Does Clang support `-g1` - clang

I found an old mailing list thread about this here, however, not any further info nor anything in the documentation.
Does Clang support that flag? I tried -g1, -g2, -g3 and -g on a sample but the result was always the same, so it looks like it does not.
Is that planned?

Currently clang does not, but it's not very well documented. Although it accepts the options you've mentioned it treats them all as though they had been -g. From the clang source code, llvm/tools/clang/lib/Driver/Tools.cpp comes this code (on about line 2825 of rev 205900):
// Use the last option from "-g" group. "-gline-tables-only" and "-gdwarf-x"
// are preserved, all other debug options are substituted with "-g".
Args.ClaimAllArgs(options::OPT_g_Group);
if (Arg *A = Args.getLastArg(options::OPT_g_Group)) {
if (A->getOption().matches(options::OPT_gline_tables_only)) {
// FIXME: we should support specifying dwarf version with
// -gline-tables-only.
CmdArgs.push_back("-gline-tables-only");
// Default is dwarf-2 for darwin.
if (getToolChain().getTriple().isOSDarwin())
CmdArgs.push_back("-gdwarf-2");
} else if (A->getOption().matches(options::OPT_gdwarf_2))
CmdArgs.push_back("-gdwarf-2");
else if (A->getOption().matches(options::OPT_gdwarf_3))
CmdArgs.push_back("-gdwarf-3");
else if (A->getOption().matches(options::OPT_gdwarf_4))
CmdArgs.push_back("-gdwarf-4");
else if (!A->getOption().matches(options::OPT_g0) &&
!A->getOption().matches(options::OPT_ggdb0)) {
// Default is dwarf-2 for darwin.
if (getToolChain().getTriple().isOSDarwin())
CmdArgs.push_back("-gdwarf-2");
else
CmdArgs.push_back("-g");
}
}
As you can see by the last few lines, any -g option which hasn't already been eliminated (e.g. -g7) gets transformed into a plain -g within this routine, Clang::ConstructJob().
I don't know of any plans to change this, but I'm also not a clang developer. You might want to ask on the mailing list.
See the online manual for the documented switches, and the source code for Tools.cpp to look at the rest of the option handling in detail.

