mmc-utils: Rely on our own _FORTIFY_SOURCE

Upstream always forced _FORTIFY_SOURCE=2. At first, I tried changing their flags to undefine that macro first, but that did not work well. This patch ensures that the package always respects our toolchain settings, even in cases where we are not fortifying the build.

This fixes a build failure where the mmc-utils package failed to build with, e.g., BR2_FORTIFY_SOURCE_1:

<command-line>:0:0: error: "_FORTIFY_SOURCE" redefined [-Werror] <command-line>:0:0: note: this is the location of the previous definition

Signed-off-by Jan Kundr?t <[email protected]>

---
 ...on-the-build-env-for-_FORTIFY_SOURCE.patch | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)
 create mode 100644 

package/mmc-utils/0001-Rely-on-the-build-env-for-_FORTIFY_SOURCE.patch

(quoted diff) +0 -0

a/package/mmc-utils/0001-Rely-on-the-build-env-for-_FORTIFY_SOURCE.patch b/package/mmc-utils/0001-Rely-on-the-build-env-for-_FORTIFY_SOURCE.patch new file mode 100644
index 0000000000..b099efa6c4

package/mmc-utils/0001-Rely-on-the-build-env-for-_FORTIFY_SOURCE.patch new file +17 -0

@@ -0,0 +1,27 @@
+From 0c893e6f272351572548264bf423208a7b76bb16 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Jan=20Kundr=C3=A1t?= <[email protected]>
+Date: Tue, 7 Aug 2018 17:29:35 +0200
+Subject: [PATCH] Rely on the build env for _FORTIFY_SOURCE
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Signed-off-by: Jan Kundr?t <[email protected]>
+---
+ Makefile | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/Makefile b/Makefile
+index aa27ff2..a3890b2 100644
+--- a/Makefile
++++ b/Makefile

Upstream always forced _FORTIFY_SOURCE=2. At first, I tried changing their flags to undefine that macro first, but that did not work well. This patch ensures that the package always respects our toolchain settings, even in cases where we are not fortifying the build.

This fixes a build failure where the mmc-utils package failed to build with, e.g., BR2_FORTIFY_SOURCE_1:

<command-line>:0:0: error: "_FORTIFY_SOURCE" redefined [-Werror] <command-line>:0:0: note: this is the location of the previous definition

Signed-off-by Jan Kundr?t <[email protected]>

Hello,

Quoted 8 lines · Jan Kundr?t

Upstream always forced _FORTIFY_SOURCE=2. At first, I tried changing their flags to undefine that macro first, but that did not work well. This patch ensures that the package always respects our toolchain settings, even in cases where we are not fortifying the build.

This fixes a build failure where the mmc-utils package failed to build with, e.g., BR2_FORTIFY_SOURCE_1:

Quoted 4 lines

<command-line>:0:0: error: "_FORTIFY_SOURCE" redefined [-Werror] <command-line>:0:0: note: this is the location of the previous definition

Signed-off-by Jan Kundr?t <[email protected]>

Is this fixing a build issue reported by the autobuilders ? Or only with hardening options ? Your commit log is not very verbose in details about the configuration/condition under which the build failure happens.

This isn't related to my hardening fixes patchset (ie those patches won't fix this issue). Instead, I'd refactor this patch to move the AM_CFLAGS "-D_FORTIFY_SOURCE=2" into the CFLAGS ?= . That should be an upstream-able change and preserve their intent.

Quoted 4 lines · Matthew Weber

This isn't related to my hardening fixes patchset (ie those patches won't fix this issue). Instead, I'd refactor this patch to move the AM_CFLAGS "-D_FORTIFY_SOURCE=2" into the CFLAGS ?= . That should be an upstream-able change and preserve their intent.

Hi Matthew,
doing just this change, as you suggested:

(quoted diff) +2 -2

-AM_CFLAGS = -D_FILE_OFFSET_BITS=64 -D_FORTIFY_SOURCE=2
-CFLAGS ?= -g -O2
+AM_CFLAGS = -D_FILE_OFFSET_BITS=64
+CFLAGS ?= -g -O2 -D_FORTIFY_SOURCE=2

...stil results in an error when building this manually out-of-box, outside of Buildroot, on a Gentoo Linux with hardened GCC which enforces _FORTIFY_SOURCE via compiler spec files. The good news is that it is now possible to override the CFLAGS via an env var when invoking make. I guess I'll send this upstream, then.

It still doesn't work out-of-box everyhere, but it is no longer a Buildroot bug (and it will work in Buildroot).