Openmoko Bug #2180: stable-tracking: 'rxserr' UART messages

[Openmoko Public Trac]

#2180: stable-tracking: 'rxserr' UART messages
-----------------------------+----------------------------------------------
 Reporter:  laforge          |          Owner:  openmoko-kernel         
     Type:  defect           |         Status:  new                     
 Priority:  high             |      Milestone:  FSO                     
Component:  System Software  |        Version:                          
 Severity:  major            |       Keywords:  gps s3x24xx_serial rxerr
 Haspatch:  0                |      Blockedby:                          
Estimated:                   |    Patchreview:                          
 Blocking:                   |   Reproducible:                          
-----------------------------+----------------------------------------------

Comment(by andy):

 Thanks for the report and dumps Sascha.

 434     Feb  8 00:53:16 gta02 kernel: [  384.800000] rxerr: port=1
 ch=0xb5, rxs=0x00000001
 435     Feb  8 00:53:16 gta02 kernel: [  384.845000] interrupts were
 disabled for 596 us !
 436     Feb  8 00:53:17 gta02 kernel: [  385.300000] rxerr: port=1
 ch=0xb5, rxs=0x00000001
 437     Feb  8 00:53:17 gta02 kernel: [  385.345000] rxerr: port=1
 ch=0x00, rxs=0x00000001
 438     Feb  8 00:53:21 gta02 kernel: [  389.825000] interrupts were
 disabled for 599 us !
 439     Feb  8 00:53:26 gta02 kernel: [  394.925000] rxerr: port=1
 ch=0x7a, rxs=0x00000001
 440     Feb  8 00:53:26 gta02 kernel: [  394.935000] interrupts were
 disabled for 593 us !
 441     Feb  8 00:53:31 gta02 kernel: [  399.970000] rxerr: port=1
 ch=0x0d, rxs=0x00000001
 442     Feb  8 00:53:31 gta02 kernel: [  399.970000] interrupts were
 disabled for 596 us !
 443     Feb  8 00:53:36 gta02 kernel: [  404.980000] rxerr: port=0
 ch=0x7e, rxs=0x00000001
 444     Feb  8 00:53:36 gta02 /usr/sbin/gsm0710muxd[1598]:
 gsm0710muxd.c:1168:gsm0710_advanced_buffer_get_frame(): Dropping frame:
 FCS doesn't match

 Well whatever else, error with b0 set (overrun) on ch0 with crtscts on is
 actually illegal, unless I miss the point somewhere.  There shouldn't be a
 way to get an overrun seen by the RX FIFO under those circumstances.

 Either the other end (GSM) is not set to use handshakes, the timing of
 using them is wrong, or the detail of the error report from the serial
 code is bogus somehow.

 The interrupt lockout period doesn't exceed 8ms anyway and doesn't
 correlate with the error presence.  There can still be (and probably is)
 trouble somewhere in terms of locking out serial interrupts by priority,
 so some other interrupt like USB is blocking lower priority serial
 service.

 I think maybe we learn something if we study the damage done to the
 received serial stream sequence by one of these events... if we can figure
 out how many chars are dropped or what corruption is happening.

-- 
Ticket URL: <https://docs.openmoko.org/trac/ticket/2180#comment:10>
docs.openmoko.org <http://docs.openmoko.org/trac/>
openmoko trac