On 09.07.2023 00:02, Steve McIntyre wrote:
Nothing should be caring about C/H/S at all in the 21st century. Using C/H/S only allows you to access 515MB of disk [1]. *Everything* these days uses LBA instead. What makes you think that the BIOS on this old machine cares about C/H/S?
My machine is old, it's a 32-bit CPU from 2006 in a machine from 2008.My BIOS reports disk CHS geometry based on the values of CHS end of boot MBR partition. Debian installer sets the amount of sectors to 2 in CHS end.
GRUB2, for some reason, follows disk geometry even for LBA reads, which results in reading only up to 1 KB of data for each int13h call. It takes about 5 seconds to show GRUB menu (GRUB loading... for 5 seconds), the kernel (4.5 MB) is loaded in 27 seconds, and initrd takes ages to load.
Relevant GRUB2 issue description here (contains incorrect MBR CHS values, I took the first mbr parser found online which was incorrect):
https://lists.gnu.org/archive/html/grub-devel/2023-07/msg00001.html My patches to ignore CHS geometry for LBA reads: https://lists.gnu.org/archive/html/grub-devel/2023-07/msg00020.htmlI found some information in Russian sources that some BIOSes do not implement out-of-geometry reads internally for int13h, but that is for very old PCs like IBM AT/XT, and GRUB is 32-bit 386+.
What happens if you tweak the partitioning by hand when installing?
Nothing happens, Debian Installer still ends with 2 sector layout. Note that I'm using disk sizes less than 8 GB, the partition sizes are respectively less than 8 GB.
I've tried to install Debian 12 i386 to a 8G disk, using qemu, with guided automatic partitioning. Testdisk data right after the installation:Disk testz.img - 8589 MB / 8192 MiB - CHS 1045 255 63 Current partition structure: Partition Start End Size in sectors 1 P Linux 0 32 33 919 199 48 14774272 Bad relative sector. 2 E extended 919 232 16 1044 52 32 1996802 Bad relative sector. No partition is bootable 5 L Linux Swap 919 232 18 1044 52 32 1996800 Bad relative sector.It seems that testdisk automatically recalculates C/H/S values and shows corrected data (in the table above). Here's what really is present in the MBR (data of the first partition entry):$ ./mbr_my.py testz.img Status: 0x0 C/H/S start: 4 4 1 Part type: 0x83 C/H/S end: 1023 254 2 LBA of first sector: 2048 Sector count: 14774272fdisk/cfdisk and parted all create partitions for which testdisk does not complain.d-i drives parted to make partitions... [1] https://forums.tomshardware.com/threads/lba-chs-bios-prob.685748/
Attachment:
OpenPGP_signature
Description: OpenPGP digital signature