* [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native
@ 2019-12-17 8:02 Mike Crowe
2019-12-17 8:56 ` ChenQi
0 siblings, 1 reply; 5+ messages in thread
From: Mike Crowe @ 2019-12-17 8:02 UTC (permalink / raw)
To: openembedded-core; +Cc: Mike Crowe
Ensure that RECIPE_SYSROOT is the same for -native recipes whether
multilib.conf is included or not.
Without this change task signatures for -native recipes change when
switching between MACHINEs that require multilib.conf and those that
don't.
This fix was one of the ones suggested by Khem Raj in
http://lists.openembedded.org/pipermail/openembedded-core/2019-December/290303.html
Add test_sstate_multilib_or_not_native_samesigs test case to
sstatetests.py to ensure that this stays fixed.
Signed-off-by: Mike Crowe <mac@mcrowe.com>
---
meta/conf/multilib.conf | 1 +
meta/lib/oeqa/selftest/cases/sstatetests.py | 40 +++++++++++++++++++++
2 files changed, 41 insertions(+)
diff --git a/meta/conf/multilib.conf b/meta/conf/multilib.conf
index cfed3fbbd0..58f2ac5c86 100644
--- a/meta/conf/multilib.conf
+++ b/meta/conf/multilib.conf
@@ -9,6 +9,7 @@ MULTILIBS ??= "multilib:lib32"
STAGING_DIR_HOST = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
STAGING_DIR_TARGET = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
RECIPE_SYSROOT = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
+RECIPE_SYSROOT_class-native = "${WORKDIR}/recipe-sysroot"
INHERIT += "multilib_global"
diff --git a/meta/lib/oeqa/selftest/cases/sstatetests.py b/meta/lib/oeqa/selftest/cases/sstatetests.py
index 6757a0ec68..9adb511960 100644
--- a/meta/lib/oeqa/selftest/cases/sstatetests.py
+++ b/meta/lib/oeqa/selftest/cases/sstatetests.py
@@ -446,6 +446,46 @@ BB_SIGNATURE_HANDLER = "OEBasicHash"
self.assertCountEqual(files1, files2)
+ def test_sstate_multilib_or_not_native_samesigs(self):
+ """The sstate checksums of two native recipes (and their dependencies)
+ where the target is using multilib in one but not the other
+ should be the same. We use the qemux86copy machine to test
+ this.
+ """
+
+ self.write_config("""
+TMPDIR = \"${TOPDIR}/tmp-sstatesamehash\"
+TCLIBCAPPEND = \"\"
+MACHINE = \"qemux86\"
+require conf/multilib.conf
+MULTILIBS = "multilib:lib32"
+DEFAULTTUNE_virtclass-multilib-lib32 = "x86"
+BB_SIGNATURE_HANDLER = "OEBasicHash"
+""")
+ self.track_for_cleanup(self.topdir + "/tmp-sstatesamehash")
+ bitbake("binutils-native -S none")
+ self.write_config("""
+TMPDIR = \"${TOPDIR}/tmp-sstatesamehash2\"
+TCLIBCAPPEND = \"\"
+MACHINE = \"qemux86copy\"
+BB_SIGNATURE_HANDLER = "OEBasicHash"
+""")
+ self.track_for_cleanup(self.topdir + "/tmp-sstatesamehash2")
+ bitbake("binutils-native -S none")
+
+ def get_files(d):
+ f = []
+ for root, dirs, files in os.walk(d):
+ for name in files:
+ f.append(os.path.join(root, name))
+ return f
+ files1 = get_files(self.topdir + "/tmp-sstatesamehash/stamps")
+ files2 = get_files(self.topdir + "/tmp-sstatesamehash2/stamps")
+ files2 = [x.replace("tmp-sstatesamehash2", "tmp-sstatesamehash") for x in files2]
+ self.maxDiff = None
+ self.assertCountEqual(files1, files2)
+
+
def test_sstate_noop_samesigs(self):
"""
The sstate checksums of two builds with these variables changed or
--
2.24.0
^ permalink raw reply related [flat|nested] 5+ messages in thread* Re: [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native
2019-12-17 8:02 [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native Mike Crowe
@ 2019-12-17 8:56 ` ChenQi
2019-12-17 13:55 ` Mike Crowe
0 siblings, 1 reply; 5+ messages in thread
From: ChenQi @ 2019-12-17 8:56 UTC (permalink / raw)
To: openembedded-core
On 12/17/2019 04:02 PM, Mike Crowe via Openembedded-core wrote:
> Ensure that RECIPE_SYSROOT is the same for -native recipes whether
> multilib.conf is included or not.
