+
+ if (flags->selector & TCA_ACT_FLAGS_NO_PERCPU_STATS)
+ print_bool(PRINT_ANY, "no_percpu", "\tno_percpu",
+ flags->value &
+ TCA_ACT_FLAGS_NO_PERCPU_STATS);
+ if (flags->selector & TCA_ACT_FLAGS_SKIP_HW) {
+ print_bool(PRINT_ANY, "skip_hw", "\tskip_hw",
+ flags->value &
+ TCA_ACT_FLAGS_SKIP_HW);
+ skip_hw = !!(flags->value & TCA_ACT_FLAGS_SKIP_HW);
+ }
+ if (flags->selector & TCA_ACT_FLAGS_SKIP_SW)
+ print_bool(PRINT_ANY, "skip_sw", "\tskip_sw",
+ flags->value &
+ TCA_ACT_FLAGS_SKIP_SW);
+ }
+ if (tb[TCA_ACT_IN_HW_COUNT] && !skip_hw) {
+ __u32 count = rta_getattr_u32(tb[TCA_ACT_IN_HW_COUNT]);
+ if (count) {
+ print_bool(PRINT_ANY, "in_hw", "\tin_hw",
+ true);
+ print_uint(PRINT_ANY, "in_hw_count",
+ " in_hw_count %u", count);
+ } else {
+ print_bool(PRINT_ANY, "not_in_hw",
+ "\tnot_in_hw", true);
+ }
+ }
- if (flags->selector & TCA_ACT_FLAGS_NO_PERCPU_STATS)
- print_bool(PRINT_ANY, "no_percpu", "\tno_percpu",
- flags->value &
- TCA_ACT_FLAGS_NO_PERCPU_STATS);
print_nl();
}
if (tb[TCA_ACT_HW_STATS])
--
1.8.3.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload
2022-01-25 10:26 [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload Baowen Zheng
@ 2022-01-25 15:22 ` Victor Nogueira
2022-01-26 8:14 ` Baowen Zheng
0 siblings, 1 reply; 5+ messages in thread
From: Victor Nogueira @ 2022-01-25 15:22 UTC (permalink / raw)
To: Baowen Zheng
Cc: David Ahern, netdev, oss-drivers, Jamal Hadi Salim, Simon Horman
Hi Baowen,
I applied your patch, ran tdc.sh and in particular the vlan tests broke.
cheers,
Victor
On Tue, Jan 25, 2022 at 7:35 AM Baowen Zheng <baowen.zheng@corigine.com> wrote:
>
> Add skip_hw and skip_sw flags for user to control whether
> offload action to hardware.
>
> Also we add hw_count to show how many hardwares accept to offload
> the action.
>
> Change man page to describe the usage of skip_sw and skip_hw flag.
>
> An example to add and query action as below.
>
> $ tc actions add action police rate 1mbit burst 100k index 100 skip_sw
>
> $ tc -s -d actions list action police
> total acts 1
> action order 0: police 0x64 rate 1Mbit burst 100Kb mtu 2Kb action reclassify overhead 0b linklayer ethernet
> ref 1 bind 0 installed 2 sec used 2 sec
> Action statistics:
> Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
> backlog 0b 0p requeues 0
> skip_sw in_hw in_hw_count 1
> used_hw_stats delayed
>
> Signed-off-by: baowen zheng <baowen.zheng@corigine.com>
> Signed-off-by: Simon Horman <simon.horman@corigine.com>
> ---
> man/man8/tc-actions.8 | 24 ++++++++++++++++++++
> tc/m_action.c | 63 +++++++++++++++++++++++++++++++++++++++++++--------
> 2 files changed, 77 insertions(+), 10 deletions(-)
>
> diff --git a/man/man8/tc-actions.8 b/man/man8/tc-actions.8
> index 6f1c201..5c399cd 100644
> --- a/man/man8/tc-actions.8
> +++ b/man/man8/tc-actions.8
> @@ -52,6 +52,8 @@ actions \- independently defined actions in tc
> .I HWSTATSSPEC
> ] [
> .I CONTROL
> +] [
> +.I SKIPSPEC
> ]
>
> .I ACTISPEC
> @@ -99,6 +101,11 @@ Time since last update.
