XO Debugging tips: Difference between revisions

From OLPC
Jump to navigation Jump to search
Line 37: Line 37:


ok t( fs-update u:\os111.zd )t
ok t( fs-update u:\os111.zd )t

== Timing reads from USB in OFW ==

To check that USB read times aren't slowing down fs-update, you

ok dev fsdisk : write-blocks-end ; : erase-blocks 2drop ; : write-blocks-start 3drop 0 ; dend fs-update u:\fs.zd

and then

ok t( fs-update u:\os111.zd )t

The first command directs redirects writes to SD into /dev/null, keeping everything else the same. fs-update tries to overlap reads and writes, so adding read time + write time doesn't give you the answer, but if you find that the null-write case exceeds your normal fs-update timings, then USB read time is a factor.

Revision as of 18:05, 4 August 2010

SD Card

Specially useful for XO-1.5

Retrieve internal SD card manufacturer

From Linux

find /sys -name oemid | xargs cat 

From OFW (Q3E49 and newer)

ok select int:0
ok show-cid

On older OFW try

ok dev /sd  patch get-cid get-csd attach-card  select int:0  " cid" $call-parent 10 dump

Interpreting the dump is not trivial. WMB dixit: "One of the fields is an enumerated hex number, another is a two-ascii character assigned ID (stored backwards), another is a 5-byte arbitrary ascii string (stored backwards), another is a BCD two-nibble version number, another is a 12-bit BCD code date, and another is a 4-byte serial number"

OEMID list

  • ADATA 0x534d or 0x5344
  • Kingston 0x544d
  • SanDisk (?) 0x4144

OFW

Timing commands in OFW

Try

ok t( yourcommand )t

For example you can time an fs-update command:

ok t( fs-update u:\os111.zd )t

Timing reads from USB in OFW

To check that USB read times aren't slowing down fs-update, you

 ok dev fsdisk  : write-blocks-end  ;  : erase-blocks 2drop ;  : write-blocks-start 3drop 0 ;  dend  fs-update u:\fs.zd

and then

ok t( fs-update u:\os111.zd )t

The first command directs redirects writes to SD into /dev/null, keeping everything else the same. fs-update tries to overlap reads and writes, so adding read time + write time doesn't give you the answer, but if you find that the null-write case exceeds your normal fs-update timings, then USB read time is a factor.