Bug 158543

great. and the y2logs dont exist. try again player one...

Created attachment 73231 [details]
bug158543.tar.bz2

still investigating

no rc1 without a fix for this.

Mmh, looks quite ok.

2006-03-15 14:44:07 <1> 1.1.1.1(2860) [YCP] bootloader/routines/lib_iface.ycp:206 Read global options: $["__exports__%global_options%activate":"bool:Change boot-device in NV-R
AM:true", "__exports__%global_options%append":"string:Append options for kernel command line", "__exports__%global_options%boot":"select:PPC Boot loader location:", "__exports
__%global_options%boot_pmac_custom":"select:HFS boot partition::/dev/hda9", "__exports__%global_options%bootfolder":"string:Bootfolder path:", "__exports__%global_options%defa
ult":"string:Default boot section:Linux", "__exports__%global_options%initrd":"path:Default initrd path", "__exports__%global_options%root":"select:Default root device:::/dev/
hda10:/dev/hda11", "__exports__%global_options%timeout":"int:Timeout in 1/10th seconds:50:0:600", "__exports__%section_options%image_append":"string:Optional kernel command li
ne parameter", "__exports__%section_options%image_copy":"bool:Copy image to boot partition:false", "__exports__%section_options%image_image":"path:Kernel image:/boot/vmlinux",
 "__exports__%section_options%image_initrd":"path:Initial RAM disk:/boot/initrd", "__exports__%section_options%image_root":"select:Root device::/dev/hda10:/dev/hda11", "__expo
rts__%section_options%other_other":"select:Boot partition of other system::", "__exports__%section_options%type_image":"bool:Kernel section", "__exports__%section_options%type
_other":"bool:Boot other system", "__exports__arch":"pmac", "lines_cache_id":"0"]
--------------
"root" can be any of( /dev/hda10, /dev/hda11 )
"boot" can be any of(  /dev/hda9 )

So all that´s left is doing a good proposal.

BTW: the selection for partitions is made the following way:

type == Apple_HFS|Apple_Bootstrap && size < 20 cyl   ==> boot
type == "Apple_HFS" and size_cyl >= 20 ==> root

cyl size is assumed to be around 8MB (the maximum)

date.suse.de has an unrecognized boot partition, empty pulldown menu

Well
y2log shows type 'unknown instead of Apple_Bootstrap:

$["boot":true, "detected_fs":`hfs, "device":"/dev/hda11", "format":true, "fsid":257, "fstype":"unknown", "name":"hda11", "nr":11, "regi
on":[2550, 8], "size_k":51200, "type":`primary, "udev_id":["ata-Maxtor_33073H3_B_L3HXJD1C-part11"], "udev_path":"pci-0.0001f000:ata-4-ide-0:0-part11", "used_fs":`hfs]

The problem might be in yast-storage:
    global list<string> not_shown_partitions = 
        [
        "Apple_partition_map",
        "Apple_Patches",
        "Apple_Driver",
        "Apple_Loader",
        "Apple_Boot",
        "Apple_Bootstrap",
        "Apple_FWDriver",
        "Apple_ProDOS" 
        ];

that list is not referenced in the sources, but I will try to remove Apple_Bootstrap

*** Bug 154550 has been marked as a duplicate of this bug. ***

yes, changing this list doesnt seem to help. hda11 is not shown, even in the expert partitioner. where is the list of partitions to hide?

I think its this code, Apple_Boot matches also Apple_Bootstrap. It may work better if HFS and UFS is probed first, and Apple_Bootstrap is handled explicit.

              if( id == Partition::ID_LINUX )
                  {
                  if( val.find( "Apple_partition" ) != string::npos ||
                      val.find( "Apple_Driver" ) != string::npos ||
                      val.find( "Apple_Loader" ) != string::npos ||
                      val.find( "Apple_Boot" ) != string::npos ||
                      val.find( "Apple_ProDOS" ) != string::npos ||
                      val.find( "Apple_FWDriver" ) != string::npos ||
                      val.find( "Apple_Patches" ) != string::npos )
                      {
                      id = Partition::ID_APPLE_OTHER;
                      }
                  else if( val.find( "Apple_HFS" ) != string::npos )
                      {
                      id = Partition::ID_APPLE_HFS;
                      }
                  else if( val.find( "Apple_UFS" ) != string::npos )
                      {
                      id = Partition::ID_APPLE_UFS;
                      }
                  }

./storage/libstorage/src/Disk.cc

The above code is relavent the one from ycp is obsolete.
Should I handle "Apple_Bootstrap" the same as "Apple_HFS"?

yes, I think that is a good assumption. Since this partition in fact has a HFS filesystem, handling it like Apple_HFS makes sense.

Currently "Apple_HFS" has "fsid":258, "fstype":"Apple_HFS".

What is the "fsid" for "Apple_Bootstrap"?

There is no extra "fsid" for "Apple_Bootstrap".
I am about to change libstorage to handle "Apple_Bootstrap" like "Apple_HFS"
which means it will how up with is 258 and text "Apple_HFS" in expert partitioner.

Is this ok?

yes, can you build me a lib for testing?

fehr@ comment #15: yes, fsid 258 is OK.

mbuild job Lagrange-fehr-2 is running if it is finished you can use it for
testing.

yes, this version works, I replaced liby2storage.so

My original defect was 154550. So, my question is still "Do we care about getting SLES10 PPC boot loader to work on G5 Mac?" I was told using G5 Mac for PPC testing is not really a valid hardware.

If we do want to get PPC boot loader to work on G5 Mac, then I will continue testing with it.

Looking closer at the defect, this defect is for SuSE 10.1. So, the question is if PPC bootloader is supported in G5 Mac, is it supported for SLES10 since my defect#154550 for SLES10 is marked duplicated of this one?

The proposal for a bootloader device on pmac systems is fixed in
yast2-bootloader-2.13.44.

ctiong@: mfrueh should be able to answer your questions about supported
features.

Taking mfrueh to NEEDINFO.

Forgot to only use partitions with size < 20 cylinder.

Fix is in yast2-bootloader-2.13.45.

mfrueh told me that we want to have a working bootloader installation for
PPC on SL 10.1.