>
> Without this change task signatures for -native recipes change when
> switching between MACHINEs that require multilib.conf and those that
> don't.
>
> This fix was one of the ones suggested by Khem Raj in
> http://lists.openembedded.org/pipermail/openembedded-core/2019-December/290303.html
>
> Add test_sstate_multilib_or_not_native_samesigs test case to
> sstatetests.py to ensure that this stays fixed.
>
> Signed-off-by: Mike Crowe <mac@mcrowe.com>
> ---
> meta/conf/multilib.conf | 1 +
> meta/lib/oeqa/selftest/cases/sstatetests.py | 40 +++++++++++++++++++++
> 2 files changed, 41 insertions(+)
>
> diff --git a/meta/conf/multilib.conf b/meta/conf/multilib.conf
> index cfed3fbbd0..58f2ac5c86 100644
> --- a/meta/conf/multilib.conf
> +++ b/meta/conf/multilib.conf
> @@ -9,6 +9,7 @@ MULTILIBS ??= "multilib:lib32"
> STAGING_DIR_HOST = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> STAGING_DIR_TARGET = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> RECIPE_SYSROOT = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> +RECIPE_SYSROOT_class-native = "${WORKDIR}/recipe-sysroot"
How about just removing MLPREFIX?
Regards,
Chen Qi
>
> INHERIT += "multilib_global"
>
> diff --git a/meta/lib/oeqa/selftest/cases/sstatetests.py b/meta/lib/oeqa/selftest/cases/sstatetests.py
> index 6757a0ec68..9adb511960 100644
> --- a/meta/lib/oeqa/selftest/cases/sstatetests.py
> +++ b/meta/lib/oeqa/selftest/cases/sstatetests.py
> @@ -446,6 +446,46 @@ BB_SIGNATURE_HANDLER = "OEBasicHash"
> self.assertCountEqual(files1, files2)
>
>
> + def test_sstate_multilib_or_not_native_samesigs(self):
> + """The sstate checksums of two native recipes (and their dependencies)
> + where the target is using multilib in one but not the other
> + should be the same. We use the qemux86copy machine to test
> + this.
> + """
> +
> + self.write_config("""
> +TMPDIR = \"${TOPDIR}/tmp-sstatesamehash\"
> +TCLIBCAPPEND = \"\"
> +MACHINE = \"qemux86\"
> +require conf/multilib.conf
> +MULTILIBS = "multilib:lib32"
> +DEFAULTTUNE_virtclass-multilib-lib32 = "x86"
> +BB_SIGNATURE_HANDLER = "OEBasicHash"
> +""")
> + self.track_for_cleanup(self.topdir + "/tmp-sstatesamehash")
> + bitbake("binutils-native -S none")
> + self.write_config("""
> +TMPDIR = \"${TOPDIR}/tmp-sstatesamehash2\"
> +TCLIBCAPPEND = \"\"
> +MACHINE = \"qemux86copy\"
> +BB_SIGNATURE_HANDLER = "OEBasicHash"
> +""")
> + self.track_for_cleanup(self.topdir + "/tmp-sstatesamehash2")
> + bitbake("binutils-native -S none")
> +
> + def get_files(d):
> + f = []
> + for root, dirs, files in os.walk(d):
> + for name in files:
> + f.append(os.path.join(root, name))
> + return f
> + files1 = get_files(self.topdir + "/tmp-sstatesamehash/stamps")
> + files2 = get_files(self.topdir + "/tmp-sstatesamehash2/stamps")
> + files2 = [x.replace("tmp-sstatesamehash2", "tmp-sstatesamehash") for x in files2]
> + self.maxDiff = None
> + self.assertCountEqual(files1, files2)
> +
> +
> def test_sstate_noop_samesigs(self):
> """
> The sstate checksums of two builds with these variables changed or
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native
2019-12-17 8:56 ` ChenQi
@ 2019-12-17 13:55 ` Mike Crowe
2019-12-18 1:19 ` ChenQi
0 siblings, 1 reply; 5+ messages in thread
From: Mike Crowe @ 2019-12-17 13:55 UTC (permalink / raw)
To: ChenQi; +Cc: openembedded-core
On Tuesday 17 December 2019 at 16:56:11 +0800, ChenQi wrote:
> On 12/17/2019 04:02 PM, Mike Crowe via Openembedded-core wrote:
> > Ensure that RECIPE_SYSROOT is the same for -native recipes whether
> > multilib.conf is included or not.
