[Bugme-new] [Bug 4187] New: I'd like please to urge the relevant powers that be to merge the 2.6.10-bk9-libata-dev1.patch

[bugme-daemon at osdl.org]

into the FC4 development kernel.
X-Bugzilla-Reason:CC
X-Bugzilla-Category: Drivers
X-Bugzilla-Component: SCSI

http://bugme.osdl.org/show_bug.cgi?id=4187

           Summary: I'd like please to urge the relevant powers that be to
                    merge the 2.6.10-bk9-libata-dev1.patch
                    into the FC4 development kernel.
    Kernel Version: vermagic:       2.6.10-1.1134_FC4 586 REGPARM 4KSTACKS
                    gcc-3.4
            Status: NEW
          Severity: high
             Owner: andmike at us.ibm.com
         Submitter: Andrew.Hamilton at colorado.edu


Distribution: FC4
Hardware Environment: Alienware Area-51m 7700, Promise SATA378 TX2+ controller
Software Environment: FC4 development
Problem Description: The 2.6.10-bk9-libata-dev1.patch patch is required for an
install to recognize a hard disk, but the installation rejects any attempt to
use patched modules.

Steps to reproduce:

Problem: the Fedora installation process does not recognize a hard disk
with a Promise SATA378 TX2+ controller, and crashes as soon as
anaconda attempts to begin an install.
lspci, after booting in rescue mode from a FC3 disk, gives:
0a:02.0 RAID bus controller: Promise Technology, Inc. PDC20378 (FastTrak
378.SATA 378) (rev 02)
                                                                                
Solution: install from a kernel patched with the above patch.
                                                                                
Unfortunately, without a disk to start from, there is no patched kernel
from which to install.
                                                                                
I tried making a Fedora driver disk with the patched modules on it,
and then installing from a cdrom containing the latest FC4 boot.iso, using
linux dd expert askmethod
The installer recognizes the driver disk and modules,
but then rejects them because:
<4>sata_promise: disagrees about version of symbol struct_module
and similar messages for the other modules on the driver disk.
                                                                                
Note that this is not a version magic problem: I took care that the modinfo
vermagic:       2.6.10-1.1134_FC4 586 REGPARM 4KSTACKS gcc-3.4
of the revised modules agreed precisely with those on the Fedora boot.iso.
However, the srcversion does not agree with the original modules,
and there's nothing I can do about that.
                                                                                
So I tried making a revised version of the Fedora boot image boot.iso,
with the kernel vmlinuz and the modules all in agreement with each other.
This works to the point that the installer appears to recognize (but not
partition) my hard disk, and for a moment it looks like all is well.
                                                                                
Unfortunately, when I then attempt a network install,
after loading base/stage2.img, the installation reports:
The Fedora Core installation tree in that directory does not seem to
match your boot media.
                                                                                
Googling what that means indicates that it is an md5sum problem.
                                                                                
I have tried a bunch of things to work around the problem, but
it seems that it is impossible.  The only feasible solution is that
the Fedora project patches the patch into the development kernel.
                                                                                
I realize that there is usually some delay between a patch appearing
and it being incorporated into the kernel, so that users can report
back potential problems.  In this case, however, it is impossible
to install the patch unless the patch is already in the kernel.
Catch 22 situation.

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.