Re: [RFA:] Fix PR optimization/15296, delayed-branch-slot bug.

[Hans-Peter Nilsson <hans-peter dot nilsson at axis dot com>]

> From: Eric Botcazou <ebotcazou@libertysurf.fr>
> Date: Wed, 5 May 2004 19:44:22 +0200
> User-Agent: KMail/1.4.3

> > I really don't know.  If you want me to check specific details
> > about that system, just ask.  I could have been more specific:
> > for instance "java doesn't build" refers to a (Sun) linker SEGV.
> 
> May I ask you to take a look at the sparc-sun-solaris2.* and 
> sparc-sun-solaris2.7 platform-specific build instructions?

Of course I had already done so and of course you still may.
(I don't know why people keep insisting that I've goofed on
basic stuff when I spot a problem.  Ah, maybe it's because I
actually *do* now and then! :-)

I checked again, and can't say I missed anything.  Last time I
read it, I basically ignored the stuff about "sun patches",
because the error descriptions didn't match what I saw, and I
didn't know how to check and shouldn't mess with system patches
anyway...  I've now STFW and found a reference to "showrev -p"
which says there's 106950-03 installed but no 107058-01, so I
don't think any of the system-specific installation items apply,
except the one about --disable-multilib.  Bootstrap compiler is
"gcc version 2.95.3 20010315 (release)" so it shouldn't matter
that 112760-07 isn't installed.

FWIW, I installed GNU make-3.79.1, bison-1.875, flex-2.5.4,
m4-1.4.

> > Oh BTW, no Ada.
> 
> Because you don't have an appropriate bootstrap compiler?

Correct.  That, and the sparc-testing not being a critical
factor for testing this patch anyway, at least not to the level
that I should jump through an extra magnitude or two of hoops
just to test Ada.  I don't do that for the local
i686-pc-linux-gnu either.

If you wish to pursue fixing these build problems, I can feed
(you and/or) bugzilla with the problem descriptions (really: the
last lines of the build log), one after another, starting with
what happens in a unified tree with configure
--disable-multilibs.  Hey, since it's TRT I'll do that anyway,
once I've tested this patch the way that works for me now.

If you "in return" or something, could do an extra round of
checking of this patch on some or all of 3.3, 3.4 and trunk, on
your favorite SPARC-based system (preferably different from
sparc-sun-solaris2.7 to cover more ground), that'd be great.

brgds, H-P