porting icedtea6 to FreeBSD

[Jason Helfman]

On Wed, Apr 18, 2012 at 09:21:36AM -0700, Jason Helfman thus spake:
>On Wed, Apr 18, 2012 at 09:23:43AM -0400, Andrew Hughes thus spake:
>>----- Original Message -----
>>> On Tue, Apr 17, 2012 at 08:08:43AM -0700, Jason Helfman thus spake:
>>> >On Tue, Apr 17, 2012 at 7:56 AM, Andrew Hughes <ahughes at redhat.com>
>>> >wrote:
>>> >
>>> >> ----- Original Message -----
>>> >> >
>>> >> >
>>> >> >
>>> >> > On Tue, Apr 17, 2012 at 4:51 AM, Andrew Hughes <
>>> >> > ahughes at redhat.com >
>>> >> > wrote:
>>> >> >
>>> >> >
>>> >> >
>>> >> > ----- Original Message -----
>>> >> > > On Mon, Apr 16, 2012 at 09:59:00AM -0400, Andrew Hughes thus
>>> >> > > spake:
>>> >> > > >> Any thoughts on any of this?
>>> >> > > >> Thanks, again.
>>> >> > > >>
>>> >> > > >> -jgh
>>> >> > > >
>>> >> > > >What are you doing here?
>>> >> > > >
>>> >> > > >/usr/bin/sed -i.bak -e 's|--dry-run||g'
>>> >> > >
>>> >> >/usr/local/workspace/redports/java/icedtea6/work/icedtea6-1.11.1/Makefile.am
>>> >> > >
>>> >> >/usr/local/workspace/redports/java/icedtea6/work/icedtea6-1.11.1/Makefile.in
>>> >> > > >/usr/bin/sed -i.bak -e 's|--check|-q|g'
>>> >> > >
>>> >> >/usr/local/workspace/redports/java/icedtea6/work/icedtea6-1.11.1/Makefile.am
>>> >> > >
>>> >> >/usr/local/workspace/redports/java/icedtea6/work/icedtea6-1.11.1/Makefile.in
>>> >> > > >
>>> >> > > >If you're removing --dry-run, you're going to be applying the
>>> >> > > >patch
>>> >> > > >twice, which is why it's failing.
>>> >> > > >--
>>> >> > > >Andrew :)
>>> >> > >
>>> >> > > I've modified this to test the patch which in FreeBSD is the
>>> >> > > "-C"
>>> >> > > switch,
>>> >> > > and the "-q" is for SHASUM's to only output the hash.
>>> >> > >
>>> >> > > I was able to get much further in the build! Any thoughts on
>>> >> > > why
>>> >> > > this
>>> >> > > patch
>>> >> > > was seen as already applied?
>>> >> > >
>>> >> >
>>> >> > Because the build tests the patch, then applies it if that
>>> >> > passes. If
>>> >> > you
>>> >> > removed --dry-run from the first invocation and didn't replace
>>> >> > it
>>> >> > with -C,
>>> >> > you'd be applying the patch twice instead.
>>> >> >
>>> >> >
>>> >> >
>>> >> > That is what I have done, though. --dry-run is being replaced
>>> >> > with -C
>>> >> > in the build.
>>> >> > Is something else showing up?
>>> >> >
>>> >> > -jgh
>>> >> >
>>> >> >
>>> >>
>>> >> And it's still failing in the same way?  I thought you said you'd
>>> >> got much
>>> >> further... :-S
>>> >> --
>>> >> Andrew :)
>>> >>
>>> >
>>> >Yes, it is failing in the same way but much further into the build.
>>> >Initially, it would fail at the first/second patch.
>>> >I attached the latest buildlog to my email last night. It should
>>> >show the
>>> >issue, and with what patch the problem
>>> >was with.
>>> >
>>> >-jgh
>>>
>>> I believe I found part of the issue. In the patchset, the file
>>> openjdk/7034464-hugepage.patch has a number of repatched instances
>>> within
>>> os_linux.cpp, which is more than likely the cause of this issue. It
>>> is being
>>> patched twice.
>>>
>>> Starting at line 40, and another starting at line 209.
>>>
>>> For instance, both are adding this information to the same file:
>>>
>>> 47 +// Define MAP_HUGETLB here so we can build HotSpot on old
>>> systems.
>>> 48 +#ifndef MAP_HUGETLB
>>> 49 +#define MAP_HUGETLB 0x40000
>>> 50 +#endif
>>>
>>> 227 +// Define MAP_HUGETLB here so we can build HotSpot on old
>>> systems.
>>> 228 +#ifndef MAP_HUGETLB
>>> 229 +#define MAP_HUGETLB 0x40000
>>> 230 +#endif
>>>
>>> -jgh
>>>
>>> --
>>> Jason Helfman         | FreeBSD Committer
>>> jgh at FreeBSD.org       | http://people.freebsd.org/~jgh
>>>
>>
>>Ah right, now that's more helpful :-)
>>
>>It looks like the patch is a bit messed up, and BSD's patch program
>>is just not as tolerant as the GNU one.  My log shows this:
>>
>>Checking patches/openjdk/7034464-hugepage.patch
>>Applying patches/openjdk/7034464-hugepage.patch
>>patching file openjdk/hotspot/src/os/linux/vm/globals_linux.hpp
>>patching file openjdk/hotspot/src/os/linux/vm/os_linux.cpp
>>patching file openjdk/hotspot/src/os/linux/vm/os_linux.cpp.orig
>>patching file openjdk/hotspot/src/os/linux/vm/os_linux.hpp
>>Checking patches/openjdk/7037939-hugepage.patch
>>Applying patches/openjdk/7037939-hugepage.patch
>>patching file openjdk/hotspot/src/os/linux/vm/os_linux.cpp
>>Hunk #1 succeeded at 4148 (offset -22 lines).
>>patching file openjdk/hotspot/src/os/solaris/vm/os_solaris.cpp
>>Hunk #2 succeeded at 5041 (offset -2 lines).
>>Checking patches/openjdk/7043564-hugepage.patch
>>Applying patches/openjdk/7043564-hugepage.patch
>>patching file openjdk/hotspot/src/os/linux/vm/os_linux.cpp
>>Hunk #1 succeeded at 5084 (offset 2234 lines).
>>
>>I'll look at fixing this up in HEAD, but for BSD you can just disabled these
>>three by deleting the references from Makefile.am.
>
>Great, thanks. I don't see what to delete though. Would you mind sending a
>diff?
>

I've got it. I ended up dropping one more patch file, and now I am getting
this error:

patches/openjdk/7103610-_NET_WM_PID_and_WM_CLIENT_MACHINE_are_not_set.patch
patch: **** misordered hunks! output would be garbled
ERROR patch
patches/openjdk/7103610-_NET_WM_PID_and_WM_CLIENT_MACHINE_are_not_set.patch
FAILED!
WARNING make clean-patch before retrying a fix

Is this patch necessary for FreeBSD?

-jgh

--
Jason Helfman         | FreeBSD Committer
jgh at FreeBSD.org       | http://people.freebsd.org/~jgh