When there is a backlog of mesh packets to be sent, the packet sender
incorrectly infers that the tx worker thread is already running
and therefore needn't be invoked. As a result, the mesh daemon will
sometimes stop broadcasting while there are still packets in the queue.
It will not resume broadcasting.
This patch will invoke the tx worker thread correctly in that case.
The logic to send packets at least twice when the transmitter is idle
was slightly modified accordingly, and should behave the same way as
before.
---
mesh/mesh-io-generic.c | 16 ++++++----------
1 file changed, 6 insertions(+), 10 deletions(-)
diff --git a/mesh/mesh-io-generic.c b/mesh/mesh-io-generic.c
index 50a2a6a86..eed77a7a8 100644
--- a/mesh/mesh-io-generic.c
+++ b/mesh/mesh-io-generic.c
@@ -725,7 +725,6 @@ static bool send_tx(struct mesh_io *io, struct mesh_io_send_info *info,
{
struct mesh_io_private *pvt = io->pvt;
struct tx_pkt *tx;
- bool sending = false;
if (!info || !data || !len || len > sizeof(tx->pkt))
return false;
@@ -739,23 +738,20 @@ static bool send_tx(struct mesh_io *io, struct mesh_io_send_info *info,
if (info->type == MESH_IO_TIMING_TYPE_POLL_RSP)
l_queue_push_head(pvt->tx_pkts, tx);
else {
- if (pvt->tx)
- sending = true;
- else
- sending = !l_queue_isempty(pvt->tx_pkts);
-
- l_queue_push_tail(pvt->tx_pkts, tx);
-
/*
* If transmitter is idle, send packets at least twice to
* guard against in-line cancelation of HCI command chain.
*/
- if (info->type == MESH_IO_TIMING_TYPE_GENERAL && !sending &&
+ if (info->type == MESH_IO_TIMING_TYPE_GENERAL && !pvt->tx &&
+ l_queue_isempty(pvt->tx_pkts) &&
tx->info.u.gen.cnt == 1)
tx->info.u.gen.cnt++;
+
+ l_queue_push_tail(pvt->tx_pkts, tx);
}
- if (!sending) {
+ /* If not already sending, schedule the tx worker */
+ if (!pvt->tx) {
l_timeout_remove(pvt->tx_timeout);
pvt->tx_timeout = NULL;
l_idle_oneshot(tx_worker, pvt, NULL);
--
2.34.1
This is automated email and please do not reply to this email!
Dear submitter,
Thank you for submitting the patches to the linux bluetooth mailing list.
This is a CI test results with your patch series:
PW Link:https://patchwork.kernel.org/project/bluetooth/list/?series=648769
---Test result---
Test Summary:
CheckPatch FAIL 1.04 seconds
GitLint PASS 0.74 seconds
Prep - Setup ELL PASS 52.08 seconds
Build - Prep PASS 0.72 seconds
Build - Configure PASS 10.12 seconds
Build - Make PASS 1878.37 seconds
Make Check PASS 12.93 seconds
Make Check w/Valgrind PASS 530.30 seconds
Make Distcheck PASS 280.29 seconds
Build w/ext ELL - Configure PASS 10.22 seconds
Build w/ext ELL - Make PASS 1835.88 seconds
Incremental Build with patchesPASS 0.00 seconds
Details
##############################
Test: CheckPatch - FAIL
Desc: Run checkpatch.pl script with rule in .checkpatch.conf
Output:
[BlueZ,1/1] Fix bug where bluetooth-meshd stops sending
WARNING:LONG_LINE: line length of 88 exceeds 80 columns
#111: FILE: mesh/mesh-io-generic.c:746:
+ l_queue_isempty(pvt->tx_pkts) &&
/github/workspace/src/12875093.patch total: 0 errors, 1 warnings, 36 lines checked
NOTE: For some of the reported defects, checkpatch may be able to
mechanically convert to the typical style using --fix or --fix-inplace.
/github/workspace/src/12875093.patch has style problems, please review.
NOTE: Ignored message types: COMMIT_MESSAGE COMPLEX_MACRO CONST_STRUCT FILE_PATH_CHANGES MISSING_SIGN_OFF PREFER_PACKED SPDX_LICENSE_TAG SPLIT_STRING SSCANF_TO_KSTRTO
NOTE: If any of the errors are false positives, please report
them to the maintainer, see CHECKPATCH in MAINTAINERS.
---
Regards,
Linux Bluetooth