[kernel-tests] 02/03: Test for nvidia modules building against current kernel
This is an automated email from the git hooks/post-receive script. jforbes pushed a commit to branch master in repository kernel-tests. commit 0cedd6705a22bb62049587b4a92be87b6c3b7299 Author: Justin M. Forbes Date: Tue Feb 28 16:51:20 2017 -0600 Test for nvidia modules building against current kernel --- thirdparty/nvidia-module/runtest.sh | 34 ++ 1 file changed, 34 insertions(+) diff --git a/thirdparty/nvidia-module/runtest.sh b/thirdparty/nvidia-module/runtest.sh new file mode 100755 index 000..b00a663 --- /dev/null +++ b/thirdparty/nvidia-module/runtest.sh @@ -0,0 +1,34 @@ +#!/bin/bash + +source ../../utils/root-check.sh +source ../../utils/build-deps.sh + +check_root +is_root=$? +if [ $is_root -ne 0 ]; then + exit 3 +fi + +# Ensure we have curl and kernel-devel +check_dep curl +check_dep kernel-devel + +baseurl='http://http.download.nvidia.com/XFree86/Linux-x86_64' +latest=`curl $baseurl/latest.txt | cut -d ' ' -f1` +installer="NVIDIA-Linux-x86_64-$latest" + + +if [ ! -f "$installer" ] +then + curl -O "$baseurl/$latest/$installer.run" + [ $? -eq 0 ] || exit 3 +fi + +sh $installer.run --extract-only --target tmp +pushd tmp/kernel +make modules > build.log +[ $? -eq 0 ] || exit 4 +popd +rm -rf tmp + +echo module version $latest built for `uname -r` -- To stop receiving notification emails like this one, please contact the administrator of this repository. ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
[kernel-tests] 03/03: Change install_dep to use dnf instead of yum
This is an automated email from the git hooks/post-receive script. jforbes pushed a commit to branch master in repository kernel-tests. commit 250f5b1a3bea2700ebf21633f59bef564f44ff11 Author: Justin M. Forbes Date: Tue Feb 28 16:52:13 2017 -0600 Change install_dep to use dnf instead of yum --- utils/build-deps.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/utils/build-deps.sh b/utils/build-deps.sh index d85eece..7608c9a 100644 --- a/utils/build-deps.sh +++ b/utils/build-deps.sh @@ -3,7 +3,7 @@ # Licensed under the terms of the GNU GPL License version 2 install_dep() { - yum install -y $1; + dnf install -y $1; rc=$?; if [ $rc -ne 0 ] then -- To stop receiving notification emails like this one, please contact the administrator of this repository. ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
[kernel-tests] 01/03: Add support for third party module tests
This is an automated email from the git hooks/post-receive script. jforbes pushed a commit to branch master in repository kernel-tests. commit 8f2ea1b2b3eb2660bd85bd7d3b50ba045663a3af Author: Justin M. Forbes Date: Tue Feb 28 16:49:38 2017 -0600 Add support for third party module tests --- config.example| 3 +++ documentation.txt | 20 +--- runtests.sh | 22 +- 3 files changed, 41 insertions(+), 4 deletions(-) diff --git a/config.example b/config.example index dd76142..6ae342e 100644 --- a/config.example +++ b/config.example @@ -14,6 +14,9 @@ submit=none # checksig=y # validsig="Fedora Secure Boot Signer" +# Test 3rd Party Modules +# thirdparty=y + # FAS User credentials. # Storing your FAS password here is technically possible, but not advisable # for security seasons diff --git a/documentation.txt b/documentation.txt index 367d639..5f8c985 100644 --- a/documentation.txt +++ b/documentation.txt @@ -30,17 +30,24 @@ It is expected that a basic set of packages are installed to run these tests. This includes glibc-devel and gcc. If those packages are not installed, please install them before running the test suite. +If you wish to test to ensure that 3rd party modules build against the current +kernel, you can add a 'thirdparty=y' line to your .config. This will run any +tests in the thirdparty directory as well. Because these are not upstream +drivers, a failure of these tests will return 4, the test suite