Re: help with grub 1.99-2
В Sat, 9 Aug 2014 15:25:50 -0400 Hernán Fernández her...@mitierra.cl пишет: thanks Jordan, I tried to analyze the outputs, without success echo $prefix return (hd0,gpt2)/EFI/Ubuntu Are you using legacy BIOS boot or EFI boot? You have both bootloaders and we do not know which one is actually used. Also I'm afraid your grub version is a bit old to show this information at runtime. bootinfo script returned (here (,gpt6)/boot/grub called my attention, I don't know what is gpt6) It is partition 6 on your drive. According to output it does contain elementary OS so it appears to be correct. Boot Info Script 0.61 [1 April 2012] = Boot Info Summary: === = Grub2 (v1.99) is installed in the MBR of /dev/sda and looks at sector 389763072 of the same hard drive for core.img. core.img is at this location and looks for (,gpt6)/boot/grub on this drive. sda1: __ File system: ntfs Boot sector type: Windows Vista/7: NTFS Boot sector info: No errors found in the Boot Parameter Block. Operating System: Boot files: sda2: __ File system: vfat Boot sector type: Unknown Boot sector info: No errors found in the Boot Parameter Block. Operating System: Boot files:/grub.cfg /efi/Boot/bootx64.efi /efi/elementary/grubx64.efi /efi/elementary/shimx64.efi /bootmgr /boot/bcd sda3: __ File system: Boot sector type: - Boot sector info: Mounting failed: mount: unknown filesystem type '' sda4: __ File system: ntfs Boot sector type: Windows Vista/7: NTFS Boot sector info: No errors found in the Boot Parameter Block. Operating System: Boot files:/bootmgr /Windows/System32/winload.exe sda5: __ File system: BIOS Boot partition Boot sector type: Grub2's core.img Boot sector info: sda6: __ File system: ext4 Boot sector type: - Boot sector info: Operating System: elementary OS Luna Boot files:/boot/grub/grub.cfg /etc/fstab /boot/grub/core.img sda7: __ File system: swap Boot sector type: - Boot sector info: Drive/Partition Info: = Drive: sda _ Disk /dev/sda: 500.1 GB, 500107862016 bytes 255 heads, 63 sectors/track, 60801 cylinders, total 976773168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 4096 bytes Partition Boot Start SectorEnd Sector # of Sectors Id System /dev/sda1 1 976,773,167 976,773,167 ee GPT GUID Partition Table detected. PartitionStart SectorEnd Sector # of Sectors System /dev/sda1 2,048 1,023,999 1,021,952 Windows Recovery Environment (Windows) /dev/sda2 1,024,000 1,638,399 614,400 EFI System partition /dev/sda3 1,638,400 1,900,543 262,144 Microsoft Reserved Partition (Windows) /dev/sda4 1,900,544 389,761,422 387,860,879 Data partition (Windows/Linux) /dev/sda5 389,763,072 389,765,119 2,048 BIOS Boot partition /dev/sda6 389,765,120 960,231,423 570,466,304 Data partition (Windows/Linux) /dev/sda7 960,231,424 976,771,07116,539,648 Swap partition (Linux) blkid output: Device UUID TYPE LABEL /dev/sda1B60E5B6E0E5B26A1 ntfs Windows RE tools /dev/sda2145D-ECAB vfat SYSTEM /dev/sda41A125FE5125FC503 ntfs /dev/sda658b9c768-d9fc-484f-a6f6-54cc25216c4a ext4 /dev/sda73acd7db7-2c97-459c-9257-880979edcedf swap Mount points: = Device Mount_Point Type Options /dev/sda6/ext4 (rw,errors=remount-ro) sda2/grub.cfg:
Re: help with grub 1.99-2
В Sun, 10 Aug 2014 01:11:12 -0400 Hernán Fernández her...@mitierra.cl пишет: The BIOS has a mode called CMS, that allow boot windows 8 and Linux. Its a Samsung notebook NP470r. You probably mean CSM. Anyway, it sounds like you actually are not using CSM. When you are in grub command line, do you have e.g. drivemap command (simply try drivemap on command line - does it return an error or anything else?) It should be the same grub version than Ubuntu. On Aug 10, 2014 12:47 AM, Andrey Borzenkov arvidj...@gmail.com wrote: В Sat, 9 Aug 2014 15:25:50 -0400 Hernán Fernández her...@mitierra.cl пишет: thanks Jordan, I tried to analyze the outputs, without success echo $prefix return (hd0,gpt2)/EFI/Ubuntu Are you using legacy BIOS boot or EFI boot? You have both bootloaders and we do not know which one is actually used. Also I'm afraid your grub version is a bit old to show this information at runtime. bootinfo script returned (here (,gpt6)/boot/grub called my attention, I don't know what is gpt6) It is partition 6 on your drive. According to output it does contain elementary OS so it appears to be correct. Boot Info Script 0.61 [1 April 2012] = Boot Info Summary: === = Grub2 (v1.99) is installed in the MBR of /dev/sda and looks at sector 389763072 of the same hard drive for core.img. core.img is at this location and looks for (,gpt6)/boot/grub on this drive. sda1: __ File system: ntfs Boot sector type: Windows Vista/7: NTFS Boot sector info: No errors found in the Boot Parameter Block. Operating System: Boot files: sda2: __ File system: vfat Boot sector type: Unknown Boot sector info: No errors found in the Boot Parameter Block. Operating System: Boot files:/grub.cfg /efi/Boot/bootx64.efi /efi/elementary/grubx64.efi /efi/elementary/shimx64.efi /bootmgr /boot/bcd sda3: __ File system: Boot sector type: - Boot sector info: Mounting failed: mount: unknown filesystem type '' sda4: __ File system: ntfs Boot sector type: Windows Vista/7: NTFS Boot sector info: No errors found in the Boot Parameter Block. Operating System: Boot files:/bootmgr /Windows/System32/winload.exe sda5: __ File system: BIOS Boot partition Boot sector type: Grub2's core.img Boot sector info: sda6: __ File system: ext4 Boot sector type: - Boot sector info: Operating System: elementary OS Luna Boot files:/boot/grub/grub.cfg /etc/fstab /boot/grub/core.img sda7: __ File system: swap Boot sector type: - Boot sector info: Drive/Partition Info: = Drive: sda _ Disk /dev/sda: 500.1 GB, 500107862016 bytes 255 heads, 63 sectors/track, 60801 cylinders, total 976773168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 4096 bytes Partition Boot Start SectorEnd Sector # of Sectors Id System /dev/sda1 1 976,773,167 976,773,167 ee GPT GUID Partition Table detected. PartitionStart SectorEnd Sector # of Sectors System /dev/sda1 2,048 1,023,999 1,021,952 Windows Recovery Environment (Windows) /dev/sda2 1,024,000 1,638,399 614,400 EFI System partition /dev/sda3 1,638,400 1,900,543 262,144 Microsoft Reserved Partition (Windows) /dev/sda4 1,900,544 389,761,422 387,860,879 Data partition (Windows/Linux) /dev/sda5 389,763,072 389,765,119 2,048 BIOS Boot partition /dev/sda6 389,765,120 960,231,423 570,466,304 Data partition (Windows/Linux) /dev/sda7 960,231,424 976,771,07116,539,648 Swap partition (Linux) blkid output: Device UUID
Re: Non-Linux, non-Unix platforms
В Fri, 1 Aug 2014 23:25:03 -0600 Dee Sharpe demetrioussha...@netscape.net пишет: Is it possible to use Grub2 on platforms other than Linux? Such platforms may not use the run level setup, which Grub2 seems to rely on. What are you talking about? grub has nothing to do with run level. Also, what about platforms that aren't Unix-based? Such platforms may not run bash scripts or have etc/ directories. Current version 2.02 beta2 switched from shell scripts to compiled binaries; e.g. Windows is supported natively, you can download prebuilt package from alpha.gnu.org. Also platform support code is sufficiently modular so adding support for new systems should be easier. Forgive me if this information is provided elsewhere. I've googled this subject I can only find dual-booting examples. I'd like to find out how to setup use Grub2 with an alternative OS, as a first class platform. Apollo Demetrious Sharpe ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Grub on a Micron Trek 2.
В Fri, 11 Jul 2014 11:10:34 -0700 pe...@easthope.ca пишет: OK, I commented everything but the menuentry command. Also removed boot from the menuentry. Still no menu and grub drops to the command interface. chainloader (hd0,5)+1enterboot starts the target system. If the --unrestricted option is removed from menuentry, then the menu appears and the target system starts from it. I don't recognize why that option blocks the menu. I can't reproduce it using current master head. I created rescue image with grub.cfg containing your single menu entry (copy-paste) and when I boot I'm presented with menu. May be something else was changed besides --unrestricted. In any case, grub is working now. Thanks for the tips,... Peter E. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: install two instance of linux in uefi mode.
On Tue, Jul 8, 2014 at 10:24 AM, Bizhan Gholikhamseh (bgholikh) bghol...@cisco.com wrote: Hi I have partitioned my disk to /dev/sda1 efi, /dev/sda2 ext4, /dev/sda3, ext4 /dev/sda4 swap. I boot my system in uefi mode, the /dev/sda1 has efi which is mounted on /boot/efi. This is ubuntu 14.04 system installed on /dev/sda2. I managed to install another instance of ubuntu 14.04 using debootstrap on /dev/sda3, now I need to install grub for that partition, what should I do please? grub2 installs bootloader in \EFI\${GRUB_DISTRIBUTOR} directory, where GRUB_DISTRIBUTOR is defined in /etc/default/grub. After installation you can manually edit this file; whether it can be changed during installation you probably need to ask on your distribution support list. I like to understand what you suggesting. If GRUB_DISTRIBUTOR is modifiable. Then you suggesting to install the grub for the new slot (/dev/sda3) \EFI\New destination\ directory ? Please let me know. Bootloader is installed in ESP (EFI System Partition) which in your case is sda1. That is why you need to different GRUB_DISTRIBUTOR, to distinguish between two different instances of the same OS. Alternatively you could create second ESP for your second OS. Linux should have no problems with it. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Access to DMI/SMBIOS data from GRUB
On Tue, Jul 8, 2014 at 6:09 AM, Michael Mohr michaelm...@hyvesolutions.com wrote: Good evening; I'm looking for a way to read strings stored in a server's DMI tables from GRUB2. In grub-core/loader/multiboot.c, there is a FIXME which states that SMBIOS tables are not supported. However, in grub-core/commands/efi/loadbios.c's fake_bios_data, it seems that there may well be a way to access this data. Is there code available somewhere which encapsulates the process of deserializing this data? If so, how do I use it? If not, how difficult would it be to implement? I'm looking at this as a solution to boot servers in a datacenter based on e.g. the system serial number as saved in DMI. Check grub-devel archives, there was proposed patch to access SMBIOS information. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Accessing UEFI variables from GRUB
В Wed, 9 Jul 2014 01:14:58 + Michael Mohr michaelm...@hyvesolutions.com пишет: Good evening; Is it possible to read EFI variables from grub-efi (i.e. grubx64.efi) via the configuration file? I would like to change the behavior of a server (i.e. boot order, kernel options, etc) based on one or more variables from EFI. No, it is not implemented so far. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Debian squeeze on raid 0 and install grub2
On Wed, Jul 2, 2014 at 10:25 AM, Damien Moity lebasqu...@hotmail.com wrote: Hello, I wrote a script that allows me to install the distribution debian squeeze on a RAID 0 but the script crashes when installing grub : --- grub-install --modules=raid mdraid part_gpt ext2 /dev/md0 Is it software MD raid or fake-raid (dmraid)? It appears pure software raid, but just to be sure ... /usr/sbin/grub-setup: warn: Attempting to install GRUB to a partitionless disk. This is a BAD idea.. /usr/sbin/grub-setup: error: embedding is not possible, but this is required when the root device is on a RAID array or LVM volume. --- Grub version : --- update-grub -v grub-mkconfig (GRUB) 1.98+20100804-14+squeeze1 grub-install -v grub-install (GRUB) 1.98+20100804-14+squeeze1 --- First I boot on a System Rescue CD and I create 'logical raid0' with 'mdadm' : --- mdadm -D /dev/md0 /dev/md0: Version : 1.2 Creation Time : Tue Jul 1 14:07:21 2014 Raid Level : raid0 Array Size : 52427776 (50.00 GiB 53.69 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Update Time : Tue Jul 1 14:07:21 2014 State : clean Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Chunk Size : 512K Name : sysresccd:0 (local to host sysresccd) UUID : 972ee921:bc2ecdb5:0c9b9350:14cbf787 Events : 0 Number Major Minor RaidDevice State 0 800 active sync /dev/sda 1 8 161 active sync /dev/sdb --- I create raid0 partitions with 'gdisk' : --- root@sysresccd /root % gdisk -l /dev/md0 GPT fdisk (gdisk) version 0.8.8 Partition table scan: MBR: protective BSD: not present APM: not present GPT: present Found valid GPT with protective MBR; using GPT. Disk /dev/md0: 10482 sectors, 50.0 GiB Logical sector size: 512 bytes Disk identifier (GUID): 45EC12F7-47B7-4339-A219-199F35A51B5F Partition table holds up to 128 entries First usable sector is 34, last usable sector is 104855518 Partitions will be aligned on 2048-sector boundaries Total free space is 2014 sectors (1007.0 KiB) Number Start (sector)End (sector) Size Code Name 12048 18431 8.0 MiB EF02 BIOS boot partition 2 1843241961471 20.0 GiBEF00 Linux filesystem 34196147244058623 1024.0 MiB 8200 Linux swap 444058624 104855518 29.0 GiB8300 Linux filesystem --- I make filesystem on created partitions : * part1 (md0p1) with mkfs.ext3 = bios_grub * part2 (md0p2) with mkfs.ext3 = / * part3 (md0p3) with mkswap = swap * part4 (md0p4) with mkswap = /srv I install debian distribution on md0p2 : * partimage -B=foo -b restore /dev/md0p2 my_image_debian * e2fsck -f -y /dev/md0p2 * resize2fs -f /dev/md0p2 I mount ressources : --- mount -l | grep md0 /dev/md0p2 on /mnt/gentoo type ext3 (rw) /dev/md0p4 on /mnt/gentoo/srv type ext3 (rw) --- I init fstab and mtab with the new partitions. And now i want to install grub. I write disk on /mnt/gentoo/boot/grub/device.map : --- cat /mnt/gentoo/boot/grub/device.map (md0) /dev/disk/by-id/md-uuid-972ee921:bc2ecdb5:0c9b9350:14cbf787 (md0p2) /dev/disk/by-id/md-uuid-972ee921:bc2ecdb5:0c9b9350:14cbf787-part2 (hd0) /dev/disk/by-id/ata-VBOX_HARDDISK_VB4b33f8ee-87039d31 (hd1) /dev/disk/by-id/ata-VBOX_HARDDISK_VBc34d645e-16ef1e20 --- I update grub with : chroot /mnt/gentoo /bin/bash update-grub and and after installation of grub crashes. Do you have any ideas on what could be wrong ? I read many forums but I have not found a solution. Thanks for your help. Damien. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Debian squeeze on raid 0 and install grub2
On Wed, Jul 2, 2014 at 12:13 PM, Damien Moity lebasqu...@hotmail.com wrote: Is it software MD raid or fake-raid (dmraid)? It appears pure software raid, but just to be sure ... It is software raid create with 'mdadm'. It's not going to work. You cannot install bootloader on software raid0. You have to install it on each individual disk. You will need to install it in MBR so it can be embedded. Le 02/07/2014 10:00, Andrey Borzenkov a écrit : On Wed, Jul 2, 2014 at 10:25 AM, Damien Moity lebasqu...@hotmail.com wrote: Hello, I wrote a script that allows me to install the distribution debian squeeze on a RAID 0 but the script crashes when installing grub : --- grub-install --modules=raid mdraid part_gpt ext2 /dev/md0 Is it software MD raid or fake-raid (dmraid)? It appears pure software raid, but just to be sure ... /usr/sbin/grub-setup: warn: Attempting to install GRUB to a partitionless disk. This is a BAD idea.. /usr/sbin/grub-setup: error: embedding is not possible, but this is required when the root device is on a RAID array or LVM volume. --- Grub version : --- update-grub -v grub-mkconfig (GRUB) 1.98+20100804-14+squeeze1 grub-install -v grub-install (GRUB) 1.98+20100804-14+squeeze1 --- First I boot on a System Rescue CD and I create 'logical raid0' with 'mdadm' : --- mdadm -D /dev/md0 /dev/md0: Version : 1.2 Creation Time : Tue Jul 1 14:07:21 2014 Raid Level : raid0 Array Size : 52427776 (50.00 GiB 53.69 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Update Time : Tue Jul 1 14:07:21 2014 State : clean Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Chunk Size : 512K Name : sysresccd:0 (local to host sysresccd) UUID : 972ee921:bc2ecdb5:0c9b9350:14cbf787 Events : 0 Number Major Minor RaidDevice State 0 800 active sync /dev/sda 1 8 161 active sync /dev/sdb --- I create raid0 partitions with 'gdisk' : --- root@sysresccd /root % gdisk -l /dev/md0 GPT fdisk (gdisk) version 0.8.8 Partition table scan: MBR: protective BSD: not present APM: not present GPT: present Found valid GPT with protective MBR; using GPT. Disk /dev/md0: 10482 sectors, 50.0 GiB Logical sector size: 512 bytes Disk identifier (GUID): 45EC12F7-47B7-4339-A219-199F35A51B5F Partition table holds up to 128 entries First usable sector is 34, last usable sector is 104855518 Partitions will be aligned on 2048-sector boundaries Total free space is 2014 sectors (1007.0 KiB) Number Start (sector)End (sector) Size Code Name 12048 18431 8.0 MiB EF02 BIOS boot partition 2 1843241961471 20.0 GiBEF00 Linux filesystem 34196147244058623 1024.0 MiB 8200 Linux swap 444058624 104855518 29.0 GiB8300 Linux filesystem --- I make filesystem on created partitions : * part1 (md0p1) with mkfs.ext3 = bios_grub * part2 (md0p2) with mkfs.ext3 = / * part3 (md0p3) with mkswap = swap * part4 (md0p4) with mkswap = /srv I install debian distribution on md0p2 : * partimage -B=foo -b restore /dev/md0p2 my_image_debian * e2fsck -f -y /dev/md0p2 * resize2fs -f /dev/md0p2 I mount ressources : --- mount -l | grep md0 /dev/md0p2 on /mnt/gentoo type ext3 (rw) /dev/md0p4 on /mnt/gentoo/srv type ext3 (rw) --- I init fstab and mtab with the new partitions. And now i want to install grub. I write disk on /mnt/gentoo/boot/grub/device.map : --- cat /mnt/gentoo/boot/grub/device.map (md0) /dev/disk/by-id/md-uuid-972ee921:bc2ecdb5:0c9b9350:14cbf787 (md0p2) /dev/disk/by-id/md-uuid-972ee921:bc2ecdb5:0c9b9350:14cbf787-part2 (hd0) /dev/disk/by-id/ata-VBOX_HARDDISK_VB4b33f8ee-87039d31 (hd1) /dev/disk/by-id/ata-VBOX_HARDDISK_VBc34d645e-16ef1e20 --- I update grub with : chroot /mnt/gentoo /bin/bash update-grub and and after installation of grub crashes. Do you have any ideas on what could be wrong ? I read many forums but I have not found a solution. Thanks for your help. Damien. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Extract memdisk from grub image
В Thu, 26 Jun 2014 13:41:08 -0500 Robert Kliewer robert.klie...@gmail.com пишет: I was wondering if there is a way to extract an embedded memdisk from a grub boot image. I am trying to recover a grub config embedded in a memdisk (this is the primary config, not the initial one used with the -c option) of a grub pxe image I created. I couldn't find what I was looking for among the current utilities and and I went off into the weeds pretty quick looking at the grub-mkimage source to figure out how to disassemble the boot image. Any help is appreciated. Thanks. Look at https://github.com/arvidjaar/bootinfoscript/raw/master/bootinfoscript It contains loop over all modules in core.img. Note that you cannot pass file to it - scripts is intended to detect bootloaders installed on a harddisk. I started at one point writing core.img parser based on it, but got distracted. Note that script expects that files starts with diskboot.img. You may need to tweak it for PXE image. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Using Grub across architectures
В Fri, 20 Jun 2014 11:00:07 +0100 David Goodenough david.goodeno...@btconnect.com пишет: If I have a host PC (say an Amd64) and I want to set up grub on a USB stick for use on another system (say an ARM or i386 system) how do I tell grub to use the version on the stick (assume that the USB stick already has a complete linux system including grub physically on the USB stick) rather than the host code? If I understand your question correctly: grub-install --directory=/path/to/grub/on/USB/arch ... e.g. grub-install --directory=/mnt/usb/usr/lib/grub/i386-pc ... Not everything is really possible to cross-platform. In particular, any actions that involve updating system NVRAM are obviously not possible unless you are on this system. You may also look at grub-mkrescue which creates bootable hybrid ISO/USB image for all platforms present in /usr/lib/grub/platform (or where it is installed). ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: PXE booting over IPv6
В Thu, 19 Jun 2014 13:10:57 -0700 Michael Mohr michaelm...@hyvesolutions.com пишет: Good afternoon; I'm attempting to set up an IPv6-based PXE boot environment with iPXE and GRUB. I've updated the option ROM on an Intel network card using the latest iPXE code (git rev 7fe0735) with IPv6 support enabled. Booting from this card, iPXE is able to auto-configure both a link-local IPv6 address as well as one from radvd (via SLAAC). iPXE also successfully requests additional parameters from a DHCPv6 server, including the bootfile-url, which allows it to retrieve a copy of GRUB generated as follows: michael@shuriken:~$ grub-mkimage --version grub-mkimage (GRUB) 2.02~beta2-9 michael@shuriken:~$ grub-mkimage --format=i386-pc-pxe --output=grub.pxe --prefix='(pxe)/boot/grub' \ pxe tftp http cpuid echo gfxterm gzio minicmd normal png vbe iPXE successfully downloads the GRUB PXE image from a local HTTP server and transfers control to it. I am then dropped into a GRUB shell and the boot process stops. I've tried running net_ipv6_autoconf, which initially reports error: couldn't autoconfigure PXE, but if I run it a few times it is eventually able to configure the network interfaces (as shown by net_ls_addr) with appropriate IPv6 addresses. grub currently does not send SLAAC solicitation, it passively waits for SLAAC advertisements, so it depends on how frequently your router sends them. Clearly GRUB isn't able to retrieve its configuration file for some reason. I've placed one on the HTTP server specified in DHCP as http://[ipv6::address]/boot/grub/grub.conf and am able to retrieve it with a standard web browser. I suspect that I'm missing something fundamental here, so any enlightenment you can provide is appreciated. As far as I can tell grub does not use square brackets when parsing IPv6 address. Try using something like (http,1:2::4:5)/path/to/file ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Updating CPU microcode possible?
