[OE-core] [oe-commits] Robert Yang : A script to clean obsolete sstate cache files

[Robert Yang]

On 03/27/2012 09:35 PM, Martin Jansa wrote:
> On Tue, Mar 27, 2012 at 09:28:09PM +0800, Robert Yang wrote:
>>
>> Hi Martin,
>>
>> Thanks for reporting this, and I'm very sorry for the inconvenience,
>> please see my comment inline ...
>>
>> On 03/27/2012 07:33 PM, Martin Jansa wrote:
>>> On Tue, Mar 27, 2012 at 01:06:33PM +0200, Martin Jansa wrote:
>>>> On Fri, Feb 24, 2012 at 12:01:37AM +0000, git at git.openembedded.org wrote:
>>>>> Module: openembedded-core.git
>>>>> Branch: master
>>>>> Commit: 7c99ef6d2173b14e1109a540ee5ae47b56d707e7
>>>>> URL:    http://git.openembedded.org/?p=openembedded-core.git&a=commit;h=7c99ef6d2173b14e1109a540ee5ae47b56d707e7
>>>>>
>>>>> Author: Robert Yang<liezhi.yang at windriver.com>
>>>>> Date:   Wed Feb 22 20:38:30 2012 +0800
>>>>>
>>>>> A script to clean obsolete sstate cache files
>>>>>
>>>>
>>>>> +# Use the "_" to substitute "-", e.g., x86-64 to x86_64
>>>>> +all_archs=`grep -r DEFAULTTUNE $tunedir | \
>>>>> +    sed -e 's/.*\"\(.*\)\"/\1/' -e 's/-/_/g' | sort -u`
>>>>> +# Add the qemu archs
>>>>> +all_archs="$all_archs qemuarm qemux86 qemumips qemuppc"
>>>>
>>>> This doesn't list any thumb variants, current list is:
>>>>
>>>> arm1136jfs arm920t arm926ejs arm9tdmi armv4 armv5 armv6 armv7 armv7a
>>>> core2 cortexa8_neon cortexa9_neon cortexm1 cortexm3 cortexr4 ep9312 i586
>>>> iwmmxt mips mips32 powerpc powerpc64 ppc603e ppce300c2 ppce500 ppce500mc
>>>> ppce500v2 ppce5500 strongarm x86 x86_64 xscale
>>>>
>>>> so thumb versions (e.g. armv4t) are not found in
>>>>
>>>>> +for arch in $all_archs; do
>>>>> +    ls | grep -q -w $arch
>>>>> +    [ $? -eq 0 ]&&   ava_archs="$ava_archs $arch"
>>>>> +done
>>>>
>>>> Cheers,
>>>
>>> And after extending list of machines in all_archs with armv4t and my
>>> machines I got only one file in sstate-cache left :/
>>>
>>> First run without armv4t found:
>>> OE crespo at shr ~/shr-core $ du -hs sstate-cache/; openembedded-core/scripts/sstate-cache-management.sh --cache-dir=sstate-cache --remove-duplicated; du -hs sstate-cache/
>>> 14G     sstate-cache/
>>> Figuring out the archs in the sstate cache dir ...
>>> The following archs have been found in the sstate cache dir:
>>> armv7a x86_64
>>> Removing the sstate-xxx_deploy-rpm.tgz ... (0 files)
>>> Removing the sstate-xxx_deploy-ipk.tgz ... (58 files)
>>> Removing the sstate-xxx_deploy-deb.tgz ... (0 files)
>>> Removing the sstate-xxx_deploy.tgz ... (0 files)
>>> Removing the sstate-xxx_package.tgz ... (62 files)
>>> Removing the sstate-xxx_populate-lic.tgz ... (46 files)
>>> Removing the sstate-xxx_populate-sysroot.tgz ... (102 files)
>>> 268 files have been removed
>>> 13G     sstate-cache/
>>>
>>> Second run with armv4t manually added
>>> -all_archs="$all_archs qemuarm qemux86 qemumips qemuppc"
>>> +all_archs="$all_archs qemuarm qemux86 qemumips qemuppc armv4t"
>>> OE crespo at shr ~/openembedded-core $ du -hs ../shr-core/sstate-cache/; scripts/sstate-cache-management.sh --cache-dir=../shr-core/sstate-cache --remove-duplicated; du -hs ../shr-core/sstate-cache/
>>> 13G     ../shr-core/sstate-cache/
>>> /OE/openembedded-core/meta/conf/machine/include
>>> Figuring out the archs in the sstate cache dir ...
>>> The following archs have been found in the sstate cache dir:
>>> armv4t armv7a x86_64
>>> Removing the sstate-xxx_deploy-rpm.tgz ... (0 files)
>>> Removing the sstate-xxx_deploy-ipk.tgz ... (460 files)
>>> Removing the sstate-xxx_deploy-deb.tgz ... (0 files)
>>> Removing the sstate-xxx_deploy.tgz ... (0 files)
>>> Removing the sstate-xxx_package.tgz ... (454 files)
>>> Removing the sstate-xxx_populate-lic.tgz ... (456 files)
>>> Removing the sstate-xxx_populate-sysroot.tgz ... (468 files)
>>> 1838 files have been removed
>>> 6.2G    ../shr-core/sstate-cache/
>>>
>>> Third run with machines I'm building manually added
>>> -all_archs="$all_archs qemuarm qemux86 qemumips qemuppc armv4t"
>>> +all_archs="$all_archs qemuarm qemux86 qemumips qemuppc armv4t om_gta01 om_gta02 om_gta04 nokia900 palmpre palmpre2 crespo spitz"
>>
>> The script figures the archs according to the 'meta/conf/machine/include',
>> this is incorrect since it only searches the oe-core layer, I will fix it.
>
> The first issue is that armv4t and other thumb variants are also in
> oe-core layer.
>
>> And where can I get the "om_gta01 om_gta02 om_gta04 ..." archs? I mean which
>> layer are they in? So that I can test them.
>
> Those are machines from meta-smartphone sublayers
> http://git.shr-project.org/git/?p=meta-smartphone.git;a=summary
>
> Maybe add some param to add extra archs/machines or just list all used
> xxx variants from 'sstate-xxx_'.
>
Thanks, add some param is a good choice. Figure out the used variants from
sstate-xxx is the best way, but it seems not easy since it uses the "-" as
the separator, and I had counted the number of field, it is not always the
same.

// Robert


> Cheers,
>