Stewart Adam
2008-Oct-29 12:46 UTC
[syslinux] Using syslinux on the PMBR of a GPT partition table
Hi, I've been working on making a USB key which is bootable via EFI and BIOS, and so far I've succeeded but only by using GRUB on the MBR. Syslinux installs correctly, but because the USB key can only have one partition and its type is EFI protective, it isn't set to bootable on the MBR partition table and so syslinux (mbr.bin) doesn't boot. I've also tried gptmbr.bin which semi-works -- I don't get "missing operating system" as I do with mbr.bin, but the OS doesn't actually load either. My BIOS just reads the USB key a bit and then gets stuck. I was wondering what was the status of syslinux and GPT, and was wondering what testing I could do to help improve hybrid GPT+MBR support. Stewart
H. Peter Anvin
2008-Oct-29 14:59 UTC
[syslinux] Using syslinux on the PMBR of a GPT partition table
Stewart Adam wrote:> Hi, > > I've been working on making a USB key which is bootable via EFI and > BIOS, and so far I've succeeded but only by using GRUB on the MBR. > Syslinux installs correctly, but because the USB key can only have one > partition and its type is EFI protective, it isn't set to bootable on > the MBR partition table and so syslinux (mbr.bin) doesn't boot. I've > also tried gptmbr.bin which semi-works -- I don't get "missing operating > system" as I do with mbr.bin, but the OS doesn't actually load either. > My BIOS just reads the USB key a bit and then gets stuck. > > I was wondering what was the status of syslinux and GPT, and was > wondering what testing I could do to help improve hybrid GPT+MBR support. >gptmbr.bin needs the UUID of the boot partition written to the MBR; I really should write a script to do that. That doesn't mean gptmbr.bin has been extensively tested by any means; yours is the first report of someone actually trying to use it. We obviously should try to make it work. Could you describe your setup a bit? -hpa
H. Peter Anvin
2008-Nov-06 01:48 UTC
[syslinux] Using syslinux on the PMBR of a GPT partition table
Stewart Adam wrote:> On 11/5/08 12:04 PM, H. Peter Anvin wrote: >> Except it should be 440 bytes. I suspect you appended the partition >> GUID as a text string, instead of in binary form. > > Ah, yup that was it. After I appended the strong using the a little-endian > byte order things worked out a little better this time. Oddly enough though, > it worked perfectly under QEMU but the USB stick refused to boot on my PC > (same "Boot partition not found" error). >Wonder how Qemu managed that... -hpa