В Mon, 02 Jun 2014 13:39:42 +0300 Nikos Chantziaras rea...@gmail.com пишет: On 01/06/14 18:12, Andrey Borzenkov wrote: В Sat, 31 May 2014 12:18:06 +0300 Nikos Chantziaras rea...@gmail.com пишет: [...] I'm using Grub 2.02 beta 2 to boot three operating systems on my machine (Gentoo, FreeBSD and Windows 8.1.) It would be nice if I could make Grub do a microcode update before booting any of those operating systems. Right now, I am only able to a microcode update in Linux, after the kernel is booted. Someone will need to write grub code to do it. I found this: http://biosbits.org This seems to be a fork of Grub. I suppose there's no plans to merge any of this in Grub? You need to contact Intel and ask them to submit code upstream. Note that they support microcode loading on PC BIOS platform only and Linux can do it on both PC BIOS and UEFI. What is wrong with loading microcode after Linux is booted? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: custom Authentication module
В Mon, 2 Jun 2014 16:08:10 + Buckeyne, Thomas tbucke...@ballytech.com пишет: On the: Do I need to request a module to be included when I run grub-mkstandalone / grub-mkimage Not sure I understand this question. Do you mean - module that you are going to build? My custom module is being included; I thought perhaps I needed to include some module (other than cipher) to get the ecc support But your other answers explain why it is not being included (it is not because it is not in the module include list; but because it is deliberately excluded) As information; I removed the exclusion of ecc.c and it compiled correctly but it did not link = it seems to need some function is a misc.c ; I removed the exclusion for misc.c but it does not compile for a variety of reasons (the util/import_gcry.py does not seem to filter it at all comes across unchanged = therefor it does not compile) With the expanding usage of ECDSA for authentication do you know if there are any plans to include the ecc module in grub build (if or when)? Not really, but if you could clean up build and offer a patch, it would be helpful. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: custom Authentication module
В Fri, 23 May 2014 16:32:57 + Buckeyne, Thomas tbucke...@ballytech.com пишет: I am creating for my company a custom Authentication module (required to meet regulatory requirements) Generally it locates file(s) on an SSD; authenticates them with ECDSA; determines if/where to boot from I have most of the module complete but I need to access the 1) gcry_pk_verify in grub-core/lib/libgcrypt/ciper/pubkey.c 2) fill_in_curve in grub-core/lib/libgcrypt/cipher/ecc.c However neither of these seem to be built when grub is compiled Nor have I been able to find an option to cause them to be built Those files are explicitly excluded when importing libgcrypt. See util/import_gcry.py. I do not know the reason. Am I missing / overlooking a configuration option No, you will need to patch util/import_gcry.py; you may need to extend it to fix any build problems with these files in grub environment. Is there a patch or setup required to have these built Not that I'm aware of. Are these supported for x86 32-bit computers That's probably more for libgcrypt community. I do not see why not. Do I need to request a module to be included when I run grub-mkstandalone / grub-mkimage Not sure I understand this question. Do you mean - module that you are going to build? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Updating CPU microcode possible?
В Sat, 31 May 2014 12:18:06 +0300 Nikos Chantziaras rea...@gmail.com пишет: (I've tried posting this before I subscribed, since http://www.gnu.org/software/grub/grub-mailinglist.html says that subscription to the list is not required to post a question. But that failed. So I'm trying again, after having subscribed...) Hello. Is it possible to update the CPU's microcode (an Intel CPU, in this case) with Grub? Meaning before the operating system starts. Not that I'm aware of. I'm using Grub 2.02 beta 2 to boot three operating systems on my machine (Gentoo, FreeBSD and Windows 8.1.) It would be nice if I could make Grub do a microcode update before booting any of those operating systems. Right now, I am only able to a microcode update in Linux, after the kernel is booted. Someone will need to write grub code to do it. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Possible to Embed GRUB Font File into Binary?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 В Tue, 27 May 2014 19:28:39 -0400 SevenBits sevenbitst...@gmail.com пишет: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/27/2014 03:48 PM, Jordan Uggla wrote: On Sun, May 25, 2014 at 9:29 AM, SevenBits sevenbitst...@gmail.com wrote: Okay, so I've tried this out and I am having difficulties. GRUB does not seem to be able to access any of the commands in the included modules. When I boot into GRUB, a bunch of errors scrolls past on the screen too quickly to read (though they appear to be about missing commands) and then I end up in the GRUB normal prompt. I've using the exact same configuration file that I'm using with grub-mkimage, except it doesn't work this time. I recall seeing something a while ago saying that modules are not automatically loaded when using grub-mkstandalone. If that's the case, then how would I activate them? When I get dropped into the prompt and type something like `insmod linux` it responds by saying that the module Linux was not found, even though I gave the command for it to be included. I've attached my grub.cfg file that I'm using. Perhaps someone could advise me as to what I'm doing wrong? From your grub.cfg: set prefix='' $prefix is the variable used by grub to find its modules, by setting it to the empty string you prevent grub from being able to find the modules. Oh, duh. Not sure how that got in there. I'm having some additional issues now. Essentially, when GRUB is loaded from a memdisk in this manner, the root is set to the memdisk. Which is fine, except for one problem. I need to be able to get the device that the EFI executable file is currently residing on, because I need to set up loopback for an ISO file located in the same directory as the GRUB EFI image. When I generated the image using grub-mkimage, the root was set to this device, so this was not a problem. This URL ( https://wiki.archlinux.org/index.php/Grub#GRUB_Standalone ) suggests using a variable called ${cmdpath} to solve this. Unfortunately, the variable is not defined in my configuration. Which grub version do you use? cmdpath is available post-2.00 (should be in 2.02beta2). I cannot hard-code the path to use (i.e hd1,msdos0) because this portable GRUB copy is meant to be put onto a USB stick and booted on users' computers. Hard-coding the path would be ineffective because my users could for example have multiple hard drives, meaning that my hard-coded path might not point to the USB that I want and instead point to some other drive entirely, or a non-existent drive. This is where my limited knowledge of GRUB internals is coming into play. Is there some easy way to get the device that the image being run is contained on? That's the reason for cmdpath - to export this information. Note that patch itself is rather trivial and can easily be backported. It is commit 1fe26ab4a0fc6ec961b661cc7fc9227db822c9be. Should I do the same for the linux command as well, e.g should linux ${var} quiet splash ... -- become linux ${var} quiet splash ... -- No. This will result in both grub and kernel getting single argument which is probably not what you expect. -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.22 (GNU/Linux) iEYEARECAAYFAlOLShQACgkQR6LMutpd94yqHQCeJ0BlGQKrB8yVSIUj5bx+9u2C 1lgAoL2l9teYcowDVV+qFfbpF9oI0AWW =p6Rg -END PGP SIGNATURE- ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: drivemap and uefi
On Sat, May 10, 2014 at 4:27 AM, Francisco Franchetti nix...@gmail.com wrote: Thanks for the clarification. I solved my problem so this is just in case you are curious. I am used to mbr and in that context it's important to start the windows installers as follows: the disk to which one will install windows, which will remain in the computer after the install is gone, has to be hd0; the installer is in another (usb) disk that must be interpreted as being at hd1; however, most bios will set the booting device to hd0, so using grub I usually flip the order with map; in a nutshell this allows for the computer to boot properly after one has installed windows; there are other workarounds equivalent to update-grub in windows with the boot.ini but what I described is my preferred way of doing it and I don't have for the future a great way to deal with the problem. Will the problem not exist under uefi/win8? I can't say how Windows will react to moving drive to different physical path, sorry. Never tried it and I do not have any UEFI based systems where I can test it. What I did was, since the disk with the win files was not gpt, and I had no flash drive of the appropriate size, I created a gpt partition at the end of the destination disk, then booted grub from the ubuntu installer, chainloaded to the gpt partition with the files, and the installer started and worked fine. But the point of this second paragraph is: the only disk was the destination one. If I had a gpt partitioned usb disk, I am not sure if the installer would like it to be sitting there in the hd0 position while installing, pushing the destination disk to hd1 during installf, but later becoming hd0. I don't know enough about uefi/win8 to know; I'll experiment in the future with that. For now, I'm out of the woods. Anyway, the most important aspect of the email is to say thank you for clarifying that drivemap is an mbr specific command. On Fri, May 9, 2014 at 3:09 PM, Andrey Borzenkov arvidj...@gmail.com wrote: On Wed, May 7, 2014 at 9:36 PM, Francisco Franchetti nix...@gmail.com wrote: I am not sure if the absence of the drivemap command is due to uefi mode booting or if it is due to the version of grub 2 that I have: 2.02beta2-9 from ubuntu installer. Anyone knows why the drivemap command is unavailable? Is it because on a normal uefi system the notion is irrelevant? Yes. drivemap manipulates BIOS drive numbers via BIOS entry points. Neither exists in case of UEFI. I am interested in the knowledge and the drivemap command in itself, but the practical application is to boot a win 8 installer and have the hd0 be the computer installer, which is showing as hd1 because hd0 is taken by the usb installer. I'm not sure I understand the problem. Could you explain it in more details? Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: drivemap and uefi
On Wed, May 7, 2014 at 9:36 PM, Francisco Franchetti nix...@gmail.com wrote: I am not sure if the absence of the drivemap command is due to uefi mode booting or if it is due to the version of grub 2 that I have: 2.02beta2-9 from ubuntu installer. Anyone knows why the drivemap command is unavailable? Is it because on a normal uefi system the notion is irrelevant? Yes. drivemap manipulates BIOS drive numbers via BIOS entry points. Neither exists in case of UEFI. I am interested in the knowledge and the drivemap command in itself, but the practical application is to boot a win 8 installer and have the hd0 be the computer installer, which is showing as hd1 because hd0 is taken by the usb installer. I'm not sure I understand the problem. Could you explain it in more details? Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Partition labels in GRUB2 menu?
On Thu, May 8, 2014 at 4:08 PM, Richard Owlett rowl...@cloud85.net wrote: Jordan Uggla wrote: On Tue, Apr 29, 2014 at 6:42 AM, Richard Owlett rowl...@cloud85.net wrote: During Debian installs I use manual partitioning. I give the partition being created a meaningful label. I would like that label to appear in Grub's menu. How? E.G. I currently have 4 flavors of Debian Wheezy installed (different desktops). Currently the menu shows long effectively meaningless string followed by cryptic partition designator (sa6, sa7, sa8, or sa9). I would like the designator to be meaningful (i.e. GNOME, KDE, LXDE, or XFCE). TIA Modify GRUB_DISTRIBUTOR= in each installation's /etc/default/grub , whatever string you give for GRUB_DISTRIBUTOR will be used in the menu entry titles. http://www.gnu.org/software/grub/manual/grub.html#Simple-configuration That doesn't accomplish my goal. I've reread .../manual/grub.html. I suspect GRUB2 cannot, perhaps by design, do what I want. I wish, that after adding/deleting an OS, update-grub Let's not mess things up. update-grub is program provided by your distribution. Any comments about this command should be addressed to your distribution, not to upstream list. yield a grub.cfg of form: ### BEGIN /etc/grub.d/10_linux ### No, /etc/grub.d/10_linux as shipped by upstream grub does not support what you want. As I already said, you can either modify it or disable and add your own script that does what you want. There are three implied restrictions: 1. the first OS listed on menu shall be the OS on /dev/sda1 First you complain that grub menu includes meaningless partition numbers and now you suddenly want grub menu to be dependent on meaningless partition number. 2. the OS on /dev/sda1 shall be the default You can set default menu entry as GRUB_DEFAULT. What is missing here? 3. all additional OS shall be in partition number order Additional OS are provided by os-prober which is not part of grub. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB_GFXMODE= 60 refresh how?
On Sat, May 3, 2014 at 12:18 PM, Felix Miata mrma...@earthlink.net wrote: On 2014-05-03 12:00 (GMT+0400) Andrey Borzenkov composed: Sat, 03 May 2014 01:22:19 -0400 Felix Miata composed: On 2014-05-03 07:51 (GMT+0400) Andrey Borzenkov composed: Fri, 02 May 2014 18:37:44 -0400 Felix Miata composed: Fedora 20 # man grub No manual entry for grub # man grub2 No manual entry for grub2 Searching with Google for GRUB_GFXMODE= where how to designate appropriate refresh rate comes up empty, always with either just HxV or HxVxBits. How does one get Grub2's boot menu to produce the Grub Legacy equivalent of having video=1024x768@60 on boot stanza cmdline? Does adding video=1024x768@60 to kernel command line in grub2 not work? It works, but what is GRUB_GFXMODE for if not to take the place either it and/or vga=791? No, grub2 does not support setting refresh rate. If neither GRUB_GFXPAYLOAD_LINUX nor GRUB_GFXMODE can provide a known *supported* video mode to suit the user, why do they even exist? Please provide reference to handover protocol that allows passing refresh rate information between bootloader and linux kernel. Any idea why there is no man page on Fedora [19,20,21] for Grub, Grub2, or grub.cfg? -- The wise are known for their understanding, and pleasant words are persuasive. Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB_GFXMODE= 60 refresh how?
