sizeof - w/ rpmbuild
poma
pomidorabelisima at gmail.com
Sat Jun 20 02:10:05 UTC 2015
On 20.06.2015 04:00, Josh Stone wrote:
> On 06/19/2015 06:47 PM, poma wrote:
>> On 20.06.2015 03:30, Josh Stone wrote:
>>> On 06/19/2015 05:51 PM, Josh Stone wrote:
>>>> On 06/19/2015 05:16 PM, poma wrote:
>>>>>
>>>>> This is exactly the opposite of the case with the kernel
>>>>> https://bugzilla.redhat.com/show_bug.cgi?id=1232206
>>>>>
>>>>>
>>>>> - Local rpmbuild builds:
>>>>>
>>>>> binutils-2.25-11 - elfutils-0.163-1
>>>>> 4.9M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>> binutils-2.25-10 - elfutils-0.163-1
>>>>> 4.9M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>> binutils-2.25-10 - elfutils-0.162-2
>>>>> 4.9M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>> binutils-2.25-10 - elfutils-0.162-1
>>>>> 4.9M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>> binutils-2.25-10 - elfutils-0.161-8
>>>>> 4.9M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>> binutils-2.25-9 - elfutils-0.161-8
>>>>> 4.9M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>>
>>>>> - Local mock build:
>>>>>
>>>>> /var/lib/mock/fedora-rawhide-x86_64/result/
>>>>> 2.0M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>>
>>>>> - Koji:
>>>>>
>>>>> https://kojipkgs.fedoraproject.org/packages/NetworkManager/1.0.4/0.1.git20150618.8cffaf3bf5.fc23/x86_64/
>>>>> 2.0M NetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>>>
>>>>>
>>>>> Any idea why this multiplying size is happening with the rpmbuild?
>>>>>
>>>>> NetworkManager is just an example of a general issue.
>>>>
>>>> I'd start with "rpm -qvlp foo.rpm" to see what exactly is bigger.
>>>> If it's an ELF file, compare "eu-readelf -S" or even "-a".
>>>
>>> FWIW, my "fedpkg local" produced 2.0M too.
>>>
>>> Perhaps you have devel packages for some optional NetworkManager
>>> features that aren't normally enabled? Or maybe you have the bad
>>> libelf.so in your path somewhere from testing the strip issue?
>>>
>>
>> - rpmbuild runs as:
>>
>> $ cat ~/.rpmmacros
>> %debug_package %{nil}
>
> Doesn't this prevent stripping?
>
> The normal %debug_package sets %global __debug_package 1, which causes
> __spec_install_post to run __debug_install_post, which runs
> find-debuginfo.sh, which is what runs eu-strip to extract debuginfo.
>
> Without that, the debug sections will be left in your files, as readelf
> shows below, so of course they will be bigger!
>
But why is this happening only recently?
I use "%debug_package %{nil}" to exclude the construction of *debuginfo* packages.
Eeeeeeeeeeeeeeeeeeeeeeeeeee
Am I really doing it the wrong way, so far!?
>> $ rpmbuild -ba NetworkManager.spec
>>
>>
>> - mock runs as:
>>
>> $ mock --verbose --root=fedora-rawhide-x86_64 --with=baseonly --without=debuginfo --rebuild NM.src.rpm
>>
>>
>> Significant differences:
>>
>> - rpm-qvlp-NM-rpmbuild-vs-mock.rpm.diff:
>>
>> < 31392 Jun 19 23:54 /usr/bin/nm-online
>> < 1729192 Jun 19 23:54 /usr/bin/nmcli
>> ---
>> > 15344 Jun 19 23:54 /usr/bin/nm-online
>> > 563256 Jun 19 23:54 /usr/bin/nmcli
>> ---
>> < 83280 Jun 19 23:54 /usr/lib64/NetworkManager/libnm-settings-plugin-ibft.so
>> < 2696 Jun 19 23:54 /usr/lib64/NetworkManager/libnm-settings-plugin-ifcfg-rh.so
>> < 38800 Jun 19 23:54 /usr/lib64/pppd/2.4.7/nm-pppd-plugin.so
>> < 18808 Jun 19 23:54 /usr/libexec/nm-avahi-autoipd.action
>> < 19736 Jun 19 23:54 /usr/libexec/nm-dhcp-helper
>> < 152336 Jun 19 23:54 /usr/libexec/nm-dispatcher
>> < 4090448 Jun 19 23:54 /usr/libexec/nm-iface-helper
>> < 7846824 Jun 19 23:54 /usr/sbin/NetworkManager
>> ---
>> > 27456 Jun 19 23:54 /usr/lib64/NetworkManager/libnm-settings-plugin-ibft.so
>> > 185104 Jun 19 23:54 /usr/lib64/NetworkManager/libnm-settings-plugin-ifcfg-rh.so
>> > 15184 Jun 19 23:54 /usr/lib64/pppd/2.4.7/nm-pppd-plugin.so
>> > 11240 Jun 19 23:54 /usr/libexec/nm-avahi-autoipd.action
>> > 11304 Jun 19 23:54 /usr/libexec/nm-dhcp-helper
>> > 48640 Jun 19 23:54 /usr/libexec/nm-dispatcher
>> > 1006808 Jun 19 23:54 /usr/libexec/nm-iface-helper
>> > 1875056 Jun 19 23:54 /usr/sbin/NetworkManager
>>
>>
>> - eu-readelf-S-NM-rpmbuild-vs-mock.diff:
>>
>> < There are 38 section headers, starting at offset 0x77b228:
>> ---
>> > There are 30 section headers, starting at offset 0x1c94f0:
>> ---
>> < [27] .comment PROGBITS 0000000000000000 001c50e0 00000058 1 MS 0 0 1
>> < [28] .debug_aranges PROGBITS 0000000000000000 001c5138 00001d80 0 0 0 1
>> < [29] .debug_info PROGBITS 0000000000000000 001c6eb8 0027f59a 0 0 0 1
>> < [30] .debug_abbrev PROGBITS 0000000000000000 00446452 000240c8 0 0 0 1
>> < [31] .debug_line PROGBITS 0000000000000000 0046a51a 0004ee66 0 0 0 1
>> < [32] .debug_str PROGBITS 0000000000000000 004b9380 0003616d 1 MS 0 0 1
>> < [33] .debug_loc PROGBITS 0000000000000000 004ef4ed 00214fcb 0 0 0 1
>> < [34] .debug_ranges PROGBITS 0000000000000000 007044b8 00023e30 0 0 0 1
>> < [35] .shstrtab STRTAB 0000000000000000 007282e8 00000177 0 0 0 1
>> < [36] .symtab SYMTAB 0000000000000000 00728460 0002baa0 24 37 4551 8
>> < [37] .strtab STRTAB 0000000000000000 00753f00 00027326 0 0 0 1
>> ---
>> > [27] .gnu_debuglink PROGBITS 0000000000000000 001c50e0 0000001c 0 0 0 4
>> > [28] .gnu_debugdata PROGBITS 0000000000000000 001c50fc 000042d0 0 0 0 1
>> > [29] .shstrtab STRTAB 0000000000000000 001c93cc 00000123 0 0 0 1
>>
>>
>
More information about the test
mailing list