examples/power: fix wrong core_id with JSON cmds

Message ID 20190107113920.37748-1-david.hunt@intel.com
State Accepted, archived
Delegated to: Thomas Monjalon
Headers show
Series
  • examples/power: fix wrong core_id with JSON cmds
Related show

Checks

Context Check Description
ci/Intel-compilation success Compilation OK
ci/mellanox-Performance-Testing success Performance Testing PASS
ci/intel-Performance-Testing success Performance Testing PASS
ci/checkpatch success coding style OK

Commit Message

David Hunt Jan. 7, 2019, 11:39 a.m.
This patch fixes a bug introduced in the 64-core limtation
enhancement where the core_id is inadvertently converted from
virtual to physical even though it may already be a physical
core_id.

We should be using the core_type field, and only converting via
hypervisor when core_type is set to CORE_TYPE_VIRTUAL

Fixes: 5776b7a371d1 ("examples/power: allow VM to use lcores over 63")

Signed-off-by: David Hunt <david.hunt@intel.com>
---
 examples/vm_power_manager/channel_monitor.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

Comments

Yao, Lei A Jan. 8, 2019, 2:02 a.m. | #1
> -----Original Message-----
> From: Hunt, David
> Sent: Monday, January 7, 2019 7:39 PM
> To: dev@dpdk.org
> Cc: Hunt, David <david.hunt@intel.com>; Yao, Lei A <lei.a.yao@intel.com>
> Subject: [PATCH] examples/power: fix wrong core_id with JSON cmds
> 
> This patch fixes a bug introduced in the 64-core limtation
> enhancement where the core_id is inadvertently converted from
> virtual to physical even though it may already be a physical
> core_id.
> 
> We should be using the core_type field, and only converting via
> hypervisor when core_type is set to CORE_TYPE_VIRTUAL
> 
> Fixes: 5776b7a371d1 ("examples/power: allow VM to use lcores over 63")
> 
> Signed-off-by: David Hunt <david.hunt@intel.com>
> ---
>  examples/vm_power_manager/channel_monitor.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/examples/vm_power_manager/channel_monitor.c
> b/examples/vm_power_manager/channel_monitor.c
> index 85622e7cb..1a3a0fa76 100644
> --- a/examples/vm_power_manager/channel_monitor.c
> +++ b/examples/vm_power_manager/channel_monitor.c
> @@ -640,7 +640,10 @@ process_request(struct channel_packet *pkt, struct
> channel_info *chan_info)
>  	if (pkt->command == CPU_POWER) {
>  		unsigned int core_num;
> 
> -		core_num = get_pcpu(chan_info, pkt->resource_id);
> +		if (pkt->core_type == CORE_TYPE_VIRTUAL)
> +			core_num = get_pcpu(chan_info, pkt->resource_id);
Hi, Dave

Now in DPDK code, only command send from VM(guest_cli sample) will set the 
pkt-<core_type to VIRTUAL, if we send JSON command with resource_id, seems
we always won't hit this branch. Because  parse_json_to_pkt() will always set the 
core_type to CORE_TYPE_PHYSICAL. 
If resource_id in instruction format JSON file will always be treated as Pcpu, it's the 
same as core_list behavior.  
> +		else
> +			core_num = pkt->resource_id;
> 
>  		switch (pkt->unit) {
>  		case(CPU_POWER_SCALE_MIN):
> --
> 2.17.1
David Hunt Jan. 8, 2019, 9:20 a.m. | #2
Hi Lei,

