Update LTTng packages to 2.8.x and Babeltrace to 1.4.0
v2 cover Update LTTng packages to 2.8.x and Babeltrace to 1.4.0
See individual patches for changes.
Tell me if it's okay to put patches found on the mailing list and from the package source of other distributions as is, only with the file names changed.
Philippe Proulx (5):
lttng-libust: bump to version 2.8.1
lttng-tools: bump to version 2.8.2
lttng-tools: remove dependency on host-lttng-babeltrace
lttng-modules: bump to version 2.8.3
lttng-babeltrace: bump to version 1.4.0
...c-remove-elfutils-version-test-for-Buildr.patch | 30 ++ .../0001-no-posix-fallocate-in-uclibc.patch | 37 -- ...ix-uuid-support-detection-on-static-build.patch | 63 --- ...b-Makefile.am-remove-unneeded-static-flag.patch | 36 ++ package/lttng-babeltrace/Config.in | 13 +- package/lttng-babeltrace/lttng-babeltrace.hash | 11 +- package/lttng-babeltrace/lttng-babeltrace.mk | 17 +- ...lf.c-define-NT_GNU_BUILD_ID-if-not-define.patch | 31 ++ package/lttng-libust/lttng-libust.hash | 12 +- package/lttng-libust/lttng-libust.mk | 10 +- package/lttng-modules/lttng-modules.hash | 8 +- package/lttng-modules/lttng-modules.mk | 27 +- ...ld-failure-when-__GLIBC_PREREQ-is-missing.patch | 55 --- ...Fix-snapshot-del-output-with-name-on-musl.patch | 28 ++ .../0002-Fix-add-missing-sys-types.h-header.patch | 29 -- ...Fix-strerror_r-behavior-is-glibc-specific.patch | 26 ++ ...et-thread-stack-size-to-ulimit-soft-value.patch | 427 +++++++++++++++++++++ package/lttng-tools/lttng-tools.hash | 8 +- package/lttng-tools/lttng-tools.mk | 17 +- 19 files changed, 654 insertions(+), 231 deletions(-) create mode 100644 package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch delete mode 100644 package/lttng-babeltrace/0001-no-posix-fallocate-in-uclibc.patch delete mode 100644 package/lttng-babeltrace/0002-configure-fix-uuid-support-detection-on-static-build.patch create mode 100644 package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch create mode 100644 package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch delete mode 100644 package/lttng-tools/0001-Fix-build-failure-when-__GLIBC_PREREQ-is-missing.patch create mode 100644 package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch delete mode 100644 package/lttng-tools/0002-Fix-add-missing-sys-types.h-header.patch create mode 100644 package/lttng-tools/0002-Fix-strerror_r-behavior-is-glibc-specific.patch create mode 100644 package/lttng-tools/0003-Set-thread-stack-size-to-ulimit-soft-value.patch
--
2.9.3
PATCH v2 1/5 lttng-libust: bump to version 2.8.1
The two current patches are removed because they were merged upstream.
The new patch is needed to build with uClibc. It was also submitted upstream, but not merged as is yet. Defining NT_GNU_BUILD_ID to 3 when it's not defined seems like it's something done in other projects. This definition's value should never change (number of the build ID note section in ELF objects).
LTTng-UST now has its man pages written in AsciiDoc and they get converted to troff at build time. This is not needed in a Buildroot image anyway.
I add the --enable-python-agent configure option to build the LTTng-UST Python agent when Python 2 or Python 3 is also part of the BR image. The LTTng-UST Python agent is a Python package which adds LTTng tracing to Python's standard logging when imported.
I tested this updated package for build and good behaviour at run time with glibc, musl, and uClibc.
---
Changes v1 -> v2:
- Detailed commit message.
- Detailed patch messages.
...lf.c-define-NT_GNU_BUILD_ID-if-not-define.patch | 31 ++++++++++++++++++++++ package/lttng-libust/lttng-libust.hash | 12 +++++---- package/lttng-libust/lttng-libust.mk | 10 +++---- 3 files changed, 42 insertions(+), 11 deletions(-) create mode 100644 package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch
package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +31 -0
@@ -0,0 +1,31 @@ +From d400314757a8d5d52bd5722d263bfd5886bb6595 Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Sat, 29 Oct 2016 13:32:57 -0400 +Subject: [PATCH] lttng-ust-elf.c: define NT_GNU_BUILD_ID if not defined + +On uClibc, NT_GNU_BUILD_ID is not defined, so we define it +manually in this case. + +Signed-off-by: Philippe Proulx <[email protected]> +--- + liblttng-ust/lttng-ust-elf.c | 4 ++++ + 1 file changed, 4 insertions(+) + +diff --git a/liblttng-ust/lttng-ust-elf.c b/liblttng-ust/lttng-ust-elf.c +index 5f27920..beaa7f3 100644 +--- a/liblttng-ust/lttng-ust-elf.c ++++ b/liblttng-ust/lttng-ust-elf.c +@@ -29,6 +29,10 @@ + + #define BUF_LEN 4096 + ++#ifndef NT_GNU_BUILD_ID ++# define NT_GNU_BUILD_ID 3 ++#endif ++ + /* + * Retrieve the nth (where n is the `index` argument) phdr (program + * header) from the given elf instance. +-- +2.9.3 +
package/lttng-libust/lttng-libust.hash +7 -5
@@ -1,6 +1,8 @@ -# http://www.lttng.org/files/lttng-ust/lttng-ust-2.7.1.tar.bz2.sha1 -sha1 597ebecb885debb58c263de57f30b1096d4eefc6 lttng-ust-2.7.1.tar.bz2 +# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.1.tar.bz2.md5 +md5 be505077245dc05f93370a565eec15f4 lttng-ust-2.8.1.tar.bz2 -# Locally computed -sha1 2e79c98e458b2e052930fb81441e6ca8377a2ed7 787364e8bac9a146f9d65c763633c2b75eb76463.patch -sha1 cf77ec415111f77573be7917d9b62d8e49dc6304 30307a67b48641875b122d8f832bbfb454e95d92.patch +# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.0.tar.bz2.sha1 +sha1 8bebdcbcb453d2643fb226dbe92de61764f1ee3b lttng-ust-2.8.1.tar.bz2 + +# Locally generated +sha256 6e41349107e83e7b43c69ed358e48788ca2fd095bad61737b850e3f3d2c0508a lttng-ust-2.8.1.tar.bz2
package/lttng-libust/lttng-libust.mk +4 -6
@@ -5,22 +5,20 @@ ################################################################################ LTTNG_LIBUST_SITE = http://lttng.org/files/lttng-ust -LTTNG_LIBUST_VERSION = 2.7.1 +LTTNG_LIBUST_VERSION = 2.8.1 LTTNG_LIBUST_SOURCE = lttng-ust-$(LTTNG_LIBUST_VERSION).tar.bz2 LTTNG_LIBUST_LICENSE = LGPLv2.1, MIT (system headers), GPLv2 (liblttng-ust-ctl/ustctl.c used by lttng-sessiond) LTTNG_LIBUST_LICENSE_FILES = COPYING - -LTTNG_LIBUST_PATCH = \ - https://github.com/lttng/lttng-ust/commit/787364e8bac9a146f9d65c763633c2b75eb76463.patch \ - https://github.com/lttng/lttng-ust/commit/30307a67b48641875b122d8f832bbfb454e95d92.patch - LTTNG_LIBUST_INSTALL_STAGING = YES LTTNG_LIBUST_DEPENDENCIES = liburcu util-linux +LTTNG_LIBUST_CONF_OPTS += --disable-man-pages ifeq ($(BR2_PACKAGE_PYTHON),y) LTTNG_LIBUST_DEPENDENCIES += python +LTTNG_LIBUST_CONF_OPTS += --enable-python-agent else ifeq ($(BR2_PACKAGE_PYTHON3),y) LTTNG_LIBUST_DEPENDENCIES += python3 +LTTNG_LIBUST_CONF_OPTS += --enable-python-agent else LTTNG_LIBUST_CONF_ENV = am_cv_pathless_PYTHON="none" endif
--
2.9.3
PATCH v2 1/5 lttng-libust: bump to version 2.8.1
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 4 lines
The two current patches are removed because they were merged upstream.
The new patch is needed to build with uClibc. It was also submitted upstream, but not merged as is yet.
Not sure I got that correct: did upstream apply another patch, or did upstream have comments that require this patch to changed before being applied?
This is a case where a pointer to the upstream submission would be handy.
Quoted 4 lines
Defining
NT_GNU_BUILD_IDto 3 when
it's not defined seems like it's something done in other projects. This definition's value should never change (number of the build ID note section in ELF objects).
Yes, it seems that '3' is the value associated with this 'note' entry.
Quoted 8 lines
LTTng-UST now has its man pages written in AsciiDoc and they get converted to troff at build time. This is not needed in a Buildroot image anyway.
I add the
--enable-python-agentconfigure option to build the LTTng-UST Python agent when Python 2 or Python 3 is also part of the BR image. The LTTng-UST Python agent is a Python package which adds LTTng tracing to Python's standard logging when imported.
Enablng the python agent should be done in a separate patch. But see below for more on this...
Quoted 2 lines
I tested this updated package for build and good behaviour at run time with glibc, musl, and uClibc.
Uncessary to state this in the commit log; it is expected that testing has been done. ;-)
You can however puth such messages as a "post-comit-log note"...
Quoted 2 lines
Signed-off-by Philippe Proulx <[email protected]>---
... here, just above (or below) the changelog.
Quoted 16 lines
Changes v1 -> v2:
- Detailed commit message.
- Detailed patch messages....lf.c-define-NT_GNU_BUILD_ID-if-not-define.patch | 31 ++++++++++++++++++++++ package/lttng-libust/lttng-libust.hash | 12 +++++---- package/lttng-libust/lttng-libust.mk | 10 +++---- 3 files changed, 42 insertions(+), 11 deletions(-) create mode 100644 package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patchpackage/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +0
-0
package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch line 7
Quoted 7 lines
package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +7
-0@@ -0,0 +1,31 @@ +From d400314757a8d5d52bd5722d263bfd5886bb6595 Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Sat, 29 Oct 2016 13:32:57 -0400 +Subject: [PATCH] lttng-ust-elf.c: define NT_GNU_BUILD_ID if not defined + +On uClibc, NT_GNU_BUILD_ID is not defined, so we define it +manually in this case.
But neither are NT_GNU_ABI_TAG nor NT_GNU_HWCAP nor NT_GNU_GOLD_VERSION which can also appear in the notes section...