В Sat, 03 May 2014 01:22:19 -0400 Felix Miata mrma...@earthlink.net пишет: On 2014-05-03 07:51 (GMT+0400) Andrey Borzenkov composed: Fri, 02 May 2014 18:37:44 -0400 Felix Miata composed: Fedora 20 # man grub No manual entry for grub # man grub2 No manual entry for grub2 Searching with Google for GRUB_GFXMODE= where how to designate appropriate refresh rate comes up empty, always with either just HxV or HxVxBits. How does one get Grub2's boot menu to produce the Grub Legacy equivalent of having video=1024x768@60 on boot stanza cmdline? Does adding video=1024x768@60 to kernel command line in grub2 not work? It works, but what is GRUB_GFXMODE for if not to take the place either it and/or vga=791? No, grub2 does not support setting refresh rate. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: PC BIOS EDD problem
В Sun, 4 May 2014 01:50:33 +0200 Alexey Orishko alexey.oris...@gmail.com пишет: Hi all, Some time ago BIOS Enhanced Disk Drive specification got a new revision where device path was changed from 8 to 16 bytes. I see that the latest grub code in git has support for the latest specification: include/grub/i386/pc/biosdisk.h: ... grub_uint8_t device_path[16]; Are old BIOS versions supported somehow or simply abandoned? The only information from INT13 0x48 used by grub is device size. Do you observe real problem or is it theoretical question? Do I need to revert commit b828fb5d9c7575700a64751efbddd738e2b7c239 of Vladimir Serbinenko to get old BIOS working? Is there any difference between 32 and 64 bit version of grub in handling BIOS? There is no 64 bit grub for BIOS platform. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB_GFXMODE= 60 refresh how?
В Fri, 02 May 2014 18:37:44 -0400 Felix Miata mrma...@earthlink.net пишет: Fedora 20 # man grub No manual entry for grub # man grub2 No manual entry for grub2 Searching with Google for GRUB_GFXMODE= where how to designate appropriate refresh rate comes up empty, always with either just HxV or HxVxBits. How does one get Grub2's boot menu to produce the Grub Legacy equivalent of having video=1024x768@60 on boot stanza cmdline? Does adding video=1024x768@60 to kernel command line in grub2 not work? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Partition labels in GRUB2 menu?
В Tue, 29 Apr 2014 08:42:13 -0500 Richard Owlett rowl...@cloud85.net пишет: During Debian installs I use manual partitioning. I give the partition being created a meaningful label. I would like that label to appear in Grub's menu. How? E.G. I currently have 4 flavors of Debian Wheezy installed (different desktops). Currently the menu shows long effectively meaningless string followed by cryptic partition designator (sa6, sa7, sa8, or sa9). I would like the designator to be meaningful (i.e. GNOME, KDE, LXDE, or XFCE). You can simply edit grub.cfg. If you want grub.cfg to be generated automatically, you can modify files under /etc/grub.d and make them do whatever you want. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Problem with USB keyboard + ehci.mod
В Mon, 28 Apr 2014 17:25:57 +0200 Richard Foltyn richard.fol...@gmail.com пишет: Dear listers, I have a USB hub/switch (ATEN US224) to share a USB keyboard / mouse between two computers. On one of these computers, the keyboard in the grub2 menu works only if I pre-load ehci.mod (independent of whether legacy USB support is enabled in BIOS or not). However, as soon as I load ehci.mod, the partitions present on the hard drive are no longer seen by grub2's ls Yes; loading ehci explicitly disables native (BIOS) disk driver. and I can no longer boot into Windows (Linux still works since it resides on LVM). I.e., BEFORE loading ehci, ls on the grub2 command line shows something like (hd0) (hd0,msdos1) (hd0,msdos2) + some LVM volumes AFTER I load ehci I see (usb0a) (usb0b) (usb0c) (usb0d) + some LVM volumes some LVM volumes is actually minor cosmetic bug here, but I'm not sure how difficult it is to track all references. I'm afraid this does not have easy solution. As I understand, native driver is disabled because driver tells BIOS to relinquish USB control and we do not know what is controlled - it could as well be hard disk or other mass storage and we lose access to it (or worse crash system when accessing it). What you could try - try loading ahci driver (better via nativedisk command if your version supports it - it will also load ehci among others). Thus you will hopefully get native access (not via BIOS) to the same disk. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Problem with USB keyboard + ehci.mod
В Mon, 28 Apr 2014 18:54:57 +0200 Richard Foltyn richard.fol...@gmail.com пишет: On Mon, Apr 28, 2014 at 5:59 PM, Andrey Borzenkov arvidj...@gmail.com wrote: What you could try - try loading ahci driver (better via nativedisk command if your version supports it - it will also load ehci among others). Thus you will hopefully get native access (not via BIOS) to the same disk. Thanks for the suggestion. Unfortunately, 'nativedisk' never returns control and the HDD LED is continuously lit (I reset the PC after ~ 5min). Loading 'ahci' by itself works, but then executing 'ls' has the same effect. Could you show full lspci -nnv? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Problem with USB keyboard + ehci.mod
В Mon, 28 Apr 2014 20:17:18 +0200 Richard Foltyn richard.fol...@gmail.com пишет: On Mon, Apr 28, 2014 at 7:17 PM, Andrey Borzenkov arvidj...@gmail.com wrote: Could you show full lspci -nnv? Here it comes: # lspci -nnv 00:11.0 SATA controller [0106]: Advanced Micro Devices, Inc. [AMD/ATI] SB7x0/SB8x0/SB9x0 SATA Controller [AHCI mode] [1002:4391] (prog-if 01 [AHCI 1.0]) 00:14.1 IDE interface [0101]: Advanced Micro Devices, Inc. [AMD/ATI] SB7x0/SB8x0/SB9x0 IDE Controller [1002:439c] (prog-if 8a [Master SecP PriP]) And which one drives your disk? You may try pata in case it is the latter. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Booting from LVM on top of LUKS
On Fri, Apr 25, 2014 at 7:22 AM, Kwesadilo X kwesad...@gmail.com wrote: Hi, I have been trying to set up GRUB to boot Linux from an LVM logical volume on top of a LUKS device. This is my system root partition, including /boot. I've read various places that GRUB2 supports booting from LVM or LUKS volumes, but nothing that seemed too official or that mentioned both at the same time. Is this a supported configuration? My current status is that I can boot GRUB, GRUB will ask for my encryption key, and show the boot menu (from grub.cfg stored in /boot) after I give my key. At this point, I can see from the GRUB shell that GRUB has mounted my LVM/LUKS volume, and I can see my kernel and initramfs in /boot. I can give the initrd and linux commands seemingly without error. But when I give the boot command, the system just hangs. Any ideas? You could try set debug=all in grub shell before doing boot. This may reveal whether it hangs in grub or after grub passed control to kernel. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Low-level documentation
В Sun, 13 Apr 2014 20:02:12 +0200 (CEST) Fredrik Tolf fred...@dolda2000.com пишет: On Sat, 12 Apr 2014, Andrey Borzenkov wrote: Grub 2, on the other hand, seems to try to go the user-friendly way, with `grub-install' wanting to do all the work for me. This is fine, I guess, so long as the circumstances aren't too strange for it to work. Could you describe your configuration where grub-install fails? I can't say I have any case readily handy, or in fresh memory, but it has usually been in cases where I'm repairing a broken system. Things like, you know, having only one leg of a broken 0.9-metadata md mirror mounted raw, or moving a disk from a broken system to another system with the intention of moving it back, or having the device order out-of-whack due to bootable USB sticks being mounted, or using unusual filesystems or device remapping layers, or the like. Sometimes when the system is broken, I just don't want grub-install to go scanning devices. grub-install does not scan devices. I do not expect grub-install to fail in any of the above cases. More importantly, however, I'd just generally like to know what is going on. It is not very seldom that I find myself doing unusual things, like experimenting with the boot structure, or setting up systems in unusual ways due to various restrictions, or repartition disks later on, moving partitions around; and I'd like to be able to know and control just what happens. For instance, as you write about grub-bios-setup: It tries to embed core.img into either post-MBR gap or reserved area on partition if filesystem is known to have one. It the worst case it gives you a rope to use file on filesystem directly. In all cases it installs boot block (MBR or PBR) which points to the installed core.img. That's all. To me, this raises such questions as: * How do I know whether and where it embeds core.img? Based on device name you used when calling it. * How can I control the same? * How can I tell it which device to address? You pass device name to grub-install. * The Texinfo manual tells me that boot.img merely loads the first sector of core.img. Does this mean that boot.img and core.img are both patched with their own sets of block lists and device numbers? Boot block only knows first sector where core.img is located (may be a couple more, do not remember). First part of core.img is disk loader which reads the rest. So yes, in this sense they both are patched to contain actual location. * What does the --allow-floppy option do, more precisely? It modifies boot.img to accept floppy as boot device when it gets from BIOS. * Maybe I'm just blind, but I still haven't found how core.img actually finds the configuration file once it boots. In most general case - by executing small script embedded in core.img on grub startup which searches for filesystem that contains configuration. There is shortcut when grub-install knows that core.img is located on the same physical disk - then it just records partition number. * Is there any more detailed information on exactly what the various choices for the --format option to grub-mkimage actually do? Probably not in the form you expect it. If you're up for simply answering those questions, that's great. The thing is that I never really needed to ask them with grub-legacy, however. The only thing you need to know about grub2 is grub-install (or probably grub-mknetdir). It figures out everything you need automatically and hides all gory details from you. It is there exactly to avoid need for every user to become grub and boot expert. Is there no documentation available other than reading through the source code? Do you volunteer to write and maintain this documentation? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Low-level documentation
В Sat, 12 Apr 2014 05:28:04 +0200 (CEST) Fredrik Tolf fred...@dolda2000.com пишет: Dear list, Grub 1 is obviously long dead by now, but one thing I always liked about it was how transparent much of its low-level workings was. It made it easy to make it work on systems that were broken in strange ways or where I had to work under weird circumstances. Grub 2, on the other hand, seems to try to go the user-friendly way, with `grub-install' wanting to do all the work for me. This is fine, I guess, so long as the circumstances aren't too strange for it to work. Could you describe your configuration where grub-install fails? What bothers me particularly is the actual installation step itself. Building a core image and all seems documented enough, but the `grub-bios-setup' utility isn't even so much as mentioned in the Texinfo manual, and even less so what it actually does, in detail. Is there any available information on this anywhere? grub-bios-setup mostly corresponds to install command of grub legacy. It tries to embed core.img into either post-MBR gap or reserved area on partition if filesystem is known to have one. It the worst case it gives you a rope to use file on filesystem directly. In all cases it installs boot block (MBR or PBR) which points to the installed core.img. That's all. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: error: embedding is not possible, but this is required for RAID and LVM install.
В Sat, 12 Apr 2014 12:55:10 +0200 bl0 bl0-...@playker.info пишет: On Thursday 10 April 2014 23:43:13 bl0 wrote: On Thursday 10 April 2014 20:02:08 Andrey Borzenkov wrote: 31467 biosdisk part_msdos lvm minix This still should fit into 62 sectors. What is full grub.core size in your case? Do you mean core.img file? Yes it's between 30.5...31 KiB so it should fit but grub-setup refuses to write it. I'll try to debug this later. grub-setup limits core size to 2 sectors less than the offset of the first partition. One sector for the MBR and one sector at the end I don't know why. your embedding area is unusually small may be displayed instead of the usual your core.img is unusually large if the first partition starts at offset 64 sectors (rather than 63). This is fixed in current master (or 2.02~beta2). It was indeed at some point set incorrectly and went this way into 2.00. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: error: embedding is not possible, but this is required for RAID and LVM install.
В Thu, 10 Apr 2014 19:07:25 +0200 bl0 bl0-...@playker.info пишет: Does this mean that grub2 will only support hard disks partitioned recently and will not support hard disks which have been in use for a longer time? bor@opensuse:~ LC_ALL=C ll /boot/grub2/i386-pc/core.img -rw-r--r-- 1 root root 26480 Jan 14 22:06 /boot/grub2/i386-pc/core.img Some info where the size comes from: I know. I was replying to will not support hard disks which have been in use for a longer time. You really cannot squeeze every conceivable feature into 32K. Not every conceivable feature, only these features already supported in past versions of grub2 such as 1.97. This could be seen as a regression. If you are really concerned you could bisect it to find out when size became too large. This could be starting point. But returning to your example $ /opt/grub2/bin/grub-mkimage -O i386-pc [module ...] | wc -c 31467 biosdisk part_msdos lvm minix This still should fit into 62 sectors. What is full grub.core size in your case? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Booting from a bootable ISO
В Sun, 6 Apr 2014 23:09:40 +0530 Rustom Mody rustompm...@gmail.com пишет: On Sun, Apr 6, 2014 at 10:34 PM, Andrey Borzenkov arvidj...@gmail.comwrote: В Sun, 6 Apr 2014 22:10:25 +0530 Rustom Mody rustompm...@gmail.com пишет: Many people -- in particular hardware vendors -- are nowadays supplying bootable ISO's. You have to 1. Download the ISO 2. Burn a CD 3. Boot with it 4. Follow the instructions (to (re)flash the motherboard/CP/Disk etc) Ive already been able to boot ubuntu from an iso on disk Next Ive done it with making a grub-bootable usb and loop-mounting the iso from there So now to boot from a general (ie non linux) bootable ISO? I do not think it is possible in general case. For a start, program that would be started from ISO may need access to CD/DVD which is possible only if system firmware (BIOS/EFI) actually see device. Sorry Maybe I focused on the wrong aspect -- hardware flashing. So to correct that: 1. I believe that Isolinux can do this You need resident driver providing emulated CD driver. grub does not include one. For BIOS systems the easiest way is to use memdisk (which is part of syslinux project indeed) which can be booted by grub. See http://www.syslinux.org/wiki/index.php/MEMDISK For other platforms supported by GRUB - does not know. 2. These ISOs are (often) under the control of FreeDos So maybe I rephrase my question as: How to boot a Freedos based system from grub2? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: error: embedding is not possible, but this is required for RAID and LVM install.
В Wed, 2 Apr 2014 12:47:33 +0200 bl0 bl0-...@playker.info пишет: On Monday 31 March 2014 18:22:57 Andrey Borzenkov wrote: В Mon, 31 Mar 2014 17:32:20 +0200 bl0 bl0-...@playker.info пишет: On Sunday 30 March 2014 19:04:55 Andrey Borzenkov wrote: Well, modern systems tend to start first partition on 1M boundary. Does this mean that grub2 will only support hard disks partitioned recently and will not support hard disks which have been in use for a longer time? bor@opensuse:~ LC_ALL=C ll /boot/grub2/i386-pc/core.img -rw-r--r-- 1 root root 26480 Jan 14 22:06 /boot/grub2/i386-pc/core.img Some info where the size comes from: I know. I was replying to will not support hard disks which have been in use for a longer time. You really cannot squeeze every conceivable feature into 32K. $ /opt/grub2/bin/grub-mkimage -O i386-pc [module ...] | wc -c 31467 biosdisk part_msdos lvm minix 29941 biosdisk part_msdos lvm It actually takes even less than I expected (note that it also pulls in diskfilter). 23047 biosdisk part_msdos 22214 biosdisk 20165 (kernel only) (minix smaller than tar or cpio on v2.00 but still doesn't fit) You may be interested in http://marc.info/?l=grub-develm=139175222350026w=2 My filesystems is mainly zfs on lvm, but grub can store its data on a separate lvm volume in any simple format it wants. I can't easily create more msdos partition table entries because of the stupid limit of 11 partitions in some linux (kernel) versions. With LVM you need just 2 partitions at most. And with zfs directly, without LVM, you have enough space to store bootloader (128K). Good. So you have solution for your problem. Yes. With this in place, I'm happy with grub2 overall. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: error: embedding is not possible, but this is required for RAID and LVM install.
