Recipes fixes for clang
cover Recipes fixes for clang
This patch series is about common fixes that came to fore due to clang compilation some of the fixes are real bugs which were latent and has been submitted upstream, some were backports and some are yet to be submitted
The following changes since commit 8402958cd2cb87b8283c8ee4e2d08e1a6717d67a:
pseudo_1.7.3.bb: New version of pseudo (2015-09-06 15:24:28 +0100)
are available in the git repository at:
git://github.com/kraj/openembedded-core clang
https://github.com/kraj/openembedded-core/tree/clang
Khem Raj (16):
irda-utils: Weak Define the build TARGETS
slang: Fix namespace conflict vis-a-vis posix_close
sudo: Include sys/types.h for id_t definition
man: Use -DSYSV when builing for linux
minicom: Fix build with musl
bdwgc: Include signal.h instead of hard asm/sigcontext.h
ccache: Dont add LDFLAGS to compiler flags
latencytop: Fix build with clang
alsa-tools: Fix compiling with clang
kexec-tools: Pass -r directly to linker
xorg-driver: Fix logical && with bitwise and operation
puzzles: Fix build with clang
sudo: Disable rsize_t deliberately
webkit-gtk: Backport fix to remove FrameDestructionObserver
mdadm: Fix bugs exposed by clang
directfb: Avoid using VLAs and printf formats
.../irda-utils/irda-utils_0.9.18.bb | 11 ++- meta/recipes-devtools/ccache/ccache.inc | 2 - meta/recipes-extended/man/man_1.6g.bb | 2 + .../mdadm/files/0001-Fix-typo-in-comparision.patch | 86 +++++++++++++++++ meta/recipes-extended/mdadm/mdadm_3.3.4.bb | 1 + ...0001-Fix-build-issus-surfaced-due-to-musl.patch | 47 +++++++++ meta/recipes-extended/minicom/minicom_2.7.bb | 1 + ...x-error-conflicting-types-for-posix_close.patch | 39 ++++++++ meta/recipes-extended/slang/slang_2.2.4.bb | 1 + ...1-Include-sys-types.h-for-id_t-definition.patch | 34 +++++++ meta/recipes-extended/sudo/sudo_1.8.14p3.bb | 2 + meta/recipes-graphics/directfb/directfb.inc | 4 +- ...direct-Aboid-usng-VLAs-and-printf-formats.patch | 61 ++++++++++++ ...put_dpms-Replace-logical-with-bitwise-ope.patch | 31 ++++++ .../xorg-driver/xf86-video-omap_0.4.3.bb | 2 + .../0001-purgatory-Pass-r-directly-to-linker.patch | 32 +++++++ meta/recipes-kernel/kexec/kexec-tools_2.0.10.bb | 1 + ...-function-signatures-to-fix-prototype-mis.patch | 64 +++++++++++++ meta/recipes-kernel/latencytop/latencytop_0.5.bb | 6 +- ...s10k1-Make-output_tram_line-static-inline.patch | 31 ++++++ meta/recipes-multimedia/alsa/alsa-tools_1.0.29.bb | 1 + .../files/0001-Use-labs-instead-of-abs.patch | 46 +++++++++ meta/recipes-sato/puzzles/puzzles_git.bb | 1 + ...ride-of-FrameDestructionObserver-willDeta.patch | 105 +++++++++++++++++++++ meta/recipes-sato/webkit/webkit-gtk_1.8.3.bb | 1 + .../bdwgc/bdwgc/0001-fix-build-with-musl.patch | 35 +++++++ meta/recipes-support/bdwgc/bdwgc_7.4.2.bb | 4 +- 27 files changed, 641 insertions(+), 10 deletions(-) create mode 100644 meta/recipes-extended/mdadm/files/0001-Fix-typo-in-comparision.patch create mode 100644 meta/recipes-extended/minicom/minicom/0001-Fix-build-issus-surfaced-due-to-musl.patch create mode 100644 meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch create mode 100644 meta/recipes-extended/sudo/sudo/0001-Include-sys-types.h-for-id_t-definition.patch create mode 100644 meta/recipes-graphics/directfb/directfb/0001-gfx-direct-Aboid-usng-VLAs-and-printf-formats.patch create mode 100644 meta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch create mode 100644 meta/recipes-kernel/kexec/kexec-tools/0001-purgatory-Pass-r-directly-to-linker.patch create mode 100644 meta/recipes-kernel/latencytop/latencytop-0.5/0001-Rectify-the-function-signatures-to-fix-prototype-mis.patch create mode 100644 meta/recipes-multimedia/alsa/alsa-tools/0001-as10k1-Make-output_tram_line-static-inline.patch create mode 100644 meta/recipes-sato/puzzles/files/0001-Use-labs-instead-of-abs.patch create mode 100644 meta/recipes-sato/webkit/webkit-gtk-1.8.3/0001-Remove-override-of-FrameDestructionObserver-willDeta.patch create mode 100644 meta/recipes-support/bdwgc/bdwgc/0001-fix-build-with-musl.patch
Signature · 1 line
2.5.1
PATCH 1/16 irda-utils: Weak Define the build TARGETS
This will help one to disable the targets via bbappends if needed e.g. musl can not compile irdaping since it uses includes net/if_packet.h, which (on GLIBC) only defines struct sockaddr_pkt but not in other libc e.g. musl that makes irdaping specific to glibc
--- meta/recipes-connectivity/irda-utils/irda-utils_0.9.18.bb | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
meta/recipes-connectivity/irda-utils/irda-utils_0.9.18.bb +7 -4
@@ -32,15 +32,18 @@ EXTRA_OEMAKE = "\ INITSCRIPT_NAME = "irattach" INITSCRIPT_PARAMS = "defaults 20" +TARGETS ??= "irattach irdaping" do_compile () { - oe_runmake -C irattach - oe_runmake -C irdaping + for t in ${TARGETS}; do + oe_runmake -C $t + done } do_install () { install -d ${D}${sbindir} - oe_runmake -C irattach ROOT="${D}" install - oe_runmake -C irdaping ROOT="${D}" install + for t in ${TARGETS}; do + oe_runmake -C $t ROOT="${D}" install + done install -d ${D}${sysconfdir}/init.d install -m 0755 ${WORKDIR}/init ${D}${sysconfdir}/init.d/${INITSCRIPT_NAME}
Signature · 1 line
2.5.1
PATCH 2/16 slang: Fix namespace conflict vis-a-vis posix_close
Fixes errors like
| /home/ubuntu/work/oe/openembedded-core/build/tmp-musl/work/core2-64-oe-linux-musl/slang/2.2.4-r12/slang-2.2.4/src/slpo sio.c:366:12: error: conflicting types for 'posix_close'
| static int posix_close (SLFile_FD_Type *f) | ^ | In file included from
/home/ubuntu/work/oe/openembedded-core/build/tmp-musl/work/core2-64-oe-linux-musl/slang/2.2.4-r1 2/slang-2.2.4/src/slinclud.h:20:0,
| from/home/ubuntu/work/oe/openembedded-core/build/tmp-musl/work/core2-64-oe-linux-musl/slang/2.2.4-r1 2/slang-2.2.4/src/slposio.c:24:
| /home/ubuntu/work/oe/openembedded-core/build/tmp-musl/sysroots/qemux86-64/usr/include/unistd.h:38:5:
note: previous de
claration of 'posix_close' was here
| int posix_close(int, int); | ^ | make[1]: ***
[/home/ubuntu/work/oe/openembedded-core/build/tmp-musl/work/core2-64-oe-linux-musl/slang/2.2.4-r12/slang- 2.2.4/src/elfobjs/slposio.o] Error 1
--- ...x-error-conflicting-types-for-posix_close.patch | 39 ++++++++++++++++++++++ meta/recipes-extended/slang/slang_2.2.4.bb | 1 + 2 files changed, 40 insertions(+) create mode 100644 meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch
meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch new file +39 -0
@@ -0,0 +1,39 @@ +From 96eb7e29822151823a66a1eb59f1fa4aead5ae08 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 31 Aug 2015 06:33:21 +0000 +Subject: [PATCH] Fix error: conflicting types for 'posix_close' + +Exposed while compiling on musl + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + src/slposio.c | 4 ++-- + 1 file changed, 2 insertions(+), 2 deletions(-) + +diff --git a/src/slposio.c b/src/slposio.c +index 9ce9697..9ff9bfe 100644 +--- a/src/slposio.c ++++ b/src/slposio.c +@@ -363,7 +363,7 @@ static int posix_close_fd (int *fd) + return 0; + } + +-static int posix_close (SLFile_FD_Type *f) ++static int posix_closex (SLFile_FD_Type *f) + { + int status = do_close (f); + +@@ -1001,7 +1001,7 @@ static SLang_Intrin_Fun_Type Fd_Name_Table [] = + MAKE_INTRINSIC_2("write", posix_write, V, F, B), + MAKE_INTRINSIC_1("dup_fd", posix_dup, V, F), + MAKE_INTRINSIC_2("dup2_fd", posix_dup2, I, F, I), +- MAKE_INTRINSIC_1("close", posix_close, I, F), ++ MAKE_INTRINSIC_1("close", posix_closex, I, F), + MAKE_INTRINSIC_1("_close", posix_close_fd, I, I), + #if defined(TTYNAME_R) + MAKE_INTRINSIC_0("ttyname", posix_ttyname, V), +-- +2.5.1 +
meta/recipes-extended/slang/slang_2.2.4.bb +1 -0
@@ -22,6 +22,7 @@ SRC_URI = "ftp://space.mit.edu/pub/davis/slang/v2.2/slang-${PV}.tar.bz2 \ file://change-char-type-to-signed-char-in-macros.patch \ file://sprintf-bug-concerning-8-bit-characters.patch \ file://slang-fix-the-iconv-existence-checking.patch \ + file://0001-Fix-error-conflicting-types-for-posix_close.patch \ " inherit autotools-brokensep
Signature · 1 line
2.5.1
PATCH 3/16 sudo: Include sys/types.h for id_t definition
In reply to Khem RajThis is exposed by musl, on glibc sys/types.h comes as indirect include from other include myriad.
