Re: Trying to build Rescue image for three different target and platform
В Sun, 16 Mar 2014 06:56:46 +0100 adrian15 пишет: > > > > To be honest, I do not understand what you are trying to do here. > > The same thing as in: > > https://forja.cenatic.es/plugins/scmgit/cgi-bin/gitweb.cgi?p=supergrub2/supergrub2.git;a=blob_plain;f=supergrub-meta-mkrescue;hb=22683e2b5fa0d65141c1be1a7487914b1d43f4dd > > (which uses Debian packages) but using source code. > Why would you want to build separate ISO for every platform? The whole point of using grub-mkrescue is to create single ISO image that supports and can be booted on all platforms. > > >> 8) What's the best way to build and install an hybrid system that > >> includes both: x86_64_efi and i386 - pc stuff? > >> > > > > Just built and install grub for each platform you want to use; > > grub-mkrescue will automatically include every platform it finds. > > Any advice for the special where I want to build x86_64_efi + x86_pc > hybrid disk? > Not really. Just build and install both platforms using the same prefix and run grub-mkrescue that is build to use the same prefix. As mentioned, when building you should distinguish between host (grub utils) and target (boot time). It is perfectly valid to build utils for x86 and target for ARM as long as you have suitable cross tools. In your case all builds should use the same host (which defaults to target which may be wrong even for x86 case - consider 32 vs. 64 bit). See INSTALL how to set various options for each environment. > > > > I prefer to build every platform in separate directory, there were > > issues with incomplete cleanup. If you hit one, please report. > > Well, as I don't want to have a separate directory for each build my > question would be: > > What is it the official commands for doing a safe cleanup? > I always used "make distclean". But IIRC there are some files that are not removed. Not sure to which extent they affect subsequent builds. I had problems when re-building on Windows but they should be fixed now. From my point of view, "make distclean" is official and if it does not work it is a bug that has to be fixed. > > > > >> 9) A) What's the best way to remove grub configuration files installed > >> on the system so that I can start from scratch so that already built > >> target+platform grub binaries or configurations are ignored? > >> > > > > There are no configuration files after "make install" (if we do not > > count /etc/grub.d as configuration). > > I meant the ones that someone makes grub-mkrescue to look for the > /usr/local/lib/grub/i386-pc/boot_hybrid.img files which I suppose are > not found in source code but in the system (Gnu/Linux filesystem I mean). > There are no special configuration files. If grub-mkrescue is called without --directory option, it will enumerate all platforms found under /usr/local/lib/grub (in your example). Just place additional platform directory there and grub-mkrescue will pick it up. ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: Trying to build Rescue image for three different target and platform
El 16/03/14 08:04, Andrey Borzenkov escribió: В Sun, 16 Mar 2014 06:56:46 +0100 adrian15 пишет: To be honest, I do not understand what you are trying to do here. The same thing as in: https://forja.cenatic.es/plugins/scmgit/cgi-bin/gitweb.cgi?p=supergrub2/supergrub2.git;a=blob_plain;f=supergrub-meta-mkrescue;hb=22683e2b5fa0d65141c1be1a7487914b1d43f4dd (which uses Debian packages) but using source code. Why would you want to build separate ISO for every platform? The whole point of using grub-mkrescue is to create single ISO image that supports and can be booted on all platforms. Because I want to support all the GRUB2 platforms but not having a single 60 MB (random size guess) download for that. The other reason is because a: i386-efi + x86_64_efi + x86_pc hybrid did not boot in a x86_64_efi only system. Right now I'm not sure if I built it from source or if I made the iso with Debian binaries but, what it's sure is that it did not boot at all. It just hung. So I want a separate ISO for every platform. 