#1376 NORM BTest-4: "Lumistar 1G" SD card not detected by Linux.

[Zarro Boogs per Child]

#1376: "Lumistar 1G" SD card not detected by Linux.
---------------------+------------------------------------------------------
  Reporter:  cjb     |       Owner:  dilinger
      Type:  defect  |      Status:  new     
  Priority:  normal  |   Milestone:  BTest-4 
 Component:  kernel  |     Version:          
Resolution:          |    Keywords:          
  Verified:  0       |  
---------------------+------------------------------------------------------
Comment (by wmb at firmworks.com):

 OFW detects timeouts as follows:

 The "wait" routine polls the interrupt status register (ISR), looking for
 a completion bitmask that depends on the command. If "wait" sees the error
 bit set, it checks the ESR to determine what kind of error it was.  In the
 usual case, all errors are fatal, resulting in an abort.  But if the
 "allow-timeout?" flag is set, command timeouts do not abort, but rather
 set another "timeout?" flag that tells the caller that a timeout occurred.
 This is used for sensing version 2 cards, since v1 cards timeout on CMD8
 and v2 cards do not.

 Note that this special behavior is for command timeouts, not data
 timeouts.  Data timeouts are always fatal.

 If I recall correctly, Linux was seeing data timeouts under stress testing
 some time ago, which is how we discovered the off-by-one-power-of-two bug
 in the timeout value register.

-- 
Ticket URL: <http://dev.laptop.org/ticket/1376#comment:7>
One Laptop Per Child <http://laptop.org/>