Re: Still seeing Failed assertion: "p[i] == 0" on armv7 buildworld

On Oct 12, 2025, at 09:11, void <void@f-m.fm> wrote:

> On Sun, Oct 12, 2025 at 07:31:15AM -0700, bob prohaska wrote:
> 
>> The machine is a Pi2 v1.1, USB mechanical hard disk, headless with
>> serial console and ssh access.
> 
> Does the problem also exist on 15-stable or 14-stable?
> 
> If so, I'll set up a machine with either of those but with
> /usr/src and /usr/obj and /var/cache/ccache nfs-mounted.


As I understand it:

) CURRENT's distributed world builds have DEBUG builds, including
   for jemalloc

) STABLE's distributed world builds have NON-DEBUG builds, including
   for jemalloc

If that is correct, jemalloc would not assert but might otherwise
misbehave for an official distribution's STABLE build. (What
matters is the running world, not the built one.)


Also, maximizing the amount of build activity may be more
likely to hit the problem sooner. That could mean that
ccache use would work against the purpose of such tests.
Also, as I understand, Bob does not use ccache for his
builds --but is using WITH_META_MODE for his more normal
activities now. However, much of this failure activity was
for a context that had not actually enabled WITH_META_MODE.
For failure reproducibility testing, WITH_META_MODE use
could work against the purpose of such tests as well.


===
Mark Millard
marklmi at yahoo.com