will pass but +with warning. + === Writing Tests === While a test can actually be any sort of executable, it is expected that these tests will follow certain basic criteria. This helps to ensure that the test suite is easy to interpret. The output is controlled by the -master script, and output is in the form of pass, fail, or skipped. All -other output is redirected to the log file. +master script, and output is in the form of pass, fail, warn, or skipped. +All other output is redirected to the log file. Return Codes: 0 - A successful test completion 3 - Test skipped +4 - Warn (this is reserved for things like out of tree modules) Anything else is interpreted as fail and the user is asked to check the log for more details. @@ -84,7 +91,6 @@ any necessary source, checking for any specific dependencies, and skipping if they are not met. At the completion of the test set, a "Test suite complete" is printed with a pass/fail result, and the appropriate return code. - Potential for harm: It is expected that these test will be run on real systems. Any tests which have increased risk of data loss or ill effects should be specified @@ -98,3 +104,11 @@ As a large number of tests are written as simple shell scripts, and many of these tests need to perform a series of the same functions, a "library" has been created to allow for reuse. source the testutil file as needed. Any functions added to testutil should be clearly commented with purpose and use. + +Thirdparty: +This directory should contain tests for out of tree drivers etc. These tests +should never return anything other than pass, skip, or warn. While it is +handy to know if these things work with the current kernel, as out of tree +modules, they are not necessarily in-step with upstream development. To +return a fail on these tests would be incorrect, but a warn does give a heads +up so that the upstream for those modules can be contacted. diff --git a/runtests.sh b/runtests.sh index f3376df..8820c2b 100755 --- a/runtests.sh +++ b/runtests.sh @@ -9,9 +9,11 @@ verbose=n testset=default cleanrun=PASS failedtests=None +warntests=None commit=n commithook=/usr/bin/true disable_retest=n +thirdparty=n if [ -f ./.config ]; then source ./.config @@ -104,6 +106,11 @@ if [ "$checksig" == "y" ]; then dirlist="secureboot $dirlist" fi +# Test Third Party Modules? +if [ "$thirdparty" == "y" ]; then +dirlist="$dirlist thirdparty" +fi + #Basic logfile headers echo "Date: $(date)" > $logfile echo "Test set: $testset" >> $logfile @@ -142,6 +149,9 @@ do 3) result=SKIP ;; + 4) + result=WARN + ;; *) result=FAIL esac @@ -154,12 +164,22 @@ do failedtests="$failedtests $testname" fi fi + if [ "$result" == "WARN" ]; then + if [ "$cleanrun" == "PASS" ]; then + cleanrun=WARN + fi + if [ "$warntests" == "None" ]; then + warntests="$testname" + else + warntests="$warntests $testname" + fi + fi popd &>/dev/null done done # Fix up logfile headers -sed -i
Re: [PATCH, kernel.spec] apply patches with build tweaks and compile fixes all the time
On 02/28/2017 03:20 AM, Thorsten Leemhuis wrote: > Hi! Find below a patch against master that slightly changes how patches are > applied in kernel.spec. It adds a section in the patch definition area of the > spec file where compile fixes can be added, to make sure they even get > applied even in the nopatch/vanilla build path. It also gets rid of the > SOURCE-approach for kbuild-AFTER_LINK.patch and makes it again a proper patch > in the spec file. > > This is a follow up to the discussion we had in > https://lists.fedoraproject.org/archives/list/kernel@lists.fedoraproject.org/thread/RRFMNKRW5JYYCDE43AJ4Y3U3JAKG3XN5/#37D7ID7XPKOCW332S4QYKDAPA2AVDNKK > I didn't consider it an important issue, but I got annoyed enough to get back > to this when yet another compile fix hit the spec file yesterday. > > FWIW: > * 0001-Work-around-for-gcc7-and-arm64.patch is strictly