Re: [PATCH v2 2/4] compat/regex: include alloca.h before undef it

From: Danh Doan <congdanhqx@gmail.com>
To: Ramsay Jones <ramsay@ramsayjones.plus.com>
Cc: git@vger.kernel.org
Subject: Re: [PATCH v2 2/4] compat/regex: include alloca.h before undef it
Date: Sun, 26 Apr 2020 07:54:51 +0700	[thread overview]
Message-ID: <20200426005451.f7pyoiijgbk4hpsj@danh.dev> (raw)
In-Reply-To: <ffcaaf4f-a43f-2d20-b70d-dfb8b1d7c687@ramsayjones.plus.com>

On 2020-04-25 21:28:05+0100, Ramsay Jones <ramsay@ramsayjones.plus.com> wrote:
> 
> 
> On 24/04/2020 23:34, Danh Doan wrote:
> [snip]
> 
> > OK, I've tried with my glibc box, it doesn't have that warning.
> > On musl, it warns:
> > 
> > 	$ make compat/regex/regex.sp
> > 	GIT_VERSION = 2.26.2
> > 	    * new build flags
> > 	    SP compat/regex/regex.c
> > 	/usr/include/alloca.h:14:9: warning: preprocessor token alloca redefined
> > 	compat/regex/regex.c:66:9: this was the original definition
> > 	compat/regex/regex_internal.c:925:1: error: symbol 're_string_context_at' redeclared with different type (originally declared at compat/regex/regex_internal.h:433) - different modifiers
> > 
> > 
> 
> OK, I had a quick look at the <alloca.h> header file on a glibc
> system (linux) and new-lib system (cygwin) and they both do
> (more or less) the same thing: first #undef alloca, and then
> if being compiled by gcc, define alloca(size) to be __builtin_alloca(size).

musl people don't do that.
They just go ahead define it, if any other header file requires
alloca, they will include alloca.h

> So, even if <alloca.h> is #included after regex.c:66, it wouldn't
> be a problem. Since I don't have access to a musl based system,
> I don't know what that system header is doing.

musl's alloca.h is available here:

https://git.musl-libc.org/cgit/musl/tree/include/alloca.h

> However, I said *even if* above, because I don't see why it is trying
> to #include <alloca.h> in the first place! ;-)

I looked into my system again. The inclusion chain is:

compat/regex/regex.c:77
`-> compat/regex/regex_internal.h:26
    `-> /usr/include/stdlib.h:138 [*1*]

[*1*]: https://git.musl-libc.org/cgit/musl/tree/include/stdlib.h#n137

I don't know why _GNU_SOURCE and/or _BSD_SOURCE is defined.

>   $ make V=1 NO_REGEX=1 compat/regex/regex.sp
>   cgcc -no-compile -Werror -Wall -Wdeclaration-after-statement -Wformat-security -Wold-style-definition -Woverflow -Wpointer-arith -Wstrict-prototypes -Wunused -Wvla -DENABLE_SHA256 -Wextra -Wmissing-prototypes -Wno-empty-body -Wno-missing-field-initializers -Wno-sign-compare -Wno-unused-parameter  -g -O2 -Wall -I. -DHAVE_SYSINFO -DGIT_HOST_CPU="\"x86_64\"" -DHAVE_ALLOCA_H  -DUSE_CURL_FOR_IMAP_SEND -DSHA1_DC -DSHA1DC_NO_STANDARD_INCLUDES -DSHA1DC_INIT_SAFE_HASH_DEFAULT=0 -DSHA1DC_CUSTOM_INCLUDE_SHA1_C="\"cache.h\"" -DSHA1DC_CUSTOM_INCLUDE_UBC_CHECK_C="\"git-compat-util.h\"" -DSHA256_BLK  -DHAVE_PATHS_H -DHAVE_DEV_TTY -DHAVE_CLOCK_GETTIME -DHAVE_CLOCK_MONOTONIC -DHAVE_GETDELIM '-DPROCFS_EXECUTABLE_PATH="/proc/self/exe"'  -DFREAD_READS_DIRECTORIES -DNO_STRLCPY -Icompat/regex -DSHELL_PATH='"/bin/sh"' -DPAGER_ENV='"LESS=FRX LV=-c"' -DGAWK -DNO_MBSUPPORT \
>   	  compat/regex/regex.c
>   $ 
>   
> ... is not being set on the command-line.