Quoted 28 lines
package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +23
-0+Signed-off-by: Philippe Proulx <[email protected]> +--- + liblttng-ust/lttng-ust-elf.c | 4 ++++ + 1 file changed, 4 insertions(+) + +diff --git a/liblttng-ust/lttng-ust-elf.c b/liblttng-ust/lttng-ust-elf.c +index 5f27920..beaa7f3 100644 +--- a/liblttng-ust/lttng-ust-elf.c ++++ b/liblttng-ust/lttng-ust-elf.c +@@ -29,6 +29,10 @@ + + #define BUF_LEN 4096 + ++#ifndef NT_GNU_BUILD_ID ++# define NT_GNU_BUILD_ID 3 ++#endif ++ + /* + * Retrieve the nth (where n is the `index` argument) phdr (program + * header) from the given elf instance. +-- +2.9.3 +package/lttng-libust/lttng-libust.hash +0
-0
package/lttng-libust/lttng-libust.hash line 4
Quoted 9 lines
package/lttng-libust/lttng-libust.hash +3
-5@@ -1,6 +1,8 @@ -# http://www.lttng.org/files/lttng-ust/lttng-ust-2.7.1.tar.bz2.sha1 -sha1 597ebecb885debb58c263de57f30b1096d4eefc6 lttng-ust-2.7.1.tar.bz2 +# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.1.tar.bz2.md5 +md5 be505077245dc05f93370a565eec15f4 lttng-ust-2.8.1.tar.bz2 -# Locally computed -sha1 2e79c98e458b2e052930fb81441e6ca8377a2ed7 787364e8bac9a146f9d65c763633c2b75eb76463.patch -sha1 cf77ec415111f77573be7917d9b62d8e49dc6304 30307a67b48641875b122d8f832bbfb454e95d92.patch +# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.0.tar.bz2.sha1
^
Typo, should be 2.8.1 --------------------------------'
But you could just write:
# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.1.tar.bz2.{md5,sha1} md5 be505077245dc05f93370a565eec15f4 lttng-ust-2.8.1.tar.bz2 sha1 8bebdcbcb453d2643fb226dbe92de61764f1ee3b lttng-ust-2.8.1.tar.bz2
package/lttng-libust/lttng-libust.hash line 8
Quoted 4 lines
package/lttng-libust/lttng-libust.hash +4
-0+sha1 8bebdcbcb453d2643fb226dbe92de61764f1ee3b lttng-ust-2.8.1.tar.bz2 + +# Locally generated +sha256 6e41349107e83e7b43c69ed358e48788ca2fd095bad61737b850e3f3d2c0508a lttng-ust-2.8.1.tar.bz2
I never remember what the policy is. We do want to add a localy computed strong hash (like sha256) when only md5 is available from upstream. But when sha1 is avaiulable, do we want to add a sha256? I'm not sure...
Quoted 20 lines
package/lttng-libust/lttng-libust.mk +1
-6@@ -5,22 +5,20 @@ ################################################################################ LTTNG_LIBUST_SITE = http://lttng.org/files/lttng-ust -LTTNG_LIBUST_VERSION = 2.7.1 +LTTNG_LIBUST_VERSION = 2.8.1 LTTNG_LIBUST_SOURCE = lttng-ust-$(LTTNG_LIBUST_VERSION).tar.bz2 LTTNG_LIBUST_LICENSE = LGPLv2.1, MIT (system headers), GPLv2 (liblttng-ust-ctl/ustctl.c used by lttng-sessiond) LTTNG_LIBUST_LICENSE_FILES = COPYING - -LTTNG_LIBUST_PATCH = \ - https://github.com/lttng/lttng-ust/commit/787364e8bac9a146f9d65c763633c2b75eb76463.patch \ - https://github.com/lttng/lttng-ust/commit/30307a67b48641875b122d8f832bbfb454e95d92.patch - LTTNG_LIBUST_INSTALL_STAGING = YES LTTNG_LIBUST_DEPENDENCIES = liburcu util-linux
package/lttng-libust/lttng-libust.mk line 23
Quoted 10 lines
package/lttng-libust/lttng-libust.mk +3
-0+LTTNG_LIBUST_CONF_OPTS += --disable-man-pages ifeq ($(BR2_PACKAGE_PYTHON),y) LTTNG_LIBUST_DEPENDENCIES += python +LTTNG_LIBUST_CONF_OPTS += --enable-python-agent else ifeq ($(BR2_PACKAGE_PYTHON3),y) LTTNG_LIBUST_DEPENDENCIES += python3 +LTTNG_LIBUST_CONF_OPTS += --enable-python-agent else LTTNG_LIBUST_CONF_ENV = am_cv_pathless_PYTHON="none"
You should explicitly disable the python agent here:
LTTNG_LIBUST_CONF_OPTS += --disable-python-agent
Regards,
Yann E. MORIN.
Quoted 8 lines
package/lttng-libust/lttng-libust.mk +0
-1endif --2.9.3
_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 1/5 lttng-libust: bump to version 2.8.1
Quoted 11 lines · Yann E. MORIN
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 4 lines
The two current patches are removed because they were merged upstream.
The new patch is needed to build with uClibc. It was also submitted upstream, but not merged as is yet.
Not sure I got that correct: did upstream apply another patch, or did upstream have comments that require this patch to changed before being applied?
Upstream is in the process of reviewing this very patch. They might have comments that require changes, but not yet. I'm not waiting for them because the BR freeze is in a few days.
Quoted 3 lines
This is a case where a pointer to the upstream submission would be handy.
Yes I can point to the GH pull request.
Phil
Quoted 169 lines
Quoted 5 lines
Defining
NT_GNU_BUILD_IDto 3 when
it's not defined seems like it's something done in other projects. This definition's value should never change (number of the build ID note section in ELF objects).Yes, it seems that '3' is the value associated with this 'note' entry.
Quoted 8 lines
LTTng-UST now has its man pages written in AsciiDoc and they get converted to troff at build time. This is not needed in a Buildroot image anyway.
I add the
--enable-python-agentconfigure option to build the LTTng-UST Python agent when Python 2 or Python 3 is also part of the BR image. The LTTng-UST Python agent is a Python package which adds LTTng tracing to Python's standard logging when imported.Enablng the python agent should be done in a separate patch. But see below for more on this...
Quoted 2 lines
I tested this updated package for build and good behaviour at run time with glibc, musl, and uClibc.
Uncessary to state this in the commit log; it is expected that testing has been done. ;-)
You can however puth such messages as a "post-comit-log note"...
Quoted 2 lines
Signed-off-by Philippe Proulx <[email protected]>---
... here, just above (or below) the changelog.
Quoted 16 lines
Changes v1 -> v2:
- Detailed commit message.
- Detailed patch messages....lf.c-define-NT_GNU_BUILD_ID-if-not-define.patch | 31 ++++++++++++++++++++++ package/lttng-libust/lttng-libust.hash | 12 +++++---- package/lttng-libust/lttng-libust.mk | 10 +++---- 3 files changed, 42 insertions(+), 11 deletions(-) create mode 100644 package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patchpackage/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +0
-0package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch line 7
Quoted 7 lines
package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +7
-0@@ -0,0 +1,31 @@ +From d400314757a8d5d52bd5722d263bfd5886bb6595 Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Sat, 29 Oct 2016 13:32:57 -0400 +Subject: [PATCH] lttng-ust-elf.c: define NT_GNU_BUILD_ID if not defined + +On uClibc, NT_GNU_BUILD_ID is not defined, so we define it +manually in this case.But neither are
NT_GNU_ABI_TAGnorNT_GNU_HWCAPnorNT_GNU_GOLD_VERSIONwhich can also appear in the notes section...Quoted 28 lines
package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch new file +23
-0+Signed-off-by: Philippe Proulx <[email protected]> +--- + liblttng-ust/lttng-ust-elf.c | 4 ++++ + 1 file changed, 4 insertions(+) + +diff --git a/liblttng-ust/lttng-ust-elf.c b/liblttng-ust/lttng-ust-elf.c +index 5f27920..beaa7f3 100644 +--- a/liblttng-ust/lttng-ust-elf.c ++++ b/liblttng-ust/lttng-ust-elf.c +@@ -29,6 +29,10 @@ + + #define BUF_LEN 4096 + ++#ifndef NT_GNU_BUILD_ID ++# define NT_GNU_BUILD_ID 3 ++#endif ++ + /* + * Retrieve the nth (where n is the `index` argument) phdr (program + * header) from the given elf instance. +-- +2.9.3 +package/lttng-libust/lttng-libust.hash +0
-0package/lttng-libust/lttng-libust.hash line 4
Quoted 9 lines
package/lttng-libust/lttng-libust.hash +3
-5@@ -1,6 +1,8 @@ -# http://www.lttng.org/files/lttng-ust/lttng-ust-2.7.1.tar.bz2.sha1 -sha1 597ebecb885debb58c263de57f30b1096d4eefc6 lttng-ust-2.7.1.tar.bz2 +# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.1.tar.bz2.md5 +md5 be505077245dc05f93370a565eec15f4 lttng-ust-2.8.1.tar.bz2 -# Locally computed -sha1 2e79c98e458b2e052930fb81441e6ca8377a2ed7 787364e8bac9a146f9d65c763633c2b75eb76463.patch -sha1 cf77ec415111f77573be7917d9b62d8e49dc6304 30307a67b48641875b122d8f832bbfb454e95d92.patch +# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.0.tar.bz2.sha1^Typo, should be 2.8.1 --------------------------------'
But you could just write:
# From http://lttng.org/files/lttng-ust/lttng-ust-2.8.1.tar.bz2.{md5,sha1} md5 be505077245dc05f93370a565eec15f4 lttng-ust-2.8.1.tar.bz2 sha1 8bebdcbcb453d2643fb226dbe92de61764f1ee3b lttng-ust-2.8.1.tar.bz2package/lttng-libust/lttng-libust.hash line 8
Quoted 4 lines
package/lttng-libust/lttng-libust.hash +4
-0+sha1 8bebdcbcb453d2643fb226dbe92de61764f1ee3b lttng-ust-2.8.1.tar.bz2 + +# Locally generated +sha256 6e41349107e83e7b43c69ed358e48788ca2fd095bad61737b850e3f3d2c0508a lttng-ust-2.8.1.tar.bz2I never remember what the policy is. We do want to add a localy computed strong hash (like sha256) when only md5 is available from upstream. But when sha1 is avaiulable, do we want to add a sha256? I'm not sure...