> .IR reclassify " | " pipe " | " drop " | " continue " | " ok
> }
>
> +.I SKIPSPEC
> +:= {
> +.IR skip_sw " | " skip_hw
> +}
> +
> .I TC_OPTIONS
> These are the options that are specific to
> .B tc
> @@ -270,6 +277,23 @@ Return to the calling qdisc for packet processing, and end classification of
> this packet.
> .RE
>
> +.TP
> +.I SKIPSPEC
> +The
> +.I SKIPSPEC
> +indicates how
> +.B tc
> +should proceed when executing the action. Any of the following are valid:
> +.RS
> +.TP
> +.B skip_sw
> +Do not process action by software. If hardware has no offload support for this
> +action, operation will fail.
> +.TP
> +.B skip_hw
> +Do not process action by hardware.
> +.RE
> +
> .SH SEE ALSO
> .BR tc (8),
> .BR tc-bpf (8),
> diff --git a/tc/m_action.c b/tc/m_action.c
> index b16882a..b9fed6f 100644
> --- a/tc/m_action.c
> +++ b/tc/m_action.c
> @@ -51,9 +51,10 @@ static void act_usage(void)
> " FL := ls | list | flush | <ACTNAMESPEC>\n"
> " ACTNAMESPEC := action <ACTNAME>\n"
> " ACTISPEC := <ACTNAMESPEC> <INDEXSPEC>\n"
> - " ACTSPEC := action <ACTDETAIL> [INDEXSPEC] [HWSTATSSPEC]\n"
> + " ACTSPEC := action <ACTDETAIL> [INDEXSPEC] [HWSTATSSPEC] [SKIPSPEC]\n"
> " INDEXSPEC := index <32 bit indexvalue>\n"
> " HWSTATSSPEC := hw_stats [ immediate | delayed | disabled ]\n"
> + " SKIPSPEC := [ skip_sw | skip_hw ]\n"
> " ACTDETAIL := <ACTNAME> <ACTPARAMS>\n"
> " Example ACTNAME is gact, mirred, bpf, etc\n"
> " Each action has its own parameters (ACTPARAMS)\n"
> @@ -210,12 +211,14 @@ int parse_action(int *argc_p, char ***argv_p, int tca_id, struct nlmsghdr *n)
> struct rtattr *tail, *tail2;
> char k[FILTER_NAMESZ];
> int act_ck_len = 0;
> + __u32 flag = 0;
> int ok = 0;
> int eap = 0; /* expect action parameters */
>
> int ret = 0;
> int prio = 0;
> unsigned char act_ck[TC_COOKIE_MAX_SIZE];
> + int skip_loop = 2;
>
> if (argc <= 0)
> return -1;
> @@ -314,13 +317,27 @@ done0:
> }
>
> if (*argv && strcmp(*argv, "no_percpu") == 0) {
> + flag |= TCA_ACT_FLAGS_NO_PERCPU_STATS;
> + NEXT_ARG_FWD();
> + }
> +
> + /* we need to parse twice to fix skip flag out of order */
> + while (skip_loop--) {
> + if (*argv && strcmp(*argv, "skip_sw") == 0) {
> + flag |= TCA_ACT_FLAGS_SKIP_SW;
> + NEXT_ARG_FWD();
> + } else if (*argv && strcmp(*argv, "skip_hw") == 0) {
> + flag |= TCA_ACT_FLAGS_SKIP_HW;
> + NEXT_ARG_FWD();
> + }
> + }
> +
> + if (flag) {
> struct nla_bitfield32 flags =
> - { TCA_ACT_FLAGS_NO_PERCPU_STATS,
> - TCA_ACT_FLAGS_NO_PERCPU_STATS };
> + { flag, flag };
>
