jfilter stuck at 100% CPU after xrun

[kylerbrown]

After an xrun, the jfilter ports are closed on the JACK server, but the process continues, consuming 100% CPU until it is killed by hand.

Ideally, an xrun shouldn't kill jfilter, as this only happens on start-up sometimes.

[kylerbrown]

Here is an output from this type of error

underfull FIFO: last frame in FIFO corrupted
RHD2K: stopping acquisition
RHD2K: starting acquisition
RHD2K: xrun of 0.000 usec
20150520T180025.068523 [jfilter] jack xrun (us): 0
xrun: 0
bad status (-2) for client mic_filter handling event (type = 8)
20150520T180027.068565 [jdetect] jack xrun (us): 0
20150520T180027.068657 [jrecord] jack xrun (us): 0
20150520T180027.068897 [jrecord] ports disconnected: mic_filter:out_1 -> record_p98:pcm_000
20150520T180027.069715 [jdetect] ports disconnected: mic_filter:out_1 -> detect_p98:in
RHD2K: delayed process cycle (35945 frames)
20150520T180027.089031 [jrecord] ERROR: xrun
RHD2K: delayed process cycle (32254 frames)
RHD2K: delayed process cycle (28648 frames)
RHD2K: delayed process cycle (24986 frames)
RHD2K: delayed process cycle (21301 frames)
RHD2K: delayed process cycle (17718 frames)
RHD2K: delayed process cycle (14073 frames)
RHD2K: delayed process cycle (10388 frames)
RHD2K: delayed process cycle (6815 frames)
RHD2K: delayed process cycle (3161 frames)  

Then from TOP:

PID USER      PR  NI  VIRT  RES  SHR S  %CPU %MEM    TIME+  COMMAND                       
31434 kjbrown   20   0 90336  46m  20m S  99.7  0.8   2:07.29 jfilter                       
  280 root     -51   0     0    0    0 S   7.6  0.0  70:36.56 irq/18-uhci_hcd               
 4343 root      20   0  175m  48m 6992 S   3.7  0.8  37:53.89 Xorg                          
  194 root     -51   0     0    0    0 S   3.0  0.0  71:06.51 irq/18-ehci_hcd               
 4417 kjbrown   20   0  216m  17m 9532 S   2.3  0.3   1:34.05 x-terminal-emul               
31276 kjbrown   20   0  240m  27m  10m S   2.3  0.5   0:06.76 mencoder       

[kylerbrown]

Here is another possibly similar issue that caused a crash in jfilter:

subgraph starting at mic_filter timed out (subgraph_wait_fd=15, status = 0, state = Finished, pollret = 0 revents = 0x0)
pp: cannot clean up byte from graph wait fd (Resource temporarily unavailable)
RHD2K: delayed process cycle (877 frames)
RHD2K: null cycle
underfull FIFO: last frame in FIFO corrupted
RHD2K: stopping acquisition
RHD2K: starting acquisition
RHD2K: xrun of 262144.000 usec
subgraph starting at record_p98 lost client
bad status (-2) for client record_p98 handling event (type = 7)
cannot send port registration notification to record_p98 (Resource temporarily unavailable)
20150521T001718.045836 [jdetect] ports disconnected: mic_filter:out_1 -> detect_p98:in
20150521T001718.050475 [jdetect] ports disconnected: detect_p98:trig_out -> record_p98:trig_in
20150521T001718.050819 [jdetect] jack xrun (us): 262144
RHD2K: delayed process cycle (10983 frames)
logic error - calling shutdown handler
20150521T001718.093218 [jfilter] the server is shutting us down: logic error
RHD2K: delayed process cycle (3641 frames)
20150521T001718.169981 [jfilter] deactivated client
20150521T001718.780834 [jdetect] signal off: frames=428680768, us=2626039755316