В Mon, 31 Mar 2014 17:32:20 +0200 bl0 bl0-...@playker.info пишет: On Sunday 30 March 2014 19:04:55 Andrey Borzenkov wrote: В Sun, 30 Mar 2014 16:49:54 +0200 bl0 bl0-...@playker.info пишет: Hello, I try to install grub2 v2.00 to my hard disk. It fails with these messages: warning: your embedding area is unusually small. core.img won't fit in it.. My first partition starts at sector 63. Is this unusual? Well, modern systems tend to start first partition on 1M boundary. Does this mean that grub2 will only support hard disks partitioned recently and will not support hard disks which have been in use for a longer time? bor@opensuse:~ LC_ALL=C ll /boot/grub2/i386-pc/core.img -rw-r--r-- 1 root root 26480 Jan 14 22:06 /boot/grub2/i386-pc/core.img error: embedding is not possible, but this is required for RAID and LVM install. So, embedding is required but not possible. From this, it's clear to me that it is not possible to use grub2 on my system configuration? Should I start looking for another bootloader? It depends on your disk configuration. If you provide more information, someone may have an idea how to use grub2 in your case. When I searched the web for this error message, most of the time the problem was solved by moving data around (sometimes large amounts of data) to accomodate grub. I expect software to accomodate the user for the user's convenience rather than have the user jump through hoops to accomodate the software for the developers convenience. From my perspective it's clear what to do. A bootloader which does not fit in the 31 KB embedding area needs to be loaded into memory by another bootloader which does fit in that area. My current setup is grub2 v2.00 loaded using 'multiboot' from an lvm volume by another bootloader, grub2 v1.99, which does fit in the embedding area (with tar module instead of ext2). Using v1.99 alone is an option but it removes the ability to install future versions of grub with new features. I prefer to keep grub2 v1.99 for the sole purpose of loading a later version of grub. Really this is the only sure way I can see to use future versions of grub which will probably continue to grow bigger if fitting into 31 KB is no longer a design goal. Good. So you have solution for your problem. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Booting iso's via grml-rescueboot gen'd GRUB2 config files
В Mon, 31 Mar 2014 12:40:17 -0700 (PDT) Firstname Lastname technoi...@yahoo.com пишет: Fundamentally, the 'iso_path' does not include the '(disk,part)' designation, which has actually been assigned to 'root' [root='hd0,msdos7']; but is not being used. So, give me some feedback. Why has the '(disk,part)' not been included into the 'iso_path', and thus the 'loop' statement identifier? Because if path name does not include device designation, value of $root is implicitly used as device. Is the current value of 'iso_path' a robust enough source identifier, otherwise? (i.e. can they be found at boot time)? If so, then what else can be wrestled with to get this feature working with newly downloaded iso's, and Grub2 tools? Sláinte, odoncaoa ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Failed to boot Linux from USB stick
В Thu, 13 Mar 2014 13:43:11 +0100 Alexey Orishko alexey.oris...@gmail.com пишет: Hi guys, I can't boot Linux on Intel Atom 32-bit from USB stick on one motherboard, but I can do it on another on the same type with different BIOS version. I've read BIOS release notes and found nothing relevant to the problem neither seen anything significantly different in BIOS menu. GRUB version 2.0 grub.cfg: set default=0 set timeout=3 insmod ext2 set root=(hd0,1) menuentry Default { linux /boot/vmlinuz-3.10.28 root=UUID=d768c1f0-79c9-45c4-b604-8d0735a71242 rootfstype=ext4 ro rootdelay=6 initrd /boot/initrd.img-3.10.28 } On the failed system grub is capable to show boot menu, but while selecting it, it fails with message: error: failure reading sector 0x57f650 from 'hd0'. if I drop to grub command prompt from boot menu (without initially selecting entry), and do some commands: grub ls (hd0) (hd0,msdos1) grub insmod ext2 grub ls (hd0,1)/boot error: failure reading sector 0x802 from 'hd0' grub ls # now ls output is empty line grub date error: no such partition. grub - If I connect the same USB stick to the motherboard with old BIOS, it boots ok. - I can boot from SATA HDD with exactly the same root fs as USB stick (I've copied root partition with cpio and updated UUID value in grub and fstab). - I can boot from USB stick only if it has FAT32, for example MSDOS boot disk or Ubuntu install disk made by Universal-USB-Installer.exe. I wonder, what else do I need to check in order to get to the bottom of this problem? Any help would be appreciated. Could you try current GIT master? I remember there were some USB related patches. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Managing multi grub installs
В Mon, 10 Mar 2014 08:40:53 +0530 Rustom Mody rustompm...@gmail.com пишет: Hi My earlier thread http://lists.gnu.org/archive/html/help-grub/2011-01/msg00017.html Thanks to help from Jordan Ugla there, in particular, the suggestion to use 'configfile', things have been mostly working for the last 2-3 years. Recently it stopped working so this mail. More details below. My blkid excerpts /dev/sda2: LABEL=Boot500G UUID=... TYPE=ext2 /dev/sda5: LABEL=Debian500G UUID=... TYPE=ext4 /dev/sda6: LABEL=Ubuntu500G UUID= TYPE=ext4 /dev/sda7: LABEL=DebOld UUID=... TYPE=ext4 In short I have multiple linuxes with their own /boots in sda5,6,7 And a /boot standalone in sda2. The mbr points to sda2. sda2 is not mounted in any OS so it does not get meddled with on upgrades It contains entries written by hand like menuentry Debian configfile chaining { search --set --label Debian500G configfile /boot/grub/grub.cfg } And this was working for about a year. IOW: the respective OSes would only upgrade their own boots under their respective filesystems. They would not stomp on each others' toes and my handwritten boot would configfile-chain to them. Why do not simply chainload each bootloader? I.e. menuentry Debian bootloader { search --set --label Debian500G multiboot /boot/grub/i386-pc/core.img } (adjust platform and paths as needed). Or simply configure each OS to install bootblock in respective partition and use chainloader +1 The problem with configfile approach is, it is not clean environment. You still get at least $prefix which refers to original grub2. Using bootloader chaining will give you clean bootloader instance for each OS. Then (small nuisance!) after upgrading the debian started introducing a load_video function that would error out when chained with configfile. No issue: I'd just delete that line after an upgrade and things kept running smoothly. But after my most recent debian-testing upgrade it just stopped working: the second configfile just does not reach and the machine keeps rebooting. As a current hack Ive made a new entry that works menuentry Debian 500 handedited { linux (hd0,5)/boot/vmlinuz-3.12-1-686-pae root=LABEL=Debian500G ro initrd (hd0,5)/boot/initrd.img-3.12-1-686-pae } How do I keep configfile working?? Thanks Rusi ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Questing on New Network Stack
В Tue, 11 Feb 2014 17:24:44 -0500 Cheng Cheng ccheng@gmail.com пишет: Hi, As noticed, there is a big change for GRUB 2.00 in the network stack from previous GRUB 1.99. I upgraded to GRUB 2.00 today, however, the PXE booting stops working. I use following command to generate a PXE bootable image: grub-mkimage --format=i386-pc-pxe --output=grub.pxe --prefix=‘(tftp)/boot/grub’ pxe tftp normal After I boot into GRUB console, I have to manually trigger network configuration by “net_bootp”. I then can use “net_ls_addr” to verify my network configuration succeeds. However, GRUB console still cannot communicate to TFTP server to fetch additional modules. For example, I tried: insmod (tftp, XXX.XXX.XXX.XXX)/boot/grub/halt.mod But returned “error: File not found”. For GRUB 1.99, I can do the same thing by: insmod (pxe:XXX.XXX.XXX.XXX)/boot/grub/halt.mod Can anyone share some hints on how to use the new network stack? I also interests in trying HTTP protocol. Please try grub-mknetdir to create grub image and directory structure suitable for netboot. Does it work? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: grub2 mdraid
On Mon, Feb 10, 2014 at 12:33 PM, Oleg lego12...@yandex.ru wrote: Hi, all. I have the problem with an installing grub on a mdraid device. I install a new system with a help of my bootable usb flash. I have x86 system on the usb flash and want to install a x86_64 system on a hdd. My steps: 1. Boot from the usb flash (/dev/sda); 1.1. Create a raid device on partitions (prepared with fdisk) with: mdadm -C /dev/md0 -l1 -n2 --bitmap=internal /dev/sdb1 /dev/sdc1 1.2. Make a raid partition and a fs on it: fdisk /dev/md0 mkfs.ext4 /dev/md0p1 1.3. Install a system: mount /dev/md0p1 /mnt debootstrap --arch=amd64 --foreign --include=grub2,mdadm wheezy /mnt http://ftp.debian.org/debian 1.4. Install grub: grub --modules=part_msdos raid mdraid1x --root-directory=/mnt /dev/sdb grub --modules=part_msdos raid mdraid1x --root-directory=/mnt /dev/sdc Upstream does not have this command, you need to ask your distribution. Also in general grub2 is intelligent enough to figure out what modules it needs; you need to have very good reasons and understanding of grub internals to manually specify them. Try using grub-install --boot-directory=/mnt/boot /dev/sdb grub-install --boot-directory=/mnt/boot /dev/sdc if it does not work, full output of grub-install --debug --boot-directory /mnt/boot /dev/sdb would be interesting. 2. Boot from hdd and get: error: no such disk. Entering rescue mode... grub rescue 2.1. I do insmod part_msdos, raid, mdraid1x. But when i do insmod normal i get: error: no such disk. Now as a workaround i install a system (x86_64) to a different single disk and booted from it i install a system (x86_64) on a mdraid and do grub-install from a chroot. And this is work. But this is a strange method. Can anyone point me to the right section of the http://www.gnu.org/software/grub/manual/html_node/index.html to resolve my problem? Thanks. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: grub2 mdraid
On Mon, Feb 10, 2014 at 1:59 PM, Oleg lego12...@yandex.ru wrote: On Mon, Feb 10, 2014 at 12:50:25PM +0400, Andrey Borzenkov wrote: grub --modules=part_msdos raid mdraid1x --root-directory=/mnt /dev/sdb grub --modules=part_msdos raid mdraid1x --root-directory=/mnt /dev/sdc Upstream does not have this command, you need to ask your distribution. Also in general grub2 is intelligent enough to figure Sorry, this is a typo - must be grub-install instead of grub. out what modules it needs; you need to have very good reasons and understanding of grub internals to manually specify them. Try using grub-install --boot-directory=/mnt/boot /dev/sdb grub-install --boot-directory=/mnt/boot /dev/sdc This commands complains: touch: cannot touch '/boot/grub/grub2-installed': Read-only file system Installation finished. No error reported. This is either downstream patch or command is wrapper around real grub-install. In upstream grub-install never attempts to touch this file. But now the system is booting ok. I test: grub-install --root-directory=/mnt /dev/sdb and the system boot again. Ugh... Now i can't understand anything. Early i tried various versions of grub2 and various command options and now i confuse myself. What is the difference between --boot-directory and --root-directory in my case? --root-directory is deprecated in current upstream, but that's basically all. There should be no differences if using upstream sources. As for your specific case, I do not know :) ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB2 Booting from ISO on LVM
On Tue, Jan 28, 2014 at 2:55 PM, Jorge Fábregas jorge.fabre...@gmail.com wrote: Hi everyone, Fedora 20 here. I'm using GRUB2 to boot an ISO stored on one of my ext4 partitions. This works perfectly fine. However, I can't make it work from an ext4 over LVM. GRUB2 boots the ISO but something is wrong with the findiso kernel parameter. When the CD boots I get a bunch of: modprobe: module LVM2_member not found in modules.dep modprobe: module swap not found in modules.dep This has nothing to do with grub. Your initrd must expect and support access to LVM. It should include necesssary kernel modules, LVM user space components as well as scripts to activate LVs. You need to ask on Fedora list whether they support it and if yes, how to activate this feature. Here's my configuration: http://fpaste.org/71588/ The first menuentry is the one that works followed by the LVM one where I even tried findiso=$isofile but didn't work. Any help will be appreciated. Thanks, Jorge ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: pxechainloader problems
On Thu, Jan 23, 2014 at 11:18 PM, Deepak Chawla dcha...@gmail.com wrote: Hi, In my current setup, I have a diskless client that PXE boots a grub2 image and then offers choices to load different ISOs. This setup is working great. I now have a need for grub to PXE boot another (custom) PXE loader and it looks like the pxechainloader command would do just that. Unfortunately, as soon as I try to PXE boot the next image (I've tried both the custom PXE loader and the pxegrub image as well), the system resets. The pxechainloader in the stock grub-2.00 refused to even load the images. The pxe grub I built from the git repo is able to load the image, but not switch to it. Here's What exactly does it mean? Any error message, what happens at all? the menuentry I have for reloading pxegrub. menuentry reload pxegrub { set kern=/boot/grub/pxegrub.0 echo -n Loading ${root}$kern: pxechainloader $kern boot } I did an extensive google search on pxechainloader and the closest info I found on this was from a 3+ year old email trail (http://lists.gnu.org/archive/html/grub-devel/2010-09/msg00049.html). Am I missing something? Any suggestions on how to proceed with this? You should report it on grub-devel and/or file GRUB bug (see on Savannah project page link to bugs). It is good time as new release is not far of and it is better fixed before it. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Booting from NVMe controller enters GRUB rescue
On Wed, Jan 22, 2014 at 4:18 PM, vinayak holikatti vinholika...@gmail.com wrote: On Tue, Jan 21, 2014 at 1:50 PM, Andrey Borzenkov arvidj...@gmail.com wrote: On Tue, Jan 21, 2014 at 12:14 PM, vinayak holikatti vinholika...@gmail.com wrote: On Mon, Jan 13, 2014 at 11:51 PM, Jordan Uggla jordan.ug...@gmail.com wrote: On Thu, Jan 9, 2014 at 10:53 PM, vinayak holikatti vinholika...@gmail.com wrote: Hi All, We have developed Legacy OptionROM for NVMe controller device. We are able to boot to all Windows 7 and later OS. But we are facing issue in booting Fedora 20 x86_64 OS. Fedora 20 gets installed successfully on NVMe device. While booting from NVMe device we are observing error as follows and enters grub rescue. We are clueless about what is happening here. We would like to know how to over come this issue and boot to Fedora 20 OS from our controller. error : no such device: 44d1bf09-4e8a-4f46-aea6-09e364abf5cb. Entering rescue mode... grub_rescue Please run ls at the rescue prompt and post the output. The ls command at grub_rescue shows as below grub_rescuels (hd0) And set command output as below grub_rescueset prefix = (hd0)/grub2 root = hd0 I think the information of prefix and root are wrongly set. It should be prefix = (hd0,1)/grub2 root = (hd0,2) /boot partition corresponds to /dev/nvme0n1p1 / paration corresponds to /dev/nvme0n1p2 Which grub version do you use? Could you test current GIT master? If it still puts prefix wrong, please show output of grub-install --verbose /dev/path-to-boot-block (/dev/nvme0n1 or /dev/nvme0n1p1 or wherever). We are using grub2. Do you want us to install grub and not grub2? grub2 is name which is used by some distributions to allow concurrent install of grub legacy and current grub. Upstream has only grub and grub legacy. As grub2-install doesn't have verbose option. Sorry, it is --debug option, not --verbose. We will try latest grub from git repo. Should we try grub or grub2? There is only grub. I would not be surprised if grub is confused by non-standard disk naming, but there were many changes recently so it may be fixed already. -- Regards, Vinayak Holikatti ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Booting from NVMe controller enters GRUB rescue
On Tue, Jan 21, 2014 at 12:14 PM, vinayak holikatti vinholika...@gmail.com wrote: On Mon, Jan 13, 2014 at 11:51 PM, Jordan Uggla jordan.ug...@gmail.com wrote: On Thu, Jan 9, 2014 at 10:53 PM, vinayak holikatti vinholika...@gmail.com wrote: Hi All, We have developed Legacy OptionROM for NVMe controller device. We are able to boot to all Windows 7 and later OS. But we are facing issue in booting Fedora 20 x86_64 OS. Fedora 20 gets installed successfully on NVMe device. While booting from NVMe device we are observing error as follows and enters grub rescue. We are clueless about what is happening here. We would like to know how to over come this issue and boot to Fedora 20 OS from our controller. error : no such device: 44d1bf09-4e8a-4f46-aea6-09e364abf5cb. Entering rescue mode... grub_rescue Please run ls at the rescue prompt and post the output. The ls command at grub_rescue shows as below grub_rescuels (hd0) And set command output as below grub_rescueset prefix = (hd0)/grub2 root = hd0 I think the information of prefix and root are wrongly set. It should be prefix = (hd0,1)/grub2 root = (hd0,2) /boot partition corresponds to /dev/nvme0n1p1 / paration corresponds to /dev/nvme0n1p2 Which grub version do you use? Could you test current GIT master? If it still puts prefix wrong, please show output of grub-install --verbose /dev/path-to-boot-block (/dev/nvme0n1 or /dev/nvme0n1p1 or wherever). I would not be surprised if grub is confused by non-standard disk naming, but there were many changes recently so it may be fixed already. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: alloc magic is broken at... error
В Mon, 20 Jan 2014 09:59:40 -0500 SevenBits sevenbitst...@gmail.com пишет: On Jan 19, 2014, at 9:23 PM, Andrey Borzenkov arvidj...@gmail.com wrote: В Sun, 19 Jan 2014 19:45:23 -0500 SevenBits sevenbitst...@gmail.com пишет: Not sure whether to post this in this bug-grub list or here, so feel free to correct me if perhaps this is in the wrong spot. I’m not sure if the issue I’m having is due to a fault in setup or a bug within GRUB. I’m the developer of a tool called Mac Linux USB Loader, and some of my users are reporting errors with my compiled copy of GRUB. Specifically, after loading the kernel, GRUB spits out the following error message: alloc magic is broken at 0x81493ca0: 207d007d68746170 Aborted. Press any key to exit. The hexadecimal values vary slightly, but generally speaking this is what the error looks like. The kernel loading operation with GRUB’s linux command is followed by the initrd command to load the RAM disc, and that operation never occurs, so I know that the operation never completes. This error never seems to occur with Ubuntu-based distributions. It also seems to occur if the kernel can’t be found - but in that case, shouldn’t the linux command fail with an error instead of something like this? I can’t post my GRUB config at the moment, as I’m not on the machine where it is stored, but any advice that you can give me without it would be appreciated, such as under what conditions this error occurs. Thanks, — SevenBits Well, the first thing to try would be upstream HEAD to verify whether problem is still present there. Sorry, just realized my previous reply didn’t get posted. The GRUB build I’m using is compiled straight from the current git repository. It might be a week or two old, but unless something really changed in the last two to three weeks I don’t think it’ll differ. In this case you need to post to gurb-devel. Is there any way to get output to serial console? I can try an older release of GRUB, like the latest stable and see if that does anything though. signature.asc Description: PGP signature ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: alloc magic is broken at... error
В Mon, 20 Jan 2014 10:30:33 -0500 SevenBits sevenbitst...@gmail.com пишет: On Jan 20, 2014, at 10:27 AM, Andrey Borzenkov arvidj...@gmail.com wrote: В Mon, 20 Jan 2014 09:59:40 -0500 SevenBits sevenbitst...@gmail.com пишет: On Jan 19, 2014, at 9:23 PM, Andrey Borzenkov arvidj...@gmail.com wrote: В Sun, 19 Jan 2014 19:45:23 -0500 SevenBits sevenbitst...@gmail.com пишет: Not sure whether to post this in this bug-grub list or here, so feel free to correct me if perhaps this is in the wrong spot. I’m not sure if the issue I’m having is due to a fault in setup or a bug within GRUB. I’m the developer of a tool called Mac Linux USB Loader, and some of my users are reporting errors with my compiled copy of GRUB. Specifically, after loading the kernel, GRUB spits out the following error message: alloc magic is broken at 0x81493ca0: 207d007d68746170 Aborted. Press any key to exit. The hexadecimal values vary slightly, but generally speaking this is what the error looks like. The kernel loading operation with GRUB’s linux command is followed by the initrd command to load the RAM disc, and that operation never occurs, so I know that the operation never completes. This error never seems to occur with Ubuntu-based distributions. It also seems to occur if the kernel can’t be found - but in that case, shouldn’t the linux command fail with an error instead of something like this? I can’t post my GRUB config at the moment, as I’m not on the machine where it is stored, but any advice that you can give me without it would be appreciated, such as under what conditions this error occurs. Thanks, — SevenBits Well, the first thing to try would be upstream HEAD to verify whether problem is still present there. Sorry, just realized my previous reply didn’t get posted. The GRUB build I’m using is compiled straight from the current git repository. It might be a week or two old, but unless something really changed in the last two to three weeks I don’t think it’ll differ. In this case you need to post to gurb-devel. Is there any way to get output to serial console? No, I don’t think so, unless there’s some GRUB trick that I don’t know about. Once this happens GRUB crashes to a stop and I can’t do anything. You could change to serial console before running this command and enable debugging. I can try an older release of GRUB, like the latest stable and see if that does anything though. signature.asc Description: PGP signature ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: alloc magic is broken at... error
В Sun, 19 Jan 2014 19:45:23 -0500 SevenBits sevenbitst...@gmail.com пишет: Not sure whether to post this in this bug-grub list or here, so feel free to correct me if perhaps this is in the wrong spot. I’m not sure if the issue I’m having is due to a fault in setup or a bug within GRUB. I’m the developer of a tool called Mac Linux USB Loader, and some of my users are reporting errors with my compiled copy of GRUB. Specifically, after loading the kernel, GRUB spits out the following error message: alloc magic is broken at 0x81493ca0: 207d007d68746170 Aborted. Press any key to exit. The hexadecimal values vary slightly, but generally speaking this is what the error looks like. The kernel loading operation with GRUB’s linux command is followed by the initrd command to load the RAM disc, and that operation never occurs, so I know that the operation never completes. This error never seems to occur with Ubuntu-based distributions. It also seems to occur if the kernel can’t be found - but in that case, shouldn’t the linux command fail with an error instead of something like this? I can’t post my GRUB config at the moment, as I’m not on the machine where it is stored, but any advice that you can give me without it would be appreciated, such as under what conditions this error occurs. Thanks, — SevenBits Well, the first thing to try would be upstream HEAD to verify whether problem is still present there. signature.asc Description: PGP signature ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: cmdline options getting \x20?