Related

Flex dropping predefined macro

I have the following flex source:
%{
#if !defined(__linux__) && !defined(__unix__)
/* Maybe on windows */
#endif
int num_chars = 0;
%}
%%
. ++num_chars;
%%
int main()
{
yylex();
printf("%d chars\n", num_chars);
return 0;
}
int yywrap()
{
return 1;
}
I generate a C file by the command flex flextest.l and compile the result with gcc -o fltest lex.yy.c
To my surprise, I get the following output:
flextest.l:2:37: error: operator "defined" requires an identifier
#if !defined(__linux__) && !defined(__unix__)
After further checking, the issue seems to be that flex has actually replaced __unix__ with the empty string, as shown by:
$ grep __linux_ lex.yy.c
#if !defined(__linux__) && !defined()
Why does this happen, and is it possible to avoid it?
It's actually m4 (the macro processor which is used by current versions of flex) which is expanding __unix__ to the empty string. The Gnu implementation of m4 defines certain symbols to empty macros so that they can be tested with ifdef.
Of course, it's (better said, it was) a bug in flex. Flex shouldn't allow m4 to expand macros within user content copied from the scanner definition file, and the current version of flex correctly arranges for the text included from the scanner description file to be quoted so that it will pass through m4 unmodified even if it happens to include a string which could be interpreted by m4 as a macro expansion.
The bug is certainly present in v2.5.39 and v2.6.1 of flex. I didn't test all previous versions, but I suppose it was introduced when flex was modified to use m4, which was v2.5.30 according to the NEWS file.
This particular quoting issue was fixed in v2.6.2 but the current version of flex (2.6.4) contains various other bug fixes, so I'd recommend you upgrade to the latest version.
If you really need a version which could work with both the buggy and the more recent versions of flex, you could use one of the two following hacks:
Find some other way to write __unix__. One possibility is the following
#define C(x,y) x##y
#define UNIX_ C(__un,ix__)
#if !defined(__linux__) && !UNIX_
That hack won't work with defined, since defined(UNIX_) tests whether UNIX_ is defined, not whether what it expands to is defined. But normally built-in symbols like __unix__ are actually defined to be 1, if they are defined, and the #if directive treats any identifier which is not #define'd as though it were 0, which means that you can usually leave use x instead of defined(x). (However, it will produce different results if there were a #define x 0 in effect, so it's not quite a perfect substitute.)
Flex, like many m4 applications, redefines m4's quote marks to be [[ and ]]. Both the buggy flex and the corrected versions substitute these quote marks with a rather elaborate sequence which effectively quotes the quote marks. However, the buggy version does not otherwise quote user-defined text, so macro substitutions will be performed in user text. (As mentioned, this is why __unix__ becomes the empty string.
In flex versions in which user-defined text is not quoted, it is possible to invoke the m4 macro which redefines quote marks. These new quote marks can then be used to quote the #if line, preventing macro substitution of __unix__. However, the quote definition must be restored, or it will completely wreck macro processing of the rest of the file. That's a bit tricky because it is impossible to write [[. (Flex will substitute it with a different string.)
The following seems to do the trick. Note that the macro invocations are placed inside C comments. The changequote macros will expand to an empty string, if they are expanded. But in flex versions since v2.6.2, user-supplied text is quoted, so the changequote macros will not be expanded. Putting them inside comments hides them from the C compiler.
%{
/*m4_changequote(<<,>>)<<*/
#if !defined(__linux__) && !defined(__unix__)
/*>>m4_changequote(<<[>><<[>>,<<]>><<]>>)*/
/* Maybe on windows */
#endif
(The m4 macro which changes quote marks is changequote but flex invokes m4 with the -P flag which changes builtins like changequote to m4_changequote. In the second call to changequote, the two [ which make up the [[ sign are individually quoted with the temporary << quote marks, which hides them from the code in flex which modifies use of [[.)
I don't know how reliable this hack is but it worked on the versions of flex which I had kicking around on my machine, including 2.5.4 (pre-M4) 2.5.39 (buggy), 2.6.1 (buggy), 2.6.2 (somewhat debugged) and 2.6.4 (more debugged).

Is it possible to keep my Nix packages in sync across machines not running NixOS?

I know with NixOS, you can simply copy over the configuration.nix file to sync your OS state including installed packages between machines.
Is it possible then, to do the same using Nix the package manager on a non-NixOS OS to sync only the installed packages?
Please note, that at least since 30.03.2017 (corresponding to 17.03 Nix/NixOS channel/release), as far as I understand the official, modern, supported and suggested solution is to use the so called overlays.
See the chapter titled "Overlays" in the nixpkgs manual for a nice guide on how to use the new approach.
As a short summary: you can put any number of files with .nix extension in $HOME/.config/nixpkgs/overlays/ directory. They will be processed in alphabetical order, and each one can modify the set of available Nix packages. Each of the files must be written as in the following pattern:
self: super:
{
boost = super.boost.override {
python = self.python3;
};
rr = super.callPackage ./pkgs/rr {
stdenv = self.stdenv_32bit;
};
}
The super set corresponds to the "old" set of packages (before the overlay was applied). If you want to refer to the old version of a package (as in boost above), or callPackage, you should reference it via super.
The self set corresponds to the eventual, "future" set of packages, representing the final result after all overlays are applied. (Note: don't be scared when sometimes using them might get rejected by Nix, as it would result in infinite recursion. Probably you should rather just use super in those cases instead.)
Note: with the above changes, the solution I mention below in the original answer seems "deprecated" now — I believe it should still work as of April 2017, but I have no idea for how long. It appears marked as "obsolete" in the nixpkgs repository.
Old answer, before 17.03:
Assuming you want to synchronize apps per-user (as non-NixOS Nix keeps apps visible on per-user basis, not system-wide, as far as I know), it is possible to do it declaratively. It's just not well advertised in the manual — though it seems quite popular among long-time Nixers!
You must create a text file at: $HOME/.nixpkgs/config.nix — e.g.:
$ mkdir -p ~/.nixpkgs
$ $EDITOR ~/.nixpkgs/config.nix
then enter the following contents:
{
packageOverrides = defaultPkgs: with defaultPkgs; {
home = with pkgs; buildEnv {
name = "home";
paths = [
nethack mc pstree #...your favourite pkgs here...
];
};
};
}
Then you should be able to install all listed packages with:
$ nix-env -i home
or:
$ nix-env -iA nixos.home # *much* faster than above
In paths you can put stuff in a similar way like in /etc/nixos/configuration.nix on NixOS. Also, home is actually a "fake package" here. You can add more custom package definitions beside it, and then include them your "paths".
(Side note: I'm hoping to write a blog post with what I learned on how exactly this works, and also showing how to extend it with more customizations. I'll try to remember to link it here if I succeed.)

