W
Sent from my iPhone ___ test mailing list -- test@lists.fedoraproject.org To unsubscribe send an email to test-le...@lists.fedoraproject.org
sizeof - w/ rpmbuild
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 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm binutils-2.25-10 - elfutils-0.163-1 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm binutils-2.25-10 - elfutils-0.162-2 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm binutils-2.25-10 - elfutils-0.162-1 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm binutils-2.25-10 - elfutils-0.161-8 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm binutils-2.25-9 - elfutils-0.161-8 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm - Local mock build: /var/lib/mock/fedora-rawhide-x86_64/result/ 2.0MNetworkManager-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/ 2.0MNetworkManager-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. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
On Sáb, 2015-06-20 at 04:10 +0200, poma wrote:
> 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
> > 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
> >
> > binutils-2.25-10 - elfutils-0.163-1
> > 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
> >
> > binutils-2.25-10 - elfutils-0.162-2
> > 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
> >
> > binutils-2.25-10 - elfutils-0.162-1
> > 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
> >
> > binutils-2.25-10 - elfutils-0.161-8
> > 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
> >
> > binutils-2.25-9 - elfutils-0.161-8
> > 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
> >
> >
> > - Local mock build:
> >
> > /var/lib/mock/fedora-rawhide-x86_64/result/
> > 2.0MNetworkManager-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/
> > 2.0MNetworkManager-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.
>
> Eee
> Am I really doing it the wrong way, so far!?
Hi,
AFAIK , We, in package, enable build with debug (./configure
--enable-debug) to create good debuginfo packages.
I think when rpmbuild begins do debuginfo packages strips the
binaries .
For example pngquant.spec
./configure --prefix=%{_prefix} --enable-debug
file /usr/bin/pngquant
/usr/bin/pngquant: ELF 64-bit LSB executable, x86-64, version 1 (SYSV),
dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for
GNU/Linux 2.6.32,
BuildID[sha1]=ef857b7ad50d1791a17a46d2a02c6a502bebdaa0, stripped
> >> $ 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
> >> > 1
Re: sizeof - w/ rpmbuild
poma wrote: > Any idea why this multiplying size is happening with the rpmbuild? Do you have redhat-rpm-config installed and -debuginfo pkgs getting generated? -- Rex -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
Rex Dieter wrote: > poma wrote: > >> Any idea why this multiplying size is happening with the rpmbuild? > > Do you have redhat-rpm-config installed and -debuginfo pkgs getting > generated? I see followups on fedora kernel list that it was a temporary (now fixed) issue with elfutils not stripping properly. -- Rex -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
On 20.06.2015 13:10, Mark Wielaard wrote:
> On Sat, 2015-06-20 at 12:17 +0200, poma wrote:
>> An exemplary example:
>>
>> $ du -h xfce4-mixer*
>> 104K xfce4-mixer-debug-before
>> 432K xfce4-mixer-debug-recent
>> 84K xfce4-mixer-strip-recent
>
> Clearly the second version has .debug sections in the binaries, and the
> others don't. Also the first seems to have a full .symtab section, but
> does not contain any reference to the .debug file (it doesn't have
> a .gnu_debuglink section) and it doesn't contain a .gnu_debugdata
> (mini-symtab) section, while the last one doesn't have a full .symtab
> section, but does have both a reference to the actual debug file
> (.gnu_debuglink section) and a mini-symtab (.gnu_debugdata section).
>
> But what exactly are you comparing? How do you build these packages
> precisely? It might help if you posted the build.logs somewhere for
> comparison.
>
> Thanks,
>
> Mark
>
Actually it was my fault lacking a better insight on this matter.
In the meantime I purge, rebuild - now without %debug_package %{nil} - and
reinstall everywhere.
We do not have anything to compare, not any more, sorry.
If rpm in standard way cannot strip without passing content to debuginfo, so
let it be.
I thank those who responded.
