fix clang warning about missing braces
PATCH 1/1 fix clang warning about missing braces
From: Lev Stipakov <[email protected]...>
A struct with subobjects should be initialized
with double braces.
--- src/openvpn/crypto.c | 2 +- src/openvpn/crypto_mbedtls.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-)
src/openvpn/crypto.c +1 -1
@@ -1857,7 +1857,7 @@ void write_pem_key_file(const char *filename, const char *pem_name) { struct gc_arena gc = gc_new(); -struct key server_key = {0 }; +struct key server_key = {{0 }};struct buffer server_key_buf =clear_buf ();struct buffer server_key_pem =clear_buf ();
src/openvpn/crypto_mbedtls.c +1 -1
@@ -304,7 +304,7 @@ mbedtls_ctr_drbg_context *rand_ctx_get (void) {static mbedtls_entropy_context ec = {0 }; -static mbedtls_ctr_drbg_context cd_ctx = {0 }; +static mbedtls_ctr_drbg_context cd_ctx = {{0 }};static bool rand_initialised = false;if (!rand_initialised)
Signature · 1 line
2.17.1
A message that is not in the archive (sent off-list or not yet received): [email protected]...
Hi
Quoted 9 lines · Steffan Karger
Quoted 2 lines · Lev Stipakov
A struct with subobjects should be initialized
with double braces.This is not true. {0} is a valid initializer for structs in C. Both clang and gcc used to have a bug where they incorrectly warned about this. GCC fixed this a while ago[0]. I thought clang had fixed it too recently, but apparently not?
Pretty much same as my thoughts.
I think the correct fix here is to remove -Werror from travis build. I have tried and failed to lobby for this earlier, but one more try can't hurt, I suppose :)
That said, it seems clang has fixed this some time after clang-7. I don't get this warning anymore after upgrading to clang-9.
Selva
hi,
Quoted 3 lines · Selva Nair
I think the correct fix here is to remove -Werror from travis build. I have tried and failed to lobby for this earlier, but one more try can't hurt, I suppose :)
Can we restrict -Werror to gcc builds, for the time being?
If not, I agree with you, and it needs to go...
Valid (and unambiguous) C statements shouldn't cause warnings, and we shouldn't add patches just because an old version of a certain compiler is silly.
gert
Signature · 6 lines
"If was one thing all people took for granted, was conviction that if you
feed honest figures into a computer, honest figures come out. Never doubted
it myself till I met a computer with a sense of humor."
Robert A. Heinlein, The Moon is a Harsh Mistress
Gert Doering - Munich, Germany [email protected]...PATCH 1/1 travis: bump clang version
From: Lev Stipakov <[email protected]...>
This should fix "suggest braces around initialization of subobject" warning.
--- .travis.yml | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-)
.travis.yml +11 -6
@@ -46,16 +46,16 @@ matrix: - env: SSLLIB="openssl" EXTRA_CONFIG="--enable-iproute2" os: linux compiler: gcc - - env: SSLLIB="openssl" CFLAGS="-fsanitize=address" + - env: SSLLIB="openssl" CFLAGS="-fsanitize=address" CC=clang-9 os: linux compiler: clang - - env: SSLLIB="openssl" OPENSSL_VERSION="1.1.1c" + - env: SSLLIB="openssl" OPENSSL_VERSION="1.1.1c" CC=clang-9 os: linux compiler: clang - env: SSLLIB="mbedtls" os: linux compiler: gcc - - env: SSLLIB="mbedtls" CFLAGS="-fsanitize=address" + - env: SSLLIB="mbedtls" CFLAGS="-fsanitize=address" CC=clang-9 os: linux compiler: clang - env: SSLLIB="openssl"
@@ -73,17 +73,21 @@ matrix: - env: SSLLIB="openssl" CHOST=i686-w64-mingw32 os: linux compiler: ": Win32 build only" - - env: SSLLIB="openssl" EXTRA_CONFIG="--disable-lzo" EXTRA_SCRIPT="make distcheck" + - env: SSLLIB="openssl" EXTRA_CONFIG="--disable-lzo" EXTRA_SCRIPT="make distcheck" CC=clang-9 os: linux compiler: clang - - env: SSLLIB="openssl" EXTRA_CONFIG="--enable-small" + - env: SSLLIB="openssl" EXTRA_CONFIG="--enable-small" CC=clang-9 os: linux compiler: clang addons: apt: + sources: + - ubuntu-toolchain-r-test + - sourceline: 'deb https://apt.llvm.org/xenial/ llvm-toolchain-xenial-9 main' + key_url: 'https://apt.llvm.org/llvm-snapshot.gpg.key' update: true - packages: [ liblzo2-dev, libpam0g-dev, liblz4-dev, linux-libc-dev, man2html, mingw-w64] + packages: [ liblzo2-dev, libpam0g-dev, liblz4-dev, linux-libc-dev, man2html, mingw-w64 ] homebrew: update: true packages: [ lzo ]
@@ -95,6 +99,7 @@ cache: - ${HOME}/Library/Caches/Homebrew install: + - if [ "$TRAVIS_OS_NAME" = "linux" ] && [ "$TRAVIS_CPU_ARCH" != "ppc64le" ]; then sudo apt-get install clang-9; fi - if [ ! -z "${CHOST}" ]; then unset CC; fi - .travis/run-build-deps.sh
Signature · 1 line
2.17.1
[Openvpn-devel] [PATCH applied] Re: travis: bump clang version
This looks reasonable - we want to see warnings-as-errors with recent compilers. Older compilers may warn all they want, but we can not consider it an *error* or spend time on fixing them (in ugly ways).
