Re: kernel-4.15.0-0.rc3.git0.1.fc28 src rpm fails to compile because of errors in ppc64p7 config file
>> On Wed, Dec 13, 2017 at 8:08 PM, Laura Abbott>> wrote: > >> > Can you share the steps you used to produce this build as well as >> > the full build log? As you noted, this was built successfully in >> > koji. > > I'm following this procedure for building from the src.rpm using > rpmbuild. > > https://fedoraproject.org/wiki/Building_a_custom_kernel/Source_RPM > > I've been using it for years, and I successfully built a 4.15 kernel > just a few weeks ago using the same procedure. > >> The ppc64p7 sub variant was removed so I'm not even sure why that's >> showing up. > > This was the key comment. It seems that the kernel picks up any kernel > config files in SOURCES. There was an old set of configs for ppc64p7 > there. Once I deleted them, everything worked fine. The previous > version of 4.15 I successfully built was in F25, and I'm now building in > rawhide, and it had the obsolete ppc64p7 config file I SOURCES from > older builds. > > Sorry for the noise, but thanks for the help. No issues, I always find "git status" is a big help here with kernel stuff, especially when it comes to stuff that might be generated. ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Re: kernel-4.15.0-0.rc3.git0.1.fc28 src rpm fails to compile because of errors in ppc64p7 config file
On Thu, 14 Dec 2017 01:00:10 + Peter Robinsonwrote: > On Wed, Dec 13, 2017 at 8:08 PM, Laura Abbott > wrote: > > Can you share the steps you used to produce this build as well as > > the full build log? As you noted, this was built successfully in > > koji. I'm following this procedure for building from the src.rpm using rpmbuild. https://fedoraproject.org/wiki/Building_a_custom_kernel/Source_RPM I've been using it for years, and I successfully built a 4.15 kernel just a few weeks ago using the same procedure. > The ppc64p7 sub variant was removed so I'm not even sure why that's > showing up. This was the key comment. It seems that the kernel picks up any kernel config files in SOURCES. There was an old set of configs for ppc64p7 there. Once I deleted them, everything worked fine. The previous version of 4.15 I successfully built was in F25, and I'm now building in rawhide, and it had the obsolete ppc64p7 config file I SOURCES from older builds. Sorry for the noise, but thanks for the help. ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
[PATCH 1/2] configs: Move base-debug and base-generic to configs/fedora
It was suggested that base-debug and base-generic were not good names to use. Further discussion led to using configs/fedora for the base config files and configs/rhel for any overrides. This patch does a plain mkdir configs/fedora git mv configs/base-{generic,debug} configs/fedora No code changes. --- ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
[PATCH 2/2] configs: Update config generation script to use configs/fedora
The previous patch moved the configs/base-{generic,debug} to configs/fedora. Now we update the scripts to reflect that change. Changing the scripts was straightforward. Handling overrides that didn't use generic names was a little trickier. To handle random override names (well rhel), I added some extra logic in the config_generation script called "ORDER". This tells the scripts which configs to lay down first and which one overrides it. Through some testing, I realized I could simplify things and just create an outer 'order' loop. This removed some duplicated code. The other change is the 'skip_if_missing' flag. The overrides directory will not mimic the baseline directory layout 100%. Ensure the baseline config files are all there, but allow the overrides to have missing files. Tested on my Fedora and my RHEL tree with success. --- configs/build_configs.sh | 62 --- configs/config_generation | 5 2 files changed, 37 insertions(+), 