--
test mailing list
test@lists.fedoraproject.org
To unsubscribe:
https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
On 22.06.2015 10:44, Mark Wielaard wrote: > On Sun, 2015-06-21 at 06:10 +0200, poma wrote: >> If rpm in standard way cannot strip without passing content to debuginfo, so >> let it be. > > It doesn't have something like that as far as I know. Should it? > It seems that if you aren't going to use the debuginfo anyway a better > way to not get it is to not generate it in the first place by not > passing -g in the build FLAGS. Maybe rpm should have an option to do > that automagically for you. Although I would discourage its use. The > debuginfo is really useful normally. > > Cheers, > > Mark > There is no real reason not to provide such functionality, that same functionality provided by the kernel's infarctstructure. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
On 23.06.2015 21:05, Mark Wielaard wrote: > On Tue, Jun 23, 2015 at 04:33:29PM +0200, poma wrote: >> On 22.06.2015 10:44, Mark Wielaard wrote: >>> On Sun, 2015-06-21 at 06:10 +0200, poma wrote: If rpm in standard way cannot strip without passing content to debuginfo, so let it be. >>> >>> It doesn't have something like that as far as I know. Should it? >>> It seems that if you aren't going to use the debuginfo anyway a better >>> way to not get it is to not generate it in the first place by not >>> passing -g in the build FLAGS. Maybe rpm should have an option to do >>> that automagically for you. Although I would discourage its use. The >>> debuginfo is really useful normally. >> >> There is no real reason not to provide such functionality, >> that same functionality provided by the kernel's infarctstructure. > > I don't know what the kernel does, but can't you do it with rpm by > filtering out -g from %__global_cflags? > > Cheers, > > Mark > "Valid values for "option" above include xen, smp, up, pae, kdump, debug and debuginfo. Specifying --without debug strips out some debugging code from the kernels, where specifying --without debuginfo disables the building of the kernel-debuginfo packages." http://fedoraproject.org/wiki/Building_a_custom_kernel#Build_the_New_Kernel "filter" out -g from %__global_cflags - how exactly? -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
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
>>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>
>>> binutils-2.25-10 - elfutils-0.163-1
>>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>
>>> binutils-2.25-10 - elfutils-0.162-2
>>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>
>>> binutils-2.25-10 - elfutils-0.162-1
>>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>
>>> binutils-2.25-10 - elfutils-0.161-8
>>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>
>>> binutils-2.25-9 - elfutils-0.161-8
>>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm
>>>
>>>
>>> - Local mock build:
>>>
>>> /var/lib/mock/fedora-rawhide-x86_64/result/
>>> 2.0MNetworkManager-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/
>>> 2.0MNetworkManager-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}
$ 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 001c50e0 0058 1
MS 0 0 1
< [28] .debug_aranges PROGBITS 001c5138 1d80 0
0 0 1
< [29] .debug_info PROGBITS 001c6eb8 0027f59a 0
0 0 1
< [30] .debug_abbrevPROGBITS 00446452 000240c8 0
0 0 1
< [31] .debug_line PROGBITS 0046a51a 0004ee66 0
0 0 1
< [32] .debug_str PROGBITS 004b9380 0003616d 1
MS 0 0 1
< [33] .debug_loc PROGBITS 004ef4ed 00214fcb 0
0 0 1
< [34] .debug_rangesPROGBITS 007044b8 00023e30 0
0 0 1
< [35] .shstrtabSTRTAB 007282e8 0177 0
0 0 1
< [36] .symtab SYMTAB 00728460 0002baa0 24
37 4551 8
< [37] .strtab STRTAB 00753f00 00027326 0
0 0 1
---
> [27] .gnu_debuglink PROGBITS 001c50e0 001c 0
>0 0 4
> [28] .gnu_debugdata PROGBITS 001c50fc 42d0 0
>0 0 1
> [29] .shstrtabSTRTAB 001c93cc 0123 0
>0 0 1
--
test mailing list
test@lists.fedoraproject.org
To
Re: sizeof - w/ rpmbuild
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 >>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm >>> >>> binutils-2.25-10 - elfutils-0.163-1 >>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm >>> >>> binutils-2.25-10 - elfutils-0.162-2 >>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm >>> >>> binutils-2.25-10 - elfutils-0.162-1 >>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm >>> >>> binutils-2.25-10 - elfutils-0.161-8 >>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm >>> >>> binutils-2.25-9 - elfutils-0.161-8 >>> 4.9MNetworkManager-1.0.4-0.1.git20150618.8cffaf3bf5.fc23.x86_64.rpm >>> >>> >>> - Local mock build: >>> >>> /var/lib/mock/fedora-rawhide-x86_64/result/ >>> 2.0MNetworkManager-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/ >>> 2.0MNetworkManager-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? > All packages are affected, not really anything special with the NM. There is no residues of the elfutils* un/installations. No worries for it. To try with some more simple package? -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: sizeof - w/ rpmbuild
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.