> >
> > Without this change task signatures for -native recipes change when
> > switching between MACHINEs that require multilib.conf and those that
> > don't.
> >
> > This fix was one of the ones suggested by Khem Raj in
> > http://lists.openembedded.org/pipermail/openembedded-core/2019-December/290303.html
> >
> > Add test_sstate_multilib_or_not_native_samesigs test case to
> > sstatetests.py to ensure that this stays fixed.
> >
> > Signed-off-by: Mike Crowe <mac@mcrowe.com>
> > ---
> > meta/conf/multilib.conf | 1 +
> > meta/lib/oeqa/selftest/cases/sstatetests.py | 40 +++++++++++++++++++++
> > 2 files changed, 41 insertions(+)
> >
> > diff --git a/meta/conf/multilib.conf b/meta/conf/multilib.conf
> > index cfed3fbbd0..58f2ac5c86 100644
> > --- a/meta/conf/multilib.conf
> > +++ b/meta/conf/multilib.conf
> > @@ -9,6 +9,7 @@ MULTILIBS ??= "multilib:lib32"
> > STAGING_DIR_HOST = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> > STAGING_DIR_TARGET = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> > RECIPE_SYSROOT = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> > +RECIPE_SYSROOT_class-native = "${WORKDIR}/recipe-sysroot"
>
> How about just removing MLPREFIX?
I'm afraid that I don't understand what you are suggesting. The above
change does remove MLPREFIX for -native only.
Thanks.
Mike.
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native
2019-12-17 13:55 ` Mike Crowe
@ 2019-12-18 1:19 ` ChenQi
2019-12-18 11:43 ` Mike Crowe
0 siblings, 1 reply; 5+ messages in thread
From: ChenQi @ 2019-12-18 1:19 UTC (permalink / raw)
To: Mike Crowe; +Cc: openembedded-core
On 12/17/2019 09:55 PM, Mike Crowe wrote:
> On Tuesday 17 December 2019 at 16:56:11 +0800, ChenQi wrote:
>> On 12/17/2019 04:02 PM, Mike Crowe via Openembedded-core wrote:
>>> Ensure that RECIPE_SYSROOT is the same for -native recipes whether
>>> multilib.conf is included or not.
>>>
>>> Without this change task signatures for -native recipes change when
>>> switching between MACHINEs that require multilib.conf and those that
>>> don't.
>>>
>>> This fix was one of the ones suggested by Khem Raj in
>>> http://lists.openembedded.org/pipermail/openembedded-core/2019-December/290303.html
>>>
>>> Add test_sstate_multilib_or_not_native_samesigs test case to
>>> sstatetests.py to ensure that this stays fixed.
>>>
>>> Signed-off-by: Mike Crowe <mac@mcrowe.com>
>>> ---
>>> meta/conf/multilib.conf | 1 +
>>> meta/lib/oeqa/selftest/cases/sstatetests.py | 40 +++++++++++++++++++++
>>> 2 files changed, 41 insertions(+)
>>>
>>> diff --git a/meta/conf/multilib.conf b/meta/conf/multilib.conf
>>> index cfed3fbbd0..58f2ac5c86 100644
>>> --- a/meta/conf/multilib.conf
>>> +++ b/meta/conf/multilib.conf
>>> @@ -9,6 +9,7 @@ MULTILIBS ??= "multilib:lib32"
>>> STAGING_DIR_HOST = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
>>> STAGING_DIR_TARGET = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
>>> RECIPE_SYSROOT = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
>>> +RECIPE_SYSROOT_class-native = "${WORKDIR}/recipe-sysroot"
>> How about just removing MLPREFIX?
