Hi Petteri,
<snip>
>>> + node->expiration = segment_stat.st_mtime;
>>> + memcpy(node->mrs, buf, sizeof(node->mrs));
>>> + memcpy(&node->total_mrs, buf + sizeof(node->mrs),
>>> + sizeof(node->total_mrs));
>>> + memcpy(&node->sent_mrs,
>>> + buf + sizeof(node->mrs) + sizeof(node->total_mrs),
>>> + sizeof(node->sent_mrs));
>>> +
>>> + memcpy(&node->deliverable, buf + sizeof(node->mrs) +
>>> + sizeof(node->total_mrs) + sizeof(node->sent_mrs),
>>> + sizeof(node->deliverable));
>>
>> To make our lives easier, I'd make the id_table_node structure packed
>> and read the data directly into it. We don't have to worry about
>> byte-ordering at this point.
>>
>
> Yep. So, we'll get rid of memcopies, and temp-buffer allocation. So I'll
> change
> to this:
>
> struct id_table_node {
> struct sms_address to; // => this will be removed
> unsigned int mrs[8];
> time_t expiration;
> unsigned char total_mrs;
> unsigned char sent_mrs;
> gboolean deliverable;
> } __attribute__((packed));
>
Yes, this seems just fine to me. I already removed the to member by the
way. :)
<snip>
>>> +static gboolean sr_assembly_add_fragment_backup(const char *imsi,
>>> + const struct id_table_node *node,
>>> + unsigned int msg_id)
>>> +{
>>> + int len = sizeof(node->mrs) + sizeof(node->total_mrs) +
>>> + sizeof(node->sent_mrs) + sizeof(node->deliverable);
>>> + unsigned char buf[len];
>>> +
>>> + if (!imsi)
>>> + return FALSE;
>>> +
>>> + memcpy(buf, node->mrs, sizeof(node->mrs));
>>> +
>>> + memcpy(buf + sizeof(node->mrs), &node->total_mrs,
>>> + sizeof(node->total_mrs));
>>> +
>>> + memcpy(buf + sizeof(node->mrs) + sizeof(node->total_mrs),
>>> + &node->sent_mrs, sizeof(node->sent_mrs));
>>> +
>>> + memcpy(buf + sizeof(node->mrs) + sizeof(node->total_mrs) +
>>> + sizeof(node->sent_mrs), &node->deliverable,
>>> sizeof(node->deliverable)); +
>>
>> Again, simply writing the id_table_node structure might be easier here,
>> especially since you're not taking care of loading the expiration member.
>>
>
> Good point:) I forgot this expiration-time member :(
> With packed-node type shouldn't be missed anymore.
>
> I have to bother you little bit more with status report-expiration logic
> soon,
> because it's not implemented yet..
Sure :)
We might also need to revisit the address / mr matching. Some networks
are stupid and rewrite the SMS address in the status report. E.g. when
I send an SMS to 555-123-4567, the status report comes back with the
address rewritten to +15551234567.
So we might need to allow some fuzzy matching based on the last n digits
or something more clever.
Regards,
-Denis
_______________________________________________
ofono mailing list
[email protected]
http://lists.ofono.org/listinfo/ofono