A gpart(8) mystery on 10.3-RELEASE

Steven Hartland killing at multiplay.co.uk
Tue Apr 5 21:32:18 UTC 2016



On 05/04/2016 20:48, Warren Block wrote:
> On Tue, 5 Apr 2016, Boris Samorodov wrote:
>
>> 05.04.16 12:30, Trond Endrestøl пишет:
>>
>>> What am I doing wrong? Can't gpart(8) write both the pmbr and the efi
>>> image as a single command? Is it an off-by-one error in gpart(8)?
>>>
>>> gpart bootcode -b /boot/pmbr -p /boot/boot1.efifat -i 1 ada0
>>> gpart: /boot/boot1.efifat: file too big (524288 limit)
>>
>> Do you try to get only UEFI boot? Then do not use "-b" option. It is
>> needed for BIOS boot.
>>
>> Do you need to get a system with both UEFI and BIOS boot? Then use two
>> different partitions for UEFI and BIOS booting schemes.
>>
>>> gpart bootcode -b /boot/pmbr ada0
>>> bootcode written to ada0
>>
>> This is needed only for BIOS boot and together with "-p /boot/gptboot"
>> option.
>
> Well... bootcode -b only writes to the PMBR and does not take a 
> partition number with -i.  So the short form version I use could be 
> refused by a very strict option parser, requiring two separate steps:
>
>   gpart bootcode -b /boot/pmbr ada0
>   gpart bootcode -p /boot/gptboot -i1 ada0
>
> The way it parses options when working on EFI partitions might be more 
> strict.
>
> Actually, the more I think about it, using bootcode -p to write the 
> entire EFI partition seems dangerous.  Unless it is surprisingly 
> smart, it will wipe out any existing stuff on that EFI partition, 
> which could be any number of important things put there by other 
> utilities or operating systems, including device drivers.
>
> The safer way is to mount that partition and copy the boot1.efi file 
> to it.
Pretty sure that's not done as you cant guarantee fat support is available.

     Regards
     Steve


More information about the freebsd-stable mailing list