On Thu, Jan 5, 2012 at 9:09 AM, Andre Robatino
<robatino(a)fedoraproject.org> wrote:
Andre Robatino <robatino <at> fedoraproject.org> writes:
> I get similar output on F16, for example "rpm -V
kernel-3.1.6-1.fc16.x86_64"
> gives
>
> ?........ /boot/System.map-3.1.6-1.fc16.x86_64
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.alias
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.alias.bin
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.builtin.bin
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.ccwmap
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.dep
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.dep.bin
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.devname
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.ieee1394map
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.inputmap
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.isapnpmap
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.ofmap
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.pcimap
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.seriomap
> ...T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.softdep
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.symbols
> 5S.T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.symbols.bin
> 5..T..... /lib/modules/3.1.6-1.fc16.x86_64/modules.usbmap
On the other hand,
[andre@compaq-pc tmp]$ rpm -V kernel-3.1.5-6.fc16.x86_64
?........ /boot/System.map-3.1.5-6.fc16.x86_64
...T..... /lib/modules/3.1.5-6.fc16.x86_64/modules.devname
...T..... /lib/modules/3.1.5-6.fc16.x86_64/modules.softdep
[andre@compaq-pc tmp]$ rpm -V kernel-3.1.5-2.fc16.x86_64
?........ /boot/System.map-3.1.5-2.fc16.x86_64
...T..... /lib/modules/3.1.5-2.fc16.x86_64/modules.devname
...T..... /lib/modules/3.1.5-2.fc16.x86_64/modules.softdep
[andre@compaq-pc tmp]$
So maybe the same issue now exists in F16 as well (I'll check whether the
deltarpm for the next F16 kernel update works).
The modules.* files (most of them anyway) are created by depmod, which
is run by new-kernel-package in the %posttrans section of the
kernel.spec. That means they are generated at RPM install time.
The 3.1.6 and rawhide kernels both contain a compat-wireless addition
to them that builds and installs newer wireless drivers. I know the
build mechanism was doing a lot of modprobe calls on those during RPM
build time and John fixed that recently. Perhaps it was causing those
files to actually get packaged from %build/%install and then the
%posttrans was overwritting the packaged files as expected. That
would cause the output you see I think.
John, any thoughts if that might be the case?
josh