Bug#717511: Kernel does not start up on Pegasos-2
On 07/22/2013 02:42 PM, Ben Hutchings wrote: On Mon, Jul 22, 2013 at 09:59:18PM +0400, Pavel Fedin wrote: Hello, Milan. 22 июля 2013 г., 5:23:34, you wrote: User did not let installer to create default partition layout but manually created incorrect one, or installed on a previously created partitions and used ext4 and ext3 file-systems. Yes, exactly. Actually i agree, there's little to blame in Installer. Only one small thing - nobody told me that my layout (previously worked) became incorrect. Perhaps Installer should warn about it ? Or, we can consider one more corner case. What if i install the whole OS to a single EXT2 partition ? Yes, again, i agree, you may ask Why, EXT2 is outdated and there's cool new EXT4. I have already said - this is corner case. Maybe the installer could warn about a likely broken partition layout after you choose manual layout. It doesn't seem right that if you want to customise layout at all you must remember any machine-specific constraints. Also, this constraint is not documented in the installation manual. Reassigning to installation-guide-powerpc and partman-ext3. Milan -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
Hello, Milan. 22 июля 2013 г., 5:23:34, you wrote: User did not let installer to create default partition layout but manually created incorrect one, or installed on a previously created partitions and used ext4 and ext3 file-systems. Yes, exactly. Actually i agree, there's little to blame in Installer. Only one small thing - nobody told me that my layout (previously worked) became incorrect. Perhaps Installer should warn about it ? Or, we can consider one more corner case. What if i install the whole OS to a single EXT2 partition ? Yes, again, i agree, you may ask Why, EXT2 is outdated and there's cool new EXT4. I have already said - this is corner case. -- Kind regards, Pavel mailto:pavel_fe...@mail.ru -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
On Mon, Jul 22, 2013 at 09:59:18PM +0400, Pavel Fedin wrote: Hello, Milan. 22 июля 2013 г., 5:23:34, you wrote: User did not let installer to create default partition layout but manually created incorrect one, or installed on a previously created partitions and used ext4 and ext3 file-systems. Yes, exactly. Actually i agree, there's little to blame in Installer. Only one small thing - nobody told me that my layout (previously worked) became incorrect. Perhaps Installer should warn about it ? Or, we can consider one more corner case. What if i install the whole OS to a single EXT2 partition ? Yes, again, i agree, you may ask Why, EXT2 is outdated and there's cool new EXT4. I have already said - this is corner case. Maybe the installer could warn about a likely broken partition layout after you choose manual layout. It doesn't seem right that if you want to customise layout at all you must remember any machine-specific constraints. Also, this constraint is not documented in the installation manual. Ben. -- Ben Hutchings We get into the habit of living before acquiring the habit of thinking. - Albert Camus -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
On Mon, Jul 22, 2013 at 09:59:18PM +0400, Pavel Fedin wrote: Perhaps Installer should warn about it ? Yes it would be nice. On 22/07/13 19:42, Ben Hutchings wrote: Maybe the installer could warn about a likely broken partition layout after you choose manual layout. It already warns if, for example, you don't create a swap partition. It also warns about using ZFS on kfreebsd-i386 (on which it is not suggested to be stable), or using ZFS on kfreebsd-amd64 with less than 512 MiB RAM. So that was probably borrowing the same code. Some other configurations can lead to a GRUB failure at the end of the install process; it would be most helpful if the installer warned about this at the partitioning stage instead: * gpt disklabel with no BIOS Boot Partition to put GRUB boot blocks; * (re)using an msdos disklabel with insufficient gap before the first partition starts (installer now does the right thing and places it after 1 MiB, but only if you're creating the partition or disklabel afresh on that media); * /boot on reiserfs (and probably others) AFAIK isn't useful for GRUB, because it doesn't know how to replay the journal / transaction log; I think that was also the case for ext3 and ext4 until recent versions, but I'd still use ext2 for it out of habit. Regards, -- Steven Chamberlain ste...@pyro.eu.org -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
Package: linux-image-3.2.0-4-powerpc Version: 3.2.46-1 Hello! I have just upgraded Linux installation on my Pegasos-II machine to Debian 7.1.0 Wheezy. The installer worked flawlessly, but the newly installed kernel refused to boot up. Attempt to boot up the kernel ends up in a text-mode screen filled up with garbage (randon characters and attributes). What's up ? Did you drop Pegasos support ? -- С уважением, Pavel mailto:pavel_fe...@mail.ru -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
