Re: Curious MySQL building problem /w GCC 3.4.2

On Thu, 2004-07-29 at 09:05, Roderick van Domburg wrote:
> First off, thumbs up for the GCC import. I was awaiting the import to 
> get a couple of sparc64 boxes up and running!
> 
> I know I should normally report this to the maintainer or file a PR, but 
> since this seems so curious and hot on the heels of the GCC import, I 
> figured I'd post it here.
> 
> I CVSupped and rebuilt a quad Xeon III on July 29th, and it runs just 
> fine. However, when recompiling MySQL, progressed just came to a halt at:
> 
> make  all-am
> Making all in strings
> cc  -DDBUG_OFF -O -pipe -march=pentium3 -c `test -f 'strings-x86.s' || 
> echo './'`strings-x86.s
> 
> 0% CPU and virtually no memory in use.
> 
> Workaround: compiling strings-x86.s without -pipe. I re-enabled it for 
> the rest of the build, which is now running in the background and seems 
> peachy.
> 
> No idea what causes it, but here's the report. Should I report it to the 
> maintainer or the GCC team instead?

Roderick, if you can try this again, what happens if you hit Ctrl-D (or
whatever your EOF character is) at the point where it hangs?

I ran into a similar problem with the www/mozilla port configure, but I
didn't get a chance to investigate in much detail.  Basically, the
configure process stops at this point:

checking for .hidden assembler directive...

At this point it's trying to compile an assembler file.  Relevant
processes running at the time are (subject to linewrap):

tomcat:bmah% ps -auxww | grep root | grep conftest
root  15871  0.0  0.0   328  232  p3  S+    7:54AM   0:00.00 cc -c -O
-pipe conftest.s
root  15872  0.0  0.2  1052  900  p3  S+    7:54AM   0:00.00 /usr/bin/as
-o conftest.o conftest.s -

Note the as process with a trailing "-" argument.  I guessed that it was
for some reason waiting for something from stdin, so I typed the EOF
character on the keyboard and that allowed the as process to continue. 
Now I have a freshly recompiled mozilla, but I'm not sure why.  :-p


Bruce.