FAKE Fsc task is writing build products to wrong directory

I'm just learning F#, and setting up a FAKE build harness for a hello-world-like application. (Though the phrase "Hell world" does occasionally come to mind... :-) I'm using a Mac and emacs (generally trying to avoid GUI IDEs by preference).
After a bit of fiddling about with documentation, here's how I'm invoking the F# compiler via FAKE:
let buildDir = #"./build-app/" // Where application build products go
Target "CompileApp" (fun _ -> // Compile application source code
!! #"src/app/**/*.fs" // Look for F# source files
|> Seq.toList // Convert FileIncludes to string list
|> Fsc (fun p -> // which is what the Fsc task wants
{p with //
FscTarget = Exe //
Platform = AnyCpu //
Output = (buildDir + "hello-fsharp.exe") }) // *** Writing to . instead of buildDir?
) //
That uses !! to make a FileIncludes of all the sources in the usual way, then uses Seq.toList to change that to a string list of filenames, which is then handed off to the Fsc task. Simple enough, and it even seems to work:
...
Starting Target: CompileApp (==> SetVersions)
FSC with args:[|"-o"; "./build-app/hello-fsharp.exe"; "--target:exe"; "--platform:anycpu";
"/Users/sgr/Documents/laboratory/hello-fsharp/src/app/hello-fsharp.fs"|]
Finished Target: CompileApp
...
However, despite what the console output above says, the actual build products go to the top-level directory, not the build directory. The message above looks like the -o argument is being passed to the compiler with an appropriate filename, but the executable gets put in . instead of ./build-app/.
So, 2 questions:
Is this a reasonable way to be invoking the F# compiler in a FAKE build harness?
What am I misunderstanding that is causing the build products to go to the wrong place?
This, or a very similar problem, was reported in FAKE issue #521 and seems to have been fixed in FAKE pull request #601, which see.
Explanation of the Problem
As is apparently well-known to everyone but me, the F# compiler as implemented in FSharp.Compiler.Service has a practice of skipping its first argument. See FSharp.Compiler.Service/tests/service/FscTests.fs around line 127, where we see the following nicely informative comment:
// fsc parser skips the first argument by default;
// perhaps this shouldn't happen in library code.
Whether it should or should not happen, it's what does happen. Since the -o came first in the arguments generated by FscHelper, it was dutifully ignored (along with its argument, apparently). Thus the assembly went to the default place, not the place specified.
Solutions
The temporary workaround was to specify --out:destinationFile in the OtherParams field of the FscParams setter in addition to the Output field; the latter is the sacrificial lamb to be ignored while the former gets the job done.
The longer term solution is to fix the arguments generated by FscHelper to have an extra throwaway argument at the front; then these 2 problems will annihilate in a puff of greasy black smoke. (It's kind of balletic in its beauty, when you think about it.) This is exactly what was just merged into the master by #forki23:
// Always prepend "fsc.exe" since fsc compiler skips the first argument
let optsArr = Array.append [|"fsc.exe"|] optsArr
So that solution should be in the newest version of FAKE (3.11.0).
The answers to my 2 questions are thus:
Yes, this appears to be a reasonable way to invoke the F# compiler.
I didn't misunderstand anything; it was just a bug and a fix is in the pipeline.
More to the point: the actual misunderstanding was that I should have checked the FAKE issues and pull requests to see if anybody else had reported this sort of thing, and that's what I'll do next time.