Quoted 20 lines
package/lttng-libust/lttng-libust.mk +1
-6@@ -5,22 +5,20 @@ ################################################################################ LTTNG_LIBUST_SITE = http://lttng.org/files/lttng-ust -LTTNG_LIBUST_VERSION = 2.7.1 +LTTNG_LIBUST_VERSION = 2.8.1 LTTNG_LIBUST_SOURCE = lttng-ust-$(LTTNG_LIBUST_VERSION).tar.bz2 LTTNG_LIBUST_LICENSE = LGPLv2.1, MIT (system headers), GPLv2 (liblttng-ust-ctl/ustctl.c used by lttng-sessiond) LTTNG_LIBUST_LICENSE_FILES = COPYING - -LTTNG_LIBUST_PATCH = \ - https://github.com/lttng/lttng-ust/commit/787364e8bac9a146f9d65c763633c2b75eb76463.patch \ - https://github.com/lttng/lttng-ust/commit/30307a67b48641875b122d8f832bbfb454e95d92.patch - LTTNG_LIBUST_INSTALL_STAGING = YES LTTNG_LIBUST_DEPENDENCIES = liburcu util-linuxpackage/lttng-libust/lttng-libust.mk line 23
Quoted 10 lines
package/lttng-libust/lttng-libust.mk +3
-0+LTTNG_LIBUST_CONF_OPTS += --disable-man-pages ifeq ($(BR2_PACKAGE_PYTHON),y) LTTNG_LIBUST_DEPENDENCIES += python +LTTNG_LIBUST_CONF_OPTS += --enable-python-agent else ifeq ($(BR2_PACKAGE_PYTHON3),y) LTTNG_LIBUST_DEPENDENCIES += python3 +LTTNG_LIBUST_CONF_OPTS += --enable-python-agent else LTTNG_LIBUST_CONF_ENV = am_cv_pathless_PYTHON="none"You should explicitly disable the python agent here:
LTTNG_LIBUST_CONF_OPTS += --disable-python-agentRegards,
Yann E. MORIN.Quoted 8 lines
package/lttng-libust/lttng-libust.mk +0
-1endif --2.9.3
_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot--
.-----------------.--------------------.------------------.--------------------.| Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. |'------------------------------^-------^------------------^--------------------'
PATCH v2 2/5 lttng-tools: bump to version 2.8.2
In reply to Philippe ProulxThe current patches are dropped because they were merged upstream.
The new patches exist to fix the build with musl. They were submitted on the lttng-dev mailing list, but not merged yet, or not merged in v2.8.2.
LTTng-tools now has its man pages written in AsciiDoc and they get converted to troff at build time. This is not needed in a Buildroot image anyway.
The --enable-lttng-ust/--disable-lttng-ust configure options are renamed to --with-lttng-ust/--without-lttng-ust in v2.8.
---
Changes v1 -> v2:
- Detailed commit message.
- Complete patches with author and date.
...ld-failure-when-__GLIBC_PREREQ-is-missing.patch | 55 --- ...Fix-snapshot-del-output-with-name-on-musl.patch | 28 ++ .../0002-Fix-add-missing-sys-types.h-header.patch | 29 -- ...Fix-strerror_r-behavior-is-glibc-specific.patch | 26 ++ ...et-thread-stack-size-to-ulimit-soft-value.patch | 427 +++++++++++++++++++++ package/lttng-tools/lttng-tools.hash | 8 +- package/lttng-tools/lttng-tools.mk | 10 +- 7 files changed, 492 insertions(+), 91 deletions(-) delete mode 100644 package/lttng-tools/0001-Fix-build-failure-when-__GLIBC_PREREQ-is-missing.patch create mode 100644 package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch delete mode 100644 package/lttng-tools/0002-Fix-add-missing-sys-types.h-header.patch create mode 100644 package/lttng-tools/0002-Fix-strerror_r-behavior-is-glibc-specific.patch create mode 100644 package/lttng-tools/0003-Set-thread-stack-size-to-ulimit-soft-value.patch
package/lttng-tools/0001-Fix-build-failure-when-__GLIBC_PREREQ-is-missing.patch deleted +0 -55
@@ -1,55 +0,0 @@ -From: Baruch Siach <[email protected]> -Date: Tue, 8 Mar 2016 14:25:34 +0200 -Subject: [PATCH] Fix: build failure when __GLIBC_PREREQ is missing - -The musl C library does not provide the __GLIBC_PREREQ macro. Instead of -relying on glibc version test, check directly for the availability of -epoll_create1(). - -Signed-off-by: Baruch Siach <[email protected]> ---- -Patch status: sent upstream rebased on master branch -(http://lists.lttng.org/pipermail/lttng-dev/2016-March/025593.html) - - configure.ac | 2 +- - src/common/compat/poll.h | 4 ++-- - 2 files changed, 3 insertions(+), 3 deletions(-) - -diff --git a/configure.ac b/configure.ac -index 66d83b60b017..4fc1160c9a08 100644 ---- a/configure.ac -+++ b/configure.ac -@@ -70,7 +70,7 @@ AC_CHECK_FUNCS([ \ - gethostbyname gethostname getpagesize localtime_r memchr memset \ - mkdir munmap putenv realpath rmdir socket strchr strcspn strdup \ - strncasecmp strndup strpbrk strrchr strstr strtol strtoul \ -- strtoull \ -+ strtoull epoll_create1 \ - ]) - - # Babeltrace viewer check -diff --git a/src/common/compat/poll.h b/src/common/compat/poll.h -index 699901848dc1..84f25d5c85aa 100644 ---- a/src/common/compat/poll.h -+++ b/src/common/compat/poll.h -@@ -73,7 +73,7 @@ enum { - LPOLLNVAL = EPOLLHUP, - LPOLLRDHUP = EPOLLRDHUP, - /* Close on exec feature of epoll */ --#if __GLIBC_PREREQ(2, 9) -+#if defined(HAVE_EPOLL_CREATE1) && defined(EPOLL_CLOEXEC) - LTTNG_CLOEXEC = EPOLL_CLOEXEC, - #else - /* -@@ -127,7 +127,7 @@ extern int compat_epoll_create(struct lttng_poll_event *events, - #define lttng_poll_create(events, size, flags) \ - compat_epoll_create(events, size, flags) - --#if __GLIBC_PREREQ(2, 9) -+#if defined(HAVE_EPOLL_CREATE1) && defined(EPOLL_CLOEXEC) - static inline int compat_glibc_epoll_create(int size __attribute__((unused)), - int flags) - { --- -2.7.0 -
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch new file +28 -0
@@ -0,0 +1,28 @@ +From: mjeanson at efficios.com Michael Jeanson +Date: Wed, 15 Jun 2016 17:18:02 -0400 +Subject: [lttng-dev] [PATCH lttng-tools 1/6] Fix: snapshot del-output with name on musl + +Some implementations of strtol() like the one in musl will +return EINVAL in perror when no valid number was found in +the string. + +Signed-off-by: Michael Jeanson <[email protected]> +--- + src/bin/lttng/commands/snapshot.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/src/bin/lttng/commands/snapshot.c b/src/bin/lttng/commands/snapshot.c +index d948226..00aa5b7 100644 +--- a/src/bin/lttng/commands/snapshot.c ++++ b/src/bin/lttng/commands/snapshot.c +@@ -350,7 +350,7 @@ static int cmd_del_output(int argc, const char **argv) + + errno = 0; + id = strtol(argv[1], &name, 10); +- if (id == 0 && errno == 0) { ++ if (id == 0 && (errno == 0 || errno == EINVAL)) { + ret = del_output(UINT32_MAX, name); + } else if (errno == 0 && *name == '\0') { + ret = del_output(id, NULL); +-- +2.7.4
package/lttng-tools/0002-Fix-add-missing-sys-types.h-header.patch deleted +0 -29
@@ -1,29 +0,0 @@ -From: Baruch Siach <[email protected]> -Date: Tue, 8 Mar 2016 14:40:49 +0200 -Subject: [PATCH] Fix: add missing sys/types.h header - -The musl C library requires inclusion of sys/types.h for mode_t. - -Signed-off-by: Baruch Siach <[email protected]> ---- -Patch status: sent upstream -(http://lists.lttng.org/pipermail/lttng-dev/2016-March/025594.html) - - src/common/runas.h | 1 + - 1 file changed, 1 insertion(+) - -diff --git a/src/common/runas.h b/src/common/runas.h -index 2c5565af3646..ac1143eecf84 100644 ---- a/src/common/runas.h -+++ b/src/common/runas.h -@@ -19,6 +19,7 @@ - * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. - */ - -+#include <sys/types.h> - #include <unistd.h> - #include <pthread.h> - --- -2.7.0 -
package/lttng-tools/0002-Fix-strerror_r-behavior-is-glibc-specific.patch new file +26 -0
@@ -0,0 +1,26 @@ +From fc743b22fc26eab9f9dbf48e4505ed2394924bba Mon Sep 17 00:00:00 2001 +From: Michael Jeanson <[email protected]> +Date: Mon, 13 Jun 2016 18:44:17 -0400 +Subject: [PATCH lttng-tools 4/4] Fix: strerror_r behavior is glibc specific + +Signed-off-by: Michael Jeanson <[email protected]> +--- + src/common/error.h | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/src/common/error.h b/src/common/error.h +index 0fbd3a2..e8c811e 100644 +--- a/src/common/error.h ++++ b/src/common/error.h +@@ -198,7 +198,7 @@ static inline void __lttng_print_check_abort(enum lttng_error_level type) + + #define _PERROR(fmt, args...) _ERRMSG("PERROR", PRINT_ERR, fmt, ## args) + +-#if !defined(__linux__) || ((_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && !defined(_GNU_SOURCE)) ++#if !defined(__GLIBC__) || ((_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && !defined(_GNU_SOURCE)) + + /* + * Version using XSI strerror_r. +-- +2.7.4 +
package/lttng-tools/0003-Set-thread-stack-size-to-ulimit-soft-value.patch new file +427 -0
427 lines not on this page · raw message
package/lttng-tools/lttng-tools.hash +7 -1
@@ -1,2 +1,8 @@ +# From http://lttng.org/files/lttng-tools/lttng-tools-2.8.2.tar.bz2.md5 +md5 cac8d64111711ae1b632d0dc12fe92c1 lttng-tools-2.8.2.tar.bz2 + +# From http://lttng.org/files/lttng-tools/lttng-tools-2.8.2.tar.bz2.sha1 +sha1 870874ba9644691de8e564a1fa1459aa5fef42ba lttng-tools-2.8.2.tar.bz2 + # Locally generated -sha256 0c799fb21dfa42475feaa1507ded934608b2e531039b46c40e944a5b81c7b21c lttng-tools-2.7.1.tar.bz2 +sha256 a07e00e9519dad2577d9330bcaf040ab3dc1b289fbcb8c702c39d520422e7565 lttng-tools-2.8.2.tar.bz2
package/lttng-tools/lttng-tools.mk +4 -6
@@ -4,14 +4,12 @@ # ################################################################################ -LTTNG_TOOLS_VERSION = 2.7.1 +LTTNG_TOOLS_VERSION = 2.8.2 LTTNG_TOOLS_SITE = http://lttng.org/files/lttng-tools LTTNG_TOOLS_SOURCE = lttng-tools-$(LTTNG_TOOLS_VERSION).tar.bz2 LTTNG_TOOLS_LICENSE = GPLv2+, LGPLv2.1+ (include/lttng/*, src/lib/lttng-ctl/*) LTTNG_TOOLS_LICENSE_FILES = gpl-2.0.txt lgpl-2.1.txt LICENSE -LTTNG_TOOLS_CONF_OPTS += --with-xml-prefix=$(STAGING_DIR)/usr -# Patching configure.ac -LTTNG_TOOLS_AUTORECONF = YES +LTTNG_TOOLS_CONF_OPTS += --with-xml-prefix=$(STAGING_DIR)/usr --disable-man-pages # The host-lttng-babeltrace technically isn't a required build # dependency. However, having the babeltrace utilities built for the
@@ -20,10 +18,10 @@ LTTNG_TOOLS_AUTORECONF = YES LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt host-lttng-babeltrace util-linux ifeq ($(BR2_PACKAGE_LTTNG_LIBUST),y) -LTTNG_TOOLS_CONF_OPTS += --enable-lttng-ust +LTTNG_TOOLS_CONF_OPTS += --with-lttng-ust LTTNG_TOOLS_DEPENDENCIES += lttng-libust else -LTTNG_TOOLS_CONF_OPTS += --disable-lttng-ust +LTTNG_TOOLS_CONF_OPTS += --without-lttng-ust endif $(eval $(autotools-package))
--
2.9.3
PATCH v2 2/5 lttng-tools: bump to version 2.8.2
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 5 lines
The current patches are dropped because they were merged upstream.
