Re: [dts] [PATCH 2/9] framework: platform independent cpu info parsing

[gowrishankar <gowrishankar.m@linux.vnet.ibm.com>]

Hi,
Thanks for this check. For yocto or any other platform which do not have
lscpu by default, we can add a function to convert cpu info into a parsable
format as given by lscpu. Any thoughts ?.

Thanks,
Gowrishankar

On Monday 29 February 2016 07:43 AM, Xu, HuilongX wrote:
> Hi Gowrishankar,
> Lscpu cmdline not default in yocto os.
> Thanks  a lot
>> -----Original Message-----
>> From: dts [mailto:dts-bounces@dpdk.org] On Behalf Of Gowrishankar
>> Sent: Friday, February 26, 2016 6:17 PM
>> To: dts
>> Subject: [dts] [PATCH 2/9] framework: platform independent cpu info
>> parsing
>>
>> To collect thread/core/socket, /proc/cpuinfo would not help in case of
>> powerpc.
>> Instead, lscpu seems to be a better alternative and platform neutral
>> approach.
>>
>> Signed-off-by: Gowrishankar <gowrishankar.m@linux.vnet.ibm.com>
>> ---
>>   framework/crb.py |   40 ++++++++++++----------------------------
>>   1 file changed, 12 insertions(+), 28 deletions(-)
>>
>> diff --git a/framework/crb.py b/framework/crb.py
>> index c6fd9fb..1711f37 100644
>> --- a/framework/crb.py
>> +++ b/framework/crb.py
>> @@ -495,41 +495,25 @@ class Crb(object):
>>
>>           cpuinfo = \
>>               self.send_expect(
>> -                "grep --color=never \"processor\\|physical id\\|core
>> id\\|^$\" /proc/cpuinfo",
>> +                "lscpu -p|grep -v \#",
>>                   "#", alt_session=True)
>>
>> -        if "processor" not in cpuinfo:
>> -            # yocto not support --color=never, but ubuntu must need --
>> color=never,
>> -            # so check cpuinfo, before parsing cpuinfo, if cpuifo get
>> error, delete --color=never
>> -            # and get cpuinfo again
>> -            cpuinfo = \
>> -                self.send_expect(
>> -                    r'grep "processor\|physical id\|core id\|^$"
>> /proc/cpuinfo',
>> -                    "#", alt_session=True)
>> -
>> -        cpuinfo = cpuinfo.split('\r\n\r\n')
>> +        cpuinfo = cpuinfo.split()
>>           # haswell cpu on cottonwood core id not correct
>>           # need addtional coremap for haswell cpu
>>           core_id = 0
>>           coremap = {}
>>           for line in cpuinfo:
>> -            m = re.search("processor\t: (\d+)\r\n" +
>> -                          "physical id\t: (\d+)\r\n" +
>> -                          "core id\t\t: (\d+)", line)
>> -
>> -            if m:
>> -                thread = m.group(1)
>> -                socket = m.group(2)
>> -                core = m.group(3)
>> -
>> -                if core not in coremap.keys():
>> -                    coremap[core] = core_id
>> -                    core_id += 1
>> -
>> -                if self.crb['bypass core0'] and core == '0' and socket ==
>> '0':
>> -                    self.logger.info("Core0 bypassed")
>> -                    continue
>> -                self.cores.append(
>> +            (thread, core, socket) = line.split(',')[0:3]
>> +
>> +            if core not in coremap.keys():
>> +                coremap[core] = core_id
>> +                core_id += 1
>> +
>> +            if self.crb['bypass core0'] and core == '0' and socket == '0':
>> +                self.logger.info("Core0 bypassed")
>> +                continue
>> +            self.cores.append(
>>                       {'thread': thread, 'socket': socket, 'core':
>> coremap[core]})
>>
>>           self.number_of_cores = len(self.cores)
>> --
>> 1.7.10.4
>