[Trac #988] EC code, putchar XON/XOFF handling and crlf conversion

[Zarro Boogs per Child]

#988: EC code, putchar XON/XOFF handling and crlf conversion
---------------------------------+------------------------------------------
 Reporter:  frief                |       Owner:  ray.tseng at quantatw.com 
     Type:  defect               |      Status:  new                    
 Priority:  normal               |   Milestone:  Untriaged              
Component:  embedded controller  |    Keywords:  EC uart serial watchdog
---------------------------------+------------------------------------------
 the EC code (at the beginning of Olpc-Q2B73.rom) seems to
 contain and use putchar code for the UART of the embedded
 controler.

 While probably not a bug itself I'm not convinced that
 a lowest level output routine should do a crlf conversion,
 and that it should do XON/XOFF handling the way it does.

 If 0x13 should have been received the code would block
 until 0x11 is received in the loop at 0xb245..0xb24e.

 A minor glitch: within the XON/XOFF handling the ri
 serial flag is conditionally reset (at 0xb244).
 I'd have expected the ri flag to be reset directly after
 reading sbuf (as it is done after reading sbuf at 0xb248).

 {{{
 -----8<---------------------------------------
 putchar:
         mov     a,r7
         cjne    a,#0ah,Xb23b
         mov     a,#0dh
         lcall   Xb23b
         mov     a,#0ah
 Xb23b:  jnb     ri,Xb24f
         mov     r0,sbuf
         cjne    r0,#13h,Xb24f
         clr     ri      ; xx not always reset!
 Xb245:  jnb     ri,Xb245
         mov     r0,sbuf
         clr     ri
         cjne    r0,#11h,Xb245
 Xb24f:  jnb     ti,Xb24f
         clr     ti
         mov     sbuf,a
         ret
 ----->8---------------------------------------
 }}}

 Would a watchdog have bitten within the
 potentially endless XOFF/XON handling?

 Greetings,

 Frieder

-- 
Ticket URL: <http://dev.laptop.org/ticket/988>
One Laptop Per Child <http://laptop.org/>