--- ...1-Include-sys-types.h-for-id_t-definition.patch | 34 ++++++++++++++++++++++ meta/recipes-extended/sudo/sudo_1.8.14p3.bb | 1 + 2 files changed, 35 insertions(+) create mode 100644 meta/recipes-extended/sudo/sudo/0001-Include-sys-types.h-for-id_t-definition.patch
meta/recipes-extended/sudo/sudo/0001-Include-sys-types.h-for-id_t-definition.patch new file +34 -0
@@ -0,0 +1,34 @@ +From 386e2c2fa2ab2e02ef71c268a57205139be329ab Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 31 Aug 2015 07:07:49 +0000 +Subject: [PATCH] Include sys/types.h for id_t definition + +/sudo_util.h:219:14: error: unknown type name 'id_t' + __dso_public id_t sudo_strtoid_v1(const char *str, const char *sep, + char **endp, const char **errstr); + ^ + make[1]: *** [preserve_fds.o] Error 1 + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + include/sudo_util.h | 2 ++ + 1 file changed, 2 insertions(+) + +diff --git a/include/sudo_util.h b/include/sudo_util.h +index 89c9f89..ac0855a 100644 +--- a/include/sudo_util.h ++++ b/include/sudo_util.h +@@ -17,6 +17,8 @@ + #ifndef SUDO_UTIL_H + #define SUDO_UTIL_H + ++#include <sys/types.h> ++ + #ifdef HAVE_STDBOOL_H + # include <stdbool.h> + #else +-- +2.5.1 +
meta/recipes-extended/sudo/sudo_1.8.14p3.bb +1 -0
@@ -3,6 +3,7 @@ require sudo.inc SRC_URI = "http://ftp.sudo.ws/sudo/dist/sudo-${PV}.tar.gz \ ${@bb.utils.contains('DISTRO_FEATURES', 'pam', '${PAM_SRC_URI}', '', d)} \ file://0001-Use-correct-path-to-init.d-and-tmpfiles.d-files.patch \ + file://0001-Include-sys-types.h-for-id_t-definition.patch \ " PAM_SRC_URI = "file://sudo.pam"
Signature · 1 line
2.5.1
PATCH 4/16 man: Use -DSYSV when builing for linux
In reply to Khem RajWe needed this define but were getting away since glibc indirectly included fcntl.h but man sources clearly guard inclusion of fcntl.h with SYSV, this is uncovered with musl
--- meta/recipes-extended/man/man_1.6g.bb | 2 ++ 1 file changed, 2 insertions(+)
meta/recipes-extended/man/man_1.6g.bb +2 -0
@@ -58,6 +58,8 @@ SRC_URI = "http://pkgs.fedoraproject.org/lookaside/pkgs/man2html/${BP}.tar.gz/ba SRC_URI[md5sum] = "ba154d5796928b841c9c69f0ae376660" SRC_URI[sha256sum] = "ccdcb8c3f4e0080923d7e818f0e4a202db26c46415eaef361387c20995b8959f" +CFLAGS += "-DSYSV" + do_configure () { ${S}/configure -default -confdir /etc +sgid +fhs +lang all }
Signature · 1 line
2.5.1
PATCH 5/16 minicom: Fix build with musl
In reply to Khem Rajmusl exposes the inherent assumption about certain header files from glibc
--- ...0001-Fix-build-issus-surfaced-due-to-musl.patch | 47 ++++++++++++++++++++++ meta/recipes-extended/minicom/minicom_2.7.bb | 1 + 2 files changed, 48 insertions(+) create mode 100644 meta/recipes-extended/minicom/minicom/0001-Fix-build-issus-surfaced-due-to-musl.patch
meta/recipes-extended/minicom/minicom/0001-Fix-build-issus-surfaced-due-to-musl.patch new file +47 -0
@@ -0,0 +1,47 @@ +From d62a5862e26ed3fc58d789efe9c40ca6c911d36b Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 31 Aug 2015 22:35:31 +0000 +Subject: [PATCH] Fix build issus surfaced due to musl + +src/getsdir.h:28:14: error: 'MAXNAMLEN' undeclared here (not in a function) + char fname[MAXNAMLEN + 1]; /* filename + terminating null */ + +src/dial.c:352:22: error: 'KIOCSOUND' undeclared (first use in this function) +| ioctl(consolefd, KIOCSOUND, k); + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + src/dial.c | 2 +- + src/getsdir.c | 1 + + 2 files changed, 2 insertions(+), 1 deletion(-) + +diff --git a/src/dial.c b/src/dial.c +index a90c1d2..bf02574 100644 +--- a/src/dial.c ++++ b/src/dial.c +@@ -39,7 +39,7 @@ + #include "intl.h" + + #ifdef VC_MUSIC +-# if defined(__GLIBC__) ++# if defined(__GLIBC__) || defined(__linux__) + # include <sys/ioctl.h> + # include <sys/kd.h> + # include <sys/time.h> +diff --git a/src/getsdir.c b/src/getsdir.c +index 2195b27..b61a361 100644 +--- a/src/getsdir.c ++++ b/src/getsdir.c +@@ -30,6 +30,7 @@ + #include <string.h> + #include <sys/types.h> + #include <sys/stat.h> ++#include <sys/param.h> + #include <errno.h> + + #include "getsdir.h" +-- +2.5.1 +
meta/recipes-extended/minicom/minicom_2.7.bb +1 -0
@@ -9,6 +9,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=420477abc567404debca0a2a1cb6b645 \ SRC_URI = "https://alioth.debian.org/frs/download.php/latestfile/3/${BP}.tar.gz \ file://allow.to.disable.lockdev.patch \ file://0001-fix-minicom-h-v-return-value-is-not-0.patch \ + file://0001-Fix-build-issus-surfaced-due-to-musl.patch \ " SRC_URI[md5sum] = "7044ca3e291268c33294f171d426dc2d"
Signature · 1 line
2.5.1
PATCH 6/16 bdwgc: Include signal.h instead of hard asm/sigcontext.h
In reply to Khem RajAll cases are about glibc and for non-glibc systems it falls back to last else choice which still is glibc's older version, ideally it should have a case where libc != glibc
--- .../bdwgc/bdwgc/0001-fix-build-with-musl.patch | 35 ++++++++++++++++++++++ meta/recipes-support/bdwgc/bdwgc_7.4.2.bb | 4 ++- 2 files changed, 38 insertions(+), 1 deletion(-) create mode 100644 meta/recipes-support/bdwgc/bdwgc/0001-fix-build-with-musl.patch
meta/recipes-support/bdwgc/bdwgc/0001-fix-build-with-musl.patch new file +35 -0
@@ -0,0 +1,35 @@ +From 829a89e7f5785239a9384e3afd46c900f9d76ada Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 31 Aug 2015 19:55:32 +0000 +Subject: [PATCH] fix build with musl + +signal.h already includes bits/sigcontext.h +Documentation seems to indicate that if you need the definitions in +sigcontext.h, you are supposed to include signal.h and not sigcontext.h. + +a good fix should be to remove glibc +specific understanding, in those ifdef's + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + os_dep.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/os_dep.c b/os_dep.c +index 08c501d..5cb7873 100644 +--- a/os_dep.c ++++ b/os_dep.c +@@ -41,7 +41,7 @@ + # else /* __GLIBC__ < 2 */ + /* libc5 doesn't have <sigcontext.h>: go directly with the kernel */ + /* one. Check LINUX_VERSION_CODE to see which we should reference. */ +-# include <asm/sigcontext.h> ++# include <signal.h> + # endif /* __GLIBC__ < 2 */ + # endif + #endif /* LINUX && !POWERPC */ +-- +2.5.1 +