Eee
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 001c50e0 0058
>> 1 MS 0 0 1
>> < [28] .debug_aranges PROGBITS 001c5138 1d80
>> 00 0 1
>> < [29] .debug_info PROGBITS 001c6eb8 0027f59a
>> 00 0 1
>> < [30] .debug_abbrevPROGBITS 00446452 000240c8
>> 00 0 1
>> < [31] .debug_line PROGBITS 0046a51a 0004ee66
>> 00 0 1
>> < [32] .debug_str
Re: sizeof - w/ rpmbuild
On 20.06.2015 04:10, poma wrote:
> 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.
>
> Eee
> Am I really doing it the wrong way, so far!?
>
>
An exemplary example:
$ du -h xfce4-mixer*
104Kxfce4-mixer-debug-before
432Kxfce4-mixer-debug-recent
84K xfce4-mixer-strip-recent
eu-readelf -S xfce4-mixer-debug-before:
- *with* %debug_package %{nil}
There are 31 section headers, starting at offset 0x18d20:
Section Headers:
[Nr] Name Type Addr Off Size ES
Flags Lk Inf Al
[ 0] NULL 0
0 0 0
[ 1] .interp PROGBITS 00400238 0238 001c 0 A
0 0 1
[ 2] .note.ABI-tagNOTE 00400254 0254 0020 0 A
0 0 4
[ 3] .note.gnu.build-id NOTE 00400274 0274 0024 0 A
0 0 4
[ 4] .gnu.hashGNU_HASH 00400298 0298 0080 0 A
5 0 8
[ 5] .dynsym DYNSYM 00400318 0318 1cc8 24 A
6 1 8
[ 6] .dynstr STRTAB 00401fe0 1fe0 1b6b 0 A
0 0 1
[ 7] .gnu.version GNU_versym 00403b4c 3b4c 0266 2 A
5 0 2
[ 8] .gnu.version_r GNU_verneed 00403db8 3db8 0050 0 A
6 2 8
[ 9] .rela.dynRELA 00403e08 3e08 0060 24 A
5 0 8
[10] .rela.pltRELA 00403e68 3e68 1bc0 24 AI
5 12 8
[11] .initPROGBITS 00405a28 5a28 001a 0 AX
0 0 4
[12] .plt PROGBITS 00405a50 5a50 1290 16 AX
0 0 16
[13] .textPROGBITS 00406ce0 6ce0 78c2 0 AX
0 0 16
[14] .finiPROGBITS 0040e5a4 e5a4 0009 0 AX
0 0 4
[15] .rodata PROGBITS 0040e5c0 e5c0 1688 0 A
0 0 32
[16] .eh_frame_hdrPROGBITS 0040fc48 fc48 03d4 0 A
0 0 4
[17] .eh_framePROGBITS 00410020 00010020 16b4 0 A
0 0 8
[18] .init_array INIT_ARRAY 00611c20
kernel w\o kernel-tools?
Didn't realise kernel-tools was not installed, until "rhgb quiet" removed. But I thougth they were joined at the hip, and an updated kernel would pull it in? kernel-3.1.0-7.fc16.i686 kernel-3.1.5-6.fc16.i686 kernel-3.2.0-0.rc7.git0.1.fc17.i686 kernel-3.2.0-0.rc7.git1.1.fc17.i686 kernel-3.2.0-0.rc7.git2.1.fc17.i686 kernel-3.2.0-0.rc7.git3.1.fc17.i686 kernel-3.2.0-0.rc7.git4.1.fc17.i686 package kernel-tools is not installed -- Regards, Frank Murphy UTF_8 Encoded Friend of fedoraproject.org -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: kernel w\o kernel-tools?