В Fri, 17 Jan 2014 10:06:06 +1000 Michael D. Setzer II mi...@kuentos.guam.net пишет: I have a project that can pass run time options via the /proc/cmdline, and it has worked fine with syslinux, grub4dos, and grub for a long time, but just run into an issue with a new install of Fedora 20? The spaces in the option are being replaced with \x20 with the grub of Fedora 20? As far as I can tell, upstream grub does not do it. You probably should ask on Fedora list. Same option of previous fedoras hadn't noticed this, though most systems where version 17, so it could have also been with 18 or 19? Was able to resolve it by having my project use sed to replace \x20 in the command line with spaces, but wondering when this change occurred with grub2, since none of the other boot loaders seem to do this. Found a mention of this in fedora 18 bugzilla, but the message was commented with EOL for 18, so there was no resolution of the issue? +--+ Michael D. Setzer II - Computer Science Instructor Guam Community College Computer Center mailto:mi...@kuentos.guam.net mailto:msetze...@gmail.com http://www.guam.net/home/mikes Guam - Where America's Day Begins G4L Disk Imaging Project maintainer http://sourceforge.net/projects/g4l/ +--+ http://setiathome.berkeley.edu (Original) Number of Seti Units Returned: 19,471 Processing time: 32 years, 290 days, 12 hours, 58 minutes (Total Hours: 287,489) BOINC@HOME CREDITS ROSETTA 10045530.779504 | SETI17491770.047084 ABC 16613838.513356 | EINSTEIN15323134.109852 ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: wrong hdX designation in grub.cfg
В Sun, 12 Jan 2014 15:59:46 -0700 Chris Murphy li...@colorremedies.com пишет: A normally working grub.cfg contains this entry: search --no-floppy --fs-uuid --set=root --hint-bios=hd0,gpt4 --hint-efi=hd0,gpt4 --hint-baremetal=ahci0,gpt4 d7bc9d0e-7706-44f9-b1a7-ff24b7c360a7 hd0,gpt4 seems to be wrong. It is just a hint anyway. At a grub prompt, there is one hd0 entry, and multiple hd1,gptY entries. These correspond to the partitionless Firewire drive, and the internal drive respectively. It seems like the firmware presents the Firewire drive to grub first and therefore is hd0 in grub. But once booted to linux, the internal drive ends up being treated as hd0 by grub-mkconfig. Firmware bug? Or grub bug? Neither. Nobody ever said firmware would enumerate devices in the same order as kernel. This is exactly why grub2 stopped relying on it and is using search instead. Is it BIOS or UEFI system? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: wrong hdX designation in grub.cfg
On Mon, Jan 13, 2014 at 7:13 AM, Chris Murphy li...@colorremedies.com wrote: On Jan 12, 2014, at 7:19 PM, Andrey Borzenkov arvidj...@gmail.com wrote: В Sun, 12 Jan 2014 15:59:46 -0700 Chris Murphy li...@colorremedies.com пишет: A normally working grub.cfg contains this entry: search --no-floppy --fs-uuid --set=root --hint-bios=hd0,gpt4 --hint-efi=hd0,gpt4 --hint-baremetal=ahci0,gpt4 d7bc9d0e-7706-44f9-b1a7-ff24b7c360a7 hd0,gpt4 seems to be wrong. It is just a hint anyway. At a grub prompt, there is one hd0 entry, and multiple hd1,gptY entries. These correspond to the partitionless Firewire drive, and the internal drive respectively. It seems like the firmware presents the Firewire drive to grub first and therefore is hd0 in grub. But once booted to linux, the internal drive ends up being treated as hd0 by grub-mkconfig. Firmware bug? Or grub bug? Neither. Nobody ever said firmware would enumerate devices in the same order as kernel. This is exactly why grub2 stopped relying on it and is using search instead. Fair enough. Is the most important reference the fs UUID? Yes Would that alone work, Yes or do the hints make finding it by UUID faster? Yes (as long as hints were guessed correctly of course). If you know boot time device order and are sure it is reliable, you can always force this particular order in device.map. This is basically the only thing this file is good for today. Is it BIOS or UEFI system? Technically neither, it's a Mac. So it's in some Apple zombie land between Intel EFI and UEFI. Chris Murphy ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Multiboot mixed width O/Ss
В Сб, 04/01/2014 в 22:25 -0600, coifed48...@mypacks.net пишет: GRUB Gurus, I realize the answer to my question ultimately may be determined by the Intel and AMD architectures. Is it possible to boot alternately to a 32bit OS and a 64bit OS with GRUB 2 and a 64 bit processor? Yes, it should be possible. http://en.wikipedia.org/wiki/GNU_GRUB mostly refers to Grub legacy and http://www.gnu.org/software/grub/grub-faq.html states The current release is working on Intel/AMD PCs, OpenFirmware-based PowerPC machines (PowerMac and Pegasos), EFI-based PC (IntelMac) and coreboot (formerly, LinuxBIOS), and is being ported to UltraSparc. without reference to processor width. I also checked https://help.ubuntu.com/community/Grub2; but found no reference to CPU width. Thanks, R ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Progress indicator on loading images
В Чт, 02/01/2014 в 04:09 +0100, Bertho Stultiens пишет: Hi, The old grub (v1) displayed a set of dots while loading kernel and initrd files from disk. However, it seems that it is impossible to get this behavior enabled in grub2. I am booting on an old machine from usb (v1.1) and it takes ages to get the 20MB loaded. An indication of activity and where it is in the process would be nice. Is there a way to configure a progress-bar or the old dots while the kernel and initrd are being loaded from disk in grub2? BTW, this is on a machine with xubuntu 12.04 lts. Current grub2 (2.02 beta2 at the moment) includes progress module that is supposed to display current percentage and estimated speed. Somehow it does not for me, but may be it just loads too fast. You could try to build it (may be it is packaged already). You will need to do insmod progress before loading files. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB2 plain dm-crypt support
On Sun, Dec 15, 2013 at 7:54 PM, joe fresh dzrdm...@gmx.com wrote: Hi, thanks for the quick response. Unfortunately, I didn't find anything related to plain dm-crypt searching through the recent grub-devel archives. OK it was not on the list but on GIT branch :p http://git.savannah.gnu.org/cgit/grub.git/log/?h=peter/devmapper Disclaimer - did not try it myself and no idea in which state it is. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Why the --skip-fs-probe option does not prevent file system check during installation?
On Wed, Nov 20, 2013 at 12:09 PM, Wang Weber mail.weber.w...@gmail.com wrote: Hi, I failed to install grub-1.99 grub 1.99 is very old. Please use at least 2.00 or preferably recent development snapshot. on a hard disk and I found the reason is because GRUB thought that there was a ext2 filesystem installed on that disk due to garbage data. The error messages are listed below. # grub-setup --directory=/mnt/root/mnt2/boot/grub /dev/sdc grub-setup: warn: Attempting to install GRUB to a disk with multiple partition labels or both partition label and filesystem. This is not supported yet.. grub-setup: warn: Embedding is not possible. GRUB can only be installed in this setup by using blocklists. However, blocklists are UNRELIABLE and their use is discouraged.. grub-setup: error: will not proceed with blocklists. I also tried the following command and used gdb to confirm this. bash-3.00# grub-probe -t fs -d /dev/sdc ext2 Then I checked the grub-setup help doc and saw that there is a --skip-fs-probe option. The description for this option is below. -s, --skip-fs-probeDo not probe for filesystems in DEVICE I tried this option but it still promoted the same error messages. Finally I checked the source code, it seems that this operation is not doing what I am expecting. It does not prevent GRUB from detecting the filesystem and GRUB refuses to do installation if some garbage data was found. Is this behavior correct? What's the workaround if I want GRUB to do the installation without checking the existence of filesystem? Thanks, Weber ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: When will the OS_PROBER be fixed
В Fri, 15 Nov 2013 15:03:15 + Dan Priest d...@wessexcomputers.com пишет: Hi just a quick one, I was wondering when GRUB's os_prober function will be fixed as when I install any linux distro with GRUB, it will boot windows 8 properly on a computer with UEFI Secure Boot with GPT partitioning, this includes Fedora and Ubuntu. The only way to get it to work is by running boot-repair which you have to disable secure boot for it to work and then this causes the shimx64.efi to stop loading. This is a real big problem at the moment for the Linux community as everyone is confusing the issue to be with secure boot when in actual fact it is GRUB's configuration that seems to be at fault. Please correct me if I am wrong. Currenty grub does not offer native implementation for EFI secure boot. Both distros you mention should carry out-of-tree patches that enable it. If it does not work, please report it to your distribution. It definitely should work in openSUSE 12.3 and above, EFI implementation quirks withstanding. Would just like to know if this is a problem you know about and if so do you know how long it will be before it is fixed??? Regards Dan Priest ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB doesn't recognize partitions on El-Torito image during EFI boot from CD
В Thu, 14 Nov 2013 19:53:10 +0100 Jacek Konieczny jaj...@jajcus.net пишет: On Thu, 14 Nov 2013 22:07:25 +0400 Andrey Borzenkov arvidj...@gmail.com wrote: В Tue, 12 Nov 2013 17:19:11 +0100 Jacek Konieczny jaj...@jajcus.net пишет: FS2: Alias(s):CD20e0a1b:;BLK8: PciRoot(0x0)/Pci(0x1F,0x2)/Sata(0x4,0x0,0x0)/CDROM(0x1)/HD(1,MBR,0x,0x1,0x855A1) FS2 is the contents of a partition of the El-Torito image on the CD. Yes, grub explicitly skips over such paths. Any good reason for such behaviour? It was before my time :) One reason I could think of is to avoid ghost hard disks appearing in grub. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB doesn't recognize partitions on El-Torito image during EFI boot from CD
On Tue, Nov 12, 2013 at 11:50 AM, Jacek Konieczny jaj...@jajcus.net wrote: Could you give example how your image is seen in EFI shell? What EFI shell command should I use for that? Using map would be a good start. You may use map -b to paginate. Could you dump it to a file (use e.g. USB stick for it) and attach? Assuming that fs2: refers to writable partition fs2: map -v map.out It is not forbidden in any way; it is just that such CD will not be directly bootable by EFI. The strange thing is – my previous 'prototype', with EFI shell and ELILO, was bootable. And GRUB loads too, it just cannot see the other files on the image. Could you please describe step by step how you build your CD? Exact command line(s) would be helpful. Would [grub] see the files on ESP when it was a proper EFI El-Torito image, as defined by the specs (unpartitioned, no emul)? If not, would it still be possible to chainload any other EFI application from this image? Not without special arrangements. grub-mkrescue is using xorriso which builds hybrid ISO image containing extra partition table(s) which results in GRUB being able to access El-Torito boot image as normal HDD partition. I wanted to be able to chain-load other EFI applications (like the EFI shell) from the GRUB, so I wanted files to be available via the EFI services – putting the files on the 'ESP' FAT file system made them available to any EFI application. I could not use the 'standard CD' this way, as EFI cannot read ISO9660 file system directly. Makes sense. Hmm ... it sounds like extending grub-mkrescue to allow adding arbitrary content to ESP will be the most easy solution in your case. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB doesn't recognize partitions on El-Torito image during EFI boot from CD
В Mon, 11 Nov 2013 18:58:20 +0100 Jacek Konieczny jaj...@jajcus.net пишет: Hi, I have a system as a disk image. It can be booted directly from an USB stick both via legacy BIOS and EFI. I wanted the same image to be usable on a CD. For EFI boot it should be straightforward – use the disk image as a hard-disk-emulation El-Torito EFI boot image. Do you mean - in addition to EFI boot image? Because EFI requires ESP in no emulation mode to boot. I have done that with other EFI boot loaders and it worked right. With GRUB there is a problem – it is unable to find the boot partition. In rescue mode I can see only: (hd0) (cd0) however, I would expect too see (hd0,msdos1) too – with the first partition of the El-Torito image. Why msdos? El-Torito CD does not have standard DOS label, so it would be wrong name. And if you place whole image of partitioned disk there, it becomes nested partition (dos label in El-Torito partition). Something like (cd,eltorito0,msdos1) probably. Is it a bug? Am I doing something wrong? No. As it stands currently grub does not expose El-Torito image as partition of CD media. It is completely invisible unless you use hybrid CD which contains other partition labels referring to it. I made a workaround by using GRUB loopback in the embedded config for the EFI loader: echo starting grub search.fs_uuid 8101-917C root set prefix=($root)/grub configfile $prefix/go_normal.cfg echo boot partition not found falling back to loopback device search.fs_uuid 2013-11-11-18-32-41-96 cd loopback loop ($cd)/pld-nr-32.img set root=(loop,msdos1) set prefix=($root)/grub It works, but I don't like this. But you need to place image on CD either way. Are you using this image for anything else, like directly booting on non-EFI? GRUB should be able to see the El-Torito contents, shouldn't it? So far there was not any use case. But yes, technically EFI does expose it as partition so grub probably should too. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: error: core image is too big (0xXXXXXX 0x78000)
В Thu, 7 Nov 2013 20:37:53 -0600 Glenn Washburn developm...@efficientek.com пишет: This only happens on i386-pc and i386 pxe. Can anyone explain or point to documentation as to why the core image can't be larger than 0x78000 for i386? It appears that this check is to prevent grub from occupying some of the upper memory area. Why can't grub use the upper memory? Why doesn't grub use unreal mode to access the extended memory? If it does use these memory regions, why is there this limit? grub does use protected mode, but core.img is loaded very early in real mode, where available memory is quite limited. For this reason core.img should be as small as possible and contain just enough to access the rest of grub modules which can be loaded later after switch to protected mode. I do not know where exact number comes from; but absolute limit is 640k anyway and part of it is already taken. Any insight is appreciated, Glenn ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Chainloading to an isofile embedded bootloader
В Mon, 04 Nov 2013 16:12:13 +0100 Arbiel (gmx) arbiel.perlacre...@gmx.fr пишет: Hi Using grub2, booting a PC from an iso file is rather straightforward when the file's bootloader is grub2 itself or that the /boot/grub/loopback.cfg file exists. Do not assume everyone knows what loopback.cfg does :) Could you explain what you are trying to do? Without referring to any third-party implementation. When this is not the case, one has to write such a /boot/grub/loopback.cfg file, and this may require a lot of expertise. A way to avoid this rather cumbersome task would be to have grub2 chainload to the loopbacked file-embedded bootloader. 1) does such a chainload operation work ? 2) in the case it works, can you please inform me which block address and length are to be used ? 3) else, would it be possible to include this development into grub's future plans ? Thanks Arbiel ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Grub is not showing Windows 7 when booting my laptop
В Fri, 25 Oct 2013 10:59:45 -0500 Daniel - Asterisk earohua...@gmail.com пишет: Hello all, I'm completely novice with grub and I have a problem, I was re-installing ArchLinux (/dev/sda2) in my laptop with existing Windows 7 (/dev/sda1), but after re-installing grub and running following commands I lost Windows at boot time. I think Windows is still there and I want to recover it. - grub-install --target=i386-pc --recheck=/dev/sda - grub-mkconfig -o /boot/grub/grub.cfg - exit - poweroff Unless Arch has something special, grub-mkconfig calls os-prober to get list of foreign OS. What is output of os-prober (as root)? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Grub is not showing Windows 7 when booting my laptop
В Fri, 25 Oct 2013 11:30:29 -0500 Daniel - Asterisk earohua...@gmail.com пишет: It shows: No volume groups found /dev/sda1: Windows 7(loader): Windows:chain OK, so what do you mean lost? Not present in grub menu? Chento On Fri, Oct 25, 2013 at 11:18 AM, Andrey Borzenkov arvidj...@gmail.comwrote: В Fri, 25 Oct 2013 10:59:45 -0500 Daniel - Asterisk earohua...@gmail.com пишет: Hello all, I'm completely novice with grub and I have a problem, I was re-installing ArchLinux (/dev/sda2) in my laptop with existing Windows 7 (/dev/sda1), but after re-installing grub and running following commands I lost Windows at boot time. I think Windows is still there and I want to recover it. - grub-install --target=i386-pc --recheck=/dev/sda - grub-mkconfig -o /boot/grub/grub.cfg - exit - poweroff Unless Arch has something special, grub-mkconfig calls os-prober to get list of foreign OS. What is output of os-prober (as root)? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB 2.0 not loading config file
В Sun, 20 Oct 2013 01:40:25 -0600 Chris Murphy li...@colorremedies.com пишет: Temporary confusion. The core.img is in gpt2, but it's looking for /boot/grub within the fs on gpt1. So it looks like the prefix is right, except there's no hdX designation before the comma, is that normal? Yes. This simply means - first partition on boot disk. It makes it more robust as it works even if disk order is changed for whatever reason. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB 2.0 not loading config file
В Sat, 19 Oct 2013 11:02:42 +0200 Ortwin Glück o...@odi.ch пишет: Hi, Symptoms: - GRUB 2.0 starts in command-line mode. The boot menu is not loaded. - on the command-line: configfile /boot/grub/grub.cfg properly loads the config file, surprisingly. - GRUB 0.9 works fine. HW: Samsung laptop that I boot in legacy (non-EFI) mode. Have not tried EFI. grub-install creates the image like so: /usr/bin/grub2-mkimage -d /usr/lib/grub/i386-pc -O i386-pc --output=/boot/grub/i386-pc/core.img --prefix=(,gpt1)/boot/grub biosdisk ext2 part_gpt prefix here is used only as fallback; grub will search for it at runtime. So check values of $root and $prefix immediately after starting grub. Your description suggests that $prefix for whatever reason is wrong but $root is correct. Partition tables: MBR: Disk /dev/sda: 126.0 GB, 126035288064 bytes, 246162672 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x3eb19e87 Device Boot Start End Blocks Id System /dev/sda1 1 246162671 123081335+ ee GPT GPT: Disk /dev/sda: 246162672 sectors, 117.4 GiB Logical sector size: 512 bytes Disk identifier (GUID): 6A7696BC-D860-44B9-87E9-6AAECD6C9E4F Partition table holds up to 128 entries First usable sector is 34, last usable sector is 246162638 Partitions will be aligned on 2-sector boundaries Total free space is 0 sectors (0 bytes) Number Start (sector)End (sector) Size Code Name 12048 246162638 117.4 GiB 8300 primary 2 342047 1007.0 KiB EF02 BIOS boot partition Any ideas? Ortwin ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB 2.0 not loading config file
В Sat, 19 Oct 2013 22:07:47 -0600 Chris Murphy li...@colorremedies.com пишет: GPT: Disk /dev/sda: 246162672 sectors, 117.4 GiB Logical sector size: 512 bytes Disk identifier (GUID): 6A7696BC-D860-44B9-87E9-6AAECD6C9E4F Partition table holds up to 128 entries First usable sector is 34, last usable sector is 246162638 Partitions will be aligned on 2-sector boundaries Total free space is 0 sectors (0 bytes) Number Start (sector)End (sector) Size Code Name 12048 246162638 117.4 GiB 8300 primary 2 342047 1007.0 KiB EF02 BIOS boot partition At the grub command line, issue the set command and report the result for prefix. Above it looks like it's setting the prefix to (,gpt1)/boot/grub which is wrong. It should be (hd0,gpt2) based on the GPT you've posted, Based on GPT it should be gpt1. but it's unusual that the partition numbers are reversed given the LBAs being used. The BIOS boot partition is first. There is no requirement that partition numbers correspond to relative partition positions on disk. At least I am not aware of it - do you have reference? It might be worth using gdisk to reorder the partitions and write out the GPT again and then re-installing grub. Chris Murphy ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Unable to boot after moving partition