> addattr_l(n, MAX_MSG, TCA_ACT_FLAGS, &flags,
> sizeof(struct nla_bitfield32));
> - NEXT_ARG_FWD();
> }
>
> addattr_nest_end(n, tail);
> @@ -396,13 +413,39 @@ static int tc_print_one_action(FILE *f, struct rtattr *arg)
> strsz, b1, sizeof(b1)));
> print_nl();
> }
> - if (tb[TCA_ACT_FLAGS]) {
> - struct nla_bitfield32 *flags = RTA_DATA(tb[TCA_ACT_FLAGS]);
> + if (tb[TCA_ACT_FLAGS] || tb[TCA_ACT_IN_HW_COUNT]) {
> + bool skip_hw = false;
> + if (tb[TCA_ACT_FLAGS]) {
> + struct nla_bitfield32 *flags = RTA_DATA(tb[TCA_ACT_FLAGS]);
> +
> + if (flags->selector & TCA_ACT_FLAGS_NO_PERCPU_STATS)
> + print_bool(PRINT_ANY, "no_percpu", "\tno_percpu",
> + flags->value &
> + TCA_ACT_FLAGS_NO_PERCPU_STATS);
> + if (flags->selector & TCA_ACT_FLAGS_SKIP_HW) {
> + print_bool(PRINT_ANY, "skip_hw", "\tskip_hw",
> + flags->value &
> + TCA_ACT_FLAGS_SKIP_HW);
> + skip_hw = !!(flags->value & TCA_ACT_FLAGS_SKIP_HW);
> + }
> + if (flags->selector & TCA_ACT_FLAGS_SKIP_SW)
> + print_bool(PRINT_ANY, "skip_sw", "\tskip_sw",
> + flags->value &
> + TCA_ACT_FLAGS_SKIP_SW);
> + }
> + if (tb[TCA_ACT_IN_HW_COUNT] && !skip_hw) {
> + __u32 count = rta_getattr_u32(tb[TCA_ACT_IN_HW_COUNT]);
> + if (count) {
> + print_bool(PRINT_ANY, "in_hw", "\tin_hw",
> + true);
> + print_uint(PRINT_ANY, "in_hw_count",
> + " in_hw_count %u", count);
> + } else {
> + print_bool(PRINT_ANY, "not_in_hw",
> + "\tnot_in_hw", true);
> + }
> + }
>
> - if (flags->selector & TCA_ACT_FLAGS_NO_PERCPU_STATS)
> - print_bool(PRINT_ANY, "no_percpu", "\tno_percpu",
> - flags->value &
> - TCA_ACT_FLAGS_NO_PERCPU_STATS);
> print_nl();
> }
> if (tb[TCA_ACT_HW_STATS])
> --
> 1.8.3.1
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* RE: [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload
2022-01-25 15:22 ` Victor Nogueira
@ 2022-01-26 8:14 ` Baowen Zheng
2022-01-26 11:57 ` Jamal Hadi Salim
0 siblings, 1 reply; 5+ messages in thread
From: Baowen Zheng @ 2022-01-26 8:14 UTC (permalink / raw)
To: Victor Nogueira
Cc: David Ahern, netdev, oss-drivers, Jamal Hadi Salim, Simon Horman
Hi Victor, thanks very much to bring this issue to us, we will make a check and fix this issue in next patch.
On Tuesday, January 25, 2022 11:23 PM, Victor wrote:
>Hi Baowen,
>
>I applied your patch, ran tdc.sh and in particular the vlan tests broke.
>
>cheers,
>Victor
>
>On Tue, Jan 25, 2022 at 7:35 AM Baowen Zheng
><baowen.zheng@corigine.com> wrote:
>>
>> Add skip_hw and skip_sw flags for user to control whether offload
>> action to hardware.