meta/recipes-support/bdwgc/bdwgc_7.4.2.bb +3 -1
@@ -21,7 +21,9 @@ SECTION = "devel" LICENSE = "MIT" LIC_FILES_CHKSUM = "file://README.QUICK;md5=55f5088f90a982fed7af9a4897047ef7" -SRC_URI = "http://www.hboehm.info/gc/gc_source/gc-${PV}.tar.gz" +SRC_URI = "http://www.hboehm.info/gc/gc_source/gc-${PV}.tar.gz \ + file://0001-fix-build-with-musl.patch \ + " SRC_URI[md5sum] = "12c05fd2811d989341d8c6d81f66af87" SRC_URI[sha256sum] = "63320ad7c45460e4a40e03f5aa4c6893783f21a16416c3282b994f933312afa2"
Signature · 1 line
2.5.1
PATCH 7/16 ccache: Dont add LDFLAGS to compiler flags
In reply to Khem RajThis is no longer required, its been carried over for a long time. As a side effect it helps compiling ccache with clang
--- meta/recipes-devtools/ccache/ccache.inc | 2 -- 1 file changed, 2 deletions(-)
meta/recipes-devtools/ccache/ccache.inc +0 -2
@@ -16,5 +16,3 @@ S = "${WORKDIR}/git" inherit autotools BBCLASSEXTEND = "native" - -TARGET_CC_ARCH += "${LDFLAGS}"
Signature · 1 line
2.5.1
PATCH 8/16 latencytop: Fix build with clang
In reply to Khem RajClang is stricter with syntax checking and flags the functions return values if they dont match with function signatures, Fixed thusly
--- ...-function-signatures-to-fix-prototype-mis.patch | 64 ++++++++++++++++++++++ meta/recipes-kernel/latencytop/latencytop_0.5.bb | 6 +- 2 files changed, 68 insertions(+), 2 deletions(-) create mode 100644 meta/recipes-kernel/latencytop/latencytop-0.5/0001-Rectify-the-function-signatures-to-fix-prototype-mis.patch
meta/recipes-kernel/latencytop/latencytop-0.5/0001-Rectify-the-function-signatures-to-fix-prototype-mis.patch new file +64 -0
@@ -0,0 +1,64 @@ +From db112739dc4f608a968b8104b382955dc3d96ca3 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 07:40:10 +0000 +Subject: [PATCH] Rectify the function signatures to fix prototype mismatches + +clang is less forgiving when it comes to coding standards, correct the +function signatures to reflect the function logic + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + fsync.c | 7 ++++--- + latencytop.h | 2 +- + 2 files changed, 5 insertions(+), 4 deletions(-) + +diff --git a/fsync.c b/fsync.c +index 82dff0e..5cefba9 100644 +--- a/fsync.c ++++ b/fsync.c +@@ -51,7 +51,7 @@ struct fsync_files { + static GList *fsync_data; + + +-static chain_file(struct fsync_process *proc, char *filename) ++static void chain_file(struct fsync_process *proc, char *filename) + { + struct fsync_files *file; + GList *item; +@@ -75,7 +75,7 @@ static chain_file(struct fsync_process *proc, char *filename) + proc->files = g_list_append(proc->files, file); + } + +-static report_file(char *process, char *file) ++static void report_file(char *process, char *file) + { + struct fsync_process *proc; + GList *item; +@@ -157,9 +157,10 @@ int enable_fsync_tracer(void) + write_to_file("/sys/kernel/debug/tracing/current_tracer", "fsync"); + write_to_file("/sys/kernel/debug/tracing/iter_ctrl", "ftrace_printk"); + write_to_file("/sys/kernel/debug/tracing/tracing_on", "1"); ++ return ret; + } + +-int disable_fsync_tracer(void) ++void disable_fsync_tracer(void) + { + write_to_file("/sys/kernel/debug/tracing/tracing_on", "0"); + } +diff --git a/latencytop.h b/latencytop.h +index 5394d73..9d107a8 100644 +--- a/latencytop.h ++++ b/latencytop.h +@@ -54,5 +54,5 @@ extern char *translate(char *line); + extern void init_translations(char *filename); + extern int fsync_display(int duration); + extern int enable_fsync_tracer(void); +-extern int disable_fsync_tracer(void); ++extern void disable_fsync_tracer(void); + extern void update_list(void); +-- +2.5.1 +
meta/recipes-kernel/latencytop/latencytop_0.5.bb +4 -2
@@ -10,7 +10,9 @@ PR = "r3" SRC_URI = "http://pkgs.fedoraproject.org/repo/pkgs/${BPN}/${BP}.tar.gz/73bb3371c6ee0b0e68e25289027e865c/${BP}.tar.gz \ file://latencytop-makefile.patch \ - file://latencytop-fsync.patch" + file://latencytop-fsync.patch \ + file://0001-Rectify-the-function-signatures-to-fix-prototype-mis.patch \ +" SRC_URI[md5sum] = "73bb3371c6ee0b0e68e25289027e865c" SRC_URI[sha256sum] = "9e7f72fbea7bd918e71212a1eabaad8488d2c602205d2e3c95d62cd57e9203ef"
@@ -21,7 +23,7 @@ PACKAGECONFIG[x11] = ",,gtk+" EXTRA_OEMAKE_X = "${@bb.utils.contains('PACKAGECONFIG', 'x11', 'HAS_GTK_GUI=1', '', d)}" -CFLAGS += "${LDFLAGS}" +#CFLAGS += "${LDFLAGS}" do_install() { oe_runmake install DESTDIR=${D} ${EXTRA_OEMAKE_X}
Signature · 1 line
2.5.1
PATCH 9/16 alsa-tools: Fix compiling with clang
In reply to Khem RajFixed a linking error occurring when building with clang, where a function reference is emitted but function is not since its marked as inline, making it static inline gives better opportunity to compiler to compile it
--- ...s10k1-Make-output_tram_line-static-inline.patch | 31 ++++++++++++++++++++++ meta/recipes-multimedia/alsa/alsa-tools_1.0.29.bb | 1 + 2 files changed, 32 insertions(+) create mode 100644 meta/recipes-multimedia/alsa/alsa-tools/0001-as10k1-Make-output_tram_line-static-inline.patch
meta/recipes-multimedia/alsa/alsa-tools/0001-as10k1-Make-output_tram_line-static-inline.patch new file +31 -0
@@ -0,0 +1,31 @@ +From 1e5926df74a35fe2cd90bc59f5264a8715c94048 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 08:16:59 +0000 +Subject: [PATCH] as10k1: Make output_tram_line() static inline + +The function is not used in any other file, its better to make it static +inline so compiler has a better chance of optimizing here + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + as10k1/as10k1.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/as10k1/as10k1.c b/as10k1/as10k1.c +index 96af0c3..d0766fd 100644 +--- a/as10k1/as10k1.c ++++ b/as10k1/as10k1.c +@@ -366,7 +366,7 @@ void as_exit(const char *message) + exit(1); + } + +-inline void output_tram_line( struct list_head *line_head, int type){ ++static inline void output_tram_line( struct list_head *line_head, int type){ + + struct tram *tram_sym; + struct list_head *entry; +-- +2.5.1 +
meta/recipes-multimedia/alsa/alsa-tools_1.0.29.bb +1 -0