8) What's the best way to build and install an hybrid system that includes both: x86_64_efi and i386 - pc stuff? Just built and install grub for each platform you want to use; grub-mkrescue will automatically include every platform it finds. Any advice for the special where I want to build x86_64_efi + x86_pc hybrid disk? Not really. Just build and install both platforms using the same prefix and run grub-mkrescue that is build to use the same prefix. Ok. As mentioned, when building you should distinguish between host (grub utils) and target (boot time). It is perfectly valid to build utils for x86 and target for ARM as long as you have suitable cross tools. In your case all builds should use the same host (which defaults to target which may be wrong even for x86 case - consider 32 vs. 64 bit). See INSTALL how to set various options for each environment. Ok,I'll check that. I prefer to build every platform in separate directory, there were issues with incomplete cleanup. If you hit one, please report. Well, as I don't want to have a separate directory for each build my question would be: What is it the official commands for doing a safe cleanup? I always used "make distclean". But IIRC there are some files that are not removed. Not sure to which extent they affect subsequent builds. I had problems when re-building on Windows but they should be fixed now. From my point of view, "make distclean" is official and if it does not work it is a bug that has to be fixed. make distclean . Written down. 9) A) What's the best way to remove grub configuration files installed on the system so that I can start from scratch so that already built target+platform grub binaries or configurations are ignored? There are no configuration files after "make install" (if we do not count /etc/grub.d as configuration). I meant the ones that someone makes grub-mkrescue to look for the /usr/local/lib/grub/i386-pc/boot_hybrid.img files which I suppose are not found in source code but in the system (Gnu/Linux filesystem I mean). There are no special configuration files. If grub-mkrescue is called without --directory option, it will enumerate all platforms found under /usr/local/lib/grub (in your example). Just place additional platform directory there and grub-mkrescue will pick it up. So I just need to purge the /usr/local/lib/grub contents so that I can start again without the fear of grub-mkrescue picking up an earlier built platform (That's what I tried to achieve before with make uninstall by the way). That's exactly what I wanted to know. Thank you very much! I do not know when I will be able to try all of this but now it's much more clear everything and I'll be able to report GRUB build errors (or runtime) more easily for my special use case. adrian15 -- Support free software. Donate to Super Grub Disk. Apoya el software libre. Dona a Super Grub Disk. http://www.supergrubdisk.org/donate/ ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: How to deal with fatal device seeks?
I'm getting the same problem ( FATAL: int13_cdrom: function 42. Can't use 64bits lba ) with hd0, hd1, hd2, hd3, and worse: * hd2,msdos5 ls (hd2,msdos5)/ Is this a bug? adrian15 El 16/03/14 06:26, adrian15 escribió: I use a for loop for detecting Operating systems like this one: for dev in (*); do echo $dev done but it hangs !!! So some clues: * Using 2.02~beta2-7 from Debian Unstable. * The disk is an hybrid one (x86_64_efi + i386_pc). * The error only happens when using Virtualbox in a non-EFI setup. So some questions: * Is it a bug that a non seekable device like (cd,apple4) or (cd,gpt1) is available in non-EFI mode? * Is it perhaps a Virtualbox bug? * I cannot catch these fatal error as if they were exceptions inside a try-catch. Or can I? I will probably improve my searchindevices functions so that it also filters these (cdsomething) devices as a workaround. Thank you. These are the minimal tests: --- Using Qemu as EFI gives no problem: echo (*) (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) ls (cd,apple4)/ error: unknown filesystem. ls (cd,gpt1)/ error: unknown filesystem. --- Using Virtualbox as BIOS (Ex. 1) (Problem): echo (*) (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) ls (cd,apple4)/ FATAL: int13_cdrom: function 42. Can't use 64bits lba --- Using Virtualbox as BIOS (Ex. 2) (No problem): echo (*) (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) ls (cd,gpt4)/ error: unknown filesystem. --- Using Virtualbox as BIOS (Ex. 3) (Problem): echo (*) (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) ls (cd,gpt1)/ FATAL: int13_cdrom: function 42. Can't use 64bits lba adrian15 -- Support free software. Donate to Super Grub Disk. Apoya el software libre. Dona a Super Grub Disk. http://www.supergrubdisk.org/donate/ ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: How to deal with fatal device seeks?
