Re: buildworld error

Am Sat, 24 Aug 2024 15:21:16 -0600
Warner Losh <imp@bsdimp.com> schrieb:

> On Sat, Aug 24, 2024 at 1:39 PM Gordon Bergling <gbe@freebsd.org> wrote:
> 
> > Hi Warner,
> >
> > On Sat, Aug 24, 2024 at 01:29:55PM -0600, Warner Losh wrote:  
> > > On Sat, Aug 24, 2024, 1:14 PM Gordon Bergling <gbe@freebsd.org> wrote:  
> > > > I got the following buildworld error a recent -CURRENT
> > > >  
> > > > ===> stand/i386/pxeldr (all)  
> > > > `kldstat.o' is up to date.
> > > > -14152 bytes available
> > > >
> > > > The same happens on stable/14:
> > > >  
> > > > ===> stand/i386/pxeldr (all)  
> > > > -22344 bytes available  
> > > > ===> share/misc (all)  
> > > > --- loader ---
> > > > *** [loader] Error code 1
> > > >
> > > > make[5]: stopped in /storage/freebsd/src/stable/14/stand/i386/pxeldr
> > > > 1 error
> > > >
> > > > src.conf looks like the following:
> > > > WITH_BEARSSL=1
> > > > WITH_RETPOLINE=1
> > > > WITHOUT_CLEAN=1
> > > >
> > > > I remove the whole obj directories and tried several full builds, but  
> > the  
> > > > error persists for a while.
> > > >
> > > > Has any one a clue how to solve this?  
> > >
> > > Either disable pxe, raise the pxe limit (though it may not work), or  
> > select  
> > > the 4th loader for pxeboot.
> > >
> > > The loader is too big on BIOS to enable all the options.  
> >
> > I looked at src.conf(5), but didn't found a switch to disable pxe. What I
> > am
> > wondering about is that no one is facing the problem, since this it is a
> > pretty clean build without and special modifications, despite the ones
> > mention
> > above in the src.conf.
> >
> > Did you have a hint on how to disable pxe?
> >  
> 
> I was sure that I'd documented everything, but it seems not:
> 
> =t
> PXEBOOT_DEFAULT_INTERP=4th
> PXEBOOTSIZE?=525000
> 
> I'll look to make sure I don't have a commit stuck in a branch somewhere....
> 
> Warner

After the introduction of

 WITHOUT_LOADER_PXEBOOT

and setting this knob to true in /etc/src.conf has mitigated for me the problem reported above
(it appears reproduceable when setting WITH_BEARSSL for me). But the problem mentione
reappeared recently (a week or two on CURRENT, building on testboxes almost daily ...) again.

Just for the record  


Kind regards,

oh

-- 
O. Hartmann