>>
>> Also we add hw_count to show how many hardwares accept to offload the
>> action.
>>
>> Change man page to describe the usage of skip_sw and skip_hw flag.
>>
>> An example to add and query action as below.
>>
>> $ tc actions add action police rate 1mbit burst 100k index 100 skip_sw
>>
>> $ tc -s -d actions list action police
>> total acts 1
>> action order 0: police 0x64 rate 1Mbit burst 100Kb mtu 2Kb action
>reclassify overhead 0b linklayer ethernet
>> ref 1 bind 0 installed 2 sec used 2 sec
>> Action statistics:
>> Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
>> backlog 0b 0p requeues 0
>> skip_sw in_hw in_hw_count 1
>> used_hw_stats delayed
>>
>> Signed-off-by: baowen zheng <baowen.zheng@corigine.com>
>> Signed-off-by: Simon Horman <simon.horman@corigine.com>
>> ---
>> man/man8/tc-actions.8 | 24 ++++++++++++++++++++
>> tc/m_action.c | 63 +++++++++++++++++++++++++++++++++++++++++++-
>-------
>> 2 files changed, 77 insertions(+), 10 deletions(-)
>>
>> diff --git a/man/man8/tc-actions.8 b/man/man8/tc-actions.8 index
>> 6f1c201..5c399cd 100644
>> --- a/man/man8/tc-actions.8
>> +++ b/man/man8/tc-actions.8
>> @@ -52,6 +52,8 @@ actions \- independently defined actions in tc .I
>> HWSTATSSPEC ] [ .I CONTROL
>> +] [
>> +.I SKIPSPEC
>> ]
>>
>> .I ACTISPEC
>> @@ -99,6 +101,11 @@ Time since last update.
>> .IR reclassify " | " pipe " | " drop " | " continue " | " ok }
>>
>> +.I SKIPSPEC
>> +:= {
>> +.IR skip_sw " | " skip_hw
>> +}
>> +
>> .I TC_OPTIONS
>> These are the options that are specific to .B tc @@ -270,6 +277,23
>> @@ Return to the calling qdisc for packet processing, and end
>> classification of this packet.
>> .RE
>>
>> +.TP
>> +.I SKIPSPEC
>> +The
>> +.I SKIPSPEC
>> +indicates how
>> +.B tc
>> +should proceed when executing the action. Any of the following are valid:
>> +.RS
>> +.TP
>> +.B skip_sw
>> +Do not process action by software. If hardware has no offload support
>> +for this action, operation will fail.
>> +.TP
>> +.B skip_hw
>> +Do not process action by hardware.
>> +.RE
>> +
>> .SH SEE ALSO
>> .BR tc (8),
>> .BR tc-bpf (8),
>> diff --git a/tc/m_action.c b/tc/m_action.c index b16882a..b9fed6f
>> 100644
>> --- a/tc/m_action.c
>> +++ b/tc/m_action.c
>> @@ -51,9 +51,10 @@ static void act_usage(void)
>> " FL := ls | list | flush | <ACTNAMESPEC>\n"
>> " ACTNAMESPEC := action <ACTNAME>\n"
>> " ACTISPEC := <ACTNAMESPEC> <INDEXSPEC>\n"
>> - " ACTSPEC := action <ACTDETAIL> [INDEXSPEC]
>[HWSTATSSPEC]\n"
>> + " ACTSPEC := action <ACTDETAIL> [INDEXSPEC] [HWSTATSSPEC]
>[SKIPSPEC]\n"
>> " INDEXSPEC := index <32 bit indexvalue>\n"
>> " HWSTATSSPEC := hw_stats [ immediate | delayed |
>disabled ]\n"
>> + " SKIPSPEC := [ skip_sw | skip_hw ]\n"
>> " ACTDETAIL := <ACTNAME> <ACTPARAMS>\n"
>> " Example ACTNAME is gact, mirred, bpf, etc\n"