Equivalent of -ftree-vectorizer-verbose for clang

The question is about how to make clang print information on which loops (or other parts of code) have been vectorized. GCC has a command line switch named -ftree-vectorizer-verbose=6 to do this (or -fopt-info-vec in newer versions of GCC), but I couldn't find anything similar for clang. Does clang support this or my only option is to peek in the disassembly ?
clang has following options to print diagnostics related to vectorization:
-Rpass=loop-vectorize identifies loops that were successfully vectorized.
-Rpass-missed=loop-vectorize identifies loops that failed vectorization and indicates if vectorization was specified.
-Rpass-analysis=loop-vectorize identifies the statements that caused vectorization to fail.
Source: http://llvm.org/docs/Vectorizers.html
Looking through the clang source code, there are a couple vectorization passes in Transforms/Vectorize:
BBVectorize
LoopVectorize
SLPVectorize
The last three don't seem to have any arguments that will print things. But in inside BBVectorize there are a couple of options for printing things when clang is built debug:
bb-vectorize-debug-instruction-examination - When debugging is enabled, output information on the instruction-examination process
bb-vectorize-debug-candidate-selection - When debugging is enabled, output information on the candidate-selection process
bb-vectorize-debug-pair-selection - When debugging is enabled, output information on the pair-selection process
bb-vectorize-debug-cycle-check - When debugging is enabled, output information on the cycle-checking process
bb-vectorize-debug-print-after-every-pair -When debugging is enabled, dump the basic block after every pair is fused
That looks like it's about it.

#line and jump to line

Do any editors honer C #line directives with regards to goto line features?
Context:
I'm working on a code generator and need to jump to a line of the output but the line is specified relative to the the #line directives I'm adding.
I can drop them but then finding the input line is even a worse pain
If the editor is scriptable it should be possible to write a script to do the navigation. There might even be a Vim or Emacs script that already does something similar.
FWIW when I writing a lot of Bison/Flexx I wrote a Zeus Lua macro script that attempted to do something similar (i.e. move from input file to the corresponding line of the output file by search for the #line marker).
For any one that might be interested here is that particular macro script.
#line directives are normally inserted by the precompiler, not into source code, so editors won't usually honor that if the file extension is .c.
However, the normal file extension for post-compiled files is .i or .gch, so you might try using that and see what happens.
I've used the following in a header file occasionally to produce clickable items in
the VC6 and recent VS(2003+) compiler ouptut window.
Basically, this exploits the fact that items output in the compiler output
are essentially being parsed for "PATH(LINENUM): message".
This presumes on the Microsoft compiler's treatment of "pragma remind".
This isn't quite exactly what you asked... but it might be generally helpful
in arriving at something you can get the compiler to emit that some editors might honor.
// The following definitions will allow you to insert
// clickable items in the output stream of the Microsoft compiler.
// The error and warning variants will be reported by the
// IDE as actual warnings and errors... which means you can make
// them occur in the task list.
// In theory, the coding standards could be checked to some extent
// in this way and reminders that show up as warnings or even
// errors inserted...
#define strify0(X) #X
#define strify(X) strify0(X)
#define remind(S) message(__FILE__ "(" strify( __LINE__ ) ") : " S)
// example usage
#pragma remind("warning: fake warning")
#pragma remind("error: fake error")
I haven't tried it in a while but it should still work.
Use sed or a similar tool to translate the #lines to something else not interpreted by the compiler, so you get C error messages on the real line, but have a reference to the original input file nearby.

Resources