В Thu, 17 Oct 2013 13:46:12 -0400 Chris Jones cjns1...@gmail.com пишет: I had to move a partion using gparted to make room for something else and I am no longer able to boot that system. I get the following error messages: | error no such device c07cc7e9-1c06-47fe-99b6-16c5145fbc4f | error HD1 cannot get C/H/S values Is it really HD1 (upper case)? GRUB names hard disks hdX (lower case). How many hard disks do you have? | error you need to load the kernel first Checked the uuid on line 1.. and that's in sync with grub.cfg. So it looks like line 2 is telling me that grub is still pointing to the old location..? Tried to run grub-install, hoping this would refresh grub's pointers but still getting the same error. Please advise. CJ ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Unable to boot after moving partition
В Thu, 17 Oct 2013 16:39:02 -0400 Chris Jones cjns1...@gmail.com пишет: On Thu, Oct 17, 2013 at 01:46:12PM EDT, Chris Jones wrote: I'm not entirely sure it was HD1 rather than hd1... since I wrote it by hand. The message apparently comes from biosdisk.mod -- but that does not have the particular disk, only a %s variable that gets replaced by the actual value before it is printed. I have two disks, known as sda sdb under linux. The error simply means that BIOS returned an error when asked about second disk. There is nothing grub can do here. Did you replace disk probably with different disk? What disk size? The partition is somewhere near the end of the second disk - i.e. /dev/sdb7. Thanks, CJ ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB 2.00 conditional structures (statements) syntax
В Tue, 15 Oct 2013 12:20:16 +0200 sas...@freenet.de пишет: Dear all, unfortunately i was not able to find any documentation on usage of conditional statements in grub configuration files. According to the “GNU GRUB Manual 2.00~rc1” found on http://www.gnu.org/ the words 'if', 'fi', 'case', 'esac', 'in' and so on are reserved. Furthermore the scripting language is referred to as “Shell-like”. Simple “if [ $condition ]; then #dosomething; fi”-statements are not problematic. But I wasn't able to find a functional syntax for the case-statements. grub does not support case statement, at least as of now. To make it short: I would like to compare strings in the grub cli. Is there any more or less exhaustive documentation about the syntax or am I just too dumb to use google? What part is unclear? case can in almost all cases be replaced with if. Actually, perl never had case too. That said, if someone implements it, I guess it will be accepted. To compare patters you can use regexp command. Current trunk contains a bit more complete list of available commands. The most recent version I used to test stuff was the 2.00 that comes shipped with arch linux. What I'm trying to do: I want to implement a grub configuration file that automatically generates menu entries. For en example: The following procedure should create entries for (small) files to be loaded with memdisk ( from the syslinux project). /boot/memsik is our kernel. Our file is the initrd. We assume that the files can be PREFIX_*.* or just *.* or README*. (e.g. README || FLOPPY_dellbiosupdate.img || freedos.img) The README*-files should not be processed. For files with a prefix, the prefix should be appended as a kernel-parameter. For files without there are no parameters to append. I supposed that this would be “shell-like” (file=$img): [code source=”/boot/grub/grub.cfg”] . .. insmod ext2 insmod regexp . .. set pathtoimages=”/boot/images” . .. for img in ${pathtoimages}/*; do set appendstr=”” ... case ${img} in */README*) continue ;; */FLOPPY_*) set appendstr=${appendstr} floppy ;; … esac menuentry “MEMDISK - $img” “$img” “$appendstr” { linux16 /boot/memdisk ${3} initrd16 ${2} } done . .. [/code] I also tried to use other quoting notations with no success. Futhermore I also tried “if [[ ${img} == */FLOPPY_* ]]; ...”. Thank you very much in advance! With kind regards, Sasha B. P.S.: Sorry if something's misspelled, English isn't my native tongue. --- Alle Postfächer an einem Ort. Jetzt wechseln und E-Mail-Adresse mitnehmen! http://email.freenet.de/basic/Informationen ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Question about LVM or RAID on /boot
В Wed, 16 Oct 2013 09:13:05 +0800 Kun Huang gar...@unitedstack.com пишет: Sorry, why does that limitation stay in BIOS stage? In whole grub progress, we have no chance to scan all disks? As long as grub is using BIOS to access them - no. It is possible to use native grub disk drivers which bypasses BIOS and directly accesses hard disk controller. On BIOS platforms that would be AHCI. On Tuesday, October 15, 2013, Chris Murphy wrote: On Oct 14, 2013, at 10:22 PM, Kun Huang gar...@unitedstack.comjavascript:_e({}, 'cvml', 'gar...@unitedstack.com'); wrote: Hi all When searching why we couldn't use LVM or RAID on /boot, the most of answers are that couldn't be support. But I would like get to know more details. In grub.conf file, we could use (hd0,0) to specify root=. In another words, grub in stage2 actually could find data in all of disks. Is this right? If grub could find all partitions of all disks, what other reasons stop we find out information of LVM or RAID about and in disks? Works fine with GRUB2 for the past ~2 years at least. Probably longer. The limitation is BIOS seeing all of the disks, not a GRUB limitation. I think there's work that needs to be done to update the GRUB LVM code when it comes to LVM integrated raid levels 10, 5, and 6; and also thin provisioning. Except for raid0, 1, 10, I think the rest is a bit esoteric for /boot, and is probably more trouble than it's worth if there are device failures. I don't actually know off hand if GRUB2's md raid code can rebuild data chunks from parity on the fly, e.g. if there's a 1 disk failure of a raid5 array. If not, yeah, all the more reason I'd keep /boot simple. Chris Murphy ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: grub2-mkconfig with root on rootfs
В Thu, 10 Oct 2013 00:27:45 +0400 Peter Volkov p...@gentoo.org пишет: OK so do I understand it correctly - you are running purely from initramfs and do not have any additional filesystems? yup. On which device do you install grub? Where is your /boot/grub located? Please show full command line and its output used to install grub. I have single hard drive /dev/sdc with two partitions. First partition is formatted with ext4 (second partition is unformatted yet). To install grub I mount it to /mnt/root/ directory This was not present in mntinfo output. How should others guess it? Well, I thought that grub2-mkconfig should work even without /boot (in my case /mnt/root) mounted. So full output is at the end of this mail. and then I run: grub2-install /dev/sdc --boot-directory=/mnt/root/ Then my /boot/grub is located at (hd0, msdos1)/grub. You did not show any output from this command, so I assume this command completes without error? Yes, without any errors. # grub2-install /dev/sdc --boot-directory /mnt/root/ Installation finished. No error reported. And even more, grub2 works correctly if I put grub.cfg into /mnt/root/grub/ and reboot. The problem is that grub2-mkconfig fails at grub2-probe: # /usr/sbin/grub2-probe --target=device / /usr/sbin/grub2-probe: error: failed to get canonical path of `rootfs'. OK, I see. Well, may be grub-mkconfig should support --boot-directory option as well. I suggest you post it to grub-devel, where someone may pick it up. If you could send patch, this would be even better :) As for you case - why do not you simply mount /dev/sda1 on /boot? It *is* your /boot after all, is not it? 32 1 8:1 / /mnt/root rw,noatime - ext4 /dev/sda1 rw,data=ordered ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: query screen resolution for font sizes and theme parameters?
В Tue, 8 Oct 2013 12:12:33 +0200 Ronny Standtke ronny.stand...@fhnw.ch пишет: Hi all I'm using grub2 as the bootloader for a live system. That means that the system is booted on a very diverse range of systems with a very diverse range of graphic boards and monitors. I'm using set gfxmode=auto in the grub configuration and it works (mostly) fine. There is only one remaining problem: On different resolutions I would like to use different font sizes or theme parameters. Is there any way to query grub for the screen resolution that was selected? As far as I know, there is no way. This was already asked recently. Exporting currently set info would not be hard. Something like gfxmode_current probably. Just needs someone to implement ... :) ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Help with custom Grub2 menu
В Sat, 05 Oct 2013 12:34:42 -0500 Richard Owlett rowl...@cloud85.net пишет: Andrey Borzenkov replied on 3 Oct 2013 but I did not receive it from the list ;( When using grub-mkconfig menu is built using scripts in /etc/grub.d. They are run sequentially and their output put into grub.cfg. You can modify or remove any of them, or do not use grub-mkconfig at all and maintain grub.cfg manually. I had understood that much. In grub.cfg, I'm not comfortable changing more than the text of the displayed menu line. After all it does say DO NOT EDIT THIS FILE ;/ Because it is autogenerated and changes will be overwritten. But if you are not using grub-mkconfig to generate it, you are free to do whatever you want. Is there more descriptive information on the files in /etc/grub.d than the included (very brief) comments? Not really; probably having high level overview in grub documentation would be helpful. But all settings that can influence behavior of these scripts are documented. There are several attempts to create different tools to maintain grub.cfg. Thank you ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Help with custom Grub2 menu
В Wed, 02 Oct 2013 14:48:55 -0500 Richard Owlett rowl...@cloud85.net пишет: I have a machine set aside for experimenting with OS installs/configuration. I'm currently experimenting with Debian Squeeze and Wheezy. Debian's installer defaults to 1. making the last install the default when booting - can be bad idea if you mess up the installation. I want the first OS installed to be loaded by default. 2. creating menu title from id of installed kernel. As multiple installs may use the same kernel, I wish to use meaningful label. Is there a tool /or documentation to make life simpler? When using grub-mkconfig menu is built using scripts in /etc/grub.d. They are run sequentially and their output put into grub.cfg. You can modify or remove any of them, or do not use grub-mkconfig at all and maintain grub.cfg manually. There are several attempts to create different tools to maintain grub.cfg. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: grub2-mkconfig with root on rootfs
On Tue, Oct 1, 2013 at 1:51 AM, Peter Volkov p...@gentoo.org wrote: Hi! We are using may be strange but rather convenient scheme of booting linux where real system root is all inside initramfs (/init is symlink to /sbin/init). Everything works fine but grub2-mkconfig that ends out the error: /usr/sbin/grub2-probe: error: failed to get canonical path of `rootfs'. Sure with root on rootfs this grub2-probe is not supposed to work. # mount | grep ' / ' rootfs on / type rootfs (rw) Could you show full /proc/mounts (or even better, /proc/self/mountinfo)? But what can we do with that? Any suggestions to fix this? I'm not CC'ed to the list. Please add me to the answers. -- Peter. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: grub2-mkconfig with root on rootfs
On Tue, Oct 1, 2013 at 3:41 PM, Peter Volkov p...@gentoo.org wrote: В Вт, 01/10/2013 в 10:59 +0400, Andrey Borzenkov пишет: On Tue, Oct 1, 2013 at 1:51 AM, Peter Volkov p...@gentoo.org wrote: We are using may be strange but rather convenient scheme of booting linux where real system root is all inside initramfs (/init is symlink to /sbin/init). Everything works fine but grub2-mkconfig that ends out the error: /usr/sbin/grub2-probe: error: failed to get canonical path of `rootfs'. Sure with root on rootfs this grub2-probe is not supposed to work. # mount | grep ' / ' rootfs on / type rootfs (rw) Could you show full /proc/mounts (or even better, /proc/self/mountinfo)? Yup, here is /proc/self/mountinfo: 1 1 0:1 / / rw - rootfs rootfs rw OK so do I understand it correctly - you are running purely from initramfs and do not have any additional filesystems? On which device do you install grub? Where is your /boot/grub located? Please show full command line and its output used to install grub. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: grub2-mkconfig with root on rootfs
В Wed, 02 Oct 2013 00:13:29 +0400 Peter Volkov p...@gentoo.org пишет: В Вт, 01/10/2013 в 17:18 +0400, Andrey Borzenkov пишет: On Tue, Oct 1, 2013 at 3:41 PM, Peter Volkov p...@gentoo.org wrote: В Вт, 01/10/2013 в 10:59 +0400, Andrey Borzenkov пишет: On Tue, Oct 1, 2013 at 1:51 AM, Peter Volkov p...@gentoo.org wrote: We are using may be strange but rather convenient scheme of booting linux where real system root is all inside initramfs (/init is symlink to /sbin/init). Everything works fine but grub2-mkconfig that ends out the error: /usr/sbin/grub2-probe: error: failed to get canonical path of `rootfs'. Sure with root on rootfs this grub2-probe is not supposed to work. # mount | grep ' / ' rootfs on / type rootfs (rw) Could you show full /proc/mounts (or even better, /proc/self/mountinfo)? Yup, here is /proc/self/mountinfo: 1 1 0:1 / / rw - rootfs rootfs rw OK so do I understand it correctly - you are running purely from initramfs and do not have any additional filesystems? yup. On which device do you install grub? Where is your /boot/grub located? Please show full command line and its output used to install grub. I have single hard drive /dev/sdc with two partitions. First partition is formatted with ext4 (second partition is unformatted yet). To install grub I mount it to /mnt/root/ directory This was not present in mntinfo output. How should others guess it? and then I run: grub2-install /dev/sdc --boot-directory=/mnt/root/ Then my /boot/grub is located at (hd0, msdos1)/grub. You did not show any output from this command, so I assume this command completes without error? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Question: skip menu, but enter by keypress
В Sun, 29 Sep 2013 08:58:20 +0400 Aleksey Midenkov mide...@gmail.com пишет: Is it possible as in LILO to make such configuration: don't show menu, don't do timeout, run default menu entry unless SHIFT key is pressed. If SHIFT menu is pressed, then enter menu. Something like this should work set timeout=0 if keystatus --shift ; then set timeout=5 fi Why I'm asking, because how it's done in Grub -- enter to menu is possible only if timeout is set to 0 is something I don't like. I don't want to spend 1 more sec on each reboot just because it is required for the ability to enter menu. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Booting to UEFI from Bios Boot Partition
В Sat, 28 Sep 2013 13:23:53 -0300 luminair lumin...@gmail.com пишет: Hi there! I'd like to know if GRUB2 installed to BBP and booted in BIOS mode can then load a UEFI boot partition No, it is not possible. EFI binary requires EFI runtime and it is not available after BIOS was loaded. (in my case pointing to an install of Windows 8) on a GPT disk. I ask because I'd like to boot a UEFI GPT disk in a Hyper-V gen1 VM which only supports BIOS mode booting. GRUB2 (or another bootloader) may provide a workaround for me! I've tried to do this already with the version of GRUB2 that comes with the latest version of Ubuntu, but the chainloader throws an error when pointed at the UEFI boot partition. Thanks! ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: xnu_kernel64: Unable to boot into Mountain Lion
В Thu, 26 Sep 2013 13:53:20 +0300 Mihai Draghicioiu mihai.draghici...@gmail.com пишет: Hello everyone! I'm trying to get a triple boot 64-bit Intel PC going correctly. There are Windows 8, Debian 7 and OSX Mountain Lion on it. Windows 8 and Debian boot fine from Grub, but the only way I got Mountain Lion to boot was via chainloading the Chameleon bootloader. The other option is using the xnu_* commands available, and update-grub generates the proper menu entries, however, when I choose the 32 bit menu entry it says there is no suitable kernel for 32 bit, and when I choose the 64 bit entry, it hangs for 1-2 seconds and then reboots. I've managed to place echo's in the boot sequence, and every command is executed before the reboot takes place. There is no output, the machine reboots before reaching the Apple logo loading screen. If this is a bug, and not a misuse on my part, please mention so I can submit it to the grub and debian bug trackers (personally I think it's already a Debian bug). Yes, apparently grub xnu loader cannot boot modern OS/X versions. I was able to boot it by directly chainloading OS/X EFI bootloader. Path was /usr/standalone/i386/boot.efi if I'm not mistaken. This avoids need to use third-party programs. Also, perhaps there are some debug flags I can turn on to see what's happening? I've found a similar situation on this webpage http://www.insanelymac.com/forum/topic/282748-grub2-xnu-kernel/ Now I'm going to provide as much info as I can, please specify if more is needed to get a proper idea why this doesn't work: OS X version is Mountain Lion 10.8 HDD is 1TB Partition table (MBR): Primary 200G NTFS (Win 8) Primary 100G HFS+ (OSX) Active Extended Linux 20G Swap 5G 600G NTFS (Storage) # fdisk /dev/sda The device presents a logical sector size that is smaller than the physical sector size. Aligning to a physical sector (or optimal I/O) size boundary is recommended, or performance may be impacted. Command (m for help): p Disk /dev/sda: 1000.2 GB, 1000204886016 bytes 255 heads, 63 sectors/track, 121601 cylinders, total 1953525168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 4096 bytes I/O size (minimum/optimal): 4096 bytes / 4096 bytes Disk identifier: 0xfa12c820 Device Boot Start End Blocks Id System /dev/sda12048 419432447 2097152007 HPFS/NTFS/exFAT /dev/sda2 * 419432448 629147647 104857600 af HFS / HFS+ /dev/sda3 629147648 1953525167 6621887605 Extended /dev/sda5 629149696 67109273520971520 83 Linux /dev/sda6 671094784 681580543 5242880 82 Linux swap / Solaris /dev/sda7 681582592 1953525167 6359712887 HPFS/NTFS/exFAT Here are a few menu entries that generate this behavior: This I got from some webpage - it seems minimal menuentry Mac OS X { set root=(hd0,2) insmod video insmod vbe gfxmode=1280x800x32 xnu_kernel /mach_kernel rd=disk0s2 if [ /System/Library/Extensions.mkext -nt /System/Library/Extensions ]; then xnu_mkext /System/Library/Extensions.mkext else xnu_kextdir /System/Library/Extensions fi } This is generated by update-grub: menuentry Mac OS X (64-bit) (on /dev/sda2) --class osx --class darwin --class os { insmod part_msdos insmod hfsplus set root='(/dev/sda,msdos2)' search --no-floppy --fs-uuid --set=root 689f9c8a6b4f3520 load_video set do_resume=0 if [ /var/vm/sleepimage -nt10 / ]; then if xnu_resume /var/vm/sleepimage; then set do_resume=1 fi fi if [ $do_resume = 0 ]; then xnu_uuid 689f9c8a6b4f3520 uuid if [ -f /Extra/DSDT.aml ]; then acpi -e /Extra/DSDT.aml fi xnu_kernel64 /mach_kernel boot-uuid=${uuid} rd=*uuid if [ /System/Library/Extensions.mkext -nt /System/Library/Extensions ]; then xnu_mkext /System/Library/Extensions.mkext else xnu_kextdir /System/Library/Extensions fi if [ -f /Extra/Extensions.mkext ]; then xnu_mkext /Extra/Extensions.mkext fi if [ -d /Extra/Extensions ]; then xnu_kextdir /Extra/Extensions fi if [ -f /Extra/devprop.bin ]; then xnu_devprop_load /Extra/devprop.bin fi if [ -f /Extra/splash.jpg ]; then insmod jpeg xnu_splash /Extra/splash.jpg fi if [ -f /Extra/splash.png ]; then insmod png xnu_splash /Extra/splash.png fi if [ -f /Extra/splash.tga ]; then insmod tga xnu_splash /Extra/splash.tga fi fi } And, finally, this is what I'm using which
Re: What is the license of grub-crypt(8) command?
