svn commit: r335244 - head

[Cy Schubert]

I have a script that invokes buildworld/kernel, installworld/kernel, then proceeds to update jails if the jails are not in use. I passed the wrong argument to my script. I hastily did control-r to find my last command and hit enter without thinking (being rushed as usual). I assumed I used the correct arguments. My assumption was incorrect. Sloppiness on my part.

My script takes a snapshot of /usr/src and /usr/obj (whatever dataset they reside in), svn up, buildworld/kernel, install*, mergemaster if requested, then finally deletes the snapshots (depending on arguments). The return code specifies whether poudeiere updates or not.

I was rushing out the door that evening, so I figured I could squeeze in a build by the time I got home. I specified to my script to build and install only the kernel but the poudeiere invocation that followed wasn't aware that I had invoked the previous script to only build and install the kernel. Not enough care for details that evening on my part. When I got home to discover the install failure I had forgotten that I passed an argument to the script to build/install only the kernel.

Not a big deal, just need to be more careful next time.

---
Sent using a tiny phone keyboard.
Apologies for any typos and autocorrect.
Also, this old phone only supports top post. Apologies.

Cy Schubert
<Cy.Schubert at cschubert.com> or <cy at freebsd.org>
The need of the many outweighs the greed of the few.
---

-----Original Message-----
From: Bryan Drewery
Sent: 19/06/2018 08:57
To: Cy Schubert
Cc: src-committers at freebsd.org; svn-src-all at freebsd.org; svn-src-head at freebsd.org
Subject: Re: svn commit: r335244 - head

On 6/15/2018 10:02 PM, Cy Schubert wrote:
> In message <201806160436.w5G4aXks039566 at slippy.cwsent.com>, Cy Schubert 
> writes:
>> In message <201806160035.w5G0ZJfd066325 at repo.freebsd.org>, Bryan 
>> Drewery writes
>> :
>>> Author: bdrewery
>>> Date: Sat Jun 16 00:35:19 2018
>>> New Revision: 335244
>>> URL: https://svnweb.freebsd.org/changeset/base/335244
>>>
>>> Log:
>>>   Assert that a build is done before an install.
>>>   
>>>   This should also catch cases where the wrong MAKEOBJDIRPREFIX is used for
>>>   install.
>>>   
>>>   MFC after:	2 weeks
>>>   Sponsored by:	Dell EMC
>>>
>>> Modified:
>>>   head/Makefile.inc1
>>>
>>> Modified: head/Makefile.inc1
>>> ===========================================================================
>> ==
>>> =
>>> --- head/Makefile.inc1	Fri Jun 15 23:42:22 2018	(r335243)
>>> +++ head/Makefile.inc1	Sat Jun 16 00:35:19 2018	(r335244)
>>> @@ -89,7 +89,10 @@ MK_GCC_BOOTSTRAP=	no
>>>  .if make(installworld) || make(install) || make(distributeworld) || \
>>>      make(stageworld)
>>>  .-include "${OBJTOP}/compiler-metadata.mk"
>>> +.if !defined(_LOADED_COMPILER_METADATA)
>>> +.error A build is required first.  You may have the wrong MAKEOBJDIRPREFIX
>>  s
>>> et.
>>>  .endif
>>> +.endif
>>>  
>>>  # Pull in COMPILER_TYPE and COMPILER_FREEBSD_VERSION early. Pull it from t
>> he
>>>  # tree to be friendlier to foreign OS builds. It's safe to do so unconditi
>> on
>>> ally
>>> @@ -195,6 +198,7 @@ compiler-metadata.mk: .PHONY .META
>>>  	@: > ${.TARGET}
>>>  	@echo ".info Using cached compiler metadata from build at $$(hostname) 
>>> on $$(date)" \
>>>  	    > ${.TARGET}
>>> +	@echo "_LOADED_COMPILER_METADATA=t" >> ${.TARGET}
>>>  .for v in ${_COMPILER_METADATA_VARS}
>>>  	@echo "${v}=${${v}}" >> ${.TARGET}
>>>  .endfor
>>>
>>
>> Another data point: This also broke the following:
>>
>> cwsys# poudriere jail -c -m src=/opt/src/svn-current -j HEADi386 -a i386
>> [00:00:00] Creating HEADi386 fs at /poudriere/amd64/jails/HEADi386... 
>> done
>> [00:00:01] Copying /opt/src/svn-current to /poudriere/amd64/jails/HEADi3
>> 86/usr/src... done
>> [00:06:01] Starting make installworld
>> --- installworld ---
>> make[1]: "/export/obj/opt/src/svn-current/i386.i386/compiler-metadata.mk
>> " line 1: Using cached compiler metadata from build at cwsys on Fri Jun 
>> 15 18:39:46 PDT 2018
>> make[1]: "/opt/src/svn-current/Makefile.inc1" line 93: A build is 
>> required first.  You may have the wrong MAKEOBJDIRPREFIX set.
>> *** [installworld] Error code 1
>>
>> make: stopped in /opt/src/svn-current
>> 1 error
>>
>> make: stopped in /opt/src/svn-current
>> [00:06:01] Error: Failed to 'make installworld'
>> [00:06:01] Error while creating jail, cleaning up.
>> [00:06:01] Removing HEADi386 jail... done
>> [00:06:03] Cleaning HEADi386 data... done
>> cwsys# 
> 
> Nevermind. I found it. Stupid error on my part.
> 
> 

I'm curious what you did? Did you svn update after a buildworld and then
tried installworld? That would fail 1 time here but not again in the
future.

(Thinking outloud now, I could probably reuse the vcsversion code in
newvers.sh to record the build revision during buildworld and check it
matches at installworld time, but the assumes the same vcs tool is on
the install machine so it probably wouldn't work.)

I'm happy to see it may have caught a subtle mistake for you.

-- 
Regards,
Bryan Drewery