Hello M-Labs people,
Just to let you know that current RTEMS master is broken for the combo
Milkymist+Flickernoise.
As soon as the RTEMS console starts, there are a few messages and then
either the M1 reboots (NULL pointer dereferenced?) or the console freezes.
After a git bisect on RTEMS master branch here is the guilty commit:
fallen@pckirox:/opt/milkymist/rtems$ git bisect bad
fed92f5f67bcaeffa6cb6ceebf5ddfdbcaff7fb8 is the first bad commit
commit fed92f5f67bcaeffa6cb6ceebf5ddfdbcaff7fb8
Author: Sebastian Huber <sebastian.hu...@embedded-brains.de
<mailto:sebastian.hu...@embedded-brains.de>>
Date: Mon Jun 10 14:36:44 2013 +0200
termios: Notify driver about inactive transmit
Returning this state in the return value leads to race conditions on
SMP. The inactive state notification must be inside the critical
section.
The commit is available over there:
http://git.rtems.org/rtems/commit/?id=fed92f5f67bcaeffa6cb6ceebf5ddfdbcaff7fb8
I think that we should track a fix RTEMS SHA-1 in xiangfu's script so
that we can ensure that the build works since we don't have the will or
the manpower to review RTEMS master branch and stay informed of the
RTEMS changes.
We can let the script track RTEMS master by using an option (cmd line
option or environment variable ?) to allow experimenting new RTEMS
versions as it is the case right now.
So I propose: make -C compile-flickernoise <target> would use the last
known working RTEMS SHA-1 hard coded in the makefile (that we can update
when needed)
and RTEMS_USE_MASTER=1 make -C compile-flickernoise <target> to use the
HEAD of master branch of RTEMS
Cheers :)
--
Yann Sionneau
_______________________________________________
Devel mailing list
Devel@lists.milkymist.org
https://ssl.serverraum.org/lists/listinfo/devel