>> " Each action has its own parameters (ACTPARAMS)\n"
>> @@ -210,12 +211,14 @@ int parse_action(int *argc_p, char ***argv_p, int
>tca_id, struct nlmsghdr *n)
>> struct rtattr *tail, *tail2;
>> char k[FILTER_NAMESZ];
>> int act_ck_len = 0;
>> + __u32 flag = 0;
>> int ok = 0;
>> int eap = 0; /* expect action parameters */
>>
>> int ret = 0;
>> int prio = 0;
>> unsigned char act_ck[TC_COOKIE_MAX_SIZE];
>> + int skip_loop = 2;
>>
>> if (argc <= 0)
>> return -1;
>> @@ -314,13 +317,27 @@ done0:
>> }
>>
>> if (*argv && strcmp(*argv, "no_percpu") == 0)
>> {
>> + flag |= TCA_ACT_FLAGS_NO_PERCPU_STATS;
>> + NEXT_ARG_FWD();
>> + }
>> +
>> + /* we need to parse twice to fix skip flag out of order */
>> + while (skip_loop--) {
>> + if (*argv && strcmp(*argv, "skip_sw") == 0) {
>> + flag |= TCA_ACT_FLAGS_SKIP_SW;
>> + NEXT_ARG_FWD();
>> + } else if (*argv && strcmp(*argv, "skip_hw") == 0) {
>> + flag |= TCA_ACT_FLAGS_SKIP_HW;
>> + NEXT_ARG_FWD();
>> + }
>> + }
>> +
>> + if (flag) {
>> struct nla_bitfield32 flags =
>> - { TCA_ACT_FLAGS_NO_PERCPU_STATS,
>> - TCA_ACT_FLAGS_NO_PERCPU_STATS };
>> + { flag, flag };
>>
>> addattr_l(n, MAX_MSG, TCA_ACT_FLAGS, &flags,
>> sizeof(struct nla_bitfield32));
>> - NEXT_ARG_FWD();
>> }
>>
>> addattr_nest_end(n, tail); @@ -396,13 +413,39
>> @@ static int tc_print_one_action(FILE *f, struct rtattr *arg)
>> strsz, b1, sizeof(b1)));
>> print_nl();
>> }
>> - if (tb[TCA_ACT_FLAGS]) {
>> - struct nla_bitfield32 *flags = RTA_DATA(tb[TCA_ACT_FLAGS]);
>> + if (tb[TCA_ACT_FLAGS] || tb[TCA_ACT_IN_HW_COUNT]) {
>> + bool skip_hw = false;
>> + if (tb[TCA_ACT_FLAGS]) {
>> + struct nla_bitfield32 *flags =
>> + RTA_DATA(tb[TCA_ACT_FLAGS]);
>> +
>> + if (flags->selector & TCA_ACT_FLAGS_NO_PERCPU_STATS)
>> + print_bool(PRINT_ANY, "no_percpu", "\tno_percpu",
>> + flags->value &
>> + TCA_ACT_FLAGS_NO_PERCPU_STATS);
>> + if (flags->selector & TCA_ACT_FLAGS_SKIP_HW) {
>> + print_bool(PRINT_ANY, "skip_hw", "\tskip_hw",
>> + flags->value &
>> + TCA_ACT_FLAGS_SKIP_HW);
>> + skip_hw = !!(flags->value & TCA_ACT_FLAGS_SKIP_HW);
>> + }
>> + if (flags->selector & TCA_ACT_FLAGS_SKIP_SW)
>> + print_bool(PRINT_ANY, "skip_sw", "\tskip_sw",
>> + flags->value &
>> + TCA_ACT_FLAGS_SKIP_SW);
>> + }
>> + if (tb[TCA_ACT_IN_HW_COUNT] && !skip_hw) {
>> + __u32 count = rta_getattr_u32(tb[TCA_ACT_IN_HW_COUNT]);
>> + if (count) {
>> + print_bool(PRINT_ANY, "in_hw", "\tin_hw",
>> + true);
>> + print_uint(PRINT_ANY, "in_hw_count",
>> + " in_hw_count %u", count);
>> + } else {
>> + print_bool(PRINT_ANY, "not_in_hw",
>> + "\tnot_in_hw", true);
>> + }
>> + }
>>
>> - if (flags->selector & TCA_ACT_FLAGS_NO_PERCPU_STATS)
>> - print_bool(PRINT_ANY, "no_percpu", "\tno_percpu",
>> - flags->value &
>> - TCA_ACT_FLAGS_NO_PERCPU_STATS);
>> print_nl();
>> }
>> if (tb[TCA_ACT_HW_STATS])
>> --
>> 1.8.3.1