The new patches exist to fix the build with musl. They were submitted on the lttng-dev mailing list, but not merged yet, or not merged in v2.8.2.
Quoted 8 lines
LTTng-tools now has its man pages written in AsciiDoc and they get converted to troff at build time. This is not needed in a Buildroot image anyway.
The --enable-lttng-ust/--disable-lttng-ust configure options are renamed to --with-lttng-ust/--without-lttng-ust in v2.8.
Signed-off-by Philippe Proulx <[email protected]>
[--SNIP--]
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch
Quoted 5 lines
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch new file +0
-0
Filename OK (as you wondered about it on your cover letter).
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch line 3
Quoted 4 lines
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch new file +3
-0@@ -0,0 +1,28 @@ +From: mjeanson at efficios.com Michael Jeanson +Date: Wed, 15 Jun 2016 17:18:02 -0400 +Subject: [lttng-dev] [PATCH lttng-tools 1/6] Fix: snapshot del-output with name on musl
Change the subject to jsut:
[PATCH] Fix: snapshot del-output with name on musl
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch line 9
Quoted 5 lines
package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch new file +5
-0+Some implementations of strtol() like the one in musl will +return EINVAL in perror when no valid number was found in +the string. + +Signed-off-by: Michael Jeanson <[email protected]>
Here you would get:
Signed-off-by: Michael Jeanson <[email protected]> [Philippe: grabbed from the mailing list; https://lists.lttng.org/pipermail/lttng-dev/2016-June/026192.html ] Signed-off-by: Philippe Proulx <[email protected]>
Similarly for the other patches....
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 3/5 lttng-tools: remove dependency on host-lttng-babeltrace
In reply to Philippe ProulxWe don't need to build Babeltrace for the host here. Any host running a major Linux distribution can get a recent version of Babeltrace.
See <http://diamon.org/babeltrace/>.
---
Changes v1 -> v2:
- Add Reviewed-by tag.
package/lttng-tools/lttng-tools.mk | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)
package/lttng-tools/lttng-tools.mk +1 -6
@@ -10,12 +10,7 @@ LTTNG_TOOLS_SOURCE = lttng-tools-$(LTTNG_TOOLS_VERSION).tar.bz2 LTTNG_TOOLS_LICENSE = GPLv2+, LGPLv2.1+ (include/lttng/*, src/lib/lttng-ctl/*) LTTNG_TOOLS_LICENSE_FILES = gpl-2.0.txt lgpl-2.1.txt LICENSE LTTNG_TOOLS_CONF_OPTS += --with-xml-prefix=$(STAGING_DIR)/usr --disable-man-pages - -# The host-lttng-babeltrace technically isn't a required build -# dependency. However, having the babeltrace utilities built for the -# host is very useful, since those tools allow to convert the binary -# trace format into an human readable format. -LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt host-lttng-babeltrace util-linux +LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt util-linux ifeq ($(BR2_PACKAGE_LTTNG_LIBUST),y) LTTNG_TOOLS_CONF_OPTS += --with-lttng-ust
--
2.9.3
PATCH v2 3/5 lttng-tools: remove dependency on host-lttng-babeltrace
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 2 lines
We don't need to build Babeltrace for the host here. Any host running a major Linux distribution can get a recent version of Babeltrace.
The LTS entreprise-class distribution may have a quite old version of babeltrace. For example, on Ubuntu Precise (14.04):
$ apt-cache policy babeltrace
babeltrace:
Installed: (none)
Candidate: 1.0.0~rc1-0ubuntu1
Version table:
1.0.0~rc1-0ubuntu1 0
500 http://archive.ubuntu.com/ubuntu/ precise/universe i386 PackagesIs this recent enough to deal with traces from today?
If not, then maybe add an entry in the "Host utilities" sub-menu?
Regards,
Yann E. MORIN.
Quoted 36 lines
See <http://diamon.org/babeltrace/>.
Signed-off-by Philippe Proulx <[email protected]>Reviewed-by Thomas Petazzoni <[email protected]>---
Changes v1 -> v2:
- Add Reviewed-by tag.package/lttng-tools/lttng-tools.mk | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)package/lttng-tools/lttng-tools.mk +1
-7@@ -10,12 +10,7 @@ LTTNG_TOOLS_SOURCE = lttng-tools-$(LTTNG_TOOLS_VERSION).tar.bz2 LTTNG_TOOLS_LICENSE = GPLv2+, LGPLv2.1+ (include/lttng/*, src/lib/lttng-ctl/*) LTTNG_TOOLS_LICENSE_FILES = gpl-2.0.txt lgpl-2.1.txt LICENSE LTTNG_TOOLS_CONF_OPTS += --with-xml-prefix=$(STAGING_DIR)/usr --disable-man-pages - -# The host-lttng-babeltrace technically isn't a required build -# dependency. However, having the babeltrace utilities built for the -# host is very useful, since those tools allow to convert the binary -# trace format into an human readable format. -LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt host-lttng-babeltrace util-linux +LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt util-linux ifeq ($(BR2_PACKAGE_LTTNG_LIBUST),y) LTTNG_TOOLS_CONF_OPTS += --with-lttng-ust --2.9.3
_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 3/5 lttng-tools: remove dependency on host-lttng-babeltrace
Quoted 17 lines · Yann E. MORIN
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 2 lines
We don't need to build Babeltrace for the host here. Any host running a major Linux distribution can get a recent version of Babeltrace.
The LTS entreprise-class distribution may have a quite old version of babeltrace. For example, on Ubuntu Precise (14.04):
$ apt-cache policy babeltrace babeltrace: Installed: (none) Candidate: 1.0.0~rc1-0ubuntu1 Version table: 1.0.0~rc1-0ubuntu1 0 500 http://archive.ubuntu.com/ubuntu/ precise/universe i386 Packages
Quoted 3 lines
Is this recent enough to deal with traces from today?
If not, then maybe add an entry in the "Host utilities" sub-menu?
v1.0.0 can read traces generated by any LTTng 2.x tracer. However it's missing some features and bug fixes, so yes, I could add it back to the host utils.
Phil
Quoted 50 lines
Reviewed-by "Yann E. MORIN" <[email protected]>Regards,
Yann E. MORIN.Quoted 32 lines
See <http://diamon.org/babeltrace/>.
Signed-off-by Philippe Proulx <[email protected]>Reviewed-by Thomas Petazzoni <[email protected]>---
Changes v1 -> v2:
- Add Reviewed-by tag.package/lttng-tools/lttng-tools.mk | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)package/lttng-tools/lttng-tools.mk +1
-7@@ -10,12 +10,7 @@ LTTNG_TOOLS_SOURCE = lttng-tools-$(LTTNG_TOOLS_VERSION).tar.bz2 LTTNG_TOOLS_LICENSE = GPLv2+, LGPLv2.1+ (include/lttng/*, src/lib/lttng-ctl/*) LTTNG_TOOLS_LICENSE_FILES = gpl-2.0.txt lgpl-2.1.txt LICENSE LTTNG_TOOLS_CONF_OPTS += --with-xml-prefix=$(STAGING_DIR)/usr --disable-man-pages - -# The host-lttng-babeltrace technically isn't a required build -# dependency. However, having the babeltrace utilities built for the -# host is very useful, since those tools allow to convert the binary -# trace format into an human readable format. -LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt host-lttng-babeltrace util-linux +LTTNG_TOOLS_DEPENDENCIES = liburcu libxml2 popt util-linux ifeq ($(BR2_PACKAGE_LTTNG_LIBUST),y) LTTNG_TOOLS_CONF_OPTS += --with-lttng-ust --2.9.3
Quoted 4 lines
_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot--
.-----------------.--------------------.------------------.--------------------.| Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. |'------------------------------^-------^------------------^--------------------'
PATCH v2 3/5 lttng-tools: remove dependency on host-lttng-babeltrace
Philippe, All,
On 2016-10-30 18:07 -0400, Philippe Proulx spake thusly:
Quoted 6 lines · Yann E. MORIN
Quoted 5 lines
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 2 lines
We don't need to build Babeltrace for the host here. Any host running a major Linux distribution can get a recent version of Babeltrace.