On Sun, 2013-07-21 at 21:28 +0400, Pavel Fedin wrote: Package: linux-image-3.2.0-4-powerpc Version: 3.2.46-1 Hello! I have just upgraded Linux installation on my Pegasos-II machine to Debian 7.1.0 Wheezy. The installer worked flawlessly, but the newly installed kernel refused to boot up. Attempt to boot up the kernel ends up in a text-mode screen filled up with garbage (randon characters and attributes). What's up ? Did you drop Pegasos support ? There was no decision not to support them, but no-one is actively working to support them either. This might be relevant: https://lists.debian.org/debian-powerpc/2013/07/msg8.html Michael, what are we supposed to do about radeon? I thought KMS still didn't work on PowerMac laptops. Can we make the default depend on the machine type? Ben. -- Ben Hutchings The most exhausting thing in life is being insincere. - Anne Morrow Lindberg signature.asc Description: This is a digitally signed message part
Bug#717511: Kernel does not start up on Pegasos-2
Hello, Ben. 21 июля 2013 г., 21:56:24, you wrote: This might be relevant: https://lists.debian.org/debian-powerpc/2013/07/msg8.html Unlikely. Looks like the kernel simply does not boot up. It does not say anything. Nor it switches video mode. I just enter boot command and see random flashing rectangles like on old good ZX Spectrum :) I have one suggestion about it. Looks like this kernel has rather big initrd attached. If so, technically it might occasionally overwrite firmware's code in RAM and kill itself even before the actual startup. I have another machine with a serial port here, i'll try to redirect the output to serial port and see what happens. Also i'm currently downloading i386 version in order to install on my AspireOne. Using it i will be able to cross-build PPC version of the kernel, and try to run it. This will also help me to either prove or reject my guess about initrd. However actually this guess comes from two facts: 1. Installed vmlinuz takes rather long time to load. 2. CD installer kernel takes much shorter time to load, and works fine. I don't believe there are many differences between CD version and working version. At least not in early startup code. -- Kind regards, Pavel mailto:pavel_fe...@mail.ru -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
On Sun, 2013-07-21 at 22:07 +0400, Pavel Fedin wrote: Hello, Ben. 21 июля 2013 г., 21:56:24, you wrote: This might be relevant: https://lists.debian.org/debian-powerpc/2013/07/msg8.html Unlikely. Looks like the kernel simply does not boot up. It does not say anything. Nor it switches video mode. I just enter boot command and see random flashing rectangles like on old good ZX Spectrum :) I have one suggestion about it. Looks like this kernel has rather big initrd attached. If so, technically it might occasionally overwrite firmware's code in RAM and kill itself even before the actual startup. If you run the installer in expert mode you can choose to create a 'targeted' initramfs which should be much smaller. I have another machine with a serial port here, i'll try to redirect the output to serial port and see what happens. Also i'm currently downloading i386 version in order to install on my AspireOne. Using it i will be able to cross-build PPC version of the kernel, and try to run it. This will also help me to either prove or reject my guess about initrd. However actually this guess comes from two facts: 1. Installed vmlinuz takes rather long time to load. 2. CD installer kernel takes much shorter time to load, and works fine. I would expect the installer initramfs to be bigger than a normal generic initramfs, but maybe I'm wrong. Ben. I don't believe there are many differences between CD version and working version. At least not in early startup code. -- Ben Hutchings The most exhausting thing in life is being insincere. - Anne Morrow Lindberg signature.asc Description: This is a digitally signed message part
Bug#717511: Kernel does not start up on Pegasos-2
Hi, Just for the record, I've installed Wheezy successfully on two Pegasos IIs, one is a rev 2B1 board, which was installed from scratch, and worked without major glitches, apart from the fact that the installer required an installation to MBR partitioned disks, because with RDB disks it said Fatal error: Out of memory, when it tried to create the main partition on an 80GB ATA disk. :) The other one is a rev 2B3 board, it was upgraded from Squeeze, required some manual fiddling, but worked in the end. Stock 3.2 kernel and initramfs in both cases, binary from the ISO (or from the net or what it builds by default anyway). Both Pegs have 1GB RAM and Radeon 9200. Everything worked, except the GBit eth controller. On Sun, 21 Jul 2013, Ben Hutchings wrote: Unlikely. Looks like the kernel simply does not boot up. It does not say anything. Nor it switches video mode. I just enter boot command and see random flashing rectangles like on old good ZX Spectrum :) I have one suggestion about it. Looks like this kernel has rather big initrd attached. If so, technically it might occasionally overwrite firmware's code in RAM and kill itself even before the actual startup. If you run the installer in expert mode you can choose to create a 'targeted' initramfs which should be much smaller. I've seen this. I can't recall if the problem was I've booted vmlinux instead of vmlinuz accidentally, or got the video= arguments wrong. (Ps: sorry I've not replied to the answer to my previous question, I had no chance to try the suggestion on the actual machine.) Regards, -- Karoly Balogh -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