>>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload
2022-01-26 8:14 ` Baowen Zheng
@ 2022-01-26 11:57 ` Jamal Hadi Salim
2022-01-27 1:44 ` Baowen Zheng
0 siblings, 1 reply; 5+ messages in thread
From: Jamal Hadi Salim @ 2022-01-26 11:57 UTC (permalink / raw)
To: Baowen Zheng, Victor Nogueira
Cc: David Ahern, netdev, oss-drivers, Simon Horman
Hi Baowen,
I didnt follow what "next patch" means because just prior to this
email you sent a v2.
As a general rule:
You have to run the tdc tests and if your iproute2 patch is breaking
something that was working fine before your patch then it needs to be
fixed first. Perhaps thats what you meant is your v2 fixes the vlan
breakage.
cheers,
jamal
On 2022-01-26 03:14, Baowen Zheng wrote:
> Hi Victor, thanks very much to bring this issue to us, we will make a check and fix this issue in next patch.
>
> On Tuesday, January 25, 2022 11:23 PM, Victor wrote:
>> Hi Baowen,
>>
>> I applied your patch, ran tdc.sh and in particular the vlan tests broke.
>>
>> cheers,
>> Victor
^ permalink raw reply [flat|nested] 5+ messages in thread
* RE: [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload
2022-01-26 11:57 ` Jamal Hadi Salim
@ 2022-01-27 1:44 ` Baowen Zheng
0 siblings, 0 replies; 5+ messages in thread
From: Baowen Zheng @ 2022-01-27 1:44 UTC (permalink / raw)
To: Jamal Hadi Salim, Victor Nogueira
Cc: David Ahern, netdev, oss-drivers, Simon Horman
Hi Jamal,
Sorry I did make myself clear, I mean our V2 patch will fix the vlan breakage issue, I should send this mail before I send the V2 patch.
Thanks for clarification about the rule, I run the tdc test before the V2 patch to be sent.
On Wednesday, January 26, 2022 7:58 PM, Jamal:
>Hi Baowen,
>I didnt follow what "next patch" means because just prior to this email you
>sent a v2.
>
>As a general rule:
>You have to run the tdc tests and if your iproute2 patch is breaking something
>that was working fine before your patch then it needs to be fixed first.
>Perhaps thats what you meant is your v2 fixes the vlan breakage.
>
>cheers,
>jamal
>
>On 2022-01-26 03:14, Baowen Zheng wrote:
>> Hi Victor, thanks very much to bring this issue to us, we will make a check
>and fix this issue in next patch.
>>
>> On Tuesday, January 25, 2022 11:23 PM, Victor wrote:
>>> Hi Baowen,
>>>
>>> I applied your patch, ran tdc.sh and in particular the vlan tests broke.
>>>
>>> cheers,
>>> Victor
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2022-01-27 1:44 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-25 10:26 [PATCH iproute2-next v1] tc: add skip_hw and skip_sw to control action offload Baowen Zheng
2022-01-25 15:22 ` Victor Nogueira
2022-01-26 8:14 ` Baowen Zheng
2022-01-26 11:57 ` Jamal Hadi Salim
2022-01-27 1:44 ` Baowen Zheng
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.