@@ -13,6 +13,7 @@ SRC_URI = "ftp://ftp.alsa-project.org/pub/tools/${BP}.tar.bz2 \ ${@bb.utils.contains('DISTRO_FEATURES', 'x11', '', \ 'file://makefile_no_gtk.patch', d)} \ file://gitcompile_hdajacksensetest \ + file://0001-as10k1-Make-output_tram_line-static-inline.patch \ " SRC_URI[md5sum] = "f339a3cd24f748c9d007bdff0e98775b"
Signature · 1 line
2.5.1
PATCH 10/16 kexec-tools: Pass -r directly to linker
In reply to Khem Raj-r is understood by gcc but not by clang, anyway its a linker option so pass it straight to linker by using -Wl explicitly
--- .../0001-purgatory-Pass-r-directly-to-linker.patch | 32 ++++++++++++++++++++++ meta/recipes-kernel/kexec/kexec-tools_2.0.10.bb | 1 + 2 files changed, 33 insertions(+) create mode 100644 meta/recipes-kernel/kexec/kexec-tools/0001-purgatory-Pass-r-directly-to-linker.patch
meta/recipes-kernel/kexec/kexec-tools/0001-purgatory-Pass-r-directly-to-linker.patch new file +32 -0
@@ -0,0 +1,32 @@ +From a1135b3170963ba956f2364c1283864c35541295 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 07:59:45 +0000 +Subject: [PATCH] purgatory: Pass -r directly to linker + +This helps compiling with clang since -r is not a known option for clang +where as gcc knows how to deal with it and passes it down to linker +unfiltered + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + purgatory/Makefile | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/purgatory/Makefile b/purgatory/Makefile +index 2b5c061..b251353 100644 +--- a/purgatory/Makefile ++++ b/purgatory/Makefile +@@ -61,7 +61,7 @@ $(PURGATORY): CPPFLAGS=$($(ARCH)_PURGATORY_EXTRA_CFLAGS) \ + -I$(shell $(CC) -print-file-name=include) + $(PURGATORY): LDFLAGS=$($(ARCH)_PURGATORY_EXTRA_CFLAGS)\ + -Wl,--no-undefined -nostartfiles -nostdlib \ +- -nodefaultlibs -e purgatory_start -r \ ++ -nodefaultlibs -e purgatory_start -Wl,-r \ + -Wl,-Map=$(PURGATORY_MAP) + + $(PURGATORY): $(PURGATORY_OBJS) +-- +2.5.1 +
meta/recipes-kernel/kexec/kexec-tools_2.0.10.bb +1 -0
@@ -6,6 +6,7 @@ SRC_URI += "file://kexec-tools-Refine-kdump-device_tree-sort.patch \ file://kexec-aarch64.patch \ file://kexec-x32.patch \ file://0002-powerpc-change-the-memory-size-limit.patch \ + file://0001-purgatory-Pass-r-directly-to-linker.patch \ " SRC_URI[md5sum] = "6cb4d22bcec71b6e070aa8e9d990a5e6"
Signature · 1 line
2.5.1
PATCH 11/16 xorg-driver: Fix logical && with bitwise and operation
In reply to Khem RajThis must have been a typo which is a bug infact we should have used bitwise & in first place.
--- ...put_dpms-Replace-logical-with-bitwise-ope.patch | 31 ++++++++++++++++++++++ .../xorg-driver/xf86-video-omap_0.4.3.bb | 2 ++ 2 files changed, 33 insertions(+) create mode 100644 meta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch
meta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch new file +31 -0
@@ -0,0 +1,31 @@ +From fc653ca3b36435aea281ae409fb90f85e1bc0d81 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 08:44:13 +0000 +Subject: [freedreno][PATCH] drmmode_output_dpms: Replace logical && with bitwise & + operation + +This is a mask calculation so bitbwise & should have been used + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Submitted + + src/drmmode_display.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/src/drmmode_display.c b/src/drmmode_display.c +index c3cf4f8..eb54b59 100644 +--- a/src/drmmode_display.c ++++ b/src/drmmode_display.c +@@ -777,7 +777,7 @@ drmmode_output_dpms(xf86OutputPtr output, int mode) + + for (i = 0; i < koutput->count_props; i++) { + props = drmModeGetProperty(drmmode->fd, koutput->props[i]); +- if (props && (props->flags && DRM_MODE_PROP_ENUM)) { ++ if (props && (props->flags & DRM_MODE_PROP_ENUM)) { + if (!strcmp(props->name, "DPMS")) { + mode_id = koutput->props[i]; + drmModeFreeProperty(props); +-- +2.5.1 +
meta/recipes-graphics/xorg-driver/xf86-video-omap_0.4.3.bb +2 -0
@@ -24,6 +24,8 @@ LICENSE = "GPLv2+" LIC_FILES_CHKSUM = "file://COPYING;md5=10ce5de3b111315ea652a5f74ec0c602" DEPENDS += "virtual/libx11 libdrm xf86driproto" +SRC_URI += "file://0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch" + SRC_URI[md5sum] = "be35daf6fa4b75092cc4a8978c437bc5" SRC_URI[sha256sum] = "db1e0e69fd4c4c8fdca5ef2cb0447bccd7518a718495876a6904bef57b39986d"
Signature · 1 line
2.5.1
PATCH 12/16 puzzles: Fix build with clang
In reply to Khem RajClang uncovered an error where abs() is used on long types, we shoud be using labs()
--- .../files/0001-Use-labs-instead-of-abs.patch | 46 ++++++++++++++++++++++ meta/recipes-sato/puzzles/puzzles_git.bb | 1 + 2 files changed, 47 insertions(+) create mode 100644 meta/recipes-sato/puzzles/files/0001-Use-labs-instead-of-abs.patch
meta/recipes-sato/puzzles/files/0001-Use-labs-instead-of-abs.patch new file +46 -0
@@ -0,0 +1,46 @@ +From 9808dc09e08937c9bffd858d3ded428225a4312a Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 09:02:38 +0000 +Subject: [PATCH] Use labs instead of abs + +cluevals is array of longs so we have to use labs() instead of abs() +since abs() returns int + +Fixes + +keen.c:1458:17: error: absolute value function 'abs' given an argument +of type 'long' but has parameter of type 'int' which may cause +truncation of value + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + keen.c | 4 ++-- + 1 file changed, 2 insertions(+), 2 deletions(-) + +diff --git a/keen.c b/keen.c +index a6a7238..1631992 100644 +--- a/keen.c ++++ b/keen.c +@@ -1043,7 +1043,7 @@ done + cluevals[j] *= grid[i]; + break; + case C_SUB: +- cluevals[j] = abs(cluevals[j] - grid[i]); ++ cluevals[j] = labs(cluevals[j] - grid[i]); + break; + case C_DIV: + { +@@ -1455,7 +1455,7 @@ static int check_errors(const game_state *state, long *errors) + cluevals[j] *= state->grid[i]; + break; + case C_SUB: +- cluevals[j] = abs(cluevals[j] - state->grid[i]); ++ cluevals[j] = labs(cluevals[j] - state->grid[i]); + break; + case C_DIV: + { +-- +2.5.1 +
meta/recipes-sato/puzzles/puzzles_git.bb +1 -0