30 deletions(-) diff --git a/configs/build_configs.sh b/configs/build_configs.sh index f345e2af..15ab6b38 100755 --- a/configs/build_configs.sh +++ b/configs/build_configs.sh @@ -6,7 +6,7 @@ PACKAGE_NAME="${1:-kernel}" # defines the package name used KVERREL="${2:-}" -SUBARCH="${3:-}" # defines a specific arch for use with rh-configs-arch-prep target +SUBARCH="${3:-}" # defines a specific arch SCRIPT="$(readlink -f $0)" OUTPUT_DIR="$PWD" SCRIPT_DIR="$(dirname $SCRIPT)" @@ -49,29 +49,31 @@ function merge_configs() archvar=$1 arch=$(echo "$archvar" | cut -f1 -d"-") configs=$2 + order=$3 name=$OUTPUT_DIR/$PACKAGE_NAME-$archvar.config echo -n "Building $name ... " touch config-merging config-merged - # apply base first - for config in $(echo $configs | sed -e 's/:/ /g') + # apply based on order + skip_if_missing="" + for o in $order do - perl merge.pl config-base-$config config-merging > config-merged - if [ ! $? -eq 0 ]; then - die "Failed to merge base" - fi - mv config-merged config-merging - done - for config in $(echo $configs | sed -e 's/:/ /g') - do - # not all override files exist - test -e config-$config || continue - - perl merge.pl config-$config config-merging > config-merged - if [ ! $? -eq 0 ]; then - die "Failed to merge configs" - fi - mv config-merged config-merging + for config in $(echo $configs | sed -e 's/:/ /g') + do + cfile="config-$o-$config" + + test -n "$skip_if_missing" && test ! -e $cfile && continue + + perl merge.pl $cfile config-merging > config-merged + if [ ! $? -eq 0 ]; then + die "Failed to merge $cfile" + fi + mv config-merged config-merging + done + + # first configs in $order is baseline, all files should be + # there. second pass is overrides and can be missing. + skip_if_missing="1" done if [ "x$arch" == "xaarch64" ]; then echo "# arm64" > $name @@ -93,22 +95,22 @@ function merge_configs() echo "done" } -glist=$(find base-generic -type d) -dlist=$(find base-debug -type d) -gllist=$(test -d generic && find generic -type d) -dllist=$(test -d debug && find debug -type d) - -for d in $glist $dlist $gllist $dllist -do - combine_config_layer $d -done - while read line do if [ $(echo "$line" | grep -c "^#") -ne 0 ]; then continue elif [ $(echo "$line" | grep -c "^$") -ne 0 ]; then continue + elif [ $(echo "$line" | grep -c "^ORDER") -ne 0 ]; then + order=$(echo "$line" | cut -f2 -d"=") + for o in $order + do + glist=$(find $o -type d) + for d in $glist + do + combine_config_layer $d + done + done else arch=$(echo "$line" | cut -f1 -d"=") configs=$(echo "$line" | cut -f2 -d"=") @@ -117,7 +119,7 @@ do continue fi - merge_configs $arch $configs + merge_configs $arch $configs "$order" fi done < $control_file diff --git a/configs/config_generation b/configs/config_generation index b88a5ea9..e8614c43 100644 --- a/configs/config_generation +++ b/configs/config_generation @@ -3,6 +3,11 @@ # the first arg is arch and variant, the second is a hierarchy of # config options, lowest priority to highest +# tells the build_configs.sh which order to build the
Re: Enabling CONFIG_REGULATOR on x86_64
On 12/13/2017 01:26 AM, Hans de Goede wrote: Hi, On 13-12-17 10:24, Hans de Goede wrote: Hi All, I would like to enable CONFIG_REGULATOR on x86_64, to fix USB PD negotiating (and charging with more the 0.5 A in general) and battery monitoring on Intel Cherry Trail devices which use an Intel PMIC + FUSB302 Type-C controller to enable USB-PD charging. Since the USB-PD negotiation in this case is handled by the kernel rather then by firmware in the Alpine Ridge controller, the kernel also needs to enable/disable the 5V boost converter feeding 5v to the Type-C port when it is in host / power-source mode and this is modeled as a regulator. One of the drivers for the complex of 4 ICs which implement the Type-C and battery charging monitoring on these devices has a Kconfig dep on CONFIG_REGULATOR because of this and without this