Re: [PATCH 1/8] lpm: Fix pointer aliasing issues

[Aaron Conole <aconole@redhat.com>]

Hi Bruce,

Bruce Richardson <bruce.richardson@intel.com> writes:

> On Thu, Feb 25, 2016 at 01:48:34PM -0500, Aaron Conole wrote:
>> The current implementation attempts to use a uint16_t to alias the lpm table
>> structures. Such aliasing can break optimizer performance. This patch uses
>> union type indirection and adds static inline functions for performing the
>> aliasing.
>> 
>> Signed-off-by: Aaron Conole <aconole@redhat.com>
>> ---
>>  lib/librte_lpm/rte_lpm.h | 53 +++++++++++++++++++++++++++++++++++++-----------
>>  1 file changed, 41 insertions(+), 12 deletions(-)
>> 
>> diff --git a/lib/librte_lpm/rte_lpm.h b/lib/librte_lpm/rte_lpm.h
>> index c299ce2..eae6ff1 100644
>> --- a/lib/librte_lpm/rte_lpm.h
>> +++ b/lib/librte_lpm/rte_lpm.h
>> @@ -157,6 +157,33 @@ struct rte_lpm {
>>  };
>>  
>>  /**
>> + * Convert from tbl_entry types to integer types
>> + */
>> +static inline uint16_t
>> +rte_lpm_tbl24_entry_to_uint16(const struct rte_lpm_tbl24_entry *entry)
>> +{
>> +	union {
>> +		uint16_t                   i;
>> +		struct rte_lpm_tbl24_entry s;
>> +	} tbl_entry_u;
>> +
>> +	tbl_entry_u.s = *entry;
>> +	return tbl_entry_u.i;
>> +}
>> +
>> +static inline uint16_t
>> +rte_lpm_tbl8_entry_to_uint16(const struct rte_lpm_tbl8_entry *entry)
>> +{
>> +	union {
>> +		uint16_t                  i;
>> +		struct rte_lpm_tbl8_entry s;
>> +	} tbl_entry_u;
>> +
>> +	tbl_entry_u.s = *entry;
>> +	return tbl_entry_u.i;
>> +}
>> +
>
> These two new functions could be reduced to one with the help of patch:
> http://dpdk.org/dev/patchwork/patch/9087/
>
> Anyone care to go back and review or ack that patch for me and simplify all
> the lpm code just that little bit?

I definitely ack that patch; I also want to apply it in my local testing
setup and test a bit so you can add my Tested-by, but it looks sane on
the surface.

Thanks so much for the review!

> /Bruce