Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Thu, 12 Jan 2017 13:29:22 +0200
From: Dmitry Golovin <dima@...ovin.in>
To: "musl@...ts.openwall.com" <musl@...ts.openwall.com>
Subject: Re: Clang warning silencing patch

I just figured out the proper test for whether the option is supported (patch attached, it just adds one line).

I understand why you don't want to separate CFLAGS from LDFLAGS. People tend to put flags for compiler, assembler and linker into a single variable which makes life much easier.

All the unused arguments warnings can possibly be silenced by adding `-Qunused-arguments`, but I don't think that this is the right method to be used.

With my patch assembler flags (like --noexecstack) are omitted. Maybe the proper thing to do is to introduce ASFLAGS variable, then parse CFLAGS and put all '-Wl,*' to LDFLAGS and '-Wa,*' to ASFLAGS, removing them from CFLAGS, but it looks like nobody is really doing such thing, so probably just leaving it as it is now and getting some harmless unused arguments warnings is okay.

Regards,
Dmitry

12.01.2017, 06:23, "Rich Felker" <dalias@...c.org>:
> On Tue, Jan 10, 2017 at 12:34:07PM +0200, Dmitry Golovin wrote:
>>  Sorry, I probably have misformatted the patch. Here it is hopefully in correct format. Please tell me if it is wrong.
>>
>>  Regards,
>>  Dmitry
>>
>>  10.01.2017, 11:59, "Dmitry Golovin" <dima@...ovin.in>:
>>  > The attached patch will reduce the amount of warnings produced when using clang compiler.
>>  >
>>  > It does two things:
>>  >
>>  >   1. Tests for excess-precision=standard and rounding-math are
>>  > disabled for clang. The problem is that those tests pass, but do
>>  > no good: every compilation commands produce two warnings about
>>  > unsupported optimization flags.
>
> We attempt to catch issues like that with:
>
> tryflag CFLAGS_TRY -Werror=unknown-warning-option
> tryflag CFLAGS_TRY -Werror=unused-command-line-argument
> tryldflag LDFLAGS_TRY -Werror=unknown-warning-option
> tryldflag LDFLAGS_TRY -Werror=unused-command-line-argument
>
> If it's not working, can you figure out why? Hard-coding them disabled
> for a particular compiler is not reasonable. If they ever are
> supported and necessary to get the correct behavior, we'll have
> hard-coded wrong values, which is a lot worse than a spurious warning.
>
>>  >   2. LDFLAGS (added to --help) variable is only used for linking,
>>  > CFLAGS is only used for compiling, none of them are used for
>>  > assembly. This suppresses all unused argument warnings.
>
> At the expense of wrong behavior; see below:
>
>>  From 2b3f03c46211ad3699e2a72f9054861ba7933d52 Mon Sep 17 00:00:00 2001
>>  From: Dmitry Golovin <dima@...ovin.in>
>>  Date: Tue, 10 Jan 2017 12:26:27 +0200
>>  Subject: [PATCH 3464/3464] improvements for building with clang
>>
>>    1. Tests for excess-precision=standard and rounding-math are disabled
>>       for clang. The problem is that those tests pass, but do no good:
>>       every compilation commands produce two warnings about
>>       unsupported optimization flags.
>>
>>    2. LDFLAGS (added to --help) variable is only used for linking,
>>       CFLAGS is only used for compiling, none of them are used for assembly.
>>       This suppresses all unused argument warnings.
>>  ---
>>   Makefile | 4 ++--
>>   configure | 9 +++++++++
>>   2 files changed, 11 insertions(+), 2 deletions(-)
>>
>>  diff --git a/Makefile b/Makefile
>>  index 8246b78..508878c 100644
>>  --- a/Makefile
>>  +++ b/Makefile
>>  @@ -142,7 +142,7 @@ CC_CMD = $(CC) $(CFLAGS_ALL) -c -o $@ $<
>>   ifeq ($(ADD_CFI),yes)
>>           AS_CMD = LC_ALL=C awk -f $(srcdir)/tools/add-cfi.common.awk -f $(srcdir)/tools/add-cfi.$(ARCH).awk $< | $(CC) $(CFLAGS_ALL) -x assembler -c -o $@ -
>>   else
>>  - AS_CMD = $(CC_CMD)
>>  + AS_CMD = $(CC) -c -o $@ $<
>>   endif
>
> Thi is definitely not acceptable as-is; it drops all of the CFLAGS
> intended to affect the assembler, including critical ones like
> -Wa,--noexecstack.
>
> Ultimately I consider what clang is doing here a bug. The conventional
> compiler driver behavior has always been for the driver to accept all
> options and only apply the ones relevant to what it's currently doing.
> The clang folks have a habit of gratuitously breaking things like this
> (see how their arm assembler is pedantic and violates the official arm
> documentation in regards to which mnemonic forms it accepts for which
> -march settings) and I really don't want to play whack-a-mole with
> them.
>
>>   obj/%.o: $(srcdir)/%.s
>>  @@ -164,7 +164,7 @@ obj/%.lo: $(srcdir)/%.c $(GENH) $(IMPH)
>>           $(CC_CMD)
>>
>>   lib/libc.so: $(LOBJS) $(LDSO_OBJS)
>>  - $(CC) $(CFLAGS_ALL) $(LDFLAGS_ALL) -nostdlib -shared \
>>  + $(CC) $(LDFLAGS_ALL) -nostdlib -shared \
>>           -Wl,-e,_dlstart -o $@ $(LOBJS) $(LDSO_OBJS) $(LIBCC)
>
> This probably doesn't currently break anything, but I can't say for
> sure. In principle there can be CFLAGS that should also be specified
> at link like (like stack protector, sanitizer, etc. stuff) that
> probably don't affect musl, but omitting CFLAGS when linking is not in
> general a good practice.
>
>>   lib/libc.a: $(AOBJS)
>>  diff --git a/configure b/configure
>>  index c2db298..9f01ae5 100755
>>  --- a/configure
>>  +++ b/configure
>>  @@ -39,6 +39,7 @@ Optional features:
>>   Some influential environment variables:
>>     CC C compiler command [detected]
>>     CFLAGS C compiler flags [-Os -pipe ...]
>>  + LDFLAGS linker flags [none]
>>     CROSS_COMPILE prefix for cross compiler and tools [none]
>>     LIBCC compiler runtime library [detected]
>
> This hunk looks ok.
>
> Rich
View attachment "0001-properly-test-for-unsupported-optimization-flags-wit.patch" of type "text/x-diff" (710 bytes)

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.