@@ -11,6 +11,7 @@ LIC_FILES_CHKSUM = "file://LICENCE;md5=33bcd4bce8f3c197f2aefbdbd2d299bc" SRC_URI = "git://git.tartarus.org/simon/puzzles.git \ file://fix-compiling-failure-with-option-g-O.patch \ + file://0001-Use-labs-instead-of-abs.patch \ " SRCREV = "c296301a06ce49b87c954c9d15452521dfeddf1a" PE = "1"
Signature · 1 line
2.5.1
PATCH 13/16 sudo: Disable rsize_t deliberately
In reply to Khem Rajgcc does not have it but clang does, problem happens when host compiler is gcc and cross compiler is clang, because autoconf detects it with clang and slaps it to host compiler as well
--- meta/recipes-extended/sudo/sudo_1.8.14p3.bb | 1 + 1 file changed, 1 insertion(+)
meta/recipes-extended/sudo/sudo_1.8.14p3.bb +1 -0
@@ -15,6 +15,7 @@ DEPENDS += " ${@bb.utils.contains('DISTRO_FEATURES', 'pam', 'libpam', '', d)}" RDEPENDS_${PN} += " ${@bb.utils.contains('DISTRO_FEATURES', 'pam', 'pam-plugin-limits pam-plugin-keyinit', '', d)}" EXTRA_OECONF += " \ + ac_cv_type_rsize_t=no \ ${@bb.utils.contains('DISTRO_FEATURES', 'pam', '--with-pam', '--without-pam', d)} \ ${@bb.utils.contains('DISTRO_FEATURES', 'systemd', '--enable-tmpfiles.d=${libdir}/tmpfiles.d', '--disable-tmpfiles.d', d)} \ "
Signature · 1 line
2.5.1
PATCH 14/16 webkit-gtk: Backport fix to remove FrameDestructionObserver
In reply to Khem RajThis helps clang compile webkit without problems. This is a backport from upstream
--- ...ride-of-FrameDestructionObserver-willDeta.patch | 105 +++++++++++++++++++++ meta/recipes-sato/webkit/webkit-gtk_1.8.3.bb | 1 + 2 files changed, 106 insertions(+) create mode 100644 meta/recipes-sato/webkit/webkit-gtk-1.8.3/0001-Remove-override-of-FrameDestructionObserver-willDeta.patch
meta/recipes-sato/webkit/webkit-gtk-1.8.3/0001-Remove-override-of-FrameDestructionObserver-willDeta.patch new file +105 -0
@@ -0,0 +1,105 @@ +From 540a792c05f0c7800f694eb39a04d8c31b685789 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 16:36:02 +0000 +Subject: [PATCH] Remove override of FrameDestructionObserver::willDetachPage() + +Reviewed by Antti Koivisto. +This callback was only used by the Chromium port. +editing/Editor.h: +editing/Editor.cpp: +Remove override of FrameDestructionObserver::willDetachPage(). +page/EditorClient.h: +Remove frameWillDetachPage(). This is repeated in every EditorClient subclass. + +Backported from https://bugs.webkit.org/show_bug.cgi?id=119950 + +Upstream-Status: Backport + +Signed-off-by: Khem Raj <[email protected]> +--- + Source/WebCore/editing/Editor.cpp | 6 ------ + Source/WebCore/editing/Editor.h | 1 - + Source/WebCore/loader/EmptyClients.h | 1 - + Source/WebCore/page/EditorClient.h | 1 - + Source/WebKit/gtk/WebCoreSupport/EditorClientGtk.h | 1 - + Source/WebKit2/WebProcess/WebCoreSupport/WebEditorClient.h | 1 - + 6 files changed, 11 deletions(-) + +diff --git a/Source/WebCore/editing/Editor.cpp b/Source/WebCore/editing/Editor.cpp +index 45c5612..2af4057 100644 +--- a/Source/WebCore/editing/Editor.cpp ++++ b/Source/WebCore/editing/Editor.cpp +@@ -3020,10 +3020,4 @@ bool Editor::unifiedTextCheckerEnabled() const + return WebCore::unifiedTextCheckerEnabled(m_frame); + } + +-void Editor::willDetachPage() +-{ +- if (EditorClient* editorClient = client()) +- editorClient->frameWillDetachPage(frame()); +-} +- + } // namespace WebCore +diff --git a/Source/WebCore/editing/Editor.h b/Source/WebCore/editing/Editor.h +index 32fc3dc..0c8cc40 100644 +--- a/Source/WebCore/editing/Editor.h ++++ b/Source/WebCore/editing/Editor.h +@@ -385,7 +385,6 @@ public: + void deviceScaleFactorChanged(); + + private: +- virtual void willDetachPage() OVERRIDE; + + OwnPtr<DeleteButtonController> m_deleteButtonController; + RefPtr<CompositeEditCommand> m_lastEditCommand; +diff --git a/Source/WebCore/loader/EmptyClients.h b/Source/WebCore/loader/EmptyClients.h +index 76e6642..7525a42 100644 +--- a/Source/WebCore/loader/EmptyClients.h ++++ b/Source/WebCore/loader/EmptyClients.h +@@ -439,7 +439,6 @@ public: + EmptyEditorClient() { } + virtual ~EmptyEditorClient() { } + virtual void pageDestroyed() { } +- virtual void frameWillDetachPage(Frame*) { } + + virtual bool shouldDeleteRange(Range*) { return false; } + virtual bool shouldShowDeleteInterface(HTMLElement*) { return false; } +diff --git a/Source/WebCore/page/EditorClient.h b/Source/WebCore/page/EditorClient.h +index f8a2106..451bbd4 100644 +--- a/Source/WebCore/page/EditorClient.h ++++ b/Source/WebCore/page/EditorClient.h +@@ -65,7 +65,6 @@ class EditorClient { + public: + virtual ~EditorClient() { } + virtual void pageDestroyed() = 0; +- virtual void frameWillDetachPage(Frame*) = 0; + + virtual bool shouldDeleteRange(Range*) = 0; + virtual bool shouldShowDeleteInterface(HTMLElement*) = 0; +diff --git a/Source/WebKit/gtk/WebCoreSupport/EditorClientGtk.h b/Source/WebKit/gtk/WebCoreSupport/EditorClientGtk.h +index 0ff9278..2533c16 100644 +--- a/Source/WebKit/gtk/WebCoreSupport/EditorClientGtk.h ++++ b/Source/WebKit/gtk/WebCoreSupport/EditorClientGtk.h +@@ -76,7 +76,6 @@ class EditorClient : public WebCore::EditorClient { + + // from EditorClient + virtual void pageDestroyed(); +- virtual void frameWillDetachPage(WebCore::Frame*) { } + + virtual bool shouldDeleteRange(WebCore::Range*); + virtual bool shouldShowDeleteInterface(WebCore::HTMLElement*); +diff --git a/Source/WebKit2/WebProcess/WebCoreSupport/WebEditorClient.h b/Source/WebKit2/WebProcess/WebCoreSupport/WebEditorClient.h +index 20818ff..b82d1ef 100644 +--- a/Source/WebKit2/WebProcess/WebCoreSupport/WebEditorClient.h ++++ b/Source/WebKit2/WebProcess/WebCoreSupport/WebEditorClient.h +@@ -42,7 +42,6 @@ public: + + private: + virtual void pageDestroyed() OVERRIDE; +- virtual void frameWillDetachPage(WebCore::Frame*) OVERRIDE { } + + virtual bool shouldDeleteRange(WebCore::Range*) OVERRIDE; + virtual bool shouldShowDeleteInterface(WebCore::HTMLElement*) OVERRIDE; +-- +2.5.1 +
meta/recipes-sato/webkit/webkit-gtk_1.8.3.bb +1 -0
@@ -30,6 +30,7 @@ SRC_URI = "\ file://0001-Enable-mips64-build.patch \ file://aarch64.patch \ file://webkit-gtk-ANGLE-doesn-t-build-with-bison-3.patch \ + file://0001-Remove-override-of-FrameDestructionObserver-willDeta.patch \ " SRC_URI[md5sum] = "dcbf9d5e2e6391f857c29a57528b32a6"
Signature · 1 line
2.5.1
PATCH 15/16 mdadm: Fix bugs exposed by clang
In reply to Khem Rajclang points to coding errors which otherwise go unnoticed
--- .../mdadm/files/0001-Fix-typo-in-comparision.patch | 86 ++++++++++++++++++++++ meta/recipes-extended/mdadm/mdadm_3.3.4.bb | 1 + 2 files changed, 87 insertions(+) create mode 100644 meta/recipes-extended/mdadm/files/0001-Fix-typo-in-comparision.patch
meta/recipes-extended/mdadm/files/0001-Fix-typo-in-comparision.patch new file +86 -0
@@ -0,0 +1,86 @@ +From 18d360d74271a066a849bc1fba4f10dbb23ed251 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Tue, 8 Sep 2015 08:17:42 +0000 +Subject: [PATCH] Fix typo in comparision + +error: comparison of array 'devnm' equal to a null pointer is always false + +User bitwise '&' operator as it is a mask + +Fixes +error: use of logical '&&' with constant operand + +Remove extraneous parens + +error: equality comparison with extraneous parentheses + +Remove dead code + +restripe.c:465:21: error: explicitly assigning value of variable of type 'int' to itself + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending + + mdmon.h | 2 +- + mdopen.c | 2 +- + restripe.c | 3 --- + super-intel.c | 2 +- + 4 files changed, 3 insertions(+), 6 deletions(-) + +diff --git a/mdmon.h b/mdmon.h +index aa750c6..0b08c3d 100644 +--- a/mdmon.h ++++ b/mdmon.h +@@ -101,7 +101,7 @@ static inline int is_resync_complete(struct mdinfo *array) + break; + case 10: + l = array->array.layout; +- ncopies = (l & 0xff) * ((l >> 8) && 0xff); ++ ncopies = (l & 0xff) * ((l >> 8) & 0xff); + sync_size = array->component_size * array->array.raid_disks; + sync_size /= ncopies; + break; +diff --git a/mdopen.c b/mdopen.c +index 28410f4..c5ee2a6 100644 +--- a/mdopen.c ++++ b/mdopen.c +@@ -318,7 +318,7 @@ int create_mddev(char *dev, char *name, int autof, int trustworthy, + else if (num < 0) { + /* need to choose a free number. */ + char *_devnm = find_free_devnm(use_mdp); +- if (devnm == NULL) { ++ if (_devnm == NULL) { + pr_err("No avail md devices - aborting\n"); + return -1; + } +diff --git a/restripe.c b/restripe.c +index 4d92190..cfe9666 100644 +--- a/restripe.c ++++ b/restripe.c +@@ -461,9 +461,6 @@ int raid6_check_disks(int data_disks, int start, int chunk_size, + curr_broken_disk = diskD; + } + +- if((Px == 0) && (Qx == 0)) +- curr_broken_disk = curr_broken_disk; +- + if(curr_broken_disk >= data_disks + 2) + broken_status = 2; + +diff --git a/super-intel.c b/super-intel.c +index 3f812d6..1838012 100644 +--- a/super-intel.c ++++ b/super-intel.c +@@ -10027,7 +10027,7 @@ enum imsm_reshape_type imsm_analyze_change(struct supertype *st, + geo->size = max_size; + } + +- if ((direction == ROLLBACK_METADATA_CHANGES)) { ++ if (direction == ROLLBACK_METADATA_CHANGES) { + /* accept size for rollback only + */ + } else { +-- +2.5.1 +
meta/recipes-extended/mdadm/mdadm_3.3.4.bb +1 -0
@@ -12,6 +12,7 @@ SRC_URI = "${KERNELORG_MIRROR}/linux/utils/raid/mdadm/${BPN}-${PV}.tar.xz \ file://mdadm-3.2.2_fix_for_x32.patch \ file://gcc-4.9.patch \ file://mdadm-3.3.2_x32_abi_time_t.patch \ + file://0001-Fix-typo-in-comparision.patch \ file://run-ptest \ " SRC_URI[md5sum] = "7ca8b114710f98f53f20c5787b674a09"
Signature · 1 line
2.5.1
PATCH 16/16 directfb: Avoid using VLAs and printf formats
In reply to Khem RajThese are not portable features and are flagged by clang
--- meta/recipes-graphics/directfb/directfb.inc | 4 +- ...direct-Aboid-usng-VLAs-and-printf-formats.patch | 61 ++++++++++++++++++++++ 2 files changed, 64 insertions(+), 1 deletion(-) create mode 100644 meta/recipes-graphics/directfb/directfb/0001-gfx-direct-Aboid-usng-VLAs-and-printf-formats.patch
meta/recipes-graphics/directfb/directfb.inc +3 -1
@@ -14,7 +14,9 @@ DEPENDS = "jpeg libpng freetype zlib tslib sysfsutils" SRC_URI = "http://www.directfb.org/downloads/Core/DirectFB-1.7/DirectFB-${PV}.tar.gz \ file://configurefix.patch \ file://fusion.patch \ - file://bashism.patch" + file://bashism.patch \ + file://0001-gfx-direct-Aboid-usng-VLAs-and-printf-formats.patch \ + " S = "${WORKDIR}/DirectFB-${PV}"
meta/recipes-graphics/directfb/directfb/0001-gfx-direct-Aboid-usng-VLAs-and-printf-formats.patch new file +61 -0
61 lines not on this page · raw message
Signature · 1 line
2.5.1
Re: [PATCH 11/16] xorg-driver: Fix logical && with bitwise and operation
In reply to Khem RajQuoted 12 lines · Khem Raj
...put_dpms-Replace-logical-with-bitwise-ope.patch | 31 ++++++++++++++++++++++ .../xorg-driver/xf86-video-omap_0.4.3.bb | 2 ++ 2 files changed, 33 insertions(+) create mode 100644 meta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patchmeta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch new file +0
-0
meta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch line 5
Quoted 5 lines
meta/recipes-graphics/xorg-driver/xf86-video-omap/0001-drmmode_output_dpms-Replace-logical-with-bitwise-ope.patch new file +5
-0@@ -0,0 +1,31 @@ +From fc653ca3b36435aea281ae409fb90f85e1bc0d81 Mon Sep 17 00:00:00 2001 +From: Khem Raj <[email protected]> +Date: Mon, 7 Sep 2015 08:44:13 +0000 +Subject: [freedreno][PATCH] drmmode_output_dpms: Replace logical && with bitwise & + operation
why are you adding a patch sent to freedreno mailing list (so I would assume against xf86-video-freedreno), in the xf86-video-omap recipe, is that a typo?
Re: [PATCH 11/16] xorg-driver: Fix logical && with bitwise and operation
Quoted 7 lines · Nicolas Dechesne
why are you adding a patch sent to freedreno mailing list (so I would assume against xf86-video-freedreno), in the xf86-video-omap recipe, is that a typo?
thats where it was sent to. since thats the same list omap driver is using as well.
Re: [PATCH 11/16] xorg-driver: Fix logical && with bitwise and operation
Quoted 7 lines · Khem Raj
Quoted 3 lines
why are you adding a patch sent to freedreno mailing list (so I would assume against xf86-video-freedreno), in the xf86-video-omap recipe, is that a typo?
thats where it was sent to. since thats the same list omap driver is using as well.
hmm. really? I can't find any info about that. how did you figure this out?