On 8/1/2019 2:02 AM, Yao, Lei A wrote:
>
>> -----Original Message-----
>> From: Hunt, David
>> Sent: Monday, January 7, 2019 7:39 PM
>> To: dev@dpdk.org
>> Cc: Hunt, David <david.hunt@intel.com>; Yao, Lei A <lei.a.yao@intel.com>
>> Subject: [PATCH] examples/power: fix wrong core_id with JSON cmds
>>
>> This patch fixes a bug introduced in the 64-core limtation
>> enhancement where the core_id is inadvertently converted from
>> virtual to physical even though it may already be a physical
>> core_id.
>>
>> We should be using the core_type field, and only converting via
>> hypervisor when core_type is set to CORE_TYPE_VIRTUAL
>>
>> Fixes: 5776b7a371d1 ("examples/power: allow VM to use lcores over 63")
>>
>> Signed-off-by: David Hunt <david.hunt@intel.com>
>> ---
>>   examples/vm_power_manager/channel_monitor.c | 5 ++++-
>>   1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/examples/vm_power_manager/channel_monitor.c
>> b/examples/vm_power_manager/channel_monitor.c
>> index 85622e7cb..1a3a0fa76 100644
>> --- a/examples/vm_power_manager/channel_monitor.c
>> +++ b/examples/vm_power_manager/channel_monitor.c
>> @@ -640,7 +640,10 @@ process_request(struct channel_packet *pkt, struct
>> channel_info *chan_info)
>>   	if (pkt->command == CPU_POWER) {
>>   		unsigned int core_num;
>>
>> -		core_num = get_pcpu(chan_info, pkt->resource_id);
>> +		if (pkt->core_type == CORE_TYPE_VIRTUAL)
>> +			core_num = get_pcpu(chan_info, pkt->resource_id);
> Hi, Dave
>
> Now in DPDK code, only command send from VM(guest_cli sample) will set the
> pkt-<core_type to VIRTUAL, if we send JSON command with resource_id, seems
> we always won't hit this branch. Because  parse_json_to_pkt() will always set the
> core_type to CORE_TYPE_PHYSICAL.
> If resource_id in instruction format JSON file will always be treated as Pcpu, it's the
> same as core_list behavior.


Yes, that's correct.  But I believe it's OK for the moment.

Currently the only way for the guest app to send commands and policies
to the host is via the virtio-serial interface, which takes a different 
code path. There is no
way currently for the guest app to send a JSON string to the host, so 
that code path will
never be hit. However, now that the JSON functionality is in the host, 
we plan to add
that same functionality to the guest, and when we do that we will ensure 
that core_type
is handled appropriately.

Regards,
Dave.

>> +		else
>> +			core_num = pkt->resource_id;
>>
>>   		switch (pkt->unit) {
>>   		case(CPU_POWER_SCALE_MIN):
>> --
>> 2.17.1
>
>
Yao, Lei A Jan. 8, 2019, 12:47 p.m. | #3
> -----Original Message-----
> From: Hunt, David
> Sent: Tuesday, January 8, 2019 5:20 PM
> To: Yao, Lei A <lei.a.yao@intel.com>; dev@dpdk.org
> Subject: Re: [PATCH] examples/power: fix wrong core_id with JSON cmds
> 
> Hi Lei,
> 
> On 8/1/2019 2:02 AM, Yao, Lei A wrote:
> >
> >> -----Original Message-----
> >> From: Hunt, David
> >> Sent: Monday, January 7, 2019 7:39 PM
> >> To: dev@dpdk.org
> >> Cc: Hunt, David <david.hunt@intel.com>; Yao, Lei A <lei.a.yao@intel.com>
> >> Subject: [PATCH] examples/power: fix wrong core_id with JSON cmds
> >>
> >> This patch fixes a bug introduced in the 64-core limtation
> >> enhancement where the core_id is inadvertently converted from
> >> virtual to physical even though it may already be a physical
> >> core_id.
> >>
> >> We should be using the core_type field, and only converting via
> >> hypervisor when core_type is set to CORE_TYPE_VIRTUAL
> >>
> >> Fixes: 5776b7a371d1 ("examples/power: allow VM to use lcores over 63")
> >>
> >> Signed-off-by: David Hunt <david.hunt@intel.com>
Reviewed-by: Lei Yao<lei.a.yao@intel.com>
Tested-by: Lei Yao<lei.a.yao@intel.com>