В Mon, 23 Sep 2013 11:11:21 +0900 Hiroshi Umehara zinrai0...@gmail.com пишет: The license is not specified in the grub-crypt(8) command. What is the license of grub-crypt(8) command? There is no such command in grub sources. You need to ask your distribution which probably provides this add-on. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Famous Your embedding area is unusually small however my embedding area looks like just the right size
В Fri, 06 Sep 2013 21:28:39 -0700 Mike Power mpo...@alumni.calpoly.edu пишет: I am trying to reinstall grub on my system. Grub was installed before but I have had hard drive troubles, so one of the drives does not have grub on it. I have a lvm on top of a mirror raid. I run the following command $sudo grub-install /dev/sda [sudo] password for mpower: /usr/sbin/grub-setup: warn: Your embedding area is unusually small. core.img won't fit in it.. /usr/sbin/grub-setup: error: embedding is not possible, but this is required when the root device is on a RAID array or LVM volume. This seems odd since it has grub on it. I really should have tried sdb first. But none the less if I look at sda it has 62 unused sectors on the front. From what I have read that is the right amount bor@opensuse:~ grub2-mkimage -O i386-pc -o /tmp/core.img ext2 mdraid1x lvm search_fs_uuid bor@opensuse:~ LC_ALL=C ll /tmp/core.img -rw-r--r-- 1 bor bor 32099 Sep 7 11:32 /tmp/core.img That exceeds 31KiB (31744B). $ sudo fdisk -l /dev/sda Disk /dev/sda: 500.1 GB, 500107862016 bytes 255 heads, 63 sectors/track, 60801 cylinders, total 976773168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00018e76 Device Boot Start End Blocks Id System /dev/sda1 63 976768064 488384001 fd Linux raid autodetect Any idea what broke? Some additional info: $ sudo pvdisplay --- Physical volume --- PV Name /dev/md0 VG Name vg PV Size 465.76 GiB / not usable 888.00 KiB Allocatable yes (but full) PE Size 4.00 MiB Total PE 119234 Free PE 0 Allocated PE 119234 PV UUID wS9qYc-jWkT-FR3Z-bi0H-cY8I-wLS6-2Z1dPl $ sudo vgdisplay --- Volume group --- VG Name vg System ID Formatlvm2 Metadata Areas1 Metadata Sequence No 3 VG Access read/write VG Status resizable MAX LV0 Cur LV2 Open LV 2 Max PV0 Cur PV1 Act PV1 VG Size 465.76 GiB PE Size 4.00 MiB Total PE 119234 Alloc PE / Size 119234 / 465.76 GiB Free PE / Size 0 / 0 VG UUID QwDeNr-aQvO-5uYv-ImDP-C8eJ-uKyi-ZHYWtk $ sudo lvdisplay --- Logical volume --- LV Name/dev/vg/swap_1 VG Namevg LV UUIDhxn1wt-EcJP-8rfO-3lCc-X3OM-gG3f-uuR3WK LV Write Accessread/write LV Status available # open 2 LV Size7.45 GiB Current LE 1907 Segments 1 Allocation inherit Read ahead sectors auto - currently set to 256 Block device 253:0 --- Logical volume --- LV Name/dev/vg/root VG Namevg LV UUIDtjmJkA-njlJ-VuDz-mBvY-5hpB-bsMD-ftKMXP LV Write Accessread/write LV Status available # open 1 LV Size458.31 GiB Current LE 117327 Segments 1 Allocation inherit Read ahead sectors auto - currently set to 256 Block device 253:1 $ sudo mdadm -D /dev/md0 /dev/md0: Version : 1.2 Creation Time : Wed Jan 11 20:30:26 2012 Raid Level : raid1 Array Size : 488383352 (465.76 GiB 500.10 GB) Used Dev Size : 488383352 (465.76 GiB 500.10 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Update Time : Fri Sep 6 21:26:31 2013 State : clean Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Name : petros:0 (local to host petros) UUID : dea84e95:1a6753ea:2982d49d:b812ff86 Events : 2204266 Number Major Minor RaidDevice State 2 810 active sync /dev/sda1 3 8 171 active sync /dev/sdb1 $ grub-install -v grub-install (GRUB) 1.99-12ubuntu5.1 $ lsb_release -a No LSB modules are available. Distributor ID:
Re: GRUB netboot with HTTP download
В Wed, 28 Aug 2013 19:38:16 +0200 Daniel Dehennin daniel.dehen...@baby-gnu.org пишет: Hello, I try to setup a DHCP/TFTP/HTTP netboot but always get the rescue shell. Are you using current trunk? I start with grub-mknetdir: /usr/sbin/grub-mknetdir --net-directory=tftp --subdir=grub --modules=http I would like to make grub use HTTP to download files, so I modify the grub image to include http and set the prefix: grub-mkimage -d /usr/lib/grub/x86_64-efi -O x86_64-efi \ --output=tftp/grub/x86_64-efi/core.efi \ --prefix='(http,10.0.0.1)/grub' \ tftp http efinet This result in the following: error: no server is specified. grub rescue set net_default_server= prefix=(http,10.0.0.1)/grub pxe_default_server= root=http,10.0.0.1 grub rescue ls (hd0) (hd1) The error message seems to come from grub-core/net/net.c[1] resulting of a miss-match[2] of my “prefix”. I looked for documentation on how to use grub modules but only found explanation on PXE[3] and device syntax[4]. I should have done something wrong but I do not see what, any hints? Regards. Footnotes: [1] http://bzr.savannah.gnu.org/lh/grub/trunk/grub/annotate/head:/grub-core/net/net.c#L1279 [2] http://bzr.savannah.gnu.org/lh/grub/trunk/grub/annotate/head:/grub-core/net/net.c#L1262 [3] https://www.gnu.org/software/grub/manual/grub.html#Network [4] https://www.gnu.org/software/grub/manual/grub.html#Device-syntax signature.asc Description: PGP signature ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: What updates the Grub2 Device map?
В Sat, 17 Aug 2013 19:14:01 -0700 Upscope upsc...@nwi.net пишет: In removed an unused disk from my system. Afterward I Looked at the partitioner and it does not show there. Also does not show in grub-menu, os-prober or system info but it still shows in grub2 device map. Ran the update for grub2. It still shows in the device map. I could just edit the device map but I may screw something up doing that. grub2 does not change device.map. It may be created or updated by your system management programs (e.g. openSUSE YaST2 will update device.map) or manually if required. grub2 will use device.map if present but will never change it itself. An ideas where to look? Thanks! Russ ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: chainloader grub1 to grub2
В Wed, 14 Aug 2013 14:02:58 -0300 robert verge robert.ver...@gmail.com пишет: The only thing old me back from upgrading from grub1 to grub2 is loading a chainloader config which loads directly from the drive. I have not found the equivalent in grub2. Any suggestions ? This is the current configuration that works in grub1 title ESXI on (hd2) chainloader (hd2)+1 root(hd2) menuentry ESXI on (hd2) { set root=hd2 chainloader ($root)+1 } ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Using grub, is it possible to use if, while during booting (before loading normal.mod)?
В Tue, 13 Aug 2013 07:33:28 -0700 Binh Minh binhminhmatt...@gmail.com пишет: It is a security feature requires my implementation to ask for the user's input and compare the input before we process information in grub.cfg. It is proprietary so I can't discuss more than that. If I include modules echo, sleep, read during grub-mkimage, I was able to have those commands in myconfig.cfg, and this works before normal.mod is loaded. But how come if statement is unknown? The grub documentation on this web site http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmlhas an example showing using if statement is possible. What kind of built-in script parser is used when processing the config file that was linked into core.img? normal.mod. Quoting documentation: To do this, include the `configfile' and `normal' modules in the core image, On Tue, Aug 13, 2013 at 2:39 AM, Andrey Borzenkov arvidj...@gmail.comwrote: On Tue, Aug 13, 2013 at 9:10 AM, Binh Minh binhminhmatt...@gmail.com wrote: I am trying to read in a user's input, compare the user's input to X. If the user's input matches X, proceed to load normal.mod; otherwise loop back to get user's input. And why you cannot do it after normal.mod is loaded? On Mon, Aug 12, 2013 at 7:39 PM, Andrey Borzenkov arvidj...@gmail.com wrote: В Mon, 12 Aug 2013 13:39:24 -0700 Binh Minh binhminhmatt...@gmail.com пишет: My computer has grub installed. During booting before normal.mod is loaded, I need to be able to run if and while commands in my config file which has been linked into core.img (using grub-mkimage -c myconfig.confg). In myconfig.config, I have an if statement and I kept getting unknown command if during booting. I saw an example in http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmland looks like I just need to include search, test, and normal modules. Am I missing something? Thanks It is not possible. Those commands are provide by normal.mod. The only purpose of core.img is normally to load normal.mod. What are you trying to do? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Using grub, is it possible to use if, while during booting (before loading normal.mod)?
В Tue, 13 Aug 2013 09:38:19 -0700 Binh Minh binhminhmatt...@gmail.com пишет: I included the normal and configfile modules in grub-mkimage and still system does not recognize if Yes, documentation apparently is wrong here. Embedded config is parsed before normal.mod is loaded so it supports only what is available in rescue mode. On Aug 13, 2013 9:15 AM, Andrey Borzenkov arvidj...@gmail.com wrote: В Tue, 13 Aug 2013 07:33:28 -0700 Binh Minh binhminhmatt...@gmail.com пишет: It is a security feature requires my implementation to ask for the user's input and compare the input before we process information in grub.cfg. It is proprietary so I can't discuss more than that. If I include modules echo, sleep, read during grub-mkimage, I was able to have those commands in myconfig.cfg, and this works before normal.mod is loaded. But how come if statement is unknown? The grub documentation on this web site http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmlhas an example showing using if statement is possible. What kind of built-in script parser is used when processing the config file that was linked into core.img? normal.mod. Quoting documentation: To do this, include the `configfile' and `normal' modules in the core image, On Tue, Aug 13, 2013 at 2:39 AM, Andrey Borzenkov arvidj...@gmail.com wrote: On Tue, Aug 13, 2013 at 9:10 AM, Binh Minh binhminhmatt...@gmail.com wrote: I am trying to read in a user's input, compare the user's input to X. If the user's input matches X, proceed to load normal.mod; otherwise loop back to get user's input. And why you cannot do it after normal.mod is loaded? On Mon, Aug 12, 2013 at 7:39 PM, Andrey Borzenkov arvidj...@gmail.com wrote: В Mon, 12 Aug 2013 13:39:24 -0700 Binh Minh binhminhmatt...@gmail.com пишет: My computer has grub installed. During booting before normal.mod is loaded, I need to be able to run if and while commands in my config file which has been linked into core.img (using grub-mkimage -c myconfig.confg). In myconfig.config, I have an if statement and I kept getting unknown command if during booting. I saw an example in http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmland looks like I just need to include search, test, and normal modules. Am I missing something? Thanks It is not possible. Those commands are provide by normal.mod. The only purpose of core.img is normally to load normal.mod. What are you trying to do? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Using grub, is it possible to use if, while during booting (before loading normal.mod)?
В Tue, 13 Aug 2013 07:33:28 -0700 Binh Minh binhminhmatt...@gmail.com пишет: It is a security feature requires my implementation to ask for the user's input and compare the input before we process information in grub.cfg. If you are concerned about tampering with on-disk file - current trunk supports (mandatory) signature verification. So you load trusted minimal grub.cfg which then loads another script for final processing. It is proprietary so I can't discuss more than that. If I include modules echo, sleep, read during grub-mkimage, I was able to have those commands in myconfig.cfg, and this works before normal.mod is loaded. But how come if statement is unknown? The grub documentation on this web site http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmlhas an example showing using if statement is possible. What kind of built-in script parser is used when processing the config file that was linked into core.img? On Tue, Aug 13, 2013 at 2:39 AM, Andrey Borzenkov arvidj...@gmail.comwrote: On Tue, Aug 13, 2013 at 9:10 AM, Binh Minh binhminhmatt...@gmail.com wrote: I am trying to read in a user's input, compare the user's input to X. If the user's input matches X, proceed to load normal.mod; otherwise loop back to get user's input. And why you cannot do it after normal.mod is loaded? On Mon, Aug 12, 2013 at 7:39 PM, Andrey Borzenkov arvidj...@gmail.com wrote: В Mon, 12 Aug 2013 13:39:24 -0700 Binh Minh binhminhmatt...@gmail.com пишет: My computer has grub installed. During booting before normal.mod is loaded, I need to be able to run if and while commands in my config file which has been linked into core.img (using grub-mkimage -c myconfig.confg). In myconfig.config, I have an if statement and I kept getting unknown command if during booting. I saw an example in http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmland looks like I just need to include search, test, and normal modules. Am I missing something? Thanks It is not possible. Those commands are provide by normal.mod. The only purpose of core.img is normally to load normal.mod. What are you trying to do? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Using grub, is it possible to use if, while during booting (before loading normal.mod)?
