[OE-core] [PATCH] toolchain-shar-extract: compare SDK and host gcc versions

[Khem Raj]

On Thu, Oct 5, 2017 at 10:43 AM, Mark Hatle <mark.hatle at windriver.com> wrote:
> I realize this reply is months after the original email.  But I think I found a
> bug with the implementation.  See below inline.
>
> On 1/25/17 7:18 AM, Ed Bartosh wrote:
>> If ext sdk is built by gcc version higher than host gcc version
>> and host gcc version is 4.8 or 4.9 the installation is known to
>> fail due to the way uninative sstate package is built.
>>
>> It's a known issue and we don't have a way to fix it for above
>> mentioned combinations of build and host gcc versions.
>>
>> Detected non-installable combinations of gcc versions and
>> print an installation error.
>>
>> [YOCTO #10881]
>>
>> Signed-off-by: Ed Bartosh <ed.bartosh at linux.intel.com>
>> ---
>>  meta/classes/populate_sdk_base.bbclass | 1 +
>>  meta/files/toolchain-shar-extract.sh   | 8 ++++++++
>>  2 files changed, 9 insertions(+)
>>
>> diff --git a/meta/classes/populate_sdk_base.bbclass b/meta/classes/populate_sdk_base.bbclass
>> index 5ad08c6..588a3cb 100644
>> --- a/meta/classes/populate_sdk_base.bbclass
>> +++ b/meta/classes/populate_sdk_base.bbclass
>> @@ -226,6 +226,7 @@ EOF
>>               -e 's#@SDK_VERSION@#${SDK_VERSION}#g' \
>>               -e '/@SDK_PRE_INSTALL_COMMAND@/d' \
>>               -e '/@SDK_POST_INSTALL_COMMAND@/d' \
>> +             -e 's#@SDK_GCC_VER@#${@oe.utils.host_gcc_version(d)}#g' \
>
> The return value from "oe.utils.host_gcc_version(d)" includes a leading '-'...
>
>>               ${SDKDEPLOYDIR}/${TOOLCHAIN_OUTPUTNAME}.sh
>>
>>       # add execution permission
>> diff --git a/meta/files/toolchain-shar-extract.sh b/meta/files/toolchain-shar-extract.sh
>> index 9295ddc..3f54c96 100644
>> --- a/meta/files/toolchain-shar-extract.sh
>> +++ b/meta/files/toolchain-shar-extract.sh
>> @@ -11,6 +11,9 @@ export PATH=`echo "$PATH" | sed -e 's/:\.//' -e 's/::/:/'`
>>  INST_ARCH=$(uname -m | sed -e "s/i[3-6]86/ix86/" -e "s/x86[-_]64/x86_64/")
>>  SDK_ARCH=$(echo @SDK_ARCH@ | sed -e "s/i[3-6]86/ix86/" -e "s/x86[-_]64/x86_64/")
>>
>> +INST_GCC_VER=$(gcc --version | sed -ne 's/.* \([0-9]\+\.[0-9]\+\)\.[0-9]\+.*/\1/p')
>
> The item above fails with the message:
>
> line 28: gcc: command not found
>
> If the host does not have gcc on it.  (I realize this is incredibly rare
> situation, but is valid in some cases.)  It might be a good idea to catch this
> and put up a warning or otherwise clean up the error situation.
>

I think if we need gcc then we should sanitize the installer to check-before-use
so end user gets a good error message.


>> +SDK_GCC_VER='@SDK_GCC_VER@'
>> +
>
> So based on the above you will end up with:
>
> INST_GCC_VER="4.8"
> SDK_GCC_VER="-4.9"
>
>>  verlte () {
>>       [  "$1" = "`printf "$1\n$2" | sort -V | head -n1`" ]
>>  }
>> @@ -112,6 +115,11 @@ fi
>>  # SDK_EXTENSIBLE is exposed from the SDK_PRE_INSTALL_COMMAND above
>>  if [ "$SDK_EXTENSIBLE" = "1" ]; then
>>       DEFAULT_INSTALL_DIR="@SDKEXTPATH@"
>> +     if [ "$INST_GCC_VER" = '4.8' -a "$SDK_GCC_VER" = '4.9' ] || [ "$INST_GCC_VER" = '4.8' -a "$SDK_GCC_VER" = '' ] || \
>> +             [ "$INST_GCC_VER" = '4.9' -a "$SDK_GCC_VER" = '' ]; then
>
> The comparisons above do not include the leading '-', so it would appear to me
> that the checked would never end up being true -- and thus the error below would
> never happen.
>
> Am I seeing what is happening right?  If so, we'll need to get this fixed...
>
> --Mark
>
>
>> +             echo "Error: Incompatible SDK installer! Your host gcc version is $INST_GCC_VER and this SDK was built by gcc higher version."
>> +             exit 1
>> +     fi
>>  fi
>>
>>  if [ "$target_sdk_dir" = "" ]; then
>>
>
> --
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core at lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core