The LTS entreprise-class distribution may have a quite old version of babeltrace.
[--SNIP--]
Quoted 3 lines
v1.0.0 can read traces generated by any LTTng 2.x tracer. However it's missing some features and bug fixes, so yes, I could add it back to the host utils.
Yes, pretty please yes! ;-)
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 4/5 lttng-modules: bump to version 2.8.3
In reply to Philippe ProulxLTTng-modules has its own Makefile to build its modules. In particular, starting from v2.8, LTTng-modules has a way to be included as part of a Linux kernel image, that is, not as loadable kernel modules. This implies that additional environment variables need to be passed to the Makefile and kernel build system.
It's always a good idea for this project to use its Makefile directly, not building it like you build your typical kernel modules, that is, using make with -C to change directory to the kernel source, and pointing the module with M=. The LTTng-modules does this internally anyway.
Using the project's Makefile directly instead of the generic BR kernel-module package ensures that future minor bumps of this project will be easier as well.
---
Changes v1 -> v2:
- Detailed commit message.
package/lttng-modules/lttng-modules.hash | 8 ++++++-- package/lttng-modules/lttng-modules.mk | 27 +++++++++++++++++++++++++-- 2 files changed, 31 insertions(+), 4 deletions(-)
package/lttng-modules/lttng-modules.hash +6 -2
@@ -1,2 +1,6 @@ -# from http://lttng.org/files/lttng-tools/lttng-modules-2.7.1.tar.bz2.sha1 -sha1 38c42ec89b4966c7a4cc04aea75f5841634d8d66 lttng-modules-2.7.1.tar.bz2 +# From http://lttng.org/files/lttng-modules/lttng-modules-2.8.3.tar.bz2.md5 +md5 424354c4834d20235dacd9a1fd7653b6 lttng-modules-2.8.3.tar.bz2 + +# From http://lttng.org/files/lttng-modules/lttng-modules-2.8.3.tar.bz2.sha1 +sha1 862705784a54962f70226004f1fd0bf77f73f10f lttng-modules-2.8.3.tar.bz2 +
package/lttng-modules/lttng-modules.mk +25 -2
@@ -4,11 +4,34 @@ # ################################################################################ -LTTNG_MODULES_VERSION = 2.7.1 +LTTNG_MODULES_VERSION = 2.8.3 LTTNG_MODULES_SITE = http://lttng.org/files/lttng-modules LTTNG_MODULES_SOURCE = lttng-modules-$(LTTNG_MODULES_VERSION).tar.bz2 LTTNG_MODULES_LICENSE = LGPLv2.1/GPLv2 (kernel modules), MIT (lib/bitfield.h, lib/prio_heap/*) LTTNG_MODULES_LICENSE_FILES = lgpl-2.1.txt gpl-2.0.txt mit-license.txt LICENSE +LTTNG_MODULES_DEPENDENCIES = linux + +ifeq ($(BR2_PACKAGE_LTTNG_MODULES),y) +LINUX_NEEDS_MODULES = y +endif + +LTTNG_MODULES_MAKE_OPTS = \ + KERNELDIR="$(LINUX_DIR)" \ + ARCH="$(KERNEL_ARCH)" \ + CROSS_COMPILE="$(TARGET_CROSS)" + +define LTTNG_MODULES_BUILD_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) -C $(@D) +endef + +define LTTNG_MODULES_INSTALL_STAGING_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) \ + INSTALL_MOD_PATH="$(STAGING_DIR)" -C $(@D) modules_install +endef + +define LTTNG_MODULES_INSTALL_TARGET_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) \ + INSTALL_MOD_PATH="$(TARGET_DIR)" -C $(@D) modules_install +endef -$(eval $(kernel-module)) $(eval $(generic-package))
--
2.9.3
PATCH v2 4/5 lttng-modules: bump to version 2.8.3
Philippe, All,
Eh, undoing my patch here. Meh... ;-]
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 5 lines
LTTng-modules has its own
Makefileto build its modules. In particular, starting from v2.8, LTTng-modules has a way to be included as part of a Linux kernel image, that is, not as loadable kernel modules. This implies that additional environment variables need to be passed to theMakefileand kernel build system.
OK, so two things here:
1- We do not support building LTTng built-in to the kernel do we? Or do
you plan in making that possible?
2- we can pass additional variables when building modules:
https://buildroot.org/downloads/manual/manual.html#_infrastructure_for_packages_building_kernel_modules
LTTNG_MODULES_MODULE_MAKE_OPTS = VARIABLE=value
Quoted 9 lines
It's always a good idea for this project to use its
Makefiledirectly, not building it like you build your typical kernel modules, that is, using make with-Cto change directory to the kernel source, and pointing the module withM=. The LTTng-modules does this internally anyway.Using the project's
Makefiledirectly instead of the generic BR kernel-module package ensures that future minor bumps of this project will be easier as well.
Using the kernel-module infra makes it sure that we pass the kernel related variables.
I'm not too fond of changing that back.
Quoted 1 line
Signed-off-by Philippe Proulx <[email protected]>
[--SNIP--]
Quoted 23 lines
package/lttng-modules/lttng-modules.mk +10
-1@@ -4,11 +4,34 @@ # ################################################################################ -LTTNG_MODULES_VERSION = 2.7.1 +LTTNG_MODULES_VERSION = 2.8.3 LTTNG_MODULES_SITE = http://lttng.org/files/lttng-modules LTTNG_MODULES_SOURCE = lttng-modules-$(LTTNG_MODULES_VERSION).tar.bz2 LTTNG_MODULES_LICENSE = LGPLv2.1/GPLv2 (kernel modules), MIT (lib/bitfield.h, lib/prio_heap/*) LTTNG_MODULES_LICENSE_FILES = lgpl-2.1.txt gpl-2.0.txt mit-license.txt LICENSE +LTTNG_MODULES_DEPENDENCIES = linux + +ifeq ($(BR2_PACKAGE_LTTNG_MODULES),y) +LINUX_NEEDS_MODULES = y +endif + +LTTNG_MODULES_MAKE_OPTS = \ + KERNELDIR="$(LINUX_DIR)" \ + ARCH="$(KERNEL_ARCH)" \
package/lttng-modules/lttng-modules.mk line 30
Quoted 10 lines
package/lttng-modules/lttng-modules.mk +10
-0+ CROSS_COMPILE="$(TARGET_CROSS)" + +define LTTNG_MODULES_BUILD_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) -C $(@D) +endef + +define LTTNG_MODULES_INSTALL_STAGING_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) \ + INSTALL_MOD_PATH="$(STAGING_DIR)" -C $(@D) modules_install +endef
Why do you need to install the modules in staging?
I stand by the fact that we should use the kernel-modules infra, if it can be made to work with specifying just the few new variables that are needed, as explained above.
If not possible, then provide an explanation why this is not sufficient.
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 4/5 lttng-modules: bump to version 2.8.3
Quoted 65 lines · Yann E. MORIN
Philippe, All,
Eh, undoing my patch here. Meh... ;-]
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 5 lines
LTTng-modules has its own
Makefileto build its modules. In particular, starting from v2.8, LTTng-modules has a way to be included as part of a Linux kernel image, that is, not as loadable kernel modules. This implies that additional environment variables need to be passed to theMakefileand kernel build system.OK, so two things here:
1- We do not support building LTTng built-in to the kernel do we? Or do
you plan in making that possible?2- we can pass additional variables when building modules:
https://buildroot.org/downloads/manual/manual.html#_infrastructure_for_packages_building_kernel_modules
LTTNG_MODULES_MODULE_MAKE_OPTS= VARIABLE=valueQuoted 9 lines
It's always a good idea for this project to use its
Makefiledirectly, not building it like you build your typical kernel modules, that is, using make with-Cto change directory to the kernel source, and pointing the module withM=. The LTTng-modules does this internally anyway.Using the project's
Makefiledirectly instead of the generic BR kernel-module package ensures that future minor bumps of this project will be easier as well.Using the kernel-module infra makes it sure that we pass the kernel related variables.
I'm not too fond of changing that back.
Quoted 1 line
Signed-off-by Philippe Proulx <[email protected]>[--SNIP--]
Quoted 27 lines
package/lttng-modules/lttng-modules.mk +14
-1@@ -4,11 +4,34 @@ # ################################################################################ -LTTNG_MODULES_VERSION = 2.7.1 +LTTNG_MODULES_VERSION = 2.8.3 LTTNG_MODULES_SITE = http://lttng.org/files/lttng-modules LTTNG_MODULES_SOURCE = lttng-modules-$(LTTNG_MODULES_VERSION).tar.bz2 LTTNG_MODULES_LICENSE = LGPLv2.1/GPLv2 (kernel modules), MIT (lib/bitfield.h, lib/prio_heap/*) LTTNG_MODULES_LICENSE_FILES = lgpl-2.1.txt gpl-2.0.txt mit-license.txt LICENSE +LTTNG_MODULES_DEPENDENCIES = linux + +ifeq ($(BR2_PACKAGE_LTTNG_MODULES),y) +LINUX_NEEDS_MODULES = y +endif + +LTTNG_MODULES_MAKE_OPTS = \ + KERNELDIR="$(LINUX_DIR)" \ + ARCH="$(KERNEL_ARCH)" \ + CROSS_COMPILE="$(TARGET_CROSS)" + +define LTTNG_MODULES_BUILD_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) -C $(@D)
Quoted 8 lines
package/lttng-modules/lttng-modules.mk line 30
Quoted 6 lines
package/lttng-modules/lttng-modules.mk +6
-0+endef + +define LTTNG_MODULES_INSTALL_STAGING_CMDS + $(TARGET_CONFIGURE_OPTS) $(MAKE) $(LTTNG_MODULES_MAKE_OPTS) \ + INSTALL_MOD_PATH="$(STAGING_DIR)" -C $(@D) modules_install +endefWhy do you need to install the modules in staging?
Oh that's a bad copy from another package. Indeed this is not needed in staging.
Quoted 6 lines
I stand by the fact that we should use the kernel-modules infra, if it can be made to work with specifying just the few new variables that are needed, as explained above.
If not possible, then provide an explanation why this is not sufficient.
It's definitely possible. I changed it because upstream said that it's better to use their own Makefile as they can add new variables without having to track them in the distribution build systems. I can convert it back to using the the kernel-module infra if you prefer. I have no other arguments.
