Fix CMake build

Fix CMake build
Clean up the LLVMBuild and CMakeLists.txt files

BUG=none
TEST=none

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-llvm.git;a=commit;h=2f43696

[Derek Schuff, 2013-04-17 23:46:08]

we aren't using cmake yet, but I'm still fiddling with trying to build llvm with
clang on mac, and this might help there. Also it might be nice to use ninja for
local development.

[eliben, 2013-04-18 14:50:16]

https://codereview.chromium.org/14328017/diff/1/lib/Target/Mips/CMakeLists.txt
File lib/Target/Mips/CMakeLists.txt (right):

https://codereview.chromium.org/14328017/diff/1/lib/Target/Mips/CMakeLists.tx...
lib/Target/Mips/CMakeLists.txt:33: # LOCALMOD-BEGIN
Meh. I wish we could encapsulate all LOCALMOD NaCl-related code in a single
directory somewhere. Is that feasible?

https://codereview.chromium.org/14328017/diff/1/tools/pnacl-abicheck/LLVMBuil...
File tools/pnacl-abicheck/LLVMBuild.txt (right):

https://codereview.chromium.org/14328017/diff/1/tools/pnacl-abicheck/LLVMBuil...
tools/pnacl-abicheck/LLVMBuild.txt:22: required_libraries = AsmParser BitReader
NaClAnalysis
Isn't there some sort of naming inconsistency here? Sometimes NaClAnalysis,
sometimes naclanalysis

[I'm not well familiar with the semantics of LLVMBuild.txt files]

[Derek Schuff, 2013-04-18 16:15:53]

https://codereview.chromium.org/14328017/diff/1/lib/Target/Mips/CMakeLists.txt
File lib/Target/Mips/CMakeLists.txt (right):

https://codereview.chromium.org/14328017/diff/1/lib/Target/Mips/CMakeLists.tx...
lib/Target/Mips/CMakeLists.txt:33: # LOCALMOD-BEGIN
On 2013/04/18 14:50:16, eliben wrote:
> Meh. I wish we could encapsulate all LOCALMOD NaCl-related code in a single
> directory somewhere. Is that feasible?
I don't know, I don't really know enough about cmake.

https://codereview.chromium.org/14328017/diff/1/tools/pnacl-abicheck/LLVMBuil...
File tools/pnacl-abicheck/LLVMBuild.txt (right):

https://codereview.chromium.org/14328017/diff/1/tools/pnacl-abicheck/LLVMBuil...
tools/pnacl-abicheck/LLVMBuild.txt:22: required_libraries = AsmParser BitReader
NaClAnalysis
On 2013/04/18 14:50:16, eliben wrote:
> Isn't there some sort of naming inconsistency here? Sometimes NaClAnalysis,
> sometimes naclanalysis
> 
> [I'm not well familiar with the semantics of LLVMBuild.txt files]

Yes, and it's intentional. If I understand correctly, there are actually 2
completely different namespaces: One is defined in the name field of
LLVMBuild.txt files and referred to by the required_libraries fields of other
LLVMBuild.txt files. The other is defined by the library_name field and referred
to by the CMakeLists.txt files (sometimes with "LLVM" prepended and sometimes
not ಠ_ಠ). The capitalization is different, to make it clearer which names go
together. Other components (e.g. Transforms/IPO and Ananlysis/IPA) are named
this way as well.

[eliben, 2013-04-18 16:34:57]

On 2013/04/18 16:15:53, Derek Schuff wrote:
> https://codereview.chromium.org/14328017/diff/1/lib/Target/Mips/CMakeLists.txt
> File lib/Target/Mips/CMakeLists.txt (right):
> 
>
https://codereview.chromium.org/14328017/diff/1/lib/Target/Mips/CMakeLists.tx...
> lib/Target/Mips/CMakeLists.txt:33: # LOCALMOD-BEGIN
> On 2013/04/18 14:50:16, eliben wrote:
> > Meh. I wish we could encapsulate all LOCALMOD NaCl-related code in a single
> > directory somewhere. Is that feasible?
> I don't know, I don't really know enough about cmake.
> 
>
https://codereview.chromium.org/14328017/diff/1/tools/pnacl-abicheck/LLVMBuil...
> File tools/pnacl-abicheck/LLVMBuild.txt (right):
> 
>
https://codereview.chromium.org/14328017/diff/1/tools/pnacl-abicheck/LLVMBuil...
> tools/pnacl-abicheck/LLVMBuild.txt:22: required_libraries = AsmParser
BitReader
> NaClAnalysis
> On 2013/04/18 14:50:16, eliben wrote:
> > Isn't there some sort of naming inconsistency here? Sometimes NaClAnalysis,
> > sometimes naclanalysis
> > 
> > [I'm not well familiar with the semantics of LLVMBuild.txt files]
> 
> Yes, and it's intentional. If I understand correctly, there are actually 2
> completely different namespaces: One is defined in the name field of
> LLVMBuild.txt files and referred to by the required_libraries fields of other
> LLVMBuild.txt files. The other is defined by the library_name field and
referred
> to by the CMakeLists.txt files (sometimes with "LLVM" prepended and sometimes
> not ಠ_ಠ). The capitalization is different, to make it clearer which names go
> together. Other components (e.g. Transforms/IPO and Ananlysis/IPA) are named
> this way as well.

According to the docs, library_name is referred to by required_libraries. And
it's optional - if not given the name is derived from the component (maybe this
is why so many components don't define it at all?). As for case, Vectorize
defines both to be the same.

So I'm not saying you should change anything, just lamenting that we're
participating in the mess :-)

LGTM

[Derek Schuff, 2013-04-18 16:51:54]

Committed patchset #2 manually as r2f43696 (presubmit successful).

[jvoung - send to chromium..., 2013-04-22 16:52:50]

On 2013/04/18 16:51:54, Derek Schuff wrote:
> Committed patchset #2 manually as r2f43696 (presubmit successful).

Currently getting from tryjobs of a deps bump
https://codereview.chromium.org/13973008/:

make[2]: *** No rule to make target
`/mnt/data/b/build/slave/nacl-toolchain/build/native_client/pnacl/build/translator-armv7/llvm-sb/Release/lib/libLLVMnaclanalysis.a',
needed by
`/mnt/data/b/build/slave/nacl-toolchain/build/native_client/pnacl/build/translator-armv7/llvm-sb/Release/bin/llc'.
 Stop.