driver these devices are limited to charging (or rather slowly discharging) at 5V 0.5A rather then charging at up to 12V 3A, also without this battery monitoring will not work and the kernel will be unable to make the port provide 5V when a Type-C to USB-A-receptacle cable gets plugged in. Downside of this is a slightly bigger kernel (the regulator framework is tiny) and some messages like these when some drivers load: [ 19.955279] rt5645 i2c-10EC5645:00: i2c-10EC5645:00 supply avdd not found, using dummy regulator [ 19.955331] rt5645 i2c-10EC5645:00: i2c-10EC5645:00 supply cpvdd not found, using dummy regulator As these drivers now actually call into the regulator framework which prints these messages, rather then calling empty stubs. These are printed at "warning" level so will not show up on the console. So if there are no objections I will push a commit enabling CONFIG_REGULATOR on x86_64 one of the coming days. p.s. I'm proposing this change for the 4.15 kernel only, as that finally has all the necessary drivers to do USB PD on these devices and enabling CONFIG_REGULATOR on x86_64 is all that is missing to get this to work. Regards, Hans This sounds okay for rawhide. The regulator framework has been well tested and I wouldn't expect too many problems. Thanks, Laura ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Re: Changing the SND_HDA_POWER_SAVE_DEFAULT option from 0 to 1
On 11/03/2017 04:59 AM, Hans de Goede wrote: > Hi, > > On 01-11-17 20:34, Laura Abbott wrote: >> I'm always a bit uneasy about enabling hardware power saving >> since it rarely seems to be done correctly but as Peter pointed >> out it's easy to revert if something goes wrong. Having the >> endorsement of upstream is also good since it means they should >> be responsive. >> >> Enabling for 4.14 sounds like a good plan. Did you intend for >> this to end up in stable releases on rebase? > > That was my original intention yes. But now that you explicitly > ask I guess that we may want to hold off on that and just > make the change for rawhide / F28. I will try to get the > change committed to the master branch today (otherwise > I will do it Monday). I will leave it up to you if you also > make the change to the stable releases on rebase. We ended up putting this in the 4.14 rebase and there were a few issues reported[0][1] on i686 and x86_64 so I've set it back to 0 on all arches in F27 and F26. Hans, do you want to leave in on in Rawhide and see if we can come up with some fixes or should I switch it to 0 there as well? [0] https://bugzilla.redhat.com/show_bug.cgi?id=1525561 [1] https://bugzilla.redhat.com/show_bug.cgi?id=1525104 Thanks, Jeremy ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Re: kernel-4.15.0-0.rc3.git0.1.fc28 src rpm fails to compile because of errors in ppc64p7 config file
On 12/13/2017 10:25 AM, stan wrote: I tried to compile this kernel from the src rpm on rawhide. I kept getting the error that two config options for ppc64p7 were set to m, and m wasn't a valid choice. So, I generated the following patch, diff --git a/configs/kernel-4.15.0-ppc64p7.config b/configs/kernel-4.15.0-ppc64p7.config index 18f4b44..ca5b410 100644 --- a/configs/kernel-4.15.0-ppc64p7.config +++ b/configs/kernel-4.15.0-ppc64p7.config @@ -1704,7 +1704,7 @@ CONFIG_HW_RANDOM_TIMERIOMEM=m CONFIG_HW_RANDOM_TPM=m CONFIG_HW_RANDOM_VIRTIO=m CONFIG_HW_RANDOM=y -CONFIG_HWSPINLOCK=m +CONFIG_HWSPINLOCK=y CONFIG_HYSDN_CAPI=y CONFIG_HYSDN=m CONFIG_HZ=100 @@ -4447,7 +4447,7 @@ CONFIG_SND_DARLA24=m # CONFIG_SND_DEBUG_VERBOSE is not set CONFIG_SND_DEBUG=y CONFIG_SND_DESIGNWARE_I2S=m -CONFIG_SND_DESIGNWARE_PCM=m +CONFIG_SND_DESIGNWARE_PCM=y CONFIG_SND_DICE=m CONFIG_SND_DMAENGINE_PCM=m CONFIG_SND_DRIVERS=y I don't know if y is a valid setting for a ppc for those options, but I presume if m was desired, y would succeed. I don't really care about ppc64p7 as I'm building for x86_64. I changed the spec file so that this architecture won't be built in nobuildarches, and this error still occurred. Unfortunately, at the time that the kernel is patched, this config file doesn't exist yet, and so the patch fails. I see a successful build in koji; I'm not sure how that can be. What is the best way to proceed from here? Can you share the steps you used to produce this build as well as the full build log? As you noted, this was built successfully in koji. Thanks, Laura ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
