Re: Possible race condition building libraries: head/amd64 r303329 -> r303379

On 7/27/2016 5:49 AM, David Wolfskill wrote:
> I track head daily on both my laptop and a "build machine;" I only saw a
> problem on the latter -- not on my laptop.
> 
> (The build machine is a bit beefier, and uses an SSD as its non-volatile
> storage; the laptop uses a hybrid disk -- in case that is useful.)
> 
> As indicated in the Subject, in each case I was performing a
> source-based upgrade-in-place from r303329 to r303379.  (And I've
> been doing this routinely for quite some time.)
> 
> The build failed (initially -- a restart worked):
> 
> ...
>>>> stage 4.2: building libraries
> ...
> --- secure/lib/libcrypto__L ---
> Building /common/S4/obj/usr/src/secure/lib/libcrypto/dso_openssl.o
> --- lib/ncurses/ncursesw__L ---
> /usr/lib/libtermlibw.so -> libncursesw.so
> /usr/lib/libtinfow.so -> libncursesw.so
> --- kerberos5/lib/libwind__L ---
> Building /common/S4/obj/usr/src/kerberos5/lib/libwind/normalize_table.So
> --- kerberos5/lib/libheimipcc__L ---
> /usr/obj/usr/src/tmp/usr/bin/ld: cannot find -lgcc_s
> cc: error: linker command failed with exit code 1 (use -v to see invocation)
> *** [libprivateheimipcc.so.11] Error code 1

Nathan Whitehorn ran into a very similar one last year that I couldn't
explain either.

> --- kerberos5/lib__L ---
> --- libkadm5srv.so.11 ---
> /usr/obj/usr/src/tmp/usr/bin/ld: cannot find -lgcc
> *** [libkadm5srv.so.11] Error code 1
> 
> make[5]: stopped in /usr/src/kerberos5/lib/libkadm5srv 


These shouldn't happen since libgcc is build in startup_libs and krb5 is
built in prebuild_libs, which waits on startup_libs. Very strange.

Can you send me the typescript please?

-- 
Regards,
Bryan Drewery