only needed on >=f26, > but for now it is simply applied everywhere. We can simply add a "%if > 0%{?fedora} >= 26" "%endif" around the patch definition > * I didn't check, but 0001-iio-Use-event-header-from-kernel-tree.patch might > only be needed on <=f25, as the headers in f26 are likely new enough by now > > CU, thl > Thanks for this. I applied it to rawhide. > commit 224aa0a6c443020f1ddd1a23fed4319a4be008ba > Author: Thorsten Leemhuis > Date: Tue Feb 28 12:01:38 2017 +0100 > > apply patches with build tweaks (build-AFTER-LINK.patch) and compile fixes > all the time > > diff --git a/kernel.spec b/kernel.spec > index 247f0d1..c6f3a15 100644 > --- a/kernel.spec > +++ b/kernel.spec > @@ -487,11 +487,23 @@ Source5000: patch-4.%{base_sublevel}-git%{gitrev}.xz > %endif > %endif > > +## patches needed for building this package > + > # build tweak for build ID magic, even for -vanilla > -Source5005: kbuild-AFTER_LINK.patch > +Patch001: kbuild-AFTER_LINK.patch > > -%if !%{nopatches} > +## compile fixes > + > +# ongoing complaint, full discussion delayed until ksummit/plumbers > +Patch849: 0001-iio-Use-event-header-from-kernel-tree.patch > + > +# Still other instances of this bug floating around > +Patch853: 0001-Work-around-for-gcc7-and-arm64.patch > > +# compile error > +Patch854: 0001-mm-x86-fix-HIGHMEM64-PARAVIRT-build-config-for-nativ.patch > + > +%if !%{nopatches} > # Git trees. > > # Standalone patches > @@ -583,15 +595,6 @@ Patch509: > MODSIGN-Don-t-try-secure-boot-if-EFI-runtime-is-disa.patch > #CVE-2016-3134 rhbz 1317383 1317384 > Patch665: netfilter-x_tables-deal-with-bogus-nextoffset-values.patch > > -#ongoing complaint, full discussion delayed until ksummit/plumbers > -Patch849: 0001-iio-Use-event-header-from-kernel-tree.patch > - > -# Still other instances of this bug floating around > -Patch853: 0001-Work-around-for-gcc7-and-arm64.patch > - > -# compile error > -Patch854: 0001-mm-x86-fix-HIGHMEM64-PARAVIRT-build-config-for-nativ.patch > - > # END OF PATCH DEFINITIONS > > %endif > @@ -1192,18 +1195,12 @@ do > done > %endif > > -# The kbuild-AFTER_LINK patch is needed regardless so we list it as a Source > -# file and apply it separately from the rest. > -git am %{SOURCE5005} > - > -%if !%{nopatches} > - > +# Note: Even in the "nopatches" path some patches (build tweaks and compile > +# fixes) will always get applied; see patch defition above for details > git am %{patches} > > # END OF PATCH APPLICATIONS > > -%endif > - > # Any further pre-build tree manipulations happen here. > > chmod +x scripts/checkpatch.pl > @@ -2162,6 +2159,10 @@ fi > # > # > %changelog > +* Tue Feb 28 2017 Thorsten Leemhuis - > 4.11.0-0.rc0.git5.2 > +- apply patches with build tweaks (build-AFTER-LINK.patch) and compile fixes > + all the time > + > * Mon Feb 27 2017 Laura Abbott - > 4.11.0-0.rc0.git5.1 > - Linux v4.10-10320-ge5d56ef > - Disable a series of s390x configuration options > ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Re: [PATCH] Drop kernel-devel virtual provide rhbz#1420754
On Tue, Feb 28, 2017 at 6:38 AM, Josh Boyer wrote: > On Tue, Feb 28, 2017 at 4:02 AM, Nicolas Chauvet > wrote: > > 2017-02-16 19:33 GMT+01:00 Nicolas Chauvet : > >> 2017-02-16 19:26 GMT+01:00 Nicolas Chauvet : > >>> --- > >>> kernel.spec | 1 - > >>> 1 file changed, 1 deletion(-) > >>> > >>> diff --git a/kernel.spec b/kernel.spec > >>> index 4363050..38968ba 100644 > >>> --- a/kernel.spec > >>> +++ b/kernel.spec > >>> @@ -815,7 +815,6 @@ Summary: Development package for building kernel > modules to match the %{?2:%{2} > >>> Group: System Environment/Kernel\ > >>> Provides: kernel%{?1:-%{1}}-devel-%{_target_cpu} = > %{version}-%{release}\ > >>> Provides: kernel-devel-%{_target_cpu} = %{version}-%{release}%{?1:+%{ > 1}}\ > >>> -Provides: kernel-devel = %{version}-%{release}%{?1:+%{1}}\ > >>> Provides: kernel-devel-uname-r = %{KVERREL}%{?variant}%{?1:+%{1}}\ > >>> Provides: installonlypkg(kernel)\ > >>> AutoReqProv: no\ > >>> -- > >>> 2.7.4 > >>> > >> > >> So, this was the "light description version" of the patch. > >> please see a full rationale here http://bugzilla.redhat.com/1420754 > >> Basically, this patch make the situation in sync with el7 kernel WRT > >> theses provides. > >> There is one virtual provides kernel-devel-uname-r > >> Other are real (packages) provides. > >> > >> That will restore the ability to prefer one kernel-devel varriant over > >> another one as (such as kernel-debug-devel). > >> In the current situation every kernel-devel varriant have this > >> kernel-devel (virtual) provide. > > > > Any answer from the kernel team about this ? > Sorry this was put off, this has been pushed to all branches and should go out with the next builds. Thanks, Justin ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Re: [PATCH] Drop kernel-devel virtual provide rhbz#1420754
On Tue, Feb 28, 2017 at 4:02 AM, Nicolas Chauvet wrote: > 2017-02-16 19:33 GMT+01:00 Nicolas Chauvet : >> 2017-02-16 19:26 GMT+01:00 Nicolas Chauvet : >>> --- >>> kernel.spec | 1 - >>> 1 file changed, 1 deletion(-) >>> >>> diff --git a/kernel.spec b/kernel.spec >>> index 4363050..38968ba 100644 >>> --- a/kernel.spec >>> +++ b/kernel.spec >>> @@ -815,7 +815,6 @@ Summary: Development package for building kernel >>> modules to match the %{?2:%{2} >>> Group: System Environment/Kernel\ >>> Provides: kernel%{?1:-%{1}}-devel-%{_target_cpu} = %{version}-%{release}\ >>> Provides: kernel-devel-%{_target_cpu} = %{version}-%{release}%{?1:+%{1}}\ >>> -Provides: kernel-devel = %{version}-%{release}%{?1:+%{1}}\ >>> Provides: kernel-devel-uname-r = %{KVERREL}%{?variant}%{?1:+%{1}}\ >>> Provides: installonlypkg(kernel)\ >>> AutoReqProv: no\ >>> -- >>> 2.7.4 >>> >> >> So, this was the "light description version" of the patch. >> please see a full rationale here http://bugzilla.redhat.com/1420754 >> Basically, this patch make the situation in sync with el7 kernel WRT >> theses provides. >> There is one virtual provides kernel-devel-uname-r >> Other are real (packages) provides. >> >> That will restore the ability to prefer one kernel-devel varriant over >> another one as (such as kernel-debug-devel). >> In the current situation every kernel-devel varriant have this >> kernel-devel (virtual) provide. > > Any answer from the kernel team about this ? I'm not on the kernel team any longer. However, I went digging and this was fixed 2 years ago in RHEL for the exact same reasons with the exact same fix. So I guess ACK from me. josh ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
[PATCH, kernel.spec] apply patches with build tweaks and compile fixes all the time
Hi! Find below a patch against master that slightly changes how patches are applied in kernel.spec. It adds a section in the patch definition area of the spec file where compile fixes can be added, to make sure they even get applied even in the nopatch/vanilla build path. It also gets rid of the SOURCE-approach for kbuild-AFTER_LINK.patch and makes it again a proper patch in the spec file. This is a follow up to the discussion we had in https://lists.fedoraproject.org/archives/list/kernel@lists.fedoraproject.org/thread/RRFMNKRW5JYYCDE43AJ4Y3U3JAKG3XN5/#37D7ID7XPKOCW332S4QYKDAPA2AVDNKK I didn't consider it an important issue, but I got annoyed enough to get back to this when yet another compile fix hit the spec file yesterday. FWIW: * 0001-Work-around-for-gcc7-and-arm64.patch is strictly only needed on >=f26, but for now it is simply applied everywhere. We can simply add a "%if 0%{?fedora} >= 26" "%endif" around the patch definition * I didn't check, but 0001-iio-Use-event-header-from-kernel-tree.patch might only be needed on <=f25, as the headers in f26 are likely new enough by now CU, thl commit 224aa0a6c443020f1ddd1a23fed4319a4be008ba Author: Thorsten Leemhuis Date: Tue Feb 28 12:01:38 2017 +0100 apply patches with build tweaks (build-AFTER-LINK.patch) and compile fixes all the time diff --git a/kernel.spec b/kernel.spec index 247f0d1..c6f3a15 100644 --- a/kernel.spec +++ b/kernel.spec @@ -487,11 +487,23 @@ Source5000: patch-4.