Note: I saw your patch on the freedreno ML, and actually the patch also exists on the -freedreno driver as well... probably that file was just copy/pasted.. in any case, from your email sent to freedreno it's not obvious at all that you are targeting -omap driver. so if you can confirm that -omap patches need to be sent on the freedreno list, then at least you might want to resend with [omap] in subject to avoid confusion.. I will test/submit the same patch for -freedreno on my side.
Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close
In reply to a message not in the archivemeta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch line 10
Quoted 5 lines
meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch new file +5
-0+Exposed while compiling on musl + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: Pending
If this has been submitted upstream (as you say in the cover letter for this patchset), please provide a link for the submission in the upstream-status. We can't afford to let the amount of 'Pending' patches in oe-core grow indefinitely.
Same goes for other patches in this set.
Regards,
Alex
Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close
Quoted 11 lines
meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch line 10
Quoted 5 lines
meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch new file +5
-0+Exposed while compiling on musl + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: PendingIf this has been submitted upstream (as you say in the cover letter for this patchset), please provide a link for the submission in the upstream-status. We can't afford to let the amount of 'Pending' patches in oe-core grow indefinitely.
Quoted 1 line
Same goes for other patches in this set.
Actually scratch the above, it was too hasty. What I really wanted to say is that, ideally, you should first submit the patches upstream, have a prooflink of that, and only then send them here with that proof in the upstream-status. There are already too many eternal 'Pending' patches in oe-core that never made it to upsream. We can't afford more of them.
Regards,
Alex
Re: [PATCH 14/16] webkit-gtk: Backport fix to remove FrameDestructionObserver
In reply to Khem RajQuoted 3 lines · Khem Raj
This helps clang compile webkit without problems. This is a backport from upstream
MUT has an upgrade to webkitgtk 2.8.5, would this version have this fix integrated?
Ross
Re: [PATCH 11/16] xorg-driver: Fix logical && with bitwise and operation
In reply to Nicolas DechesneQuoted 11 lines · Nicolas Dechesne
Quoted 7 lines · Khem Raj
Quoted 3 lines
why are you adding a patch sent to freedreno mailing list (so I would assume against xf86-video-freedreno), in the xf86-video-omap recipe, is that a typo?
thats where it was sent to. since thats the same list omap driver is using as well.
hmm. really? I can't find any info about that. how did you figure this out?
Quoted 8 lines
Note: I saw your patch on the freedreno ML, and actually the patch also exists on the -freedreno driver as well... probably that file was just copy/pasted.. in any case, from your email sent to freedreno it's not obvious at all that you are targeting -omap driver. so if you can confirm that -omap patches need to be sent on the freedreno list, then at least you might want to resend with [omap] in subject to avoid confusion.. I will test/submit the same patch for -freedreno on my side.
i talked to Rob, he merged the patch in both -omap and -freedreno. 2 for the price of 1..
nico
Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close
In reply to [email protected]Quoted 14 lines · On Sep 9, 2015, at 3:23 AM, [email protected] wrote:
Quoted 12 lines
meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch line 10
Quoted 6 lines
meta/recipes-extended/slang/slang/0001-Fix-error-conflicting-types-for-posix_close.patch new file +5
-0+Exposed while compiling on musl + +Signed-off-by: Khem Raj <[email protected]> +--- +Upstream-Status: PendingIf this has been submitted upstream (as you say in the cover letter for this patchset), please provide a link for the submission in the upstream-status. We can't afford to let the amount of 'Pending' patches in oe-core grow indefinitely.
Quoted 8 lines
Quoted 1 line
Same goes for other patches in this set.
Actually scratch the above, it was too hasty. What I really wanted to say is that, ideally, you should first submit the patches upstream, have a prooflink of that, and only then send them here with that proof in the upstream-status. There are already too many eternal 'Pending' patches in oe-core that never made it to upsream. We can't afford more of them.
There is process for that. The upstreaming of a patch is an independent thing thats why pending status
is agreed upon probably it predates you in OE contributor's community.
since it deals with different community, and usually we carry a patch for a release or two
and then it gets merged or redone and we drop it. Many a times with recipe upgrades you will
see patches get dropped for these reasons. As integrators, the engagement of rule is a bit different for us. since every component community has its own interworking and nature, this way we make progress while upstreaming takes its own course. This process is followed by all major distro’s as well including OE. Now if we want to step ahead and change a policy to first submit patch upstream then it will be accepted in OE
thats a fine idea, however that requires a wider discussion community wide and also needs tools to reject
patches with out upstream submission. It also has downside where a distro developer may not participate upstream
and hence with not enough momentum things will stall and people may prefer to carry own forks, so we have
to make it less burdensome. We can not go by grey area where we say this patch should be first submitted upstream and this we can accept without submission since that will put extra burden on maintainers of OE for deciding on code thats they may not be experts in.
So I encourage you to formulate a policy and make a proposal to community and get everyone on board in changing the policy OE wide.
Cheers
-Khem
Re: [PATCH 11/16] xorg-driver: Fix logical && with bitwise and operation
In reply to Nicolas DechesneQuoted 8 lines · Nicolas Dechesne
Quoted 4 lines · Khem Raj
Quoted 3 lines
why are you adding a patch sent to freedreno mailing list (so I would assume against xf86-video-freedreno), in the xf86-video-omap recipe, is that a typo?
Quoted 5 lines
Quoted 3 lines
thats where it was sent to. since thats the same list omap driver is using as well.
hmm. really? I can't find any info about that. how did you figure this out?
well I wanted to get Rob’s attention doesnt matter which mailing list goes :)
Quoted 9 lines
Note: I saw your patch on the freedreno ML, and actually the patch also exists on the -freedreno driver as well... probably that file was just copy/pasted.. in any case, from your email sent to freedreno it's not obvious at all that you are targeting -omap driver. so if you can confirm that -omap patches need to be sent on the freedreno list, then at least you might want to resend with [omap] in subject to avoid confusion.. I will test/submit the same patch for -freedreno on my side.
Re: [PATCH 11/16] xorg-driver: Fix logical && with bitwise and operation
In reply to Nicolas DechesneQuoted 25 lines · Nicolas Dechesne
Quoted 11 lines · Nicolas Dechesne
Quoted 7 lines · Khem Raj
Quoted 3 lines
why are you adding a patch sent to freedreno mailing list (so I would assume against xf86-video-freedreno), in the xf86-video-omap recipe, is that a typo?
thats where it was sent to. since thats the same list omap driver is using as well.
hmm. really? I can't find any info about that. how did you figure this out?
Quoted 8 lines
Note: I saw your patch on the freedreno ML, and actually the patch also exists on the -freedreno driver as well... probably that file was just copy/pasted.. in any case, from your email sent to freedreno it's not obvious at all that you are targeting -omap driver. so if you can confirm that -omap patches need to be sent on the freedreno list, then at least you might want to resend with [omap] in subject to avoid confusion.. I will test/submit the same patch for -freedreno on my side.
Quoted 2 lines
i talked to Rob, he merged the patch in both -omap and -freedreno. 2 for the price of 1..
ah there we go as hoped :)
Quoted 2 lines
nico
Re: [PATCH 14/16] webkit-gtk: Backport fix to remove FrameDestructionObserver
In reply to Burton, RossHi Ross
Quoted 9 lines · Ross
On 9 September 2015 at 00:07, Khem Raj <[email protected] <mailto:[email protected]>> wrote:
This helps clang compile webkit without problems. This is a backport
from upstreamMUT has an upgrade to webkitgtk 2.8.5, would this version have this fix integrated?
yes its been fixed in 2.8.x. point me to MUT commit I can cherry-pick locally here and try it out to make it sure.