В Mon, 12 Aug 2013 13:39:24 -0700 Binh Minh binhminhmatt...@gmail.com пишет: My computer has grub installed. During booting before normal.mod is loaded, I need to be able to run if and while commands in my config file which has been linked into core.img (using grub-mkimage -c myconfig.confg). In myconfig.config, I have an if statement and I kept getting unknown command if during booting. I saw an example in http://www.gnu.org/software/grub/manual/html_node/Embedded-configuration.htmland looks like I just need to include search, test, and normal modules. Am I missing something? Thanks It is not possible. Those commands are provide by normal.mod. The only purpose of core.img is normally to load normal.mod. What are you trying to do? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: detecting disks and disk size
В Thu, 8 Aug 2013 20:33:57 + Loving, Kent kent.lov...@boeing.com пишет: Hello, I am using GRUB on a CD to boot a specialized live linux on the CD. I would like to customize the GRUB menu based on the number of hard disks that are installed in the machine that is booting the CD. I would also like to customize based on the size of the installed disks. I've searched GUB manual, and tried every command at the GRUB prompt that I hope would work. But nothing I've tried is helping. I tried to use ls to check (hd0) and (hd1), but ls always returns 0, even if it prints an error. Once the linux is booted I can use something like cat /sys/block/sda/size but that's no help in GRUB. Is there a command I am missing, or maybe a 'diskinfo' module? It sounds similar to http://lists.gnu.org/archive/html/help-grub/2013-06/msg00039.html ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Format option for chainloading GRUB from BOOTMGR
В Wed, 7 Aug 2013 13:23:18 -0400 Alexei Podtelezhnikov apodt...@gmail.com пишет: On Wed, Aug 7, 2013 at 12:16 PM, Andrey Borzenkov arvidj...@gmail.comwrote: Alexei Podtelezhnikov apodt...@gmail.com пишет: How is location of /boot/grub relevant here? To change it you update grub2.img and if I understand you correctly, BCD does not need to be updated for it? Correct. BCD needs to know the location of grub2.img only, not what's inside. It is too bad that UUID cannot be given in --prefix. The search_fs_uuid stuff is not for the end user. grub-mkimage (or grub-probe for that matter) are not end user tools. End user should use grub-install which does search for UUID and creates image that looks for it. How about making this a part of i386-pc-lnx format execution together with automatic appending lnxboot.img? I think if you send patch to grub-devel we'll see. It should likely go into grub-install. It may even be generalized by allowing multiple boot methods for a single platform. I have no idea how the interface should look like so that grub-install puts the image onto an NTFS bootmgr partition root rather than its boot sector. I would say, just grub-install --target i386-pc-lnx /path/to/image My initial post was to fill the gap: while there is a lot of information on how to chainload BOOTMGR from GRUB, there is no information on chainloading GRUB from BOOTMGR. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Format option for chainloading GRUB from BOOTMGR
В Mon, 5 Aug 2013 13:30:04 -0400 Alexei Podtelezhnikov apodt...@gmail.com пишет: Hi all, Thanks to ideas in [1], I managed to chainload GRUB2 core.img from Windows 7 BOOTMGR. Here is the three steps. 1) Create core.img with **full** direct unmounted path to grub directory starting from partition, for example grub2-mkimage --output=./core.img --format=i386-pc --prefix=*(hd0,1)*/grub2 biosdisk part_msdos ext2 More robust is to use embedded config to search for partition UUID. Hard disk enumeration may change between reboots. 2) Concatenate lnxboot.img cat /boot/grub2/i386-pc/lnxboot.img ./core.img ./grub2.img 3) Place ./grub2.img onto windows boot partition and configure BCD with bcdedit.exe under Windows to load the file as a BOOTSECTOR application. It turns out that ./grun2.img can be larger than 512 bytes. BOOTMGR will load it just fine. Now, to be honest, the second step is just annoying. Do you have a format option for grub2-mkimage that would automatically include lnxboot.img? It would make the whole process so much smoother and more logical. Just a suggestion. Adding something like i386-pc-lnx is relatively trivial from technical PoV. The main problem is that we need to keep boot code and /boot/grub directory in think - i.e. both must be updated in one step. Does BCD configuration need update after grub2.img gets updated? Thanks, Alexei [1] http://blog.mudy.info/2010/08/boot-grub2-stage2-directly-from-windows-bootmgr-with-grub4dos-stage1/ ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: new install not booting - solved, but questions remain
В Sat, 03 Aug 2013 15:34:12 -0400 Miles Fidelman mfidel...@meetinghouse.net пишет: Hi Folks, Just installed Debian Wheezy onto a remote server, and encountered some problems getting grub to boot properly. Figured it out through trial and error, but an explanation of what's going on would be very much appreciated. Note, most of my familiarity is with grub-legacy, this is my first time working through things with grub 2. Basic setup: - remote install via IPMI console and later via ssh - PXEboot into the installer - partition and configure 4 disk drives into md devices for boot, swap, root, LVM - USB drive shows up as /dev/sda during install - with hard drives as /dev/sdb-e - after install hard drives show up as /dev/sda-d, usb drive as /dev/sde - pretty standard install process until the end - had to fiddle a bit to install MBR into hard drives (default would have put it on the USB) - attempt to boot, end up at grub prompt on the remote console - type boot and get error: no kernel loaded - did some googling, based on what I found, tried the following (from the grub prompt) -- set prefix=(md/0)/boot/grub -- set root=(md/2) -- boot -- still get error: no kernel loaded -- noted that ls (md/0)/ showed System.map, config-, vmlinuz, initrd as well as grub -- tried the command linux (md/0)/vmlinuz... - let it autocomplete the full file name - got error: file not found #THIS IS CONFUSING - SINCE LS AND AUTOCOMPLETE FOUND THE FILE -- insmod normal -- insmod -- now I get a boot screen, and after the timeout, everything booted normally - reboot brought be back to the grub prompt - this time, after booting: -- ran update-grub and grub-install on all four drives, and and on /dev/md0 (boot) and /dev/md2 (root) for good measure -- rebooted - everything came up fine So problem is solved but 2 questions But I'm left with two questions: - What exactly is going on? (WHY did the above fix the problem?) core.img generated during installation contained wrong reference to /boot/grub directory (most likely incorrect disk designation). It is impossible to say more with information you provided. When you rebuilt it, it apparently picked up the working one (again, I won't say correct one because we do not know it). - Why didn't the install set things up properly? (This one is really a question for the debian-boot list, and perhaps a bug report - but I'd kind of like to understand what's going on with grub, and what should be installed, before focusing on the installer.) Thanks very much, Miles Fidelman ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB2, GPT, mdadm LVM
В Fri, 19 Jul 2013 11:25:54 +0200 Sander Smeenk ssme...@freshdot.net пишет: Quoting Andrey Borzenkov (arvidj...@gmail.com): Yes. MD array will appear in the list as soon as at least one member was seen, but we need enough members so it becomes ready and can be scanned for LVM. May be extending ls so it can show diskfilter volume states would be useful. So the first think to check is how many physical disks are visible at grub prompt. Only the first 8 disk (on the first controller?) show up in grub: OK that explains why LVM never appears. grub rescue ls (md/0) (hd0) (hd0,gpt2) (hd0,gpt1) (hd1) (hd1,gpt2) (hd1,gpt1) (hd2) (hd2,gpt2) (hd2,gpt1) (hd3) (hd3,gpt2) (hd3,gpt1) (hd4) (hd4,gpt2) (hd4,gpt1) (hd5) (hd5,gpt2) (hd5,gpt1) (hd6) (hd6,gpt2) (hd6,gpt1) (hd7) (hd7,gpt2) (hd7,gpt1) There should be another 8 disks. Now i wonder if this is GRUB at fault or the controller not publishing its disks properly(?). I'll look into the controller settings... grub is at mercy of (controller) BIOS here. It can only access devices that are presented by BIOS. Are they all on the same controller or on different ones? If you have two different controllers, may be you need to enable BIOS (option ROM or however is it called) on the second one. If your controller is AHCI compliant (I assume it is unlikely to be IDE) you could try to use ahci instead of biosdisk as physical disk driver. You need to build different image for it. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB2, GPT, mdadm LVM
В Thu, 18 Jul 2013 13:56:21 +0200 Sander Smeenk ssme...@freshdot.net пишет: Hi, I'm having issues getting GRUB to boot from my setup. It drops to an rescue shell stating it can't find the LVM root for my system. Similar 'stacked' setups (md raid5/6 lvm, root on lvm) work fine: difference seems to be GPT. Setup is: 16x 3TB disks with GPT partitions: | 1 2048 4095 1024.0 KiB EF02 BIOS boot partition | 2 4096 5860533134 2.7 TiB FD00 Linux RAID Next i created an mdadm /dev/md0 with raid6: | md0 : active raid6 sda2[0] [ .. ] sdp2[2] | 41021865472 blocks super 1.2 level 6, 64k chunk, algorithm 2 [16/16] [] | [==..] resync = 10.1% (296747044/2930133248) finish=985.3min speed=44540K/sec The entire /dev/md0 is a Physical Volume for LVM on which two Logical Volumes were created, one for / and one for swap. This setup works just fine when activated from within a LiveCD like GRML: | # lvdisplay | grep LV Name | LV Namehost_root | LV Namehost_swap | | # mount /dev/mapper/fdi-host_root /tgt | | # ls /tgt/boot/ | System.map-3.8.0-26-generic abi-3.8.0-26-generic config-3.8.0-26-generic | grub/ initrd.img-3.8.0-26-generic vmlinuz-3.8.0-26-generic During installation no errors are logged. GRUB embeds fine in the BIOS boot partition and the generated grub.cfg shows 'insmod lvm' and 'insmod raid6rec' amongst others. The BIOS starts GRUB just fine, it's GRUB where it seems to fail: During boot, GRUB starts and drops to 'grub rescue' stating Error: disk lvm/fdi-host_root not found. Typing 'ls' in the rescue shell shows the '(md/0)' device, but LVM is never started by GRUB. Much more than 'ls' isn't available in the rescue shell. Typing 'insmod lvm' just returns without error, typing 'insmod thisdoesnotexist' complains GRUB can't find the disk 'lvm/fdi-host_root'. Could someone please advise where i am going wrong? I tried (re)creating the core.img like so: | # grub-mkimage -o /boot/grub/core_lvm.img -O i386-pc \ | part_gpt part_msdos mdraid09 mdraid1x raid6rec \ | diskfilter lvm ext2 Your image is lacking physical disk driver. This will be biosdisk on PC BIOS platform. And put that in all the disks: | # grub-bios-setup -b i386-pc/boot.img -c core_lvm.img /dev/sdX Alas, to no avail. Thanks! -Sndr ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Obtaining the UUID of the system for a PXE boot
I guess further discussion should be really moved to grub-devel В Thu, 18 Jul 2013 12:39:44 +0200 Holger Goetz holg...@hgsys.de пишет: On 17.07.2013 16:31, Andrey Borzenkov wrote: On Wed, Jul 17, 2013 at 3:19 PM, Holger Goetzholg...@hgsys.de wrote: Hello all, i've started to use grub.efi as boot loader for UEFI based systems and am getting slowly familiar w/ grub2.Now i've come across a problem w/ assigning individual boot configurations to target system identification. The BIOS line of systems gets booted through pxelinux and get their PXE configuration through files w/ names based on the individual UUID of the system. The same should be possible for UEFI booted systems, but i couldn't find a way to get hold of the system-UUID yet. The MAC addresse(s) is/are not available and possibly have changed since the definition of the configuration, so even the lately introduced net_default_mac variable doesn't work here. Having the UUID as unique system identifyer across NICs which can be changed would be of big help. Any pointer/help would be greatly appreciated. So, you need to fetch system GUID used by PXE, right? To be honest, I do not even know how (if it is possible at all) to obtain it. The only words about PXE GUID in EFI allow to switch between using GUID and MAC for client identification, but that's all: === SendGUIDThis field is used to change the Client Hardware Address (chaddr) field in the DHCP and Discovery packets. Set to TRUE to send the SystemGuid (if one is available). === Do you know how to get this information? Hi Andrey, the system UUID is part of the SMBIOS structure which can be searched between 0xf000 and 0x10 in memory. It has a signature string _SM_. With in the SMBIOS table (memory block) the table type 1 identifys the UUID table and it contains a 16byte binary number which typically is returned in a form like 00112233-4455-6677-8899-AABBCCDDEEFF. See the SMBIOS specs here: http://www.dmtf.org/standards/smbios. David Michael posted patch to grub-devel which provides access to SM BIOS information. You may be interested in checking to which extent it could be used in your case. Thread title is [PATCH/RFC] cpuid.c: Provide CPU model information Also in the spec is some discussions around byte swappig parts of the uuid, for consistency probably a byte by byte representation as a string would be best. One implementation can be found in eg. syslinux's core/dmi.c - including a rudimentary validation check. Other implementations under Linux (eg. dmidecode) often take /proc/efi/systab or /sys/firmware/efi/systab as starting points to the SMBIOS structure, but that's obviously not available at the time grub starts ;-) I'm still not sure what exactly you need. There are three steps in booting grub2 over network 1. Firmware loads core.img using PXE. At this point it may provide MAC or GUID as client identification 2. core.img loads normal.mod which loads grub.cfg. This is done using location hardcoded in core.img. 3. Finally grub.cfg loads other files, loads OS kernel and executes it. It is using whatever features and commands grub2 provides and may decide location dynamically. Now, at which step exactly you need to know and use system GUID from SM BIOS? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB2, GPT, mdadm LVM
В Thu, 18 Jul 2013 20:35:43 +0200 Sander Smeenk ssme...@freshdot.net пишет: Quoting Chris Murphy (li...@colorremedies.com): During boot, GRUB starts and drops to 'grub rescue' stating Error: disk lvm/fdi-host_root not found. This means it's not finding normal.mod or grub.cfg. Yep. ;) [..] So it's a bit curious that it's looking for its files on an lvm root rather than on /boot. I could have stated that clearer indeed. /boot and everyting else is on the /dev/mapper/fdi-host_root LV. I checked with 'set', the prefix= and root= settings seem correct: root='lvm/fdi-host_root' prefix=(lvm/fdi-host_root)/boot/grub But please note that doing 'ls' in the rescue shell does NOT show a 'device' named lvm/fdi-host_root or anything that would indicate it might be the LVM LV i need. Only (md/0) and all the (hdX/gptY) 'devices' show. Typing 'ls' in the rescue shell shows the '(md/0)' device, but LVM is never started by GRUB. Only if /boot/grub files are also on LVM is that needed, in which case lvm.mod needs to have been baked into the core.img. But the fact that you're getting this message sounds like it's not loading lvm.mod. So it seems. Yet, if i type 'insmod lvm' in the grub rescue shell, there is no error returned. This would indicate to me that the .mod was (already?) loaded correctly, because if i type 'insmod foobar' instead, grub complains it can't read/find the host_root LV, which is correct since 'foobar' isn't in core.img so grub would have to load it from disk.. I forget, but lsmod might work in the grub rescue shell to list what modules are available in core.img. It seems my core.img doesn't support lsmod. :( I tried (re)creating the core.img like so: | # grub-mkimage -o /boot/grub/core_lvm.img -O i386-pc \ | part_gpt part_msdos mdraid09 mdraid1x raid6rec \ | diskfilter lvm ext2 And put that in all the disks: | # grub-bios-setup -b i386-pc/boot.img -c core_lvm.img /dev/sdX You might be better off using grub-install --debug and figuring out how it identifies the various parts. Meh. Why didn't i think of doing that ;) 'No errors reported.', okay then, that's not my problem! :P But all seems fine from the start: Please paste exact command invocation and full output. | + abstractions=diskfilter mdraid1x raid6rec lvm | + grub_device=/dev/mapper/fdi-host_root | + devabstraction_module=diskfilter mdraid1x raid6rec lvm | + prefix_drive=(lvm/fdi-host_root) There is a small bug in the process where part_gpt gets added to the 'modules=' list for each disk in my system, resulting in: | + /usr/bin/grub-mkimage -d /usr/lib/grub/i386-pc -O i386-pc \ | --output=/boot/grub/i386-pc/core.img --prefix=(lvm/fdi-host_root)/boot/grub \ | biosdisk ext2 part_gpt part_gpt part_gpt part_gpt part_gpt part_gpt \ | part_gpt part_gpt part_gpt part_gpt part_gpt part_gpt part_gpt \ | part_gpt part_gpt part_gpt diskfilter mdraid1x raid6rec lvm being called. However i would assume this is harmless... Then grub-bios-setup runs. It scans all disks and logs: | info: Inserting hostdisk//dev/sdX into md/0 (mdraid1) for each of my disks. It's not actually raid1, but raid6, i'm not sure wether 'mdraid1' is an all purpose mdraid module? This *might* be an issue? grub-setup continues nonetheless: | info: Scanning for lvm devices on disk md/0. | info: Found array fdi. | info: Inserting md/0 into fdi (lvm) | [..] | info: guessed root_dev `lvm/fdi-host_root' from dir `/boot/grub/i386-pc'. | info: setting the root device to `lvm/fdi-host_root'. | info: the first sector is 2048,0,512. perfectly true. It saves the sectors from 2048 to 2183, which seems the size of core.img and logs 'Installation finished. No error reported.' I'm at a loss :) Thanks for your time, -Sndr. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: GRUB2, GPT, mdadm LVM
В Thu, 18 Jul 2013 20:37:49 +0200 Sander Smeenk ssme...@freshdot.net пишет: Quoting Andrey Borzenkov (arvidj...@gmail.com): I tried (re)creating the core.img like so: | # grub-mkimage -o /boot/grub/core_lvm.img -O i386-pc \ | part_gpt part_msdos mdraid09 mdraid1x raid6rec \ | diskfilter lvm ext2 Your image is lacking physical disk driver. This will be biosdisk on PC BIOS platform. Unfortunately, adding biosdisk to the list doesn't give me a working GRUB either. Device (md/0) shows but (lvm/fdi-host_root) doesn't. Could you test current trunk? There were some changes in how grub scans for multilayer devices; it may have fixed it. ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub
Re: Obtaining the UUID of the system for a PXE boot
On Wed, Jul 17, 2013 at 3:19 PM, Holger Goetz holg...@hgsys.de wrote: Hello all, i've started to use grub.efi as boot loader for UEFI based systems and am getting slowly familiar w/ grub2.Now i've come across a problem w/ assigning individual boot configurations to target system identification. The BIOS line of systems gets booted through pxelinux and get their PXE configuration through files w/ names based on the individual UUID of the system. The same should be possible for UEFI booted systems, but i couldn't find a way to get hold of the system-UUID yet. The MAC addresse(s) is/are not available and possibly have changed since the definition of the configuration, so even the lately introduced net_default_mac variable doesn't work here. Having the UUID as unique system identifyer across NICs which can be changed would be of big help. Any pointer/help would be greatly appreciated. So, you need to fetch system GUID used by PXE, right? To be honest, I do not even know how (if it is possible at all) to obtain it. The only words about PXE GUID in EFI allow to switch between using GUID and MAC for client identification, but that's all: === SendGUIDThis field is used to change the Client Hardware Address (chaddr) field in the DHCP and Discovery packets. Set to TRUE to send the SystemGuid (if one is available). === Do you know how to get this information? ___ Help-grub mailing list Help-grub@gnu.org https://lists.gnu.org/mailman/listinfo/help-grub