Ticket27678 #323
Ticket27678 #323
Conversation
This commit gives us a control port call to emit a CIRC_BW event for a single circuit instead of all circuits. This commit only moves code. No functionality has been changed.
We determine that a cell was dropped by inspecting CIRC_BW fields. If we did not update the delivered or overhead fields, the cell was dropped/not processed. Also emit CIRC_BW events for cases where we decide to close the circuit in this function, so vanguards can print messages about dropped cells in those cases, too.
src/core/or/command.c
Outdated
|
|
||
| /* Stash the original delivered and overhead values, so we can inform the | ||
| * control port about dropped cells immediately after processing below. */ | ||
| orig_delivered_bw = ocirc->n_delivered_read_circ_bw; |
What parts of the code can modify these values and we are keeping the originals? Perhaps we can put it in the comment.
Clarified comment in 02b5a6c
src/core/or/command.c
Outdated
| @@ -535,6 +542,7 @@ command_process_relay_cell(cell_t *cell, channel_t *chan) | |||
| (unsigned)cell->circ_id); | |||
| if (CIRCUIT_IS_ORIGIN(circ)) { | |||
| circuit_log_path(LOG_WARN, LD_OR, TO_ORIGIN_CIRCUIT(circ)); | |||
| control_event_circ_bandwidth_used_for_circ(TO_ORIGIN_CIRCUIT(circ)); | |||
We seem to be calling this function from three places. How can we know that these are the only places we need to call it from?
We call it when we get an error to close the circuit, or when the codepaths called from circuit_receive_relay_cell() fails to update bw values (because the cell was not valid).
Comment 02b5a6c also clarifies this.
src/core/or/command.c
Outdated
| * so the controller can react quickly. */ | ||
| if (orig_delivered_bw == ocirc->n_delivered_read_circ_bw && | ||
| orig_overhead_bw == ocirc->n_overhead_read_circ_bw) { | ||
| control_event_circ_bandwidth_used_for_circ(ocirc); |
Does it matter if we double-call this new function and emit two control events? Maybe we can add a guard in the function to make sure we don't double or triple call it?
It does not matter if we call this function multiple times. It already has a guard to prevent redundant events. See updated comments in 66b05e3.
| @@ -480,6 +480,8 @@ command_process_relay_cell(cell_t *cell, channel_t *chan) | |||
| const or_options_t *options = get_options(); | |||
| circuit_t *circ; | |||
| int reason, direction; | |||
| uint32_t orig_delivered_bw = 0; | |||
Nitpick: Commit msg is wrong, references 27679 instead of 27678.
Ok I will fix this in a new branch. I don't like that commit summary anyway, since it doesn't change code.
Clarify comments about stored bw fields.
https://trac.torproject.org/projects/tor/ticket/27678
The text was updated successfully, but these errors were encountered: