[dpdk-dev] [PATCH v2 11/18] devargs: simplify implementation

[Shreyansh Jain]

Hello Gaetan,

Please ignore this email (reason inline)

On Monday 16 October 2017 05:09 PM, Shreyansh Jain wrote:
> Hello Gaetan,
> 
> On Thursday 12 October 2017 01:51 PM, Gaetan Rivet wrote:
>> Re-use existing code, remove incorrect comments.
>>
>> Signed-off-by: Gaetan Rivet <gaetan.rivet at 6wind.com>
>> ---
>>   lib/librte_eal/common/eal_common_devargs.c | 8 +++-----
>>   1 file changed, 3 insertions(+), 5 deletions(-)
>>
>> diff --git a/lib/librte_eal/common/eal_common_devargs.c 
>> b/lib/librte_eal/common/eal_common_devargs.c
>> index 49cc3b8..1d87cd9 100644
>> --- a/lib/librte_eal/common/eal_common_devargs.c
>> +++ b/lib/librte_eal/common/eal_common_devargs.c
>> @@ -153,21 +153,19 @@ rte_eal_devargs_insert(struct rte_devargs *da)
>>       return 0;
>>   }
> 
> While trying to work on this patch, I noticed that the complete series 
> (including "Move PCI away from EAL") is not cleanly applicable on 
> current master (17.11 RC1). I thought it would be some tiny issues.
> 
> But there are some issues which I couldn't pass, Like...
> 
>> -/* store a whitelist parameter for later parsing */
>>   int
> 
> In the this function
> 
>> -rte_eal_devargs_add(const char *devargs_str)
>> +rte_eal_devargs_add(const char *dev)
>>   {
>>       struct rte_devargs *devargs = NULL;
>> -    const char *dev = devargs_str;
>> -    /* use calloc instead of rte_zmalloc as it's called early at init */
>>       devargs = calloc(1, sizeof(*devargs));
>>       if (devargs == NULL)
>>           goto fail;
>>       if (rte_eal_devargs_parse(devargs, "%s", dev))
>>           goto fail;
> 
> These lines don't exist in your patch
> 
> ---
> 59c2ba6c 172)   if (bus->conf.probe_mode == RTE_BUS_PROBE_UNDEFINED) {
> b631f3b0 173)           if (devargs->policy == RTE_DEV_WHITELISTED)
> 59c2ba6c 174)                   bus->conf.probe_mode = 
> RTE_BUS_PROBE_WHITELIST;
> b631f3b0 175)           else if (devargs->policy == RTE_DEV_BLACKLISTED)
> 59c2ba6c 176)                   bus->conf.probe_mode = 
> RTE_BUS_PROBE_BLACKLIST;
> 02823c1d 177)   }
> bf6dea0e 178)   TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
> bf6dea0e 179)   return 0;
> 0215a4c6 180)
> ---
> (Some introduced by the move PCI series, but others like b631f3b0 are 
> very old ~17.08)
> 
> 
>> -    TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
>> +    if (rte_eal_devargs_insert(devargs))
>> +        goto fail;
> 
> And hence, I don't know whether you intend to insert the above line 
> after or before checking PROBE.
> 
>>       return 0;
>>   fail:
>>
> 
> Maybe I am doing something wrong here - any ideas? Can you send an 
> updated/rebased version on current master HEAD?

Just after sending this, I noticed that I had not applied the "Bus 
control framework" patch set which the "devargs..." cover letter talks 
about.
I will try with that and confirm if there is still any issue.

> 
> -
> Shreyansh
>