> I'm afraid that I don't understand what you are suggesting. The above
> change does remove MLPREFIX for -native only.
>
> Thanks.
>
> Mike.
>
I mean removing all these MLPREFIX, so that STAGING_DIR_HOST,
STAGING_DIR_TARGET and RECIPE_SYSROOT will all be ${WORKDIR}/recipe-sysroot.
Best Regards,
Chen Qi
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native
2019-12-18 1:19 ` ChenQi
@ 2019-12-18 11:43 ` Mike Crowe
0 siblings, 0 replies; 5+ messages in thread
From: Mike Crowe @ 2019-12-18 11:43 UTC (permalink / raw)
To: ChenQi, openembedded-core
On Wednesday 18 December 2019 at 09:19:20 +0800, ChenQi wrote:
> On 12/17/2019 09:55 PM, Mike Crowe wrote:
> > On Tuesday 17 December 2019 at 16:56:11 +0800, ChenQi wrote:
> > > On 12/17/2019 04:02 PM, Mike Crowe via Openembedded-core wrote:
> > > > Ensure that RECIPE_SYSROOT is the same for -native recipes whether
> > > > multilib.conf is included or not.
> > > >
> > > > Without this change task signatures for -native recipes change when
> > > > switching between MACHINEs that require multilib.conf and those that
> > > > don't.
> > > >
> > > > This fix was one of the ones suggested by Khem Raj in
> > > > http://lists.openembedded.org/pipermail/openembedded-core/2019-December/290303.html
> > > >
> > > > Add test_sstate_multilib_or_not_native_samesigs test case to
> > > > sstatetests.py to ensure that this stays fixed.
> > > >
> > > > Signed-off-by: Mike Crowe <mac@mcrowe.com>
> > > > ---
> > > > meta/conf/multilib.conf | 1 +
> > > > meta/lib/oeqa/selftest/cases/sstatetests.py | 40 +++++++++++++++++++++
> > > > 2 files changed, 41 insertions(+)
> > > >
> > > > diff --git a/meta/conf/multilib.conf b/meta/conf/multilib.conf
> > > > index cfed3fbbd0..58f2ac5c86 100644
> > > > --- a/meta/conf/multilib.conf
> > > > +++ b/meta/conf/multilib.conf
> > > > @@ -9,6 +9,7 @@ MULTILIBS ??= "multilib:lib32"
> > > > STAGING_DIR_HOST = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> > > > STAGING_DIR_TARGET = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> > > > RECIPE_SYSROOT = "${WORKDIR}/${MLPREFIX}recipe-sysroot"
> > > > +RECIPE_SYSROOT_class-native = "${WORKDIR}/recipe-sysroot"
> > > How about just removing MLPREFIX?
> >
> > I'm afraid that I don't understand what you are suggesting. The above
> > change does remove MLPREFIX for -native only.
>
> I mean removing all these MLPREFIX, so that STAGING_DIR_HOST,
> STAGING_DIR_TARGET and RECIPE_SYSROOT will all be
> ${WORKDIR}/recipe-sysroot.
The prefix is necessary for RECIPE_SYSROOT because without it a single
recipe can't depend on multiple multilib variants of the same recipe. We do
this because our bootloader requires both a 32-bit and a 64-bit compiler.
When I changed RECIPE_SYSROOT to always be "${WORKDIR}/recipe-sysroot" I
get failures like:
do_prepare_recipe_sysroot: The file /usr/include/mtd/mtd-user.h is installed by both linux-libc-headers and lib32-linux-libc-headers, aborting
Thanks.
Mike.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2019-12-18 11:43 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2019-12-17 8:02 [PATCH] multilib.conf: Ensure that RECIPE_SYSROOT is unchanged for native Mike Crowe
2019-12-17 8:56 ` ChenQi
2019-12-17 13:55 ` Mike Crowe
2019-12-18 1:19 ` ChenQi
2019-12-18 11:43 ` Mike Crowe
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox