[Skiboot] [PATCH v1 5/5] external/opal-prd: opal-prd to support htmgt passthru and attribute override

[Neelesh Gupta]

Hi Patrick,

Thanks for the review.

On 08/14/2015 05:05 AM, Patrick Williams wrote:
> On Wed, Aug 12, 2015 at 09:36:32PM +0530, Neelesh Gupta wrote:
>> +static void handle_prd_control_htmgt_passthru(struct control_msg
>> *send_msg,
>> +                       struct control_msg *recv_msg)
> ...
>> + pr_debug("CTRL: calling htmgt_pass_thru");
>> + send_msg->response =3D call_htmgt_pass_thru(recv_msg->data_len,
>> +                       recv_msg->data, &rsp_len,
>> +                       send_msg->data);
>> + send_msg->data_len =3D be16toh(rsp_len);
> Is recv_msg->data_len in the correct endianness prior to calling
> call_htmgt_pass_thru?  I don't see any obvious hto16be here, nor
> in handle_prd_control.

'external/opal-prd/thunk.S' takes care of the endianness of the 
arguments before
invoking HBRT by branching to 'call_be'. It is good as long as there is 
direct reference
to the arguments but if the callee dereference them, then 'opal-prd' 
needs to take
care of the byte ordering and use appropriate accessor..

>
>> +static void handle_prd_control_attr_override(struct control_msg
>> *send_msg,
>> +                      struct control_msg *recv_msg)
> ...
>> + pr_debug("CTRL: calling apply_attr_override");
>> + send_msg->response =3D call_apply_attr_override(
>> +         recv_msg->data, recv_msg->data_len);
>> + send_msg->data_len =3D 0;
>> +}
> Same endianness question here.
>
>> +
>> +/* TODO usage */
>>   static void usage(const char *progname)
>>   {
> Still TODO?  Looks like you did something.

It was my local TODO .. :) and forgot to remove it before posting,
will correct this.

Thanks,
Neelesh.

>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/skiboot/attachments/20150814/4b4855c2/attachment.html>