> >> ---
> >>   examples/vm_power_manager/channel_monitor.c | 5 ++++-
> >>   1 file changed, 4 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/examples/vm_power_manager/channel_monitor.c
> >> b/examples/vm_power_manager/channel_monitor.c
> >> index 85622e7cb..1a3a0fa76 100644
> >> --- a/examples/vm_power_manager/channel_monitor.c
> >> +++ b/examples/vm_power_manager/channel_monitor.c
> >> @@ -640,7 +640,10 @@ process_request(struct channel_packet *pkt,
> struct
> >> channel_info *chan_info)
> >>   	if (pkt->command == CPU_POWER) {
> >>   		unsigned int core_num;
> >>
> >> -		core_num = get_pcpu(chan_info, pkt->resource_id);
> >> +		if (pkt->core_type == CORE_TYPE_VIRTUAL)
> >> +			core_num = get_pcpu(chan_info, pkt->resource_id);
> > Hi, Dave
> >
> > Now in DPDK code, only command send from VM(guest_cli sample) will set
> the
> > pkt-<core_type to VIRTUAL, if we send JSON command with resource_id,
> seems
> > we always won't hit this branch. Because  parse_json_to_pkt() will always
> set the
> > core_type to CORE_TYPE_PHYSICAL.
> > If resource_id in instruction format JSON file will always be treated as Pcpu,
> it's the
> > same as core_list behavior.
> 
> 
> Yes, that's correct.  But I believe it's OK for the moment.
> 
> Currently the only way for the guest app to send commands and policies
> to the host is via the virtio-serial interface, which takes a different
> code path. There is no
> way currently for the guest app to send a JSON string to the host, so
> that code path will
> never be hit. However, now that the JSON functionality is in the host,
> we plan to add
> that same functionality to the guest, and when we do that we will ensure
> that core_type
> is handled appropriately.
> 
> Regards,
> Dave.
> 
> >> +		else
> >> +			core_num = pkt->resource_id;
> >>
> >>   		switch (pkt->unit) {
> >>   		case(CPU_POWER_SCALE_MIN):
> >> --
> >> 2.17.1
> >
> >
Thomas Monjalon Jan. 14, 2019, 11:15 p.m. | #4
> > >> This patch fixes a bug introduced in the 64-core limtation
> > >> enhancement where the core_id is inadvertently converted from
> > >> virtual to physical even though it may already be a physical
> > >> core_id.
> > >>
> > >> We should be using the core_type field, and only converting via
> > >> hypervisor when core_type is set to CORE_TYPE_VIRTUAL
> > >>
> > >> Fixes: 5776b7a371d1 ("examples/power: allow VM to use lcores over 63")
> > >>
> > >> Signed-off-by: David Hunt <david.hunt@intel.com>
> Reviewed-by: Lei Yao<lei.a.yao@intel.com>
> Tested-by: Lei Yao<lei.a.yao@intel.com>

Applied, thanks

Patch

diff --git a/examples/vm_power_manager/channel_monitor.c b/examples/vm_power_manager/channel_monitor.c
index 85622e7cb..1a3a0fa76 100644
--- a/examples/vm_power_manager/channel_monitor.c
+++ b/examples/vm_power_manager/channel_monitor.c
@@ -640,7 +640,10 @@  process_request(struct channel_packet *pkt, struct channel_info *chan_info)
 	if (pkt->command == CPU_POWER) {
 		unsigned int core_num;
 
-		core_num = get_pcpu(chan_info, pkt->resource_id);
+		if (pkt->core_type == CORE_TYPE_VIRTUAL)
+			core_num = get_pcpu(chan_info, pkt->resource_id);
+		else
+			core_num = pkt->resource_id;
 
 		switch (pkt->unit) {
 		case(CPU_POWER_SCALE_MIN):