On Wed, Feb 10, 2010 at 10:16 PM, David Brownell <[email protected]> wrote:
> On Wednesday 10 February 2010, Řyvind Harboe wrote:
>> > I would like to commit this arm11 lockup:
>> >
>> > https://lists.berlios.de/pipermail/openocd-development/2010-February/014665.html
>
> This patch addressees a different issue than the bug report you
> filed (which seems to me to be a clear case of flakey config
> scripts).

I did not report on what the bug was, I submitted a report on
how to reproduce it.

Cranking up the khz was just a robust way to exercise the infinite
loop code path. Easily spotted in the debugger....

> The patch scans OK, except that "i" is a bad name. "loop_count"
> would be more clear.  (And the description is weak, since
> the problem isn't coupled to reset or jtag clocks.)

I pushed the code changes I tested without modifications, but I tried
to make the commit comment clearer on what was fixed vs. how
to reproduce it.

> While you're thinking about it, you might consider a quick audit
> of  at least the ARM11 code to catch any other such "spin forever"
> loops to fix...

Did that a long time ago. This bug was one that slipped through the net...

Did a fresh review now, and I think that was the last one in arm11.

> It's a general rule that *ALL* "spin until hardware reports
> success" loops need timeouts.  Exceptions are rare, and mostly
> seem to revolve around "if the hardware is that broken there
> is no recovery possible".  If the loop is in OpenOCD, then a
> recovery is almost certainly possible.

Mainly I expect this failure when trying to author config scripts when
lots of incorrect settings & khz rates are fed to the target....


>> so bugfixes weigh in more heavily than
>> regression concerns for this target, IMO.
>
> If you mean "potential regressions",

I meant I'm more concerned about fixing bugs(regressions
or not) than the risk of introducing regressions when fixing
bugs for arm11 since it is a relatively fresh target.

For arm7/9 I would be more paranoid about these sort of changes
at this point. Better one obscure bug left in for 0.4 than chancing
breaking some common code path for arm7/9.



-- 
Øyvind Harboe

Visit us at Embedded World, March 2nd-4th. IS2T's stand, HALL 10 - 118
http://www.zylin.com/events_embeddedworld.html

US toll free 1-866-980-3434 / International +47 51 63 25 00
http://www.zylin.com/zy1000.html
ARM7 ARM9 ARM11 XScale Cortex
JTAG debugger and flash programmer
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to