Re: [OLPC Engineering] [Techteam] 11.3.1 build 10 released for XO-1.75

[John Watlington]

Actually, I just realized that I've been seeing the "stuck in suspend"
crash on B1s, which don't have hardware signaling of the CPU status
to the EC.   Can an EC hacker confirm that the EC will continue to
think the system is in suspend until the EC talks to it ?

If this is true, then the "stuck in suspend" issue may actually
be the same as "hung in early resume".

Cheers,
wad

On Nov 9, 2011, at 3:55 PM, C. Scott Ananian wrote:

>  On Wed, Nov 9, 2011 at 3:43 PM, Martin Langhoff
>  <martin.langhoff(a)gmail.com&gt; wrote:
> > The "suspense resume" build, featuring suspend resume that may or may
> > not resume.
>  
>  IIRC, some of the resume crashes are in fact suspend crashes.  So it
>  may not even suspend.  More suspense!
>  --scott
>  
>  -- 
>        ( http://cscott.net )
>  _______________________________________________
>  Techteam mailing list
>  Techteam(a)lists.laptop.org
>  http://lists.laptop.org/listinfo/techteam
>  _______________________________________________
>  Engineering mailing list
>  Engineering(a)laptop.org
>  http://lists.laptop.org/listinfo/engineering
>

[OLPC Engineering] [Techteam] 11.3.1 build 10 released for XO-1.75

[Paul Fox]

john wrote:
>  
>  Actually, I just realized that I've been seeing the "stuck in suspend"
>  crash on B1s, which don't have hardware signaling of the CPU status
>  to the EC.   Can an EC hacker confirm that the EC will continue to
>  think the system is in suspend until the EC talks to it ? 
                         
                  ^^ if you mean SoC, then yes.

because of the lack of a SOC_SLEEP signal on B3 and earlier, the EC
relies on a hint from the SoC that it will be in suspend soon.  the
command is sent some time before suspend actually occurs.  during
resume, any SoC command to the EC will let the EC know suspend is
finished -- again, a command is sent sometime soon after suspend,
but we've made no effort to make it "absolutely right away".  because
the EC doesn't have a critical need to know, and because these are
B units we're talking about, "close" was considered good enough.
(the EC uses the information to blink the LED, and to save power
by sleeping, itself.)

paul

>  
>  If this is true, then the "stuck in suspend" issue may actually
>  be the same as "hung in early resume".
>  
>  Cheers,
>  wad
>  
>  On Nov 9, 2011, at 3:55 PM, C. Scott Ananian wrote:
>  
>  > On Wed, Nov 9, 2011 at 3:43 PM, Martin Langhoff
>  > <martin.langhoff(a)gmail.com&gt; wrote:
>  >> The "suspense resume" build, featuring suspend resume that may or may
>  >> not resume.
>  > 
>  > IIRC, some of the resume crashes are in fact suspend crashes.  So it
>  > may not even suspend.  More suspense!
>  > --scott
>  > 
>  > -- 
>  >       ( http://cscott.net )
>  > _______________________________________________
>  > Techteam mailing list
>  > Techteam(a)lists.laptop.org
>  > http://lists.laptop.org/listinfo/techteam
>  > _______________________________________________
>  > Engineering mailing list
>  > Engineering(a)laptop.org
>  > http://lists.laptop.org/listinfo/engineering
>  > 
>  
>  _______________________________________________
>  Techteam mailing list
>  Techteam(a)lists.laptop.org
>  http://lists.laptop.org/listinfo/techteam
>  _______________________________________________
>  Engineering mailing list
>  Engineering(a)laptop.org
>  http://lists.laptop.org/listinfo/engineering 
=---------------------
 paul fox, pgf(a)laptop.org