Phil
Quoted 11 lines
Regards,
Yann E. MORIN.--
.-----------------.--------------------.------------------.--------------------.| Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. |'------------------------------^-------^------------------^--------------------'
PATCH v2 4/5 lttng-modules: bump to version 2.8.3
Philippe, All,
On 2016-10-30 18:13 -0400, Philippe Proulx spake thusly:
Quoted 7 lines · Yann E. MORIN
Quoted 6 lines
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 5 lines
LTTng-modules has its own
Makefileto build its modules. In particular, starting from v2.8, LTTng-modules has a way to be included as part of a Linux kernel image, that is, not as loadable kernel modules. This implies that additional environment variables need to be passed to theMakefileand kernel build system.
[--SNIP--]
Quoted 10 lines
Quoted 5 lines
I stand by the fact that we should use the kernel-modules infra, if it can be made to work with specifying just the few new variables that are needed, as explained above.
If not possible, then provide an explanation why this is not sufficient.
It's definitely possible. I changed it because upstream said that it's better to use their own
Makefileas they can add new variables without having to track them in the distribution build systems. I can convert it back to using the the kernel-module infra if you prefer. I have no other arguments.
Well, what is different from a package that adds, removes or renames a configure option? We downstreams have to deal with those changes every bump we do anyway.
All we need is that upstreams provide such information in the the docs.
So, I still prefer we keep using the kernel-module infra.
Thanks for the feedback! :-)
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 5/5 lttng-babeltrace: bump to version 1.4.0
In reply to Philippe ProulxCurrent patches are dropped because they were merged upstream.
Official Babeltrace's website changed from
<http://www.efficios.com/babeltrace> to <http://diamon.org/babeltrace>.
Official upstream site of Babeltrace for the distribution of tarballs changed from lttng.org to efficios.com (see
<http://diamon.org/babeltrace>).
I'm removing the host package build, as users can get Babeltrace on major Linux distributions using their package manager.
Since Babeltrace 1.4, there is support for printing debug information along with compatible traces and event records. Babeltrace needs elfutils in this case. If elfutils exists in the BR image, then this feature is enabled.
---
Changes v1 -> v2:
- Detailed commit message.
- Detailed patch messages.
- Add comment above LTTNG_BABELTRACE_AUTORECONF = YES in lttng-babeltrace.mk.
...c-remove-elfutils-version-test-for-Buildr.patch | 30 +++++++++++ .../0001-no-posix-fallocate-in-uclibc.patch | 37 ------------- ...ix-uuid-support-detection-on-static-build.patch | 63 ---------------------- ...b-Makefile.am-remove-unneeded-static-flag.patch | 36 +++++++++++++ package/lttng-babeltrace/Config.in | 13 ++--- package/lttng-babeltrace/lttng-babeltrace.hash | 11 ++-- package/lttng-babeltrace/lttng-babeltrace.mk | 17 +++--- 7 files changed, 88 insertions(+), 119 deletions(-) create mode 100644 package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch delete mode 100644 package/lttng-babeltrace/0001-no-posix-fallocate-in-uclibc.patch delete mode 100644 package/lttng-babeltrace/0002-configure-fix-uuid-support-detection-on-static-build.patch create mode 100644 package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch new file +30 -0
@@ -0,0 +1,30 @@ +From d3cce60dfde30bd2fd67caa3fbc5f111131c243a Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Fri, 28 Oct 2016 01:48:25 -0400 +Subject: [PATCH] configure.ac: remove elfutils version test for Buildroot + +The AX_LIB_ELFUTILS macro builds an executable file and executes it to +discover the version of elfutils. Obviously we cannot execute a +cross-compiled program here, hence this test is removed because we know +that elfutils's version is greater than 0.154 anyway in Buildroot. + +Signed-off-by: Philippe Proulx <[email protected]> +--- + configure.ac | 1 - + 1 file changed, 1 deletion(-) + +diff --git a/configure.ac b/configure.ac +index 0b7bde2..f1bfe52 100644 +--- a/configure.ac ++++ b/configure.ac +@@ -268,7 +268,6 @@ AS_IF([test "x$_enable_debug_info" = xyes], [ + # Check if libelf and libdw are present + AC_CHECK_LIB([elf], [elf_version], [], [AC_MSG_ERROR(Missing libelf (from elfutils >= 0.154) which is required by debug info. You can disable this feature using --disable-debug-info.)]) + AC_CHECK_LIB([dw], [dwarf_begin], [], [AC_MSG_ERROR(Missing libdw (from elfutils >= 0.154) which is required by debug info. You can disable this feature using --disable-debug-info.)]) +- AX_LIB_ELFUTILS([0], [154], [], [AC_MSG_ERROR(elfutils >= 0.154 is required to use the debug info feature. You can disable this feature using --disable-debug-info.)]) + AC_DEFINE([ENABLE_DEBUG_INFO], [1], [Define to 1 if you enable the 'debug info' feature]) + ], []) + +-- +2.9.3 +
package/lttng-babeltrace/0001-no-posix-fallocate-in-uclibc.patch deleted +0 -37
@@ -1,37 +0,0 @@ -Do not call posix_fallocate() on uClibc - -uClibc does not implement posix_fallocate(), and posix_fallocate() is -mostly only an hint to the kernel that we will need such or such -amount of space inside a file. So we just don't call posix_fallocate() -when building against uClibc. - -Signed-off-by: Thomas Petazzoni <[email protected]> -[Peter: add #include <features.h> ---- - formats/ctf/ctf.c | 3 +++ - 1 file changed, 3 insertions(+) - -Index: lttng-babeltrace-0.8/formats/ctf/ctf.c -=================================================================== ---- lttng-babeltrace-0.8.orig/formats/ctf/ctf.c -+++ lttng-babeltrace-0.8/formats/ctf/ctf.c -@@ -32,6 +32,7 @@ - #include <sys/stat.h> - #include <fcntl.h> - #include <dirent.h> -+#include <features.h> - #include <glib.h> - #include <unistd.h> - #include <stdlib.h> -@@ -384,9 +385,11 @@ - } - pos->content_size = -1U; /* Unknown at this point */ - pos->packet_size = WRITE_PACKET_LEN; -+#ifndef __UCLIBC__ - off = posix_fallocate(pos->fd, pos->mmap_offset, - pos->packet_size / CHAR_BIT); - assert(off >= 0); -+#endif - pos->offset = 0; - } else { - read_next_packet:
package/lttng-babeltrace/0002-configure-fix-uuid-support-detection-on-static-build.patch deleted +0 -63
63 lines not on this page · raw message
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch new file +36 -0
@@ -0,0 +1,36 @@ +From 27e37d2621317d25e08ba0683bf8540b05b9347d Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Fri, 28 Oct 2016 02:09:20 -0400 +Subject: [PATCH] tests/lib/Makefile.am: remove unneeded -static flag + +Having those -static flags breaks the build in shared-only builds. + +Upstream does not even know why those flags are passed here, and they +are about to remove them as of this date. + +Signed-off-by: Philippe Proulx <[email protected]> +--- + tests/lib/Makefile.am | 2 -- + 1 file changed, 2 deletions(-) + +diff --git a/tests/lib/Makefile.am b/tests/lib/Makefile.am +index 33ada16..c685e77 100644 +--- a/tests/lib/Makefile.am ++++ b/tests/lib/Makefile.am +@@ -31,13 +31,11 @@ check_SCRIPTS = test_seek_big_trace \ + test_ctf_writer_complete + + if ENABLE_DEBUG_INFO +-test_dwarf_LDFLAGS = -static + test_dwarf_LDADD = $(LIBTAP) \ + $(top_builddir)/lib/libbabeltrace.la \ + $(top_builddir)/lib/libdebug-info.la + test_dwarf_SOURCES = test_dwarf.c + +-test_bin_info_LDFLAGS = -static + test_bin_info_LDADD = $(LIBTAP) \ + $(top_builddir)/lib/libbabeltrace.la \ + $(top_builddir)/lib/libdebug-info.la +-- +2.9.3 +
package/lttng-babeltrace/Config.in +7 -6
@@ -14,18 +14,19 @@ config BR2_PACKAGE_LTTNG_BABELTRACE to/from another trace format. Babeltrace is the reference implementation of the Common - Trace Format (see <http://diamon.org/docs/ctf/>). Babeltrace + Trace Format (see <http://diamon.org/ctf/>). Babeltrace can decode and display LTTng traces, since the latter are written using the CTF format. + The babeltrace utility is built with debug information + printing support when the elfutils package is selected. + Having this tool on the target is not mandatory. You can transfer the binary trace output from the target to your - development workstation, and use babeltrace here. Buildroot - automatically builds a host babeltrace in - $(O)/host/usr/bin/babeltrace as a dependency of the - lttng-tools package. + development workstation, and use babeltrace here. Babeltrace + is packaged for all major Linux distributions. - http://www.efficios.com/babeltrace + http://diamon.org/babeltrace comment "lttng-babeltrace needs a toolchain w/ wchar, threads" depends on BR2_USE_MMU
package/lttng-babeltrace/lttng-babeltrace.hash +5 -6
@@ -1,7 +1,6 @@ -# From http://lttng.org/files/babeltrace/babeltrace-1.2.4.tar.bz2.md5 -md5 20a806fa7570e69a91707855a98f3ca6 babeltrace-1.2.4.tar.bz2 -# From http://lttng.org/files/babeltrace/babeltrace-1.2.4.tar.bz2.sha1 -sha1 c72a80e8fd905cb1ae3b0cf88885517faafe0fa1 babeltrace-1.2.4.tar.bz2 +# From http://www.efficios.com/files/babeltrace/babeltrace-1.4.0.tar.bz2.md5 +md5 fa99064048af5c9c6ff00e489ac809c2 babeltrace-1.4.0.tar.bz2 +# From http://www.efficios.com/files/babeltrace/babeltrace-1.4.0.tar.bz2.sha1 +sha1 3b162480ee883625e8581bf163817f78a4b6487b babeltrace-1.4.0.tar.bz2 # Locally generated -sha256 666e3a1ad2dc7d5703059963056e7800f0eab59c8eeb6be2efe4f3acc5209eb1 babeltrace-1.2.4.tar.bz2 - +sha256 9469eeb22617cd12668683b04c27003e5337e9ac66ade914988df3642fc0d0e4 babeltrace-1.4.0.tar.bz2
package/lttng-babeltrace/lttng-babeltrace.mk +10 -7
@@ -4,18 +4,21 @@ # ################################################################################ -LTTNG_BABELTRACE_SITE = http://lttng.org/files/babeltrace -LTTNG_BABELTRACE_VERSION = 1.2.4 +LTTNG_BABELTRACE_SITE = http://www.efficios.com/files/babeltrace +LTTNG_BABELTRACE_VERSION = 1.4.0 LTTNG_BABELTRACE_SOURCE = babeltrace-$(LTTNG_BABELTRACE_VERSION).tar.bz2 LTTNG_BABELTRACE_LICENSE = MIT, LGPLv2.1 (include/babeltrace/list.h), GPLv2 (test code) LTTNG_BABELTRACE_LICENSE_FILES = mit-license.txt gpl-2.0.txt LICENSE - LTTNG_BABELTRACE_DEPENDENCIES = popt util-linux libglib2 host-pkgconf -HOST_LTTNG_BABELTRACE_DEPENDENCIES = \ - host-popt host-util-linux host-libglib2 host-pkgconf -# for 0002-configure-fix-uuid-support-detection-on-static-build.patch +# For 0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch LTTNG_BABELTRACE_AUTORECONF = YES +ifeq ($(BR2_PACKAGE_ELFUTILS),y) +LTTNG_BABELTRACE_DEPENDENCIES += elfutils +LTTNG_BABELTRACE_CONF_OPTS += --enable-debug-info +else +LTTNG_BABELTRACE_CONF_OPTS += --disable-debug-info +endif + $(eval $(autotools-package)) -$(eval $(host-autotools-package))
--
2.9.3
PATCH v2 5/5 lttng-babeltrace: bump to version 1.4.0
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 8 lines
Current patches are dropped because they were merged upstream.