On Tue, Jan 3, 2012 at 11:15 AM, Frank Murphy wrote: > Didn't realise kernel-tools was not installed, > until "rhgb quiet" removed. > > But I thougth they were joined at the hip, > and an updated kernel would pull it in? Nope. Unless you have something it provides already installed. On rawhide, it only provides the cpupower stuff, which isn't needed by default. josh -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
dbus failure w/ NetworkManager-0.8.1-4git20100817 (x86_64)
Running updated F14-branched w/ LXDE x86_64 desktop. Updated NetworkManager to 0.8.1-4git20100817, but network connection failed upon reboot. Tried calling up nm-applet manually and got the following error (taken from /var/log/messages) Aug 19 14:55:48 Pyrope kernel: init[1]: D-Bus activation failed for NetworkManager-by-dbus.service: Invalid argument Reverting to NetworkManager-0.8.1-1.fc14.x86_64 gets me a working NetworkManager again Known issue, or bugzilla time? -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: dbus failure w/ NetworkManager-0.8.1-4git20100817 (x86_64)
On Thu, 2010-08-19 at 13:33 -0700, Steven I Usdansky wrote: > Running updated F14-branched w/ LXDE x86_64 desktop. Updated > NetworkManager to 0.8.1-4git20100817, but network connection failed upon > reboot. Tried calling up nm-applet manually and got the following error > (taken from /var/log/messages) > > Aug 19 14:55:48 Pyrope kernel: init[1]: D-Bus activation failed for > NetworkManager-by-dbus.service: Invalid argument > > Reverting to NetworkManager-0.8.1-1.fc14.x86_64 gets me a working > NetworkManager again > For me a simple # service NetworkManager restart was enough to get it working again. So there might be some connection to systemd... Martin signature.asc Description: This is a digitally signed message part -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: dbus failure w/ NetworkManager-0.8.1-4git20100817 (x86_64)
On 08/19/2010 10:33 PM, Steven I Usdansky wrote: > Running updated F14-branched w/ LXDE x86_64 desktop. Updated > NetworkManager to 0.8.1-4git20100817, but network connection failed upon > reboot. Tried calling up nm-applet manually and got the following error > (taken from /var/log/messages) > > Aug 19 14:55:48 Pyrope kernel: init[1]: D-Bus activation failed for > NetworkManager-by-dbus.service: Invalid argument > > Reverting to NetworkManager-0.8.1-1.fc14.x86_64 gets me a working > NetworkManager again > > Known issue, or bugzilla time? > > > > Yup, it's an known problem and it's already fixed in 0.8.1-5 version of NetworkManager. Kind regards Vaclav -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Anyone running F15 on Sandy-Bridge w/ Intel-Gfx?
My sandy-bridge F15 desktop with intel graphics locks up really hard a few seconds after X-Startup. As I cannot find any traces of the lockup in the logfiles I'm having a hard time filing this is a bug - is anyone running F15 successfully on similar hardware? -- sven === jabber/xmpp: s...@lankes.net -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: Emergency F16 blocker review meeting Minutes -> Adam W.
Hi Adam W., I went through the bug review meeting log [1]. I agree with the final decision which was made during the meeting (#748272); the last problem is probably specific to the EFI implementation on the netbook. However, if the reporter is described as "a bit of a pita", who needs "shepherding", it doesn't make any sense to bring it up, regardless the bug is well-founded or not. Please, next time, if you have some objections, let me know directly, I am not deaf and blind and any feedback is welcomed. Best Regards Vaclav M. [1] http://meetbot.fedoraproject.org/fedora-bugzappers/2011-11-01/f16-blocker-bug-review-meeting-5-and-a-half.2011-11-01-17.01.log.html -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
Re: Emergency F16 blocker review meeting Minutes -> Adam W.
On Wed, 2011-11-02 at 14:57 +0100, little.owl wrote: > Hi Adam W., > > I went through the bug review meeting log [1]. I agree with the final > decision which was made during the meeting (#748272); the last problem is > probably specific to the EFI implementation on the netbook. > > However, if the reporter is described as "a bit of a pita", who needs > "shepherding", it doesn't make any sense to bring it up, regardless the bug > is well-founded or not. > > Please, next time, if you have some objections, let me know directly, I am > not deaf and blind and any feedback is welcomed. > > Best Regards really sorry about that, I agree it was inappropriate, all I can say is it's the end of the cycle and I'm feeling somewhat frazzled! it was just feeling like one of those bugs where we could never quite establish what was going on. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora http://www.happyassassin.net -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