Your patch has been applied to the master branch.
commit 4a2c8dab5dffff7a09d794d9b2afc4bf710db2b9
Author: Lev Stipakov Date: Fri Nov 29 12:43:35 2019 +0200
travis: bump clang version
Signed-off-by: Lev Stipakov <[email protected]...>
Acked-by: Gert Doering <[email protected]...>
Message-Id: <[email protected]...>
URL: https://www.mail-archive.com/[email protected]/msg19203.html
Signed-off-by: Gert Doering <[email protected]...>--
kind regards,
Gert Doering
Quoted 7 lines · Gert Doering
hi,
Quoted 3 lines · Selva Nair
I think the correct fix here is to remove -Werror from travis build. I have tried and failed to lobby for this earlier, but one more try can't hurt, I suppose :)
Quoted 7 lines
Can we restrict -Werror to gcc builds, for the time being?
If not, I agree with you, and it needs to go...
Valid (and unambiguous) C statements shouldn't cause warnings, and we shouldn't add patches just because an old version of a certain compiler is silly.
With GCC-4.3.8, I see this warning:
crypto.c: In function ‘write_pem_key_file’:
crypto.c:1860:12: warning: missing braces around initializer [-Wmissing-braces]
struct key server_key = { 0 };Perhaps we should just add -Wno-missing-braces? A quick patch is attached, this silences this warning with GCC-4.3.8 at least. That said, I'm not sure this is the best approach; it may hide other missing braces warnings we should see.
Signature · 4 lines
kind regards, David Sommerseth OpenVPN Inc
From 1166081219e8623e32aeea93ddabd2937df36ef8 Mon Sep 17 00:00:00 2001
From: David Sommerseth <[email protected]...>
Date: Fri, 29 Nov 2019 11:43:34 +0100
Subject: [PATCH] build: Add -Wno-missing-braces to CFLAGS
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Some older GCC and clang compilers complains about missing braces when initaling structs.
crypto.c: In function ‘write_pem_key_file’:
crypto.c:1860:12: warning: missing braces around initializer [-Wmissing-braces]
struct key server_key = { 0 };Use of the ACL_CHECK_ADD_COMPILE_FLAGS() macro to check if the compiler support -Wno-missing-braces flag, which ends up in CFLAGS if the compiler accepts it.
--- configure.ac | 1 + 1 file changed, 1 insertion(+)
configure.ac +1 -0
@@ -1289,6 +1289,7 @@ AC_DEFUN([ACL_CHECK_ADD_COMPILE_FLAGS], [ ACL_CHECK_ADD_COMPILE_FLAGS([-Wno-stringop-truncation]) ACL_CHECK_ADD_COMPILE_FLAGS([-Wno-unused-function]) ACL_CHECK_ADD_COMPILE_FLAGS([-Wno-unused-parameter]) +ACL_CHECK_ADD_COMPILE_FLAGS([-Wno-missing-braces]) ACL_CHECK_ADD_COMPILE_FLAGS([-Wall]) if test "${enable_pedantic}" = "yes"; then
Signature · 1 line
2.17.1
Hi,
Quoted 1 line · David Sommerseth
With GCC-4.3.8, I see this warning:
This is about as old as you :-) - do we care about suppressing warnings in old gcc versions that might suppress a *relevant* warning when compiling with a current gcc version?
gert
Signature · 6 lines
"If was one thing all people took for granted, was conviction that if you
feed honest figures into a computer, honest figures come out. Never doubted
it myself till I met a computer with a sense of humor."