Official Babeltrace's website changed from
<http://www.efficios.com/babeltrace> to <http://diamon.org/babeltrace>.Official upstream site of Babeltrace for the distribution of tarballs changed from lttng.org to efficios.com (see
<http://diamon.org/babeltrace>).
Also, the package has changed name from "lttng-babeltrace" to just "babeltrace". Maybe we could reflect this change in Buildroot as well?
If so, do not forget to add an entry in Config.in.legacy.
Quoted 2 lines
I'm removing the host package build, as users can get Babeltrace on major Linux distributions using their package manager.
This should be a separate patch (but see my other comment in my previous reply to patch 3/5).
Quoted 4 lines
Since Babeltrace 1.4, there is support for printing debug information along with compatible traces and event records. Babeltrace needs elfutils in this case. If elfutils exists in the BR image, then this feature is enabled.
So, we discussed this with Philippe on IRC, but elfutils is not always available. It needs glibc or uClibc, and is not availa for musl. Also, it needs shared libs and wchar.
However, lttng-babeltrace by itself does not have all of those requirements, so Philippe was wondering how to inform the user that debugging info would not be available in those cases.
I think adding a kconfig comment in that case is appropriate (see below).
Quoted 1 line
Signed-off-by Philippe Proulx <[email protected]>
[--SNIP--]
Quoted 6 lines
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch new file +0
-0
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch line 9
Quoted 9 lines
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch new file +9
-0@@ -0,0 +1,30 @@ +From d3cce60dfde30bd2fd67caa3fbc5f111131c243a Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Fri, 28 Oct 2016 01:48:25 -0400 +Subject: [PATCH] configure.ac: remove elfutils version test for Buildroot + +The AX_LIB_ELFUTILS macro builds an executable file and executes it to +discover the version of elfutils. Obviously we cannot execute a +cross-compiled program here, hence this test is removed because we know +that elfutils's version is greater than 0.154 anyway in Buildroot.
It's a pity we can't feed it with the result, with one of the ac_cv_* cache variables. :-(
Maybe you could push such a feature upstream, so we can use it at the next release?
[--SNIP--]
Quoted 6 lines
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch new file +0
-0
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch line 9
Quoted 9 lines
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch new file +9
-0@@ -0,0 +1,36 @@ +From 27e37d2621317d25e08ba0683bf8540b05b9347d Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Fri, 28 Oct 2016 02:09:20 -0400 +Subject: [PATCH] tests/lib/Makefile.am: remove unneeded -static flag + +Having those -static flags breaks the build in shared-only builds. + +Upstream does not even know why those flags are passed here, and they +are about to remove them as of this date.
Please, add a pointer to the upstream discussion about those flags.
Quoted 5 lines
package/lttng-babeltrace/Config.in +0
-0
package/lttng-babeltrace/Config.in config BR2_PACKAGE_LTTNG_BABELTRACE · line 22
Quoted 10 lines
package/lttng-babeltrace/Config.in +3
-1@@ -14,18 +14,19 @@ config BR2_PACKAGE_LTTNG_BABELTRACE to/from another trace format. Babeltrace is the reference implementation of the Common - Trace Format (see <http://diamon.org/docs/ctf/>). Babeltrace + Trace Format (see <http://diamon.org/ctf/>). Babeltrace can decode and display LTTng traces, since the latter are written using the CTF format. + The babeltrace utility is built with debug information + printing support when the elfutils package is selected.
Drop this sentence, but add a config option for it:
if BR2_PACKAGE_LTTNG_BABELTRACE
config BR2_PACKAGE_LTTNG_BABELTRACE_DEBUG_INFO
bool
default y
depends on !BR2_bfin # elfutils
depends on !BR2_STATIC_LIBS # elfutils
depends on BR2_TOOLCHAIN_USES_UCLIBC || BR2_TOOLCHAIN_USES_GLIBC # elfutils select BR2_PACKAGE_ELFUTILS
comment "handling debug info in traces needs a glibc or uClibc toolchain w/ dynamic library"
depends on !BR2_bfin # elfutils
depends on BR2_STATIC_LIBS \
|| !(BR2_TOOLCHAIN_USES_UCLIBC || BR2_TOOLCHAIN_USES_GLIBC)
endif
Then, in the .mk, see below...
package/lttng-babeltrace/Config.in config BR2_PACKAGE_LTTNG_BABELTRACE · line 27
Quoted 8 lines
package/lttng-babeltrace/Config.in +2
-4Having this tool on the target is not mandatory. You can transfer the binary trace output from the target to your - development workstation, and use babeltrace here. Buildroot - automatically builds a host babeltrace in - $(O)/host/usr/bin/babeltrace as a dependency of the - lttng-tools package. + development workstation, and use babeltrace here. Babeltrace + is packaged for all major Linux distributions.
To be adapted if you make it a host-package in the "Host utilities" sub-menu (which I'd like you do, pretty please! ;-) ).
[--SNIP--]
Quoted 15 lines
package/lttng-babeltrace/lttng-babeltrace.mk +2
-2@@ -4,18 +4,21 @@ # ################################################################################ -LTTNG_BABELTRACE_SITE = http://lttng.org/files/babeltrace -LTTNG_BABELTRACE_VERSION = 1.2.4 +LTTNG_BABELTRACE_SITE = http://www.efficios.com/files/babeltrace +LTTNG_BABELTRACE_VERSION = 1.4.0 LTTNG_BABELTRACE_SOURCE = babeltrace-$(LTTNG_BABELTRACE_VERSION).tar.bz2 LTTNG_BABELTRACE_LICENSE = MIT, LGPLv2.1 (include/babeltrace/list.h), GPLv2 (test code) LTTNG_BABELTRACE_LICENSE_FILES = mit-license.txt gpl-2.0.txt LICENSE
package/lttng-babeltrace/lttng-babeltrace.mk line 17
Quoted 10 lines
package/lttng-babeltrace/lttng-babeltrace.mk +2
-4- LTTNG_BABELTRACE_DEPENDENCIES = popt util-linux libglib2 host-pkgconf -HOST_LTTNG_BABELTRACE_DEPENDENCIES = \ - host-popt host-util-linux host-libglib2 host-pkgconf -# for 0002-configure-fix-uuid-support-detection-on-static-build.patch +# For 0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch LTTNG_BABELTRACE_AUTORECONF = YES +ifeq ($(BR2_PACKAGE_ELFUTILS),y)
Here, use:
ifeq ($(BR2_PACKAGE_LTTNG_BABELTRACE_DEBUG_INFO),y)
package/lttng-babeltrace/lttng-babeltrace.mk line 21
Quoted 8 lines
package/lttng-babeltrace/lttng-babeltrace.mk +6
-1+LTTNG_BABELTRACE_DEPENDENCIES += elfutils +LTTNG_BABELTRACE_CONF_OPTS += --enable-debug-info +else +LTTNG_BABELTRACE_CONF_OPTS += --disable-debug-info +endif + $(eval $(autotools-package)) -$(eval $(host-autotools-package))
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
PATCH v2 5/5 lttng-babeltrace: bump to version 1.4.0
Quoted 8 lines · Yann E. MORIN
Philippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 5 lines
Current patches are dropped because they were merged upstream.
Official Babeltrace's website changed from
<http://www.efficios.com/babeltrace> to <http://diamon.org/babeltrace>.
Quoted 8 lines
Quoted 3 lines
Official upstream site of Babeltrace for the distribution of tarballs changed from lttng.org to efficios.com (see
<http://diamon.org/babeltrace>).Also, the package has changed name from "lttng-babeltrace" to just "babeltrace". Maybe we could reflect this change in Buildroot as well?
If so, do not forget to add an entry in Config.in.legacy.
I remember trying to do this once, and you guys nacked me because "there always has been pushback from Peter against gratuitous renaming". My patch did not include an entry in Config.in.legacy, however.
If you're willing, I can rename it in another patch after bumping to v1.4.0.
Quoted 35 lines
Quoted 3 lines
I'm removing the host package build, as users can get Babeltrace on major Linux distributions using their package manager.
This should be a separate patch (but see my other comment in my previous reply to patch 3/5).
Quoted 4 lines
Since Babeltrace 1.4, there is support for printing debug information along with compatible traces and event records. Babeltrace needs elfutils in this case. If elfutils exists in the BR image, then this feature is enabled.
So, we discussed this with Philippe on IRC, but elfutils is not always available. It needs glibc or uClibc, and is not availa for musl. Also, it needs shared libs and wchar.
However, lttng-babeltrace by itself does not have all of those requirements, so Philippe was wondering how to inform the user that debugging info would not be available in those cases.
I think adding a kconfig comment in that case is appropriate (see below).