kernel-4.15.0-0.rc3.git0.1.fc28 src rpm fails to compile because of errors in ppc64p7 config file
I tried to compile this kernel from the src rpm on rawhide. I kept getting the error that two config options for ppc64p7 were set to m, and m wasn't a valid choice. So, I generated the following patch, diff --git a/configs/kernel-4.15.0-ppc64p7.config b/configs/kernel-4.15.0-ppc64p7.config index 18f4b44..ca5b410 100644 --- a/configs/kernel-4.15.0-ppc64p7.config +++ b/configs/kernel-4.15.0-ppc64p7.config @@ -1704,7 +1704,7 @@ CONFIG_HW_RANDOM_TIMERIOMEM=m CONFIG_HW_RANDOM_TPM=m CONFIG_HW_RANDOM_VIRTIO=m CONFIG_HW_RANDOM=y -CONFIG_HWSPINLOCK=m +CONFIG_HWSPINLOCK=y CONFIG_HYSDN_CAPI=y CONFIG_HYSDN=m CONFIG_HZ=100 @@ -4447,7 +4447,7 @@ CONFIG_SND_DARLA24=m # CONFIG_SND_DEBUG_VERBOSE is not set CONFIG_SND_DEBUG=y CONFIG_SND_DESIGNWARE_I2S=m -CONFIG_SND_DESIGNWARE_PCM=m +CONFIG_SND_DESIGNWARE_PCM=y CONFIG_SND_DICE=m CONFIG_SND_DMAENGINE_PCM=m CONFIG_SND_DRIVERS=y I don't know if y is a valid setting for a ppc for those options, but I presume if m was desired, y would succeed. I don't really care about ppc64p7 as I'm building for x86_64. I changed the spec file so that this architecture won't be built in nobuildarches, and this error still occurred. Unfortunately, at the time that the kernel is patched, this config file doesn't exist yet, and so the patch fails. I see a successful build in koji; I'm not sure how that can be. What is the best way to proceed from here? Thanks. ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Enabling CONFIG_REGULATOR on x86_64
Hi All, I would like to enable CONFIG_REGULATOR on x86_64, to fix USB PD negotiating (and charging with more the 0.5 A in general) and battery monitoring on Intel Cherry Trail devices which use an Intel PMIC + FUSB302 Type-C controller to enable USB-PD charging. Since the USB-PD negotiation in this case is handled by the kernel rather then by firmware in the Alpine Ridge controller, the kernel also needs to enable/disable the 5V boost converter feeding 5v to the Type-C port when it is in host / power-source mode and this is modeled as a regulator. One of the drivers for the complex of 4 ICs which implement the Type-C and battery charging monitoring on these devices has a Kconfig dep on CONFIG_REGULATOR because of this and without this driver these devices are limited to charging (or rather slowly discharging) at 5V 0.5A rather then charging at up to 12V 3A, also without this battery monitoring will not work and the kernel will be unable to make the port provide 5V when a Type-C to USB-A-receptacle cable gets plugged in. Downside of this is a slightly bigger kernel (the regulator framework is tiny) and some messages like these when some drivers load: [ 19.955279] rt5645 i2c-10EC5645:00: i2c-10EC5645:00 supply avdd not found, using dummy regulator [ 19.955331] rt5645 i2c-10EC5645:00: i2c-10EC5645:00 supply cpvdd not found, using dummy regulator As these drivers now actually call into the regulator framework which prints these messages, rather then calling empty stubs. These are printed at "warning" level so will not show up on the console. So if there are no objections I will push a commit enabling CONFIG_REGULATOR on x86_64 one of the coming days. Regards, Hans ___ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-le...@lists.fedoraproject.org