Robert A. Heinlein, The Moon is a Harsh Mistress
Gert Doering - Munich, Germany [email protected]...Quoted 8 lines · Gert Doering
Hi,
Quoted 1 line · David Sommerseth
With GCC-4.3.8, I see this warning:
This is about as old as you :-) - do we care about suppressing warnings in old gcc versions that might suppress a *relevant* warning when compiling with a current gcc version?
This is the default compiler on stock RHEL-7 ... which goes EOL August 2024.
Signature · 4 lines
kind regards, David Sommerseth OpenVPN Inc
Hi,
Quoted 9 lines · David Sommerseth
Quoted 6 lines · Gert Doering
Quoted 1 line · David Sommerseth
With GCC-4.3.8, I see this warning:
This is about as old as you :-) - do we care about suppressing warnings in old gcc versions that might suppress a *relevant* warning when compiling with a current gcc version?
This is the default compiler on stock RHEL-7 ... which goes EOL August 2024.
We're not going to break it.
I'm just not going to care very much about the warnings it might throw (and I might object to any patches that are "just for the benefit of suppressing warnings from gcc-4").
gert
Signature · 6 lines
"If was one thing all people took for granted, was conviction that if you
feed honest figures into a computer, honest figures come out. Never doubted
it myself till I met a computer with a sense of humor."
Robert A. Heinlein, The Moon is a Harsh Mistress
Gert Doering - Munich, Germany [email protected]...Quoted 11 lines · Gert Doering
Hi,
Quoted 8 lines · David Sommerseth
Quoted 6 lines · Gert Doering
Quoted 1 line · David Sommerseth
With GCC-4.3.8, I see this warning:
This is about as old as you :-) - do we care about suppressing warnings in old gcc versions that might suppress a *relevant* warning when compiling with a current gcc version?
Quoted 7 lines
Quoted 1 line
This is the default compiler on stock RHEL-7 ... which goes EOL August 2024.
We're not going to break it.
I'm just not going to care very much about the warnings it might throw (and I might object to any patches that are "just for the benefit of suppressing warnings from gcc-4").
Fair enough. But that will actually restrain us from adding -Werror by default for some time forward, unless we want to make package maintenance a bit more tricky - needing to revert a -Werror chage or add -Wno-missing-braces. If we have status quo for the moment, I can live with that until RHEL-7 is EOL.
Signature · 4 lines
kind regards, David Sommerseth OpenVPN Inc
Hi,
Quoted 9 lines · David Sommerseth
Quoted 3 lines
I'm just not going to care very much about the warnings it might throw (and I might object to any patches that are "just for the benefit of suppressing warnings from gcc-4").
Fair enough. But that will actually restrain us from adding -Werror by default for some time forward, unless we want to make package maintenance a bit more tricky - needing to revert a -Werror chage or add -Wno-missing-braces. If we have status quo for the moment, I can live with that until RHEL-7 is EOL.
I don't think we can add -Werror as "on by default" for any platform in the foreseeable future - we'll always hit "old" or "too new" compilers that cause build breakage and lots of extra effort to work around it.
Now, well-defined environments ("travis, clang-9, Linux") I'm all for it :)
gert
Signature · 6 lines
"If was one thing all people took for granted, was conviction that if you
feed honest figures into a computer, honest figures come out. Never doubted
it myself till I met a computer with a sense of humor."
Robert A. Heinlein, The Moon is a Harsh Mistress
Gert Doering - Munich, Germany [email protected]...I upgraded clang to 9.0 and warning disappeared.
https://travis-ci.org/lstipakov/openvpn/builds/618527918
Patch is on the list.
to 28. marrask. 2019 klo 21.44 Gert Doering ([email protected]...) kirjoitti:
Quoted 32 lines
hi,
Quoted 3 lines · Selva Nair
I think the correct fix here is to remove -Werror from travis build. I have tried and failed to lobby for this earlier, but one more try can't hurt, I suppose :)
Can we restrict -Werror to gcc builds, for the time being?
If not, I agree with you, and it needs to go...
Valid (and unambiguous) C statements shouldn't cause warnings, and we shouldn't add patches just because an old version of a certain compiler is silly.
gert
--
"If was one thing all people took for granted, was conviction that if you
feed honest figures into a computer, honest figures come out. Never
doubted
it myself till I met a computer with a sense of humor."
Robert A. Heinlein, The Moon is a Harsh
MistressGert Doering - Munich, Germany
[email protected]...
_______________________________________________
Openvpn-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openvpn-devel
Signature · 1 line
-Lev