%{base_sublevel}-git%{gitrev}.xz %endif %endif +## patches needed for building this package + # build tweak for build ID magic, even for -vanilla -Source5005: kbuild-AFTER_LINK.patch +Patch001: kbuild-AFTER_LINK.patch -%if !%{nopatches} +## compile fixes + +# ongoing complaint, full discussion delayed until ksummit/plumbers +Patch849: 0001-iio-Use-event-header-from-kernel-tree.patch + +# Still other instances of this bug floating around +Patch853: 0001-Work-around-for-gcc7-and-arm64.patch +# compile error +Patch854: 0001-mm-x86-fix-HIGHMEM64-PARAVIRT-build-config-for-nativ.patch + +%if !%{nopatches} # Git trees. # Standalone patches @@ -583,15 +595,6 @@ Patch509: MODSIGN-Don-t-try-secure-boot-if-EFI-runtime-is-disa.patch #CVE-2016-3134 rhbz 1317383 1317384 Patch665: netfilter-x_tables-deal-with-bogus-nextoffset-values.patch -#ongoing complaint, full discussion delayed until ksummit/plumbers -Patch849: 0001-iio-Use-event-header-from-kernel-tree.patch - -# Still other instances of this bug floating around -Patch853: 0001-Work-around-for-gcc7-and-arm64.patch - -# compile error -Patch854: 0001-mm-x86-fix-HIGHMEM64-PARAVIRT-build-config-for-nativ.patch - # END OF PATCH DEFINITIONS %endif @@ -1192,18 +1195,12 @@ do done %endif -# The kbuild-AFTER_LINK patch is needed regardless so we list it as a Source -# file and apply it separately from the rest. -git am %{SOURCE5005} - -%if !%{nopatches} - +# Note: Even in the "nopatches" path some patches (build tweaks and compile +# fixes) will always get applied; see patch defition above for details git am %{patches} # END OF PATCH APPLICATIONS -%endif - # Any further pre-build tree manipulations happen here. chmod +x scripts/checkpatch.pl @@ -2162,6 +2159,10 @@ fi # # %changelog +* Tue Feb 28 2017 Thorsten Leemhuis - 4.11.0-0.rc0.git5.2 +- apply patches with build tweaks (build-AFTER-LINK.patch) and compile fixes + all the time + * Mon Feb 27 2017 Laura Abbott - 4.11.0-0.rc0.git5.1 - Linux v4.10-10320-ge5d56ef - Disable a series of s390x configuration options ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Re: [PATCH] Drop kernel-devel virtual provide rhbz#1420754
2017-02-16 19:33 GMT+01:00 Nicolas Chauvet : > 2017-02-16 19:26 GMT+01:00 Nicolas Chauvet : >> --- >> kernel.spec | 1 - >> 1 file changed, 1 deletion(-) >> >> diff --git a/kernel.spec b/kernel.spec >> index 4363050..38968ba 100644 >> --- a/kernel.spec >> +++ b/kernel.spec >> @@ -815,7 +815,6 @@ Summary: Development package for building kernel modules >> to match the %{?2:%{2} >> Group: System Environment/Kernel\ >> Provides: kernel%{?1:-%{1}}-devel-%{_target_cpu} = %{version}-%{release}\ >> Provides: kernel-devel-%{_target_cpu} = %{version}-%{release}%{?1:+%{1}}\ >> -Provides: kernel-devel = %{version}-%{release}%{?1:+%{1}}\ >> Provides: kernel-devel-uname-r = %{KVERREL}%{?variant}%{?1:+%{1}}\ >> Provides: installonlypkg(kernel)\ >> AutoReqProv: no\ >> -- >> 2.7.4 >> > > So, this was the "light description version" of the patch. > please see a full rationale here http://bugzilla.redhat.com/1420754 > Basically, this patch make the situation in sync with el7 kernel WRT > theses provides. > There is one virtual provides kernel-devel-uname-r > Other are real (packages) provides. > > That will restore the ability to prefer one kernel-devel varriant over > another one as (such as kernel-debug-devel). > In the current situation every kernel-devel varriant have this > kernel-devel (virtual) provide. Any answer from the kernel team about this ? Thx -- - Nicolas (kwizart) ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org