Quoted 1 line
Ross
Re: [PATCH 14/16] webkit-gtk: Backport fix to remove FrameDestructionObserver
Quoted 3 lines · Khem Raj
yes its been fixed in 2.8.x. point me to MUT commit I can cherry-pick locally here and try it out to make it sure.
The webkitgtk upgrade is poky-contrib:ross/mut (a01c056).
Ross
Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close
In reply to Khem RajQuoted 2 lines · Khem Raj
So I encourage you to formulate a policy and make a proposal to community and get everyone on board in changing the policy OE wide.
Actually, first we need some kind of tool that keeps track of custom patches and their upstream-statuses and minimizes the amount of manual work.
Then we can formulate a policy on top of that: for example, if a patch has been 'Pending' for 3 months, the person who added it gets a reminder to update the status; if the patch has been 'Pending' for 6 months and the patch person has been unresponsive, no more patches are accepted from that person until he does the upstreaming of the ones he already got into oe-core. Something like that.
Regards,
Alex
Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close
Quoted 11 lines · Alexander Kanavin
Quoted 3 lines · Khem Raj
So I encourage you to formulate a policy and make a proposal to community and get everyone on board in changing
the policy OE wide.Actually, first we need some kind of tool that keeps track of custom patches and their upstream-statuses and minimizes the amount of manual work.
Quoted 6 lines
Then we can formulate a policy on top of that: for example, if a patch has been 'Pending' for 3 months, the person who added it gets a reminder to update the status; if the patch has been 'Pending' for 6 months and the patch person has been unresponsive, no more patches are accepted from that person until he does the upstreaming of the ones he already got into oe-core. Something like that.
With so many maintainers on so many different layers/etc we would need some sort of billboard somewhere listing the names of people currently on the "naughty" list ;-)
Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close
Quoted 4 lines · Trevor Woerner
With so many maintainers on so many different layers/etc we would need some sort of billboard somewhere listing the names of people currently on the "naughty" list ;-)
I prefer the concept of a naughty list to blanket rejecting... and I have a script for internal tracking this sort of metric so when I have a spare
hour I'll see if I can extend to do this. "pending" lines added per
person or something.
Ross
Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
In reply to Trevor WoernerQuoted 10 lines · Trevor Woerner
Quoted 6 lines
Then we can formulate a policy on top of that: for example, if a patch has been 'Pending' for 3 months, the person who added it gets a reminder to update the status; if the patch has been 'Pending' for 6 months and the patch person has been unresponsive, no more patches are accepted from that person until he does the upstreaming of the ones he already got into oe-core. Something like that.
With so many maintainers on so many different layers/etc we would need some sort of billboard somewhere listing the names of people currently on the "naughty" list ;-)
Yes, something similar to http://recipes.yoctoproject.org
And the autobuilder could automatically reject or flag a patch that adds custom 'Pending' patches, if the patch author is determined to be on the 'naughty' list :-)
Alex
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
Quoted 9 lines · Alexander Kanavin
Quoted 7 lines · Trevor Woerner
Quoted 6 lines
Then we can formulate a policy on top of that: for example, if a patch has been 'Pending' for 3 months, the person who added it gets a reminder to update the status; if the patch has been 'Pending' for 6 months and the patch person has been unresponsive, no more patches are accepted from that person until he does the upstreaming of the ones he already got into oe-core. Something like that.
Quoted 7 lines
Quoted 3 lines
With so many maintainers on so many different layers/etc we would need some sort of billboard somewhere listing the names of people currently on the "naughty" list ;-)
Yes, something similar to http://recipes.yoctoproject.org
And the autobuilder could automatically reject or flag a patch that adds custom 'Pending' patches, if the patch author is determined to be on the 'naughty' list :-)
What would you do to the poor sod that sends a patch upstream (i.e. Pending) that is not being moved on by the upstream holders? Punish him by making his patch (and potentially all others he has created) invalid? Seems a bit harsh...
Signature · 4 lines
------------------------------------------------------------ Gary Thomas | Consulting for the MLB Associates | Embedded world ------------------------------------------------------------
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
Quoted 5 lines · Gary Thomas
What would you do to the poor sod that sends a patch upstream (i.e. Pending) that is not being moved on by the upstream holders? Punish him by making his patch (and potentially all others he has created) invalid? Seems a bit harsh...
Send upstream is Upstream-Status: Submitted.
Pending means "I wrote a patch but cannot be bothered to send it upstream".
Ross
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
Quoted 9 lines · Ross
On 10 September 2015 at 13:12, Gary Thomas <[email protected] <mailto:[email protected]>> wrote:
What would you do to the poor sod that sends a patch upstream (i.e. Pending) that is not being moved on by the upstream holders? Punish him by making his patch (and potentially all others he has created) invalid? Seems a bit harsh...
Quoted 3 lines
Send upstream is Upstream-Status: Submitted.
Pending means "I wrote a patch but cannot be bothered to send it upstream".
Fair enough, that's a different case which does seem to warrant this treatment.
Signature · 4 lines
------------------------------------------------------------ Gary Thomas | Consulting for the MLB Associates | Embedded world ------------------------------------------------------------
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
In reply to Burton, RossQuoted 10 lines · Ross
On 10 September 2015 at 13:12, Gary Thomas <[email protected] <mailto:[email protected]>> wrote:
What would you do to the poor sod that sends a patch upstream
(i.e. Pending) that is not being moved on by the upstream holders?
Punish him by making his patch (and potentially all others he
has created) invalid? Seems a bit harsh...
Quoted 3 lines
Send upstream is Upstream-Status: Submitted.
Pending means "I wrote a patch but cannot be bothered to send it upstream”.
Or more politely, I am working on upstreaming it on my own schedule.
Quoted 7 lines
Ross
Signature · 4 lines
_______________________________________________ Openembedded-core mailing list [email protected] http://lists.openembedded.org/mailman/listinfo/openembedded-core
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
In reply to Alexander KanavinQuoted 13 lines · Alexander Kanavin
Quoted 10 lines · Trevor Woerner
Quoted 6 lines
Then we can formulate a policy on top of that: for example, if a patch has been 'Pending' for 3 months, the person who added it gets a reminder to update the status; if the patch has been 'Pending' for 6 months and the patch person has been unresponsive, no more patches are accepted from that person until he does the upstreaming of the ones he already got into oe-core. Something like that.
With so many maintainers on so many different layers/etc we would need some sort of billboard somewhere listing the names of people currently on the "naughty" list ;-)
Quoted 5 lines
Yes, something similar to http://recipes.yoctoproject.org
And the autobuilder could automatically reject or flag a patch that adds custom 'Pending' patches, if the patch author is determined to be on the 'naughty' list :-)
That might just alter peoples behavior and change them to use "Submitted" and still not do anything. It does not address the issue.
Having the ability to report a patch's status does have some merit. What is done with it is a different story.
- armin
Quoted 3 lines
Alex
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
In reply to Khem RajQuoted 4 lines
Quoted 2 lines
Pending means "I wrote a patch but cannot be bothered to send it upstream�.
Or more politely, I am working on upstreaming it on my own schedule.
Allowing "own schedule" doesn't work out in reality. Here's how many 'Pending' patches the yocto releases have:
master: 712
fido: 681
dizzy: 683
daisy: 598
dora: 611
dylan: 555
danny: 548
denzil: 518
edison: 426
There should be measures to keep this number from growing indefinitely.
Regards,
Alex
Re: Policy for 'Pending' patches (was Re: [PATCH 02/16] slang: Fix namespace conflict vis-a-vis posix_close)
In reply to akuster808Quoted 6 lines
Quoted 3 lines
And the autobuilder could automatically reject or flag a patch that adds custom 'Pending' patches, if the patch author is determined to be on the 'naughty' list :-)
That might just alter peoples behavior and change them to use "Submitted" and still not do anything. It does not address the issue.
Of course 'Submitted' must include a proof link of the submission, or a notice that the only way to submit upstream is by private email (which is very rare). Such link is necessary anyway to follow up later what happened with the submission.
Regards,
Alex