Quoted 1 line
Signed-off-by Philippe Proulx <[email protected]>[--SNIP--]
Quoted 10 lines
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch new file +4
-0@@ -0,0 +1,30 @@ +From d3cce60dfde30bd2fd67caa3fbc5f111131c243a Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Fri, 28 Oct 2016 01:48:25 -0400 +Subject: [PATCH] configure.ac: remove elfutils version test for Buildroot
Quoted 8 lines
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch line 9
Quoted 5 lines
package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch new file +5
-0+ +The AX_LIB_ELFUTILS macro builds an executable file and executes it to +discover the version of elfutils. Obviously we cannot execute a +cross-compiled program here, hence this test is removed because we know +that elfutils's version is greater than 0.154 anyway in Buildroot.It's a pity we can't feed it with the result, with one of the ac_cv_* cache variables. :-(
Yes it scks.
Quoted 3 lines
Maybe you could push such a feature upstream, so we can use it at the next release?
I will. I'll make sure the next BR release does not need to patch any of the LTTng/Babeltrace projects.
Quoted 11 lines
[--SNIP--]
Quoted 9 lines
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch new file +3
-0@@ -0,0 +1,36 @@ +From 27e37d2621317d25e08ba0683bf8540b05b9347d Mon Sep 17 00:00:00 2001 +From: Philippe Proulx <[email protected]> +Date: Fri, 28 Oct 2016 02:09:20 -0400
Quoted 8 lines
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch line 9
Quoted 6 lines
package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch new file +6
-0+Subject: [PATCH] tests/lib/Makefile.am: remove unneeded -static flag + +Having those -static flags breaks the build in shared-only builds. + +Upstream does not even know why those flags are passed here, and they +are about to remove them as of this date.Please, add a pointer to the upstream discussion about those flags.
It happened on IRC but I will create a PR and point to it.
Quoted 19 lines
Quoted 6 lines
package/lttng-babeltrace/Config.in +0
-0package/lttng-babeltrace/Config.in config BR2_PACKAGE_LTTNG_BABELTRACE · line 22
Quoted 10 lines
package/lttng-babeltrace/Config.in +3
-1@@ -14,18 +14,19 @@ config BR2_PACKAGE_LTTNG_BABELTRACE to/from another trace format. Babeltrace is the reference implementation of the Common - Trace Format (see <http://diamon.org/docs/ctf/>). Babeltrace + Trace Format (see <http://diamon.org/ctf/>). Babeltrace can decode and display LTTng traces, since the latter are written using the CTF format. + The babeltrace utility is built with debug information + printing support when the elfutils package is selected.Drop this sentence, but add a config option for it:
Quoted 8 lines
if
BR2_PACKAGE_LTTNG_BABELTRACEconfig
BR2_PACKAGE_LTTNG_BABELTRACE_DEBUG_INFO
bool
default y
depends on !BR2_bfin # elfutils
depends on !BR2_STATIC_LIBS # elfutils
depends onBR2_TOOLCHAIN_USES_UCLIBC||BR2_TOOLCHAIN_USES_GLIBC# elfutils
How are we supposed to track the dependencies of elfutils in all the projects that need it? Copying the depends lines like this seems precarious.
Quoted 20 lines
select
BR2_PACKAGE_ELFUTILScomment "handling debug info in traces needs a glibc or uClibc toolchain w/ dynamic library"
depends on !BR2_bfin # elfutils
depends onBR2_STATIC_LIBS\
|| !(BR2_TOOLCHAIN_USES_UCLIBC ||BR2_TOOLCHAIN_USES_GLIBC)endif
Then, in the .mk, see below...
package/lttng-babeltrace/Config.in config BR2_PACKAGE_LTTNG_BABELTRACE · line 27
Quoted 8 lines
package/lttng-babeltrace/Config.in +2
-4Having this tool on the target is not mandatory. You can transfer the binary trace output from the target to your - development workstation, and use babeltrace here. Buildroot - automatically builds a host babeltrace in - $(O)/host/usr/bin/babeltrace as a dependency of the - lttng-tools package. + development workstation, and use babeltrace here. Babeltrace + is packaged for all major Linux distributions.
Quoted 2 lines
To be adapted if you make it a host-package in the "Host utilities" sub-menu (which I'd like you do, pretty please! ;-) ).
Will do.
Quoted 26 lines
[--SNIP--]
Quoted 24 lines
package/lttng-babeltrace/lttng-babeltrace.mk +3
-6@@ -4,18 +4,21 @@ # ################################################################################ -LTTNG_BABELTRACE_SITE = http://lttng.org/files/babeltrace -LTTNG_BABELTRACE_VERSION = 1.2.4 +LTTNG_BABELTRACE_SITE = http://www.efficios.com/files/babeltrace +LTTNG_BABELTRACE_VERSION = 1.4.0 LTTNG_BABELTRACE_SOURCE = babeltrace-$(LTTNG_BABELTRACE_VERSION).tar.bz2 LTTNG_BABELTRACE_LICENSE = MIT, LGPLv2.1 (include/babeltrace/list.h), GPLv2 (test code) LTTNG_BABELTRACE_LICENSE_FILES = mit-license.txt gpl-2.0.txt LICENSE - LTTNG_BABELTRACE_DEPENDENCIES = popt util-linux libglib2 host-pkgconf -HOST_LTTNG_BABELTRACE_DEPENDENCIES = \ - host-popt host-util-linux host-libglib2 host-pkgconf -# for 0002-configure-fix-uuid-support-detection-on-static-build.patch +# For 0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch LTTNG_BABELTRACE_AUTORECONF = YES
Quoted 5 lines
package/lttng-babeltrace/lttng-babeltrace.mk line 17
Quoted 1 line
package/lttng-babeltrace/lttng-babeltrace.mk +1
-0+ifeq ($(BR2_PACKAGE_ELFUTILS),y)Here, use:
ifeq ($(BR2_PACKAGE_LTTNG_BABELTRACE_DEBUG_INFO),y)
Yes, that was my first solution. I changed it to the depends before submitting v1 because I didn't have such a complex BR2_PACKAGE_LTTNG_BABELTRACE_DEBUG_INFO block.
Thanks for the tips,
Phil
Quoted 20 lines
package/lttng-babeltrace/lttng-babeltrace.mk line 21
Quoted 9 lines
package/lttng-babeltrace/lttng-babeltrace.mk +6
-1+LTTNG_BABELTRACE_DEPENDENCIES += elfutils +LTTNG_BABELTRACE_CONF_OPTS += --enable-debug-info +else +LTTNG_BABELTRACE_CONF_OPTS += --disable-debug-info +endif + $(eval $(autotools-package)) -$(eval $(host-autotools-package))Regards,
Yann E. MORIN.--
.-----------------.--------------------.------------------.--------------------.| Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. |'------------------------------^-------^------------------^--------------------'
v2 cover Update LTTng packages to 2.8.x and Babeltrace to 1.4.0
In reply to Philippe ProulxPhilippe, All,
On 2016-10-30 15:56 -0400, Philippe Proulx spake thusly:
Quoted 5 lines
See individual patches for changes.
Tell me if it's okay to put patches found on the mailing list and from the package source of other distributions as is, only with the file names changed.
When you carry patches from upstream, or patches already submitted upstream (e.g. in a PR or a mailing list), state so in the patch. For example:
[PATCH] foo/bar: blablabla
This blablabla the foo-bar.
If it was from a mailing list, add a URL to the list archive with the patch.
Use your own imagination on how to refer to other upstream locations, like github PRs and such. ;-)
Also, as Thomas already replied, try not to have the patch number in the titles, because the paches we cary in Buildroot may change numbering (e.g. we only carry patches 2 and 5 of a 7-patch series). So, the original numbers do not make sense.
Regards,
Yann E. MORIN.
Quoted 45 lines
Philippe Proulx (5):
lttng-libust: bump to version 2.8.1
lttng-tools: bump to version 2.8.2
lttng-tools: remove dependency on host-lttng-babeltrace
lttng-modules: bump to version 2.8.3
lttng-babeltrace: bump to version 1.4.0...c-remove-elfutils-version-test-for-Buildr.patch | 30 ++ .../0001-no-posix-fallocate-in-uclibc.patch | 37 -- ...ix-uuid-support-detection-on-static-build.patch | 63 --- ...b-Makefile.am-remove-unneeded-static-flag.patch | 36 ++ package/lttng-babeltrace/Config.in | 13 +- package/lttng-babeltrace/lttng-babeltrace.hash | 11 +- package/lttng-babeltrace/lttng-babeltrace.mk | 17 +- ...lf.c-define-NT_GNU_BUILD_ID-if-not-define.patch | 31 ++ package/lttng-libust/lttng-libust.hash | 12 +- package/lttng-libust/lttng-libust.mk | 10 +- package/lttng-modules/lttng-modules.hash | 8 +- package/lttng-modules/lttng-modules.mk | 27 +- ...ld-failure-when-__GLIBC_PREREQ-is-missing.patch | 55 --- ...Fix-snapshot-del-output-with-name-on-musl.patch | 28 ++ .../0002-Fix-add-missing-sys-types.h-header.patch | 29 -- ...Fix-strerror_r-behavior-is-glibc-specific.patch | 26 ++ ...et-thread-stack-size-to-ulimit-soft-value.patch | 427 +++++++++++++++++++++ package/lttng-tools/lttng-tools.hash | 8 +- package/lttng-tools/lttng-tools.mk | 17 +- 19 files changed, 654 insertions(+), 231 deletions(-) create mode 100644 package/lttng-babeltrace/0001-configure.ac-remove-elfutils-version-test-for-Buildr.patch delete mode 100644 package/lttng-babeltrace/0001-no-posix-fallocate-in-uclibc.patch delete mode 100644 package/lttng-babeltrace/0002-configure-fix-uuid-support-detection-on-static-build.patch create mode 100644 package/lttng-babeltrace/0002-tests-lib-Makefile.am-remove-unneeded-static-flag.patch create mode 100644 package/lttng-libust/0001-lttng-ust-elf.c-define-NT_GNU_BUILD_ID-if-not-define.patch delete mode 100644 package/lttng-tools/0001-Fix-build-failure-when-__GLIBC_PREREQ-is-missing.patch create mode 100644 package/lttng-tools/0001-Fix-snapshot-del-output-with-name-on-musl.patch delete mode 100644 package/lttng-tools/0002-Fix-add-missing-sys-types.h-header.patch create mode 100644 package/lttng-tools/0002-Fix-strerror_r-behavior-is-glibc-specific.patch create mode 100644 package/lttng-tools/0003-Set-thread-stack-size-to-ulimit-soft-value.patch--
2.9.3_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'