Here's the invocation of cc and cgcc:

	$ make V=1  compat/regex/regex.o
	cc -o compat/regex/regex.o -c -MF compat/regex/.depend/regex.o.d -MQ compat/regex/regex.o -MMD -MP -Werror -Wall -Wdeclaration-after-statement -Wformat-security -Wold-style-definition -Woverflow -Wpointer-arith -Wstrict-prototypes -Wunused -Wvla -DENABLE_SHA256 -Wextra -Wmissing-prototypes -Wno-empty-body -Wno-missing-field-initializers -Wno-sign-compare -Wno-unused-parameter  -g -O2 -Wall -I. -DHAVE_SYSINFO -DGIT_HOST_CPU="\"x86_64\"" -DUSE_LIBPCRE2 -DHAVE_ALLOCA_H -DUSE_CURL_FOR_IMAP_SEND -DICONV_OMITS_BOM -DSHA1_DC -DSHA1DC_NO_STANDARD_INCLUDES -DSHA1DC_INIT_SAFE_HASH_DEFAULT=0 -DSHA1DC_CUSTOM_INCLUDE_SHA1_C="\"cache.h\"" -DSHA1DC_CUSTOM_INCLUDE_UBC_CHECK_C="\"git-compat-util.h\"" -DSHA256_BLK  -DHAVE_PATHS_H -DHAVE_DEV_TTY -DHAVE_CLOCK_GETTIME -DHAVE_CLOCK_MONOTONIC -DHAVE_GETDELIM '-DPROCFS_EXECUTABLE_PATH="/proc/self/exe"' -DFREAD_READS_DIRECTORIES -DNO_STRLCPY -Icompat/regex -DSHELL_PATH='"/bin/sh"' -DPAGER_ENV='"LESS=FRX LV=-c"' -DGAWK -DNO_MBSUPPORT compat/regex/regex.c
	$ make V=1  compat/regex/regex.sp
	cgcc -no-compile -Werror -Wall -Wdeclaration-after-statement -Wformat-security -Wold-style-definition -Woverflow -Wpointer-arith -Wstrict-prototypes -Wunused -Wvla -DENABLE_SHA256 -Wextra -Wmissing-prototypes -Wno-empty-body -Wno-missing-field-initializers -Wno-sign-compare -Wno-unused-parameter  -g -O2 -Wall -I. -DHAVE_SYSINFO -DGIT_HOST_CPU="\"x86_64\"" -DUSE_LIBPCRE2 -DHAVE_ALLOCA_H -DUSE_CURL_FOR_IMAP_SEND -DICONV_OMITS_BOM -DSHA1_DC -DSHA1DC_NO_STANDARD_INCLUDES -DSHA1DC_INIT_SAFE_HASH_DEFAULT=0 -DSHA1DC_CUSTOM_INCLUDE_SHA1_C="\"cache.h\"" -DSHA1DC_CUSTOM_INCLUDE_UBC_CHECK_C="\"git-compat-util.h\"" -DSHA256_BLK  -DHAVE_PATHS_H -DHAVE_DEV_TTY -DHAVE_CLOCK_GETTIME -DHAVE_CLOCK_MONOTONIC -DHAVE_GETDELIM '-DPROCFS_EXECUTABLE_PATH="/proc/self/exe"' -DFREAD_READS_DIRECTORIES -DNO_STRLCPY -Icompat/regex -DSHELL_PATH='"/bin/sh"' -DPAGER_ENV='"LESS=FRX LV=-c"' -DGAWK -DNO_MBSUPPORT \
	          compat/regex/regex.c

> Hmm, do you have this set in config.mak, config.mak.autogen, or some other
> source? puzzled! ;-)

I don't have `config.make.autogen`,
Here is config.mak

	$ cat config.mak
	USE_ASCIIDOCTOR=Yes
	DEVELOPER=1
	DEFAULT_TEST_TARGET=prove
	prefix = /home/danh/.local
	USE_LIBPCRE2=YesPlease
	ICONV_OMITS_BOM=Yes
	NO_REGEX=YesPlease

> BTW, why are you compiling with NO_REGEX set anyway?

Because I use musl-libc, and musl-libc doesn't have StartEnd

-- 
Danh