Hello, Karoly. 21 июля 2013 г., 22:34:12, you wrote: Just for the record, I've installed Wheezy successfully on two Pegasos IIs, one is a rev 2B1 board, which was installed from scratch, and worked without major glitches, apart from the fact that the installer required an installation to MBR partitioned disks, because with RDB disks it said Fatal error: Out of memory, when it tried to create the main partition on an 80GB ATA disk. :) I am sorry for the false alarm... After posting this message i decided to reinstall the system... The story begins with the fact that until now i used single EXT partition for Linux. In the far past it used to be EXT2, then when i upgraded to Etch it became EXT3. There was no separate /boot partition because Pegasos firmware was perfectly able to read that partition. But now, first i decided to use EXT4 (yes, i performed clean reinstall). Firmware did not know what is EXT4 and could not read /boot directory. Fault. I reinstalled again, this time i used EXT3. But... Hah! This EXT3 appears to be somehow incompatible with my older EXT3. The firmware was not able to read it. But, fortunately, MorphOS was able to. I quickly extracted vmlinux file from /boot directory, copied it to SFS partition (where MorphOS lives), and booted it from there. But this time i decided to do everything in a good way. I cut a small 120MB piece from my old Linux partition and decided to use it as EXT2 /boot. The rest became EXT4 again. I reinstalled in this configuration, and... In /boot i saw two images. One had vmlinux name and another had vmlinuz name. I ran vmlinuz and everything worked fine! :) So, it's not kernel's fault. It seems to be installer fault which does not generate correct vmlinuz image when not using separate /boot partition. vmlinux, i know, is another thing. It's a plain image meant to be ran from GRUB or yaboot. It cannot be run from OFW. So, we can close this, probably with a notification to Installer guys. -- Kind regards, Pavel mailto:pavel_fe...@mail.ru -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#717511: Kernel does not start up on Pegasos-2
Control: reassign -1 yaboot-installer 1.1.26 On Mon, 2013-07-22 at 00:38 +0400, Pavel Fedin wrote: [...] So, it's not kernel's fault. It seems to be installer fault which does not generate correct vmlinuz image when not using separate /boot partition. vmlinux, i know, is another thing. It's a plain image meant to be ran from GRUB or yaboot. It cannot be run from OFW. So, we can close this, probably with a notification to Installer guys. There is no need to close it. I'm reassigning to the relevant part of the installer. Ben. -- Ben Hutchings The most exhausting thing in life is being insincere. - Anne Morrow Lindberg signature.asc Description: This is a digitally signed message part
Bug#717511: Kernel does not start up on Pegasos-2
On Sun, 2013-07-21 at 20:36 -0400, Milan Kupcevic wrote: Control: notfound -1 yaboot-installer/1.1.26 On 07/21/2013 04:55 PM, Ben Hutchings wrote: Control: reassign -1 yaboot-installer 1.1.26 On Mon, 2013-07-22 at 00:38 +0400, Pavel Fedin wrote: [...] So, it's not kernel's fault. It seems to be installer fault which does not generate correct vmlinuz image when not using separate /boot partition. vmlinux, i know, is another thing. It's a plain image meant to be ran from GRUB or yaboot. It cannot be run from OFW. So, we can close this, probably with a notification to Installer guys. There is no need to close it. I'm reassigning to the relevant part of the installer. The particular partition/file-system layout reported in this bug report cannot work. This platform does not have a boot loader at all. Sorry, I misunderstood. [...] Installer correctly creates ext2 boot partition and ext4 root partition. [...] Then how did this go wrong? Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part
Bug#717511: Kernel does not start up on Pegasos-2
Control: notfound 717511 yaboot-installer/1.1.26 On 07/21/2013 08:49 PM, Ben Hutchings wrote: On Sun, 2013-07-21 at 20:36 -0400, Milan Kupcevic wrote: Control: notfound -1 yaboot-installer/1.1.26 On 07/21/2013 04:55 PM, Ben Hutchings wrote: Control: reassign -1 yaboot-installer 1.1.26 On Mon, 2013-07-22 at 00:38 +0400, Pavel Fedin wrote: [...] So, it's not kernel's fault. It seems to be installer fault which does not generate correct vmlinuz image when not using separate /boot partition. vmlinux, i know, is another thing. It's a plain image meant to be ran from GRUB or yaboot. It cannot be run from OFW. So, we can close this, probably with a notification to Installer guys. There is no need to close it. I'm reassigning to the relevant part of the installer. The particular partition/file-system layout reported in this bug report cannot work. This platform does not have a boot loader at all. Sorry, I misunderstood. [...] Installer correctly creates ext2 boot partition and ext4 root partition. [...] Then how did this go wrong? User did not let installer to create default partition layout but manually created incorrect one, or installed on a previously created partitions and used ext4 and ext3 file-systems. Milan -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org