[rabbitmq-discuss] RabbitMQ-c @ ARM Targets

Wed Aug 17 22:59:43 BST 2011

Some more debug information:

Just tried to run on both x86 and ARM. There's a difference in execution 
flow, even though they connect to the same server from the same subnet.

Similar flow for both:
- goes into while(amqp_data_in_buffer(state))
- receives result from server
- goes to the tick of while(1) loop
- enters while(amqp_data_in_buffer(state)) for the second time

ARM:
- tries to receive result from server again, but recv returns 0, which 
indicates that peer has performed a shutdown

x86:
- (decoded_frame->frame_type != 0) condition is met, so it returns 0 from 
function and moves on to sending log in credentials

If I understand correctly, that comment is relevant to that condition: 
decoded frame type == 0: Returning frame_type of zero indicates either 
insufficient input, or a complete, ignored frame was read.

Also, what I noticed, on ARM machine, inside of amqp_handle_input if 
(state->inbound_offset < state->target_size) is true...

But I'm not sure where to dig further.
Could it be a bug in rabbitmq-c? Or it's more likely something regarding 
cross-compiled lib itself or processor spec ?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20110817/a185b538/attachment.htm>