В Sun, 16 Mar 2014 17:05:53 +0100 adrian15 пишет: > > I'm getting the same problem > ( FATAL: int13_cdrom: function 42. Can't use 64bits lba ) > This message does not exist in grub sources, so it likely comes from firmware. 64 bit LBA would mean size over 2TB. So the first question - what size of disks do you have? > with hd0, hd1, hd2, hd3, and worse: > > * hd2,msdos5 > > ls (hd2,msdos5)/ > > Is this a bug? > Hard to tell. Do you actually have working filesystem(s) on these partitions? The message itself simply means that grub attempts to read very high offset. I wonder if it can be somehow byte order related. > adrian15 > > El 16/03/14 06:26, adrian15 escribió: > >I use a for loop for detecting Operating systems like this one: > > > > for dev in (*); do > > > >echo $dev > > > > done > > > > but it hangs !!! > > > > So some clues: > > > > * Using 2.02~beta2-7 from Debian Unstable. > > * The disk is an hybrid one (x86_64_efi + i386_pc). > > * The error only happens when using Virtualbox in a non-EFI setup. > > > > So some questions: > > > > * Is it a bug that a non seekable device like (cd,apple4) or (cd,gpt1) > > is available in non-EFI mode? > > * Is it perhaps a Virtualbox bug? > > * I cannot catch these fatal error as if they were exceptions inside a > > try-catch. Or can I? > > > > I will probably improve my searchindevices functions so that it also > > filters these (cdsomething) devices as a workaround. > > > > Thank you. > > > > > > These are the minimal tests: > > > > --- Using Qemu as EFI gives no problem: > > > > echo (*) > > (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) > > ls (cd,apple4)/ > > error: unknown filesystem. > > ls (cd,gpt1)/ > > error: unknown filesystem. > > > > --- Using Virtualbox as BIOS (Ex. 1) (Problem): > > > > echo (*) > > (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) > > ls (cd,apple4)/ > > FATAL: int13_cdrom: function 42. Can't use 64bits lba > > > > --- Using Virtualbox as BIOS (Ex. 2) (No problem): > > > > echo (*) > > (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) > > ls (cd,gpt4)/ > > error: unknown filesystem. > > > > --- Using Virtualbox as BIOS (Ex. 3) (Problem): > > > > echo (*) > > (fd0) ... (cd) ... (cd,apple2),... (cd,gpt2) > > ls (cd,gpt1)/ > > FATAL: int13_cdrom: function 42. Can't use 64bits lba > > > > > > adrian15 > ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: How to deal with fatal device seeks?
El 16/03/14 17:43, Andrey Borzenkov escribió: В Sun, 16 Mar 2014 17:05:53 +0100 adrian15 пишет: I'm getting the same problem ( FATAL: int13_cdrom: function 42. Can't use 64bits lba ) This message does not exist in grub sources, so it likely comes from firmware. 64 bit LBA would mean size over 2TB. So the first question - what size of disks do you have? SATA 0: 200 GB (Windows 7) SATA 1: 1 GB (Rescatux USB. E.g. Grub mkrescue disk) SATA 2: 20 GB (SteamOS E.g. Debian) with hd0, hd1, hd2, hd3, and worse: * hd2,msdos5 ls (hd2,msdos5)/ Is this a bug? Hard to tell. Do you actually have working filesystem(s) on these partitions? The message itself simply means that grub attempts to read very high offset. I wonder if it can be somehow byte order related. From GRUB point of view: * (hd2,msdos1) is Gnu/Linux root filesystem * (hd2,msdos2) No such partition * (hd2,msdos3) No such partition * (hd2,msdos4) No such partition * (hd2,msdos5) FATAL: int13_cdrom: function 42. Can't use 64bits lba From GNU/Linux point of view: Disk /dev/sdc: 21.5 GB, 21474836480 bytes 255 heads, 63 sectors/track, 2610 cylinders, total 41943040 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: 0x00068a9f Device Boot Start End Blocks Id System /dev/sdc1 *20484013670320067328 83 Linux /dev/sdc24013875041940991 9011215 Extended /dev/sdc54013875241940991 901120 82 Linux swap / Solaris If I try: ls (hd2,msdos5)/ from a 2.00-15 system mkrescue image (Super Grub2 Disk 2.00s1-beta6) I get: error: unknown filesystem. So, it's ok, because it is not a fatal error. It would seem that somewhere between 2.00-15 and 2.02~beta2-7 (Debian version) the bug arises. Or maybe the error was before because grub did actually to arise a FATAL error and it didn't. Thank you for any indication on how to make more tests. adrian15 -- Support free software. Donate to Super Grub Disk. Apoya el software libre. Dona a Super Grub Disk. http://www.supergrubdisk.org/donate/ ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: How to deal with fatal device seeks?
El 16/03/14 18:16, adrian15 escribió: If I try: ls (hd2,msdos5)/ from a 2.00-15 system mkrescue image (Super Grub2 Disk 2.00s1-beta6) I get: error: unknown filesystem. So, it's ok, because it is not a fatal error. It would seem that somewhere between 2.00-15 and 2.02~beta2-7 (Debian version) the bug arises. Or maybe the error was before because grub did actually to arise a FATAL error and it didn't. 2.00-22 does not have the error. So it narrows the search between 2.00-22 and 2.02~beta2-7. adrian15 Thank you for any indication on how to make more tests. adrian15 -- Support free software. Donate to Super Grub Disk. Apoya el software libre. Dona a Super Grub Disk. http://www.supergrubdisk.org/donate/ ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: How to deal with fatal device seeks?
I have digged the git log for partition related commits just in case you might think that they have something to do with my problem (From 2.00 release to current head on master branch). I have tried to filter them. So I have removed the ones related with strange table partitions or emulation code. Here there are the commits: http://git.savannah.gnu.org/cgit/grub.git/commit/?id=e88f0420b90c2565637962754cc26fa8a4ed9256 http://git.savannah.gnu.org/cgit/grub.git/commit/?id=4bad23a15fc129218f611f51dcb268c246b207f1 http://git.savannah.gnu.org/cgit/grub.git/commit/?id=25fc51a87929262c1cc132bc29cc083ce98f0e0e http://git.savannah.gnu.org/cgit/grub.git/commit/?id=df6da5a52dc2ec424203c0f8001903435b177fa8 http://git.savannah.gnu.org/cgit/grub.git/commit/?id=b7b78edb1ca05f30dd07ebed4bcb3d5a39aa5358 http://git.savannah.gnu.org/cgit/grub.git/commit/?id=258f43b7d7bf4b03799b6cd3004b5372e082d01b http://git.savannah.gnu.org/cgit/grub.git/commit/?id=86d08fdb18b0142c1ce1b95db1aae989502956c5 Just hope that you, that deal with GRUB source code every day, can find what might be the faulty commit so that I can test that the special commit source code and reproduce it (or not to reproduce it). Thank you for any idea! adrian15 El 16/03/14 19:05, adrian15 escribió: El 16/03/14 18:16, adrian15 escribió: If I try: ls (hd2,msdos5)/ from a 2.00-15 system mkrescue image (Super Grub2 Disk 2.00s1-beta6) I get: error: unknown filesystem. So, it's ok, because it is not a fatal error. It would seem that somewhere between 2.00-15 and 2.02~beta2-7 (Debian version) the bug arises. Or maybe the error was before because grub did actually to arise a FATAL error and it didn't. 2.00-22 does not have the error. So it narrows the search between 2.00-22 and 2.02~beta2-7. adrian15 Thank you for any indication on how to make more tests. adrian15 -- Support free software. Donate to Super Grub Disk. Apoya el software libre. Dona a Super Grub Disk. http://www.supergrubdisk.org/donate/ ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
Re: How to deal with fatal device seeks?
On 17.03.2014 05:05, adrian15 wrote: > El 16/03/14 18:16, adrian15 escribió: >> If I try: >> >> ls (hd2,msdos5)/ >> >> from a 2.00-15 system mkrescue image (Super Grub2 Disk 2.00s1-beta6) I >> get: >> >> error: unknown filesystem. >> >> So, it's ok, because it is not a fatal error. >> It would seem that somewhere between 2.00-15 and 2.02~beta2-7 (Debian >> version) the bug arises. >> >> Or maybe the error was before because grub did actually to arise a FATAL >> error and it didn't. > > 2.00-22 does not have the error. > > So it narrows the search between 2.00-22 and 2.02~beta2-7. > Use git bisect to find exact commit please. > adrian15 > >> Thank you for any indication on how to make more tests. >> >> adrian15